' CUMENT RESUME
ED 305 894 IR 013 747
AUTHOR Barrett, John, Ed.; Hedberg, John, Ed. TITLE Using Computers Intelligently in Tertiary Education. A Collection of Papers Presented to the Australian Society for Computers in Learning (Sydney, New South Wales, Australia, November 29-December 3, 1987). INSTITUTION Australian Society for Computers in Learning. REPORT NO ISBN-0-949088-31-5 PUB DATE 87 NOTE 477p. PUB TYPE Collected Works Conference Proceedings (021) -- Viewpoints (120) Reports - Research/Technical (143)
EDRS PRICE MF01 Plus Postage. PC Not Available from EDRS. DESCRIPTORS *Artificial Intelligence; *Computer Assisted Instruction; *Computer Managed Instruction; Computers; Computer Simulation; Courseware; *Distance Education; Foreign Countries; Higher Education; Instructional Systems; Interactive Video; *Media Research; Programing IDENTIFIERS *Australia; Intelligent CAI Systems; Intelligent Tutoring Systems
ABSTRACT The 63 papers i.n this collection .iclude two keynote addresses: r..Datient Simulation Using Interactive Video: An Application" (Joseph V. Henderson), and "Intelligent Tutoring Systems: Practice Opportunities and Explanatory Models" (Alan Lesgold). The remaining papers are groupedunder five topics:(1) Artificial Intelligence, including intelligent computerassisted learning, problem solving, artificial intelligence, andprogramming (15 papers);(2) Delivery Systems, including distance learning, communications, and hardware (9 papers); (3) Developments,including interactive video, simulation, authoring, computer managed learning, and computer based training (12 papers); (4) Research/Evaluationand Future Directions, including research, policy/planning,and philosophical aspects (21 papers); and (5) Software Tutorials, including computer assisted learning tools and commercialproduct applications (4 papers). The text is supplemented byvarious figures, and references are provided for each paper. (EW)
*********************************************************************** * Reproductions supplied by EDRS are the best thatcan be made * * from the original document. *********************************************************************** -
Il 40
Ant /On
< or, , * t. r tral t:',:i'r:"771-11..'F:::t4::k!..'tc, 4 n' 4 1V-, , < 4.. t to: ttst*ty , 4, " % *,t4,,,ott*ot, oc, 4'4 .o* '-ev4111*, to'tot., 4444 444444;wworro St oar 4 1 : O:. '" - " " Using computers intelligently in Tertiary Education A collection of papers presented to the Australian Society for Computers in Learning, Sydney, November 29th to December 3rd, 1987 Edited by John Barrett and John Hedberg © 1987 ASCILITE and individual authors 3 Acknowledgements Part of the funding for this publicationhas been provided by IBM Australia Organizing Committee Dr John Barrett, Dr Ken Wade Mr Bill Chia, Dr John Hedberg Ms Bronwyn Hewlett, Dr Rod Nason Mr Rod Sims, Mr Phillip Staines Mr Bill Stewart Editorial Assistance Anita de Sousa, Bill Chia, Bill Stewart Layout and Design John Hedberg assisted by Glen Sorensen Conference Secretariat Robyn Hall, Judy Layton, Mollie Lenthall Published by the Australian Society for Computers in Learning in Tertiary Education ISBN 0 949088 31 5 © 1987 ASCILITE and individual authors Distributed by Continuing Education, University of New South Wales, P.O. Box 1 Kensington, N.S.W. 2033. Contents Using Computers Intelligently in Tertiary Education Keynote Address 1 Henderson, Joseph. V. Patient Simulation Using Interactive Video: An Application 1 Lesgold, Alan Intelligent Tutoring Systems: Practice opportunities and explanatory models. 7 Artificial Intelligence Intelligent CAL, ProblemSolving, A.I., Programming 25 Bailey, D., P. Logan., P.Walker & I.Walsh The Physics Literacy Project - 1987. 25 Burden, Frank R The Logichem Organic Inference Program. 31 Fielden, Kay Grabbing Knowledge Graphically. 34 Foster, C,?off What Do You Say After They Press 'Return'?- Intelligent Feedback in CAL. 40 Harris, Marshall Some Lazy Thoughts on Applying AI to CAL. 47 Newby, Michael Teaching Introductory Programming using a LanguageIntelligent Programming Environment. 52 Stern, Neville The Powerful Ideas of Educational Knowledge Engineering. 58 Stern, Neville Bandwagons, Hearses and Other Vehicles: The Useful Fate of PROLOG in Education. 69 Stewart W.A. & J. Flanagan Increasing Active Learning in a CAL Environment. 76 Waugh, David A. & K.P. Stark Applying Intelligent CAL Principles to Help Subsystems for Interactive Software. 81 Waugh, David A. MacTutor: An Intelligent Novice Programming Environment and Tutor. 90 Webb, Geoffrey I. Generative CAL andCourseware Abstraction. 100 r pi Using Computers Intelligently Pars low, G.R. & T. R. Haynes Profile of a Successful CAI facility- identifying useful technology. 404 Staples, Rodney Intelligent CAL, or Intelligent CAL Users? 409 Richards, T., G. Webb & N. Craske Object-Oriented Control for Intelligent C.A.L. Systems. 447 Delivery Systems Distance Learning, Communications, Hardware 110 Castro, Angela A.ft 'AESNET' (Australian External Studies Network)? 110 Clarke, R. J. & L. Athanasiadis Automated Library And Borrowing System (ALABS): Implications for CAL. 117 Cornish, B., M. Johnston & M. Whitelaw Bringing the Distance Student On Campus the Riverina Way. 125 Fleming, Peter Computerised Formative Assessment- A Powerful New Tool. 134 Latham, S., W. Moore, G. Ritchie, B. Rothwell & L. Wilde. Computer Enhanced Distance Education. 138 Munro, Donald Quality Distance Education= Computer Based Feedback + Electronic Mail. 147 Palmer, J., K. Pierce & G. Pervan. A Case Study in Computer education by external study. 153 Coldwell, R.A. CBT: Networking of autistic & other retarded students in sheltered training establishments. 415 Weidenbach, Vanda Decentralized Learning on the Tertiary Campus: CA.( to increase student music literacy skills. 420 Developments Interactive Video, Simulation, Authoring, CML, CBT 160 Carthew, Douglas CM- A Necessity not a Choice. 160 Christie, Allan Interactive Videodisc in the Teaching of Orthopaedics in Psychotherapy 164 Duncan, Kelvin Using Computers Intelligently - What happens when computers are removed from the classroom? 170 r. t, 6 Using Computers Intelligently Edwards, David T. & Robert G. Craig. Software Simulation of Microprocessor Hardware. 176 Farrow, Margaret Innovative Applications of Computer Assisted Learning. 184 Kemm, R.E. Computer Managed Learning in the Physiology Laboratory. 18d Se linger, Benjamin Lateral Teaching Software: A Programmer's Perspective. 204 Lord, Trevor Interactive Laser Disc in Community Medicine. 210 Shaw, Neil A. Interactive Audio... A Challenge or Companion to Interactive Video? 212 Byard S.M. & N.A. Shaw Applications of Interactive Video. 222 Read, W.W. & David A.Waugh Simulation and Queuing Theory: A Combined Approach.229 Sims, Rod & P. Grant Instructional System Design: The hidden'agenda. 235 Research/ Evaluation & Future Directions Research, policy/planning, Philosophical aspects 243 Arnold, Anne The Use of CAL in Remedial Mathematics for Economics Students. 243 Barrett, John F. The Subject Matter of CAL Technology: Towards a development methodology. 248 Carss, Brian W. Computer Assisted Learning and the Future of Education.257 Clark, Neil J. Demystifying Tertiary Education. 262 Clayton, D., P. Farrands & M. Kennedy Computer Enhancing using a Fu action Graph Plotter. 271 Crock, Michael Updating Mathematics Skills: An evaluation of a computer based mathematics course. 278 Durham, Marsha Are Computers the Right Stuff? Computers and writing instruction. 296 Hasan, Helen Assessment of Student Computer Programming Assignments. 305 7 Using Computers Intelligently Higgins, Phill Easing the Load of Laboratory Assessment. 310 Isaacs, Geoff Athena, Andrew and Stanford: a look at implementation and evaluation in three large projects. 313 McCrae, Michael The Computer Based Education System: A process of integration? 322 O'Brien, G.C. Education Resource Planning. 331 Page, Michael. If It Is Useful Then Use It. Some experiences in theuse of computers in the teaching of Chemistry. 339 Parsons, D. & R. Hunter. Computer Managed Learning ht Electronics, 345 Price, John A. Formative Evaluation Systems for CAI. 352 Sampath, S. & A. J. Quaine Automated Storyboards- A plausible authoring aid. 358 Stanford, Jon D. & Howard P. CA CML-Its Applications to Increase Student Achievement Using Formative Self Assessment. 367 Stevens, Jean A. Computers in Tertiary Education in the Northern Territory379 Tao, Li., L. Fang, L. Christensen & C. Yu. Computer Assisted Examination for Fundamental Programming Courses. 383 Vlugter, Marjan. Computer Ass sted Learning at the University of Otago. 391 Chia, W. & M. Khamis A Computer Based Testing and Evaluation System. 398 Software Tutorials CAL tools and commercial product applications 431 Carnegie, U., & M. Gerrard. Keyboarding for Information: Don't let your thoughts get away from your fingers. 431 Mc Sherry, Tony Development of CBT Using Author. 436 Rundle, John Diverse CBT Applications using Author. 441 McArthur, Liz The Advanced Features of Author. 443 Preface This collection has been a learning experience for the editors, both in assembly of papers electronically and understanding the varied skills of academics in pro- ducing publishable work directly from a word proces- sor. As we move towards a time when computers allow greater access to cheap publication, it is instruc- tive to realize how much staff development work is still required for the writer to develop their abilities in preparing papers. Maybe one day the story of editing this collection will be told. Under the constraints of time we have attempted the best we could, we hope the combined work of a small team of enthusiasts will have caught the errors and make interesting reading. John Ba. rett and John Hedberg Sydney November 1987 9 The Marshall Harris keynote address It takes a great deal of hard work to establish and maintain an organization suchas ASCILITE. Many people have devoteda lot of energy to the success of our society. However, none of this effort orsuccess would have been possible were it not for the initial work of Marshall Harris. Marshall was the driving force behind the first CALITE conference. It was he who took the untested concept of a tertiary oriented CAL conference and brought it to life. Many have laboured on our behalf since Marshall led the way, but they have had the advantage of the knowledge that what they strove forwas viable. In organizing that first conference Marshall had no such assurance. ASCILITE has emerged as a society L,n his initial input. Marshall Harris, the mern-Jers of ASCILTI E thank you. The address in 1987 was presented by Professor Alan Lesgold and a text of his presentation is included in this volume on page seven. 10 Patient simulation using interactive video: An application Joseph V. Henderson, MD Department of PreventiVe Medicine and Biometrics Uniformed Services University of the Health Sciences Training military physicians in trauma manage- plans call for non-surgeon physicians to be ment is a dilemma in peacetime, since there are assigned to resuscitate and stabilize few opportunities to gain clinical experience within the military care system. We decided to trauma patients. Naturally, these physi- attempt to use interactive video as a means of cians must also get training in trauma providing some experience in clinical decision management, but clinical opportunities for making through patient simulation. Rye cases, this are even less likely than for surgeons' of increasing difficulty, are presented on a single videodisc side. The program is implemented on training. a system based on the IBM PC, and is written in the C programming language. The program As a partial solution to this problem, we are emphasizes clinical realism by providing many diniad options at each decision point, and by now providing Advanced Trauma Life audiovisually depicting combat clinical care in Support (ATLS) training to active duty and very realistic ways. The user interface is flexible reserve medical officers. ATLS was devel- and, though complex, is relatively easy to use; it oped and is administered by the American is supported by a narrated, on line tutorial. College of Surgeons (American College of Surgeons, 1984). It's designed for surgeons Providing surgical care to combat casual- and other physicians, civilian or military, ties is a problem for our armed forces. who are likely to care for trauma patients. Between wars there are few militaryTechniques of surgery are not taught; trauma patients and, consequently, fewATLS stresses early management of chances for military surgeons to get experi- trauma patients: initial resuscitation, stabi- ence in trauma care. During wartime thislization, and disposition. The course uses a shortage of experienced surgeons will be combination of lecture, small group discus- acute, and even conservative casualty pro-sion, and practical experience with ani- jections indicate that a physician draft will mals, manikins, and volunteers simulating not resolve the shortage. Care of combat injury.Although physicians sometimes casualties may suffer, and lives and limbs disagree with some of its recommenda- of servicemen may be jeopardized. tions, ATLS establishes a standard method of managing trauma patients, a standard- Part of the solution is to free experiencedized curriculum and content, and stan- surgeons to do surgery, and to permit non- dards for certifying physicians in trauma surgeon physicians to take care of traumamanagement. patients before and after they are in the operating room. At the early echelons ofSimulation is particularly useful for teach- care, dose to the battle front, non specialist ing and testing the ability of students to physicians and Hospital Corpsmen areintegrate the facts and rules learned in the already designated to treat patients. At the classroom and make patient management later echelons, such as combat zone field decisions In addition to animal and and fleet hospitals and hospital ships, our anatomic :models, and live actors already 11 2 Using Computers Intelligently used in ATLS teaching, computer-aided and in management ofemergency medical instruction (CAI) could also be usedto conditions.The most recent program, simulate. CAI can instruct in an interactivedeveloped for the Navy at the Uniformed way that is intrinsically motivating, and the Services University of the Health Sciences, pace and content of instruction can be tai- teaches ATLS principles to physicians who lored to the student. Using text and limited may be responsible for the care of combat graphics, conventional CAIcan presentcasualties. problems in a fairly realisticway. Some advantages over animal and human simu- Hardware lations are decreased cost andpres'nta- tions of uniform content and quality. The student version of the system presently uses an IBM PC XT with 640 Kbytes of Over the past few years, CAI developers random access memory, a 20 Mbyte hard have improved the realism of simulations by using optical videodisc technologydisk and single 360 Kbyte floppy disk drives, a 12" Sony RGB monitor (PVM (Leveridge, 1981; Abdulla and Henke, 1271Q) with light pen, a Pioneer 1984; Woods et al, 1984). LDV-1000 High-fidelity laser videodisc, and a genlock graphics audiovisual sequences or still framescan card with videodisc controller(GL-512/ be quickly accessed, dependingon the in- structional design and the student's VDC 100, Online Products Corp).The courseware authoring version of the sys- choices. Computer-generated graphicscan tem is as above, except that the computer is be superimposed over the video image. an IBM PC-AT with 20 Mbyte hard disc and When used in medical education, the result 1.2 Mbyte floppy, witha monochrome can be a very realistic clinical simulation,monitor for programming especially useful when interaction with use. Systems were purchased from Online Products genuine patients is undesirableor impos- sible. Corporation, Germantown, MD. The Sony monitor is equipped with Interactive videodisc (ND) is likely a light soon to pen, allowing students to enter responses be the preferred method for generaland without the keyboard. Thiscan ease use of specialty board medical examinations. the system by less experienced students. Within five years ND will be the major testing medium for Part III of the NatiokialThe laser videodisc unit is capable Board Examinations for physician certifi- of ran- domly accessing any of 54,000 framesper cation. Recently, the American College of videodisc side, with maximum seek time of Cardiology has endorsed theuse of com- about three seconds. Single frames suchas puter-based ND instruction for teachingxrays can be shown with adequate resolu- and certification in Advanced Cardiac Life tion. Motion sequences canrun forward or Support (ACLS) (Abdulla and Henke,backward, at normal, fast, and 1984). ATLS has so far made slow no use ofspeeds. There are two audio tracks which computer-based learning techniques, with or without videodisc. can be included or mixed, for example, to include narrated instructionsor ambient sounds to enhance realism. These features The Naval Health Sciences Education andare entirely under computer control: the Training Command is now using IVD tech- student's choice.. dictate what hesees and nology for medical education to trainhears. health professio la's at every level. The first programs were for training Hospital The GL512 'Genlock' card allows place- Corpsmen in anat 'imy and physiology, ment of computer-generated graphicsover 12 Using Computers Intelligently 3 the video image. Text or pictures, with or 3.Indicate an understanding of the pri- without animation, can be used to empha- mary and secondary management nec- size, simulate, and produce special video essary within the first hour of emer- effects such as windows and scrolling. The gency care for acute life threatening card also provides an interface giving the emergencies. PC control over the videodiplayer. 4.Demonstrate an understanding of proper disposition of stabilized trauma Software patients. Online Products Corporation provided usProgram. description with IVD a mtrol routines written in assem- bly code. These tools drive the GL512 inter-The program stresses realism: military, faces, and can be incorporated into adthor-clinical, and interpersonal. The inherent developed computer programs&ten indrama of wartime medicine and casualty the C programming language (Microsoft C,care is exploited to involve the learner, and Version 4.0), to control the videodisc anddraw him in a personal way into the situ- computer graphics. A result is computerations presented. As in real life, the student code that is compact and readily main-cannot predict in advance what case he will tained, that executes quickly, and that is see, even if he has used the program before. portable to other computer systems. Also, doing the same thing in apparently similar situations will not always have the Unlike the PILOT authoring language, intended.result. Finally, time is an impor- which is designed for educators who needtant element, indicated by a clock dis- not be computer professionals, C is a com-played on the screen. puter language designed for professional programmers: it requires detailed xnowl-The student is presented five different edge simply to understand a program list- cases with various combinations of airway, ing. For this reason, we have developed anbreathing, circulation, and neurological authoring environment written in C whichproblems.Successive cases increase in allows a course author to write and displaycomplexity and difficulty, and they build content, designate correct answers andon the experience of previous cases. The their sequence, design screens and menus,last case is a kind of final exam, having and to control the video and audio outputseveral pr Ablems. of the videodisc player. In this way new programs can be developed, and old onesAt the usual 30 frames per second, a single updated, with little or no intervention by a videodisc side allows only 30 minutes. of programmer. motion video. Disc economy requires that cases having common attributes share Instructional objectives audiovisual sequences where possible. This means that care be taken with action, Our program is designed to implementdialogue, setting, makeup, and props, to ATLS goals suitable to the interactive videoensure continuity. medium. In working through the simu- lated cases, the physician user will: The student interacts with the program by 1. Demonstrate an understanding of the selecting from menus either using light pen concepts and principles of primary and or, since all choices have unique first letters, secondary patient assessment. with the keyboard.Clinical realism re- 2. Show that he can establish managementquires that the learner be able to select from priorities in a trauma case. a very wide number of choices at each 13 4 Using Computers Intelligently decision point; it would be artificial topres- Meade, MD, using the Army's 10th MASH ent a complex situation and provide onlya and Marine helicopter units from Quan- few choices. The menu system provides 80 tico, VA.This establishes the combat to 90 choices, each of which results ina ambience, and introduces the main charac- more-or-less tailored feedback messageters and situation, while presenting titles. indicating the correctness (within thecon-Action movas from the general to thespe- text of ATLS) of the choice, a reason, and- cific, e.g, snowing groups of casualties if incorrect - a suggestion toward thecor-being delivered to a field hospital by heli- rect choice. The menu also conveys con-copter or ambulance, litter-bearers scram- tent, since it is structured in the same waybling for casualties, a triage area, then into ATLS structures management of trauma; it the CRA. forces the learner to think in ATLS terms in order to do the simulation. A final featureAll the simulated cases are the same charac- is 'windowing' and 'tiling' ofmenus: ter and have the same wound, a small menus in a hierarchy are superimposed onpenetrating injury to the left anterior tho- the screen, permitting the learner to keep rax. Possible problems associated with this his orientation to that hierarchy. wound are presented to the student, who must recognize what they are, and decide Before any titles or action, there isa shortwhat treatment to use. Problems include tutorial to orient the student to the touch-upper airway obstruction, simple and ten- sensitive sawn and keyboard, and thesion pneumothorax, major and minor menus. If the learner ls working independ-hemothorax, pericardial tamponade,an- ently of a learning centre, this is alwaysdomenal hemorrhage with shock, and cer- shown on the student's first use, and elec-vical spine fracture. tively thereafter. The tutorialuses com- puter graphics and animation, and is sup-Students receive final feedback on their ported by narration on the videodisc'ssec- performance in three forms: audiovisual ond audio track. communication delivered by the triage officer, numeric scores, and a medical re- The setting is the Casualty Receiving Areasources impact statement. The score is (CRA) of a 250-bed Combat Zone Fleetsubjective and intended only to rank this Hospital. The CRA is in a large tent whichstudent against others. The impact state- there are a several casualties,some injuredment assesses the student's efficiency in severely, some not. There are litters andusing medical resources: if unnecessary ;fitter stands, and medical equipment and tests or procedures were ordered, the stu- supplies needed to monitor and supportdent is told how misuse of scarce resources the patients (N's, AMBU bags and masks,can result in supplies not being available endotracheal and chest tubes, portablefor the care of other patients. Xray machines, blood pressure cuffs, and ECG monitors). The major feedback to a care-provider is how well his patient does. Death of the Main charactersare a non-surgeon medical patient is not presented, however. Instead, officer (protagonist), an experienced fe-dangerous decisions by the student result male nurse, a corpsman, and the triagein the triage officer taking over the case, officer, an experienced trauma surgeon.which the student must then repeat. The Extras are used in the background to dmu-program concludes by offering another late a busy CRA. case to the student. Computer-generated text and graphicsWe feel this program is a good example of overlie video footage shot on location at Ft. what can be done with interactive video to 14 Using Computers Intelligently 5 simulate patient care situations. The mar- Navy, the Navy Medical Department, the riage of computer and audiovisual in pre- Naval Health Sciences Education and senting content in realistic, clinically accu- Training Command, or the Uniformed rate, and dramatically absorbing ways Services University of the Health Sciences. makes for continuing medical education thal can be worthwhile, rewarding, and References enjoyable for the student. Abdulla, A. M and Henke, J. S. (1984). The Acknowledgements use of natural language and a laser videodisc player in high-fidelity micro- Our appreciation for outstanding produc- computer-based clinical simulations. tion support provided by Metamedia Sys- Proceedings, Eighth Annual IEEE Sympo- tems, Germantown, MD, contractors for sium on Computer Applications in Medical the audiovisual portion of the program, to Care, Washington, DC, Nov, 1984, p. Michael H. Spencer, Director of the audio- 936. Committee on Trauma, American visual production, and to Adam R. Galper, College of Surgeons (1984). Advances Yale University, who designed and devel- Trauma Life Support Course Instructor oped our computer programs. Our thanks Manual. American College of Surgeons. to Mr. Phil Strub of the Naval Health Sci- Leveridge, L L. (1981). Electronic process- ences Education and Training Command, ing of medical visual information.J. CAMIS project manager, Mr. Hank Cham- Medical Systems, 5, pp. 321-335. pagne, Production Coordinator, NavalWoods, J. W.,*Jones, R. R., Kuenz, M. A., School of Health Sciences, and to Dr. and Moore, R. L. (1985). The optical Llewellyn Legters, Chairman of the De- videodisc in computer based education. partment of Preventive Medicine and Bio Proceedings, Eighth Annual IEEE Sympo- metrics, for his support and encourage- sium on Computer Applications in Medical ment in the project. Thanks also to COL Care, Washington, DC, Nov, 1984, pp. Brendel, Commander of the 10th MASH, 937-940. Ft. Meade, MD, and MAJ Ginnie Boyle, Head Nurse, for major assistance and tech- nical advice in the audiovisual production. Dr. Henderson is Associate Professor Finally, we thank our board of subject at the Uniformed Services University matter experts, who helped to assure the of the Health Sciences, the U.S. federal clinical validity of our program: Mark medical school located in Bethesda, Buchly, MD, FACS, MEDSTAR (Shock- Maryland. He has joint appointments Trauma Department), Washington Hospi- in the Department of Preventive Medi- tal Centre, Washington, DC; and, at the cine and Biometrics and the Depart- Uniformed Services University of the ment of Military Medicine. He trained Health Sciences: Norman Rich, MD, FACS, at the State University of New York at Chairman, Department of Surgery, CAPT Buffalo and Yale University.Dr. Donald Sturtz, MD, FAGS, COL James Sal- Henderson is currently responsible for ander, MD, FACS, Department of Surgery, developing interactive videodisc Eran Dolev, MD, Department of Military medical training courseware for the History, and MAJ Raymond Ten Eyck, MD, U.S. Navy and Army. He is also con- Department of Emergency Medicine. ducting epidemiological research in casualty care, including development Disclaimer of TRAUMABASE, a multimedia data The opinions expressed by the authors do base system for trauma research and not necessarily reflect those of the U. S. teaching which incorporates optical 1 5 6 Using Computers Intelligently 4rArliniiirKif..,w9t1"92' (Aajtisratec) CLIPSONEI 111011ESCE MINIPAINAJW WOWSAAA ASESchle,,AINIAY CASES NAJIASIF CENTACCOM AAO SEAPIRATORTIY ANINST entestrir CASE 1 111011 CASE& SNOW 1USSIVI muaroionu unctrAllat TE16101 11DICIDNINX INELN0001101 1 Meant CASEErbtow3CATIAC var..2,41011 Ima SINSIOARS 111NACI sots 1 CASE C.SPI1E CHEST, Afe FRACTIZilir CK. ak Dere. PERITONEAL SWISH OTSPOSIT/ON Pas I TO 01 CASE TO PI CAIN a TO OR disc technology and the concept of the 1987 Nebraska Interactive Vide- hypermedia. Dr. Henderson was proj- odisc Competition. He lives on his ect director, designer, and writer for sailboat, 'Green Dolphin,' with wife the 'Advanced Combat Trauma Life and two children in Annapolis, Mary- Support' program, which received the land. Best Overall Achievement award at 16 Intelligent tutoring systems: Practice opportunities and explanatory models Alan Lesgoid Learning Research and Development Center University of Pittsburgh I present examples of a focused practice system My purpose in this presentation is to show and an exploratory environment, both imple- ways in which some of the methods of mented using artifidal inf elligence and graphics interface tools now becoming available. Then I artificial intelligence programming and describe a troubleshooting tutor that combines screen interface design can be applied'to these approaches. Finally, I discuss how the these two instructional approaches of prac- loose coupling of the two approaches affords tice and exploration. Then I will describe a opportunities for instruction in metacognitive skills. specialized troubleshooting tutor that combines some of the properties of each Two of the ways in which computers can beapproach. I will be presenting work that is, used for education are (a) as vehicles for to a large extent, the work of colleagues and practice, and (b) as exploratory laborato-associate.They deserve the credit for ries. Both of these modes of teaching andmaking these systems happen, though they learning have a long history. Schools as-may not necessarily agree with the asser- sign homework, which often consists of tions I am making about their products. practice opportunities for refining knowl- edge that may have been presented for-Focused Practice Systems mally in a class. Science courses, among others, often include laboratory sessions inI turn first to an example of what I will call which students learn basic principles and a focused practice system. Such systems acquire a "feel" for a domain by exploring are perhaps the oldest form of computer- some of the systematic phenomena withinassisted education we have. Their purpose that domain. is to present the student with opportunities to practice material that has been intro- Both practice and exploration have posedduced previously, providing feedback serious problems for educators. The prob-about the student's performance. In more lem with practice is that students, left torecent systems, there has been an increas- their own devices, may end up practicinging effort to provide coaching or help as doing a procedure incorrectly -- this oftenwell as post hoc feedback. In an era of happens when children are given arithme-management by objectives and emphasis tic homework, for example. The problemon the bottom line, focused practice sys- with exploration is that not all students are tems seem to reflect an important compo- equally facile at making the principled nent of educational philoc,hy. The idea is discoveries that would justify laboratory to optimize students' time by assuring that experiences. Many of our colleagues have they are practicing material that they are found it possible to overcome these prob-capable of mastering but have not yet lems, in whole or in part, by using themastered. computer to present and monitor practice or exploration opportunities. Early focused practice systems concen- 1 7 Amiwom=1 8 Using Computers Intelligently trated on selecting the rr ,,ht level of prob-a. In the first phase, the student constructs lems for the student and provided onlya verbal statement describing what the binary feedback (corrector incorrect). Such program will have to do. The statement is systems, given sufficiently powerfulpre-constructed from a hierarchicalmenu of scription rules for deciding which exercisesphrases which is based on extensive data to present, have been shown to be effectiveon how students describe algorithms. The for simple skill areas suchas arithmetic. range of statements available allows for the More recent efforts, such as the BIP systememergence of common misconceptions developed about a decade ago by Atkin-about programing and the difference be- son, have moved to more complex practicetween a statement that might sufficiently tasks, such as programming assignments,guide a person and a description ofan and deeper, more knowledge- driven rules algorithm. For example, studentscan de- for problem prescription. For the teaching scribe a loop by describing its first iteration of procedural skills in relativelycon-and then saying "and so on..." strained domains, focused practice envi- ronments are very powerful. Further,asThe student's product is analyzed accord- you will see in the following example, theying to the level of detail it provides and the can be quite rich in the instruction theyspecific components of description thatare provide. present.This two-dimensional analysis allows the tutor to say to the student that he An Example: Bridge or she has a good basic account but needs to refine it, for example, rather than simply Bridge is a practice environment for the having to list what is missing. Bridgecan skill of programming in Pascal (Bonar,infer the specific naive model of the student Cunningham, Beatty, St Riggs, 1987). from the description he or she constructs. Rather than simply giving programming Sonar's most recent work ison a formal assignments, however, Bridge focuses the semantics for computer program plans that task of writing a program by guiding thewill permit the procedures needed for student through three phases. modeling a student's knowledge withre- In the first phase, Bridge asks the studenttospect to a programming problem to be state, in natural language usingmenus, angenerated automatically (Bonar & Liffick, 1987). informal plan for the program. In thesec- ond stage, the plan is formalized byex- pressing it in a visual programming lan- b. In the second phase, the verbal account guage. Finally, in the last phase, the stu-of the student must elaborate theaccount dent uses a structure editor to writeaninto a plan or model of the program's func- actual program. tion. The components of this planare graphic objects on the screen. Each object Figure 1 shows the screen display from therepresents a basic building block ofpro- middle of the first phase. The problem isongrams, such as a loop that terminates when the left, and the emerging student plan ison a specific control value is detected for a the right. Figure 2 shows thescreen duringcontrol variable or an input routine that the third phase.The visual plan con- fetches one integer. The overall plan for the structed by the student is on the left, and theprogram can be thought of as a system task is to build on the right an actualpro- whose components are the low-level build- gram that realizes the plan. ing blocks. In order for the overall plan to be complete, the components must becon- Let me describe Bridge's three phases nected in two ways, indicating control flow briefly. and data flow respectively. As can be seen Ve, YiOwi4 rt..1 7 .4,s 4 4 . 10 Using Computers Intelligently in Figure 2, control flow is marked by inter-the student. It is easy to label certainper- locking the basic building blocks pieces,as formances as wrong, but it is not soeasy to in a jigsaw puzzle. Data flow is indicatedcommunicate sufficient information to the by pointing first to a datasource, such as an student to assure that he will know howto icon for a computed value, and then toado anything better next time. A good in- data port, such as the slot in an output planstructor will ask the student what hewas that indicates what is to be output. trying to do and then offer advice of the form, "If you want to accomplish X then do The second phase plan can berun as if itY." Intermediate planning languagesserve were a machine. When this is done, the data the same function in Bridge. They allow the values actually move around, and thestudent to overtly express plans and inten- building blocks reverse coloras they exe- tions. With such languages, diagnosis inan cute, so both data and control floware intelligent tutor is enormously simplified. illustrated. So, students finishing thesec- ond phase have a very specific mental pic-There is a second benefit to this approach. ture, expressed overtly on the screen, ofA basic characteristic of novice problem- what their program will do. solving performance is that it is too much centered on the specific tactics thatare c. In the third phase, the student actuallyavail...1,1e and too little concerned with writes code. Two things are importanttounderstanding the problem and develop- note here. First, the student doesn't get to ing a plan for its solution. The intermediate this point until he has a clear plan, until helanguages of Bridge are planning lan- and the computer agree on what the goalsguages, so the student is being channeled of the program are. Further, the studentinto a more expert approach to problem can point to various parts of his plan and solving, in which the problem anda plan thereby indicate what piece of planany for its solution are both represented clearly given line of code corresponds to. before coding starts. Second, the usual frustrations of beginningOverall, Bridge exemplifies three issues programming, syntax mistakes,can bethat are worthy of consideration in devel- eliminated by providing structure editorsoping focused practice environments: (a) that automatically assure the correct loca- focusing on parts of the performance that tions of semicolons and other Pascal exot-are usually internal, such as an initial prob- ica. In an old- fashioned classroom, syntaxlem representation and a plan for solution; errors may betray either low-level me-(b) using overt expressions by the student chanical failings or serious misconcep- of plans and intentions to guide the coach- tions. With the Bridge approach, sucher- ing process; and (c) providing the student rors are inevitably mechanical.With awith tools, such as the structured editor, structure editor that allows building block that maximize the proportion of time spent shells to be selected from a menu, thesyn- in significant problem solving practice. tax barrier is completely overcome. Exploratory Systems There are important reasons for using the intermediate representations in naturalI turn now to an entirely different form of language and then in the visualprogram-computer-based instruction, the labora- ming language. Consider the task faced bytory environment.Laboratory environ- a programming instructor in a course. Thements, whether on computers or in more student has a program that is incorrect, and traditional form, are meant to deal witha the instructor must figure out what to say to specific problem. This is the largegap 20 Using Computers Intelligently 11 between the ability to verbally state a prin- some coaching that helped students not to ciple and any understanding of that prin-miss important events and to be systematic ciple. It is generally believed that if stu-in their explorations. A computer might dents discover a principle themselves, then also be able to provide coached laboratory they will understand it, oecause itis experiences, and there are several good grounded in their own personal experi-reason for using a computer for this pur- ence, whereas if they are simply told the pose, in addition to the serious interna- rule, they may net clearly understand whattional shortage of good science teachers. phenomena the rule refers to. There are Computer-simulated experiments are eas- many experiences in education that seem toily repeated and can be observed from support our faith in discovery learning. For many different viewpoints. They are safe, example, there are now quite a large num-and they can be made relatively independ- ber of demonstrations that students whoent of the motor skills of the student. are quite able to solve textbook problems in physics are not able to apply the same prin-How can an effet. Ave computer-based ciples they use for those problems to every- coached laboratory environment be devel- day situations. oped? We can get some good ideas by considering an exemplary piece of work Discovery learning does have a problem,done by some other colleagues at the though. For the slower student, the wait Learning Research and Development Cen- until a discovery occurs can be very long, ter, a tutor for a simulated economics labo- Regrettably, it is also the slower student ratory they have called Smithtown. who is less likely to expand beyond rote learning to actually understanding what An Example: Smithtown principles of science or technology really mean. If we could somehow make discov- Smithtown simulates a small town's econ- ery learning wore efficient for the slower omy (Shute, Glaser & Raghavan, in press). student, this would surely be a very -vorth- The student can do experiments in which while thing to do. any of a set of factors (price, consumer preference, town population, mean in- In order to tackle this goal, we need to view come, minimum wage, weather, and num- discovery learning as a form of problember of suppliers) can be manipulated for solving. The problem to be solved is to any of a set of commodities. The effect on specify the regularities present in some quantity supplied and quantity demanded constrained environment, taking account is then simulated. A variety of tools are of any adclit,onal constraints supplied by provided to the student, including a note- the teacher. Laboratory manuals generally book/database, graph plotting tools, and a consist of suggested activities followed by means of expressing hypotheses about the questions that are asked about those activi- effects of various manipulations. ties. Implicitly, we are hoping that students will carry out the activity and then try to My colleagues have used Smithtown as a understand what they observed, using thelaboratory themselves, for studying how questions of the lab manual as clues. Ofstudents use such laboratory environ- course, not all students do this, so labs arements. By comparing students who seem not always effective. to discover principles in their experimenta- tion in Smithtown with those who do not, A good laboratory instructor might watch and also by comparing students of differ- what the students are doing, examine the ing intelligence scores, they have been able entries in their lab notebooks, and provide to specify a set of behaviors that the better 21 12 Using Computers Intelligently students show when they experiment in pelted if many variablesare manipulated in this economics laboratory. They havetheneach experiment.So, Smithtown hasan translated these findings into a set of ex- internal watchdog or critic that watchesto plicit rules for coaching explorationin this.3ee if the student repeatedly manipulates laboratory setting. The current versionof multiple variables in a single experiment. Smithtown, then, can watch for evidenceWhen this happens, the following pieceof that a student is missingsome of the ex- advice is offered: ploratory dispositions shown byexpert learners and can then coach the student,To really understand causal relationship suggesting more systematic approachesto (how a change affects the market), it is best experimentation. to vary only one variable ata time. You have been manipulating toomany vari- I will provide three examples of the kinds of ables at a time. Modify this behaviorto rules that Smithtown uses (Raghavan,per-make things easier for yourself. sonal communication, 1987). Eachdeals with a different kind of experimental skill.A second example reflectsa mixture of The first is the most straightforward.economic and more general strategic Smithtown, like many laboratory situ- kiii.,-wledge.If the student manipulates ations, has a number of variables thatcan price of a good and then changes toa differ- be manipulated. However, it is virtually ent commodity before seeing the effects of impossible to figure out what has hap-the prior price manipulation, there isa Figure 3: A Sample Screen fromSmithtown Wt.atI14 C.1%tnlerPrat, Marko{ 3666 16 l S410141 14 14 311111414164 1.1.1 el 01.1 01 01 .1101 1.1,13 .1.11.1.1.1,'? Jan V.1. 014,1 I.t-nCalst No of Smigawrt 2tilt tre tuir Isin. voel 163 46 WO 14 hi 44 44106 1 .181 .1 .16 1 .1 .1 .1 el. 1 3 211 36 3 SO 7. Si 6340 1.0;13 26 40 66 64 70 66 66166 1 / 11011 00101.1 01011 01 1.1.1.1 1 101 01.1 0/ .1 Inlorpo liatna 414515SIs111.11.1 3:00441 45 33 761416014W MI al 0 0, 0 F A:41.1777T=it".nr:114.0 t.114iiii4..i. Hat e tt 22 Using Computers Intelligently 13 possibility that the price change in the first can begin to examine transfer between good might influence the price of the sec-domains. ond good without the student even realiz- ing this. So, the student doing somethingGetting the Best of Both Worlds like this 2s given an appropriate warning. The third kind of criticism is providedWe've now considered two very different when the student's verbal hypothesescomputer-based instruction paradigms, don't match well with his experimental focused problem solving practice and labo- behavior. For example, a good hypothesisratory exploration. Each contained a little should state an effect on a dependent vari- bit of the other. For example, the second able that is produced by some manipula-phase of Bridge, the visual programming tion of one or more independent variables.phase, is partly a laboratory. One builds If a student claims that changes in a de-and runs program plans. Similarly, one can pendent variable produce some effect, thisimagine making the exploratory econom- is worthy of coaching comment, and ics environment quite directive, with prob- Smithtown provides this coaching. Over- lems like "What happens to quantity sup- all, the coaching aims at differences of the plied as price rises?" However, we have following kinds, that were found betweennot yet seen a systematic effort to combine faster and slower learners who usedexploratory environments with focused Smithtown: problem solving, and the design issues raised by such it combination have not been 1. Generating and testing hypotheses. fully explored. 2. Engaging in richer experimentation within a single market setting ratherAn instructional system my own labora- than randomly changing markets. tory at LRDC has been working on pro- 3. Replicating experiments to be sure thatvides some opportunities for refection on they knew what produced the effects. how the two approaches might be com- 4. Changing one variable at a time, hold-bined and on what concerns arise in such ing everything else constant. hybrid designs. We have been developing 5. Making big enough changes so that ef-a tutor for a specialized electronics trou- fects could easily be noticed. bleshooting job in the U. S. Air Force. The 6. Generalizing hypothesized effects people our tutor will train are responsible across markets. for repairing navigation equipment for a 7. Having systematic notebook skills. fighter plane. They have lots of trou- bleshooting guides and test equipment for Overall,.then, Smithtown has provideddoing this, and they do it well. evidence that skills associated with lesser learning ability can be tutored by a com- However, the test station, forty cubic feet of puter-based laboratory.An importantelectronics, that they use in their work also question is whether the effects are local or breaks from time to time, and fixing it is a general. That is, does the coaching merely lot harder than using it to fix parts from improve the learning of economics fromplanes. Our tutor provides a partial simu- the economics tutor, or do the coachedlation of this test station and also poses exploratory skills generalize to other labo-specific diagnosis problems.It provides ratory situations. To test for this, my col- advice and focuses problem solving activ- leagues have developed two other explora- ity considerably, but it is also a reasonably tory environments, for simple electrical rich exploratory environment, and we principles and for geometric optics, so they could make it richer. For example, within limit'', we could probably produce an in- 14 Using Computers Intelligently structional system that would allowa stu-take just a minute or two to describe the dent to ask for a simulation of the effects ofstructured interview technique, since I any of a large set of component failures onthink you will find it useful. overall test station performance. Structured Interviews during Problem Solv- This project started when we were askedto ing: The Concept of Effective Problem Space take on a primary role in developing and testing methodology for a large Air ForceIn our first effort to develop a task analysis effort to improve the quality of training andscheme that would be sufficient to inform performance measurement fora number ofdevelopment of an intelligent instructional their more technical occupational special- system, Drew Gitomer developed a trou- ties. Specifically, we were asked to explore -leshooting task that involved detection of the feasibility of using cognitive psycho- complex faults in the test station used by logical approaches to task analysis our subjects prior Air Force data sug- (Lesgold, Lajoie, et a1.,1986). As that work proceeded and cognitive task analysisgested that this was their hardest job. Asa first formative approach, he simply vide- approaches were developed, we were alsootaped subjects attempting to solve such asked to demonstrate the efficacy ofour fault detection problems. He thenexam- analytic approaches by showing that skills ined the protocols (transcriptions of the we identified as critical and not uniformly tapes) and attempted to count acquired would, if taught, make a variety of a differ- activities that seemed relevant to metacog- ence in airmen's job performance. Thisnitive as well as more tactical aspects of latter effort is now underway and involves problem solving in this domain. While the development and field testing of an intelli- gent tutoring system. results, published in his thesis (Gitomer, 1984), were of great interest,we wanted to move toward a testing approach that was To carry out this project, we have had to less del,. endent upon skilled cognitive develop cognitive task analysis techniques psy- chological training. That, after all, is that we believe are of broad general inter- one aspect of what test development is largely est. In particular, we have developed a aboutrendering explicit the procedures combination of structured interview and that insightful researchers first apply in think-aloud protocol that seems quitegen- their laboratories to study learning and erally useful. Several other approacheswe thinking. Also, we felt that ina technical have developed (Cf. Lesgold, Lajoie et al., skill, much of a person's capability is 1986) also seem likely to be useful. Further, very we have found it beneficial to comparemuch grounded in the specifics of the task (c.f. Scribner, 1984). Thus, we expecteda subjects at multiple levels of competence, method more heavily driven by domain so that we can describe competence as the expertise to be more productive thana presence or absence of certain key features "weak method." during realistic (job-like) problem solving performances. The features, in turn, signal In fact, our breakthrough came notso much specific procedural knowledge thatseems from our psychological expertise but rather to be differentially acquired by those who from interactions of the three ofus, who learn well on the job. Our current task is to had substantial cognitive psychological try, via intelligent coached practice in a training, with an electronics expert, who simulated job environment, to provide thishad extensive experience watching novice procedural knowledge and related concep- troubleshooting performances. He tual (declarative) knowledge to those air-pointed out that it was quite possible to men who do not already have it. Let me specify the entire effective problem space, 24 Using Computers Intelligently 15 even for very complex troubleshootingpurpose is.Finally, when a subject was problems. That is, he could pretty com- headed well away from a reasonable solu- pletely tell the rest of us all of the steps that tion path, we would, at preplanned points, an expert would take as well as all of theredirect his efforts back to more fertile steps that any novice was at all likely to take ground. in solving even very complex troubleshoot- ing problems. In this case, the task was to In more recent work, we have standardized find the source of a failure in a test station a set of questions that are asked at each that contained perhaps 40 cubic feet ofnode of the problem space. These ques- printed circuit boards, cables, and connec- tions address the precursor assumptions tors, but various specific aspects of the job implicit in reaching a given point of the situation constrained the task sufficiently problem space, the action to be taken at that so that the effective problem space could be point, the expected results of that action, mapped out. and the interpretation of the results both in terms of a model of the device and in terms This then created the possibility that weof appropriate next steps. We call this the could specify in advance a set of probe PARI technology, because it focuses on the questions that would get us the informa- Precursor, Action, Result, and Interpreta- tion we wanted about subjects' planning tion at each node of the problem space. and other metacognitive activity in the troubleshooting task. For what L erobablyChristopher Roth and Gary Eggan have the most complex troubleshooting task we taken our methods and applied them to a have ever seen, there are perhaps 55 to 60 slight variant of the domain we studied. different nodes in the problem space, and However, they spent more of the total ef- we have specific metacognitive probefort comparing the specific problem-space questions for perhaps 45. paths of experts and novices. The next few figures illustrate the yield of this work. Figure 4 provides an example of a small Roth worked with Eggan to develop a col- piece of the problem space and the ques-lection of diagnosis probiems that were tions we have developed for it. characteristic of real problems faced in the work environment and that were likely to An examination of the questions in thebe solved correctly by the best technicians Figure reveals that some are aimed at veryand not by many of the less-skilled, first- specific knowledge (e.g., How would you term airmen. People with experience in do this?), while others help elaborate thehandling the tasks on which many airmen subject's plan for troubleshooting (con- have difficulty tend to have a good sense of sider Why would you do this? or What dothe kinds of problems that are in this class you plan to do next?). Combined with(representative of the domain, solvable by information about the order in which the the better technicians, and not solvable by a subject worked in different parts of thesubstantial number of workers), and problem space, this probe informationEggan's advice has proven itself empiri- permits reconstruction of the subject's plancally. for finding the fault in the circuit and even provides some information about the Figure 5 illustrates the effective problem points along the way at which different space that Eggan helped Roth construct for aspects of the planning occurred. In fact,the problem. It was felt that the bulk of we went a step further and also asked aproblem solving activity would fall within number of specific questions about howthis space, for both experts and less-skilled critical components work and what theirworkers. 25 16 Using Computers Intelligently Rerun Test * Why did you rerunthe test? Tighten < Cables * Why would you do this? V Single * Why would you do this? Step * How would you do it? Check * What does it tell you? Reseat/ Rerun * How do you do this? Swap < Test * Why would you do this? A2A3A1 1j Check/ * How do you do this? * Why would you do this? Swap * Is this what you would do DMM in the shop? V Swap * How do you do this? * Why would you do this? Cables * Is this what you would do in the shop? Fails Again: * What do you think the problem is? * What do ycu plan to do next? Same Reading Figure 4:Effective Problem Space Methodology First Version 26 Using Computers Intaigently 17 Effective Problem Space Hypothese: Not hardware fault .------"--'* w Ragusa second OU. Olen thet Oil corm:sand Check Oat owltdses Iseult Can't OU. mitered property on TN set correctly Result: Can't OU. Result Can't DU. as se r hypothesis: Syrnmom- due to hardware fault Novicespace i am am Select troubleshooting strategy +-- Run diagnostics Usesploblwitilstrategy t,.- sunctIonel anelysis 1 4.-----" lissutt Falb Ouock Check a Check. 7 byte cause hne by OtLen with swot byte count mother console. of no dim into console intoMut of MUX a adaptor in SAUX Result Cm Result gad ResultBad 6 Result: SdN Reset: Can't byte count byte count an OU. OU. a/consols / we i .11 I i Conclude: Conclude: Conclude: Conclude: Condude: Fault Fault NOT Fault Fault Fault M components upstream upstream irs amnion tOMMOO to upstream components both MU Figure 5: Effective Problem space Segment, Showing Expert-Novice Differences Roth and Eggan then proceeded to run specific steps they would take in attacking subjects on this task. The problem wasthe problem. Whenever an overt step was posed verbally to airmen who had access tostated by a subject, he was immediately the full set of Technical Orders (documen-informed of the outcome of that step. tation fc,r the devices involved). Techni- cians were encouraged to offer their hy-Figure 5 also shows the basic results Roth potheses, state their plans, and announceand Eggan obtained. The experts, whose 18 Using Computers Intelligently performance is outlined with a solid line,to teach come from comparisons between used a subset of the effective problemeffective and less effective workers in the space, avoiding some of the steps thr t were job environment. In the case of our tutor, less likely to be productive. The novices,such comparisons showed us that we whose empirical problem space is outlinedneeded to focus on tracing of signals in heavy dashes, failed to make some of the through schematic diagrams, the more expert moves and made some moves thatsophisticated uses of test equipment, men- experts would not make, such as swappingtal models of the process, within the test a disc pack (see bottom left of Figure 5). station, of carrying out a test of an aircraft Further, their empirical problem space was component, and domain-specific customi- discontinuous, a set of islands. To L.; 'terzations of some general troubleshooting understand what knowledge separated trte strategies. We then put our subject matter experts from the novices, Roth and Eggan experts to work generating problems that then tried to determine the specific knowl- treated the issues we had identified and edge that experts used at critical points inthat involved various parts of the test sta- the problem space. Knowledge involved in tion. For each problem, we have an effec- making a move that experts make andtive problem space data structure that novices do not is a clear candidate forcontains all the information needed to additional training. interact with the trainees as they attempt to solve the problem. A final issue that Roth and Eggan ad- dressed was the matter of the discontinui-Each node of the effective problem space ties in novice problem spaces. In essence, for a problem is represented as an instance the gaps represent impasse points whereof a computational object.When the the novice does not know what to do. Attrainee chooses to move to a node of the those gaps, we see relatively random be-problem space, its corresponding object havior, driven by the snippets of knowl-receives a message telling it to take charge edge the novice has. For example, in theof interactions with the trainee. The object case shown in Figure 5, there is initially ahas the ability to generate certain kinds of period of board swapping that is not di-hints and contains specific text for other rected specifically at the problem. Then, hints that might be needed at that point. It there is an effort to think about the pathknows what actions can be performed at involved in a download. The multiplexorthis point, what prerequisites those actions (MUX) somehow becomes the focus of at- might require (e.g., shutting off power tention, and thought about the MUX con- before removing a card from the station, tinues until the novice is redirected toward getting the right schematic for circuit trac- the disc drive, whereupon the right possi- ing, etc.), what the results of those actions bilities are addressed, but without concernare and what they mean, and what options about the relative costs ana benefits of the to present to the trainee as the next step. five alternatives. Each node also contains a list of the curricu- lum goals relevant to it. From Testing to Tutor Development This curriculum goal information lays the The Effective Problem Space PART technol- basis for the student modeling activity that ogy gives us two things we need in order to is needed to assign problems to the trainee build an intelligent tutor: a technology forand to decide how supportive to be in offer- deciding what to teach; and a technologying help at various points in the trainee's for constructing individual problems forefforts to solve a problem. The student the tutor to present. The decisions on what modeling process we use is still very primi- 28 Using Computers Intelligently 19 tive. When a student starts a problem, annext step was to add an exploratory compo- estimate is made of his likelihood of suc-nent that permitted the student to try out cessfully negotiating each node of the prob-some of his ideas. We did this by adding a lem space. This estimate is made on thesimulation of the front panels of the test basis of a long-term student model thatstation and an ability to conduct simulated shows the learning state of each curriculumtests of the test station circuitry by placing issue If the student has mastered the issuesmeter probe icons at various points in sche- relevant to a node, then we can expect himmatic diagrams of the innards of the test to do well at that node and can thereforestation. give only subtle hints. On the other hand, if we know a node requires knowledge theA few pictures are worth many words here. student has not exhibited, we should pro- Figure 6 shows the screen in the midst of a vide more explicit help. Our categories areproblem. A representation of the front of currently simple: Good, OK, and Bad arethe test station is shown. Pressing the the possible levels of expectation for anymouse button in any of the 12 boxes on the given node. We call the expectation infer-left would cause the front panel corre- mati on that is generated fora specific prob- sponding to that part of the test station to lem when a trainee attempts it the perform-appear. The choice of how to proceed from ance model of the student for that problem.the current problem space node is made by After the problem is completed, deviationsselecting from the menu on the right. The of the performance model from actual stu-picture was taken right after the student dent performance (e.g., did he breezeasked fora hint, which can be seen in the through a spot where he was expected toinformation box in the upper left. have trouble, or did he ask for lots of hints at a spot he should have been able toFigure 7 shows the screen after one of the handle) are used to update the ratings as-front panels has been requested. The stu- signed to each curriculum issue for thatdent can move knobs by mousing the set- student. We call these ratings the compe-ting he wants. When knobs are moved, tence model of the student. meters automatically update. The student can also request documentation on the So, we assign a problem by matching prob-screen. lem characteristics with cur current assess- ment of what the student knows how to do,Figure 8 shows a page of the Technical we generate an expectation of his perform-Orders, which are the documentation for ance at each point in the problem space bythe test station. They are accessed via an noting what curricular issues are relevantindex on which the student can mouse- at that point and what we know about thebutton for the entry he desires. trainee's competence on those issues, and then, after the problem is completed, weFinally, Figure 9 shows a schematic dia- update the competence model to reflectgram. The student can ask for measure- unexpected aspects of trainee performancements of voltage, current, or resistance as he went through the problem. values by calling up an icon of a meter, setting its front panel, and then indicating Adding an Exploratory Component where its probes should be placed in the schematic diagram. The upper left of the So far, what I have described is a focusedschematic has above it the icon for a problem solving tutor. We give the studentSimpson multimeter, and its red and black a problem and, as necessary, provide hintsprobes have been placed near the middle of that keep him focused on its solution. Thethe schematic on the left (the red probe 29 Using Computers Intelligently OM et the seism pintails beim Imago, the Test Station is in the Meesuremeet hatesienal am& Mhapt you eta &mew where the 14TO 54 shalt *Menthe Tea Station mot teen yp el Int st gut pas* Figure 6: Sherlock Screen Showinga Hint Being Given Figure 7: Sherlock Screen Showinga Manipulable Panel vowA:ttm.1.4:40044084..v....wk.t.roos s 1.44...14441,4 w*Ins oz.> ra .4tit titilg405 liaaret w Centro Mionu QUIT PAINT @MOM, HELP . tt, '041e,..X.,WAA161. k .1.'111101IIiIR Table 1.1 54014.$4 4.44.1 silnia POS. some'WV IND. 0.41.4t 461110111U I Ii MOO SPAT lip 411N IWO 0 1t..ad iv, 130 IIIMOOMNI111 !MAI SIOCAON 10441101161 / Lff AllffiooMb hos es 40114101.100 Iiffas ON in%via tfi..;aowl 0;4.;.t.pp ! (MORE) insefientlik 40.4$ ..! 4,5Fiet p, .,;* It Ck« 'Ri P' ,t1 "4-j. 4` PE 51".."1! Figure 8: Sherlock Screen Showing a Page of Documentation looks like a meter probe, and the blackeral exploration or restricting exploration probe looks like the symbol for ground). to a specific circuit compcnent. When the Obviously, there are test point combina- trainee tries to perform a specific test, he tions for which we will not be able to pro-must, of necessity, have in front of him the vide result values, but the number of testscheme:: that shows the points at which he values seems to be adequate so far. wants to make a measurement. When he tries to make a measurement, the object in Coupling the Exploratory Environment and the charge of that schematic can impose special Focused Problem Solving Tutor conditions, such as turning off the power, before the test is permitted. Similarly, the The remaining task is to couple the explora-object representing the test instrument may tory environment, the simulation of the testimpose conditions on how it can be set. station, with the menu-driven focusedWhen all conditions are satisfied and the problem solving tutor. This is done in aprobes are placed, the meter icon is auto- very simple way. At some, but not all, matically updated to show the test value. points in the course of problem solution, the menu includes the option of doing aThe system keeps a record of all tests it test. When the trainee takes that option, heperforms. When the student indicates that gains access to the exploratory environ- he is done performing tests, a miniproduc- ment. The tutor can control what explora- tion system attached to the sche.aatics on tions he is permitted, allowing either gen-which tests have been performed is al- 3l 22 Using Computers Intelligently 11A.VV,1 13,4 Selman hem law the Genetic Miens menu of tit. Asian mow when yell an nasly Y tonsinut frd C e n tt s n u ^r; 111.1 Uhrm 141A y Alf Prh Grr,a hrtIonIxt.rOtagtamStvnt Ih Ir s. ---1s AS. J2 -t1It moo MU AVA Om AS IIICSTPOW ma AWN,. *WIN Ile LI Nit Ilurt VII Is OMR IS/ IS lbseeA eAll Casa nor1110/11.4 VX41 $1042/I Itilm Sw6,1186 Cards (411oaf nit Jill law* RAG tank to tuccitte YR Inn et4164* Hut* taito6111 DI6666 ....poiII it 4.4 4 a , .'*1 jt, . ";* )e. 4 s. grg . A).701,0 N.17,6,80. vaen.M.nAnn.s>,41,) s. Awe.. . .K ....a9.0A Mol.**, v;))6*).7*,>A Vt 4)4.v) 14.)(7,1!.....4v). 4X4,71,4 Figure 9: Sherlock Screen Showinga Testable Schematic lowed to execute. It contains rules thatSo, with a relatively simple convention,a show the relationship between variousproduction system that inter-relatesex- problem-space activities and various tests. ploratory actions with the representation A rule might say that if certain parts of theof the problem space, we can couple the problem space have already been negoti- two approaches. We currently write pro- ated and certain tests are performed, thena ductions by hand to do this, but it is proba- large portion of the test station is ruled outbly possible to build intelligentcompo- as the cause of the fault being diagnosed.nents to do this automatically, assuming The problem space nodes relevant to workthat the semantics of each problemspace in the ruled-out portion are notified thatobject are part of its knowledge. One they are no longer necessary steps. Later, ifwould simply ask the objects relevant to the the student enters ruled-out portions of the part of the test station that had been ex- problem space, he may be allowed to pro- plored to decide whether the tests that were ceed but will eventually be told that hedone were relevant to themselves. When really had ruled out those possibilities an object decided that a test was relevant to when he was performed a test earlier (theitself,it would, given an appropriate test can be described, of course). method, then decide what the effect of 32 Using Computers Intelligently 23 conducting the test should be on the rele-carries out, and thereby demonstrates, a vance of that point of the problem space for systematic set of metacognitive strategies further search. that the trainee would do well to model. On the Value of Not Being Perfect Heidegger observed that for the most part we are controlled by our recurrent, every- I conclude by noting that the techniques we day, expected environment. Only when have used are provably incomplete. Theo- breakdowns occur in the expected situ- retically, there are actions that a trainee ation do we have to think about what to do. could want to take that are not representedOnly then can we learn. An interesting in our effective problem spaces, and thereaccident of the approach we have taken in are sensible tests one could perform that designing our tutor is that we cannot avoid our experts might not anticipate.The some breakdowns of the tutoring process. hypothesis we are now testing by taking Upon reflection, these breakdowns may be ou.: tutor to the field is that such imperfect the best opportunities for teaching that our approaches can still provide lots of useful tutor has, and the default techniques it can instruction. It is not clear that a completeapply may be useful techniques for the simulation of a device as complex as the test trainee to encounter and to see demon- station we are working with can be devel-strated. oped at an affordable price. Further, per- fection, while it may be comforting to theIn the complex interdisciplinary world of designer, may not be necessary to effectivecomputer technologies for education, I instruction. We see ourselves as involvedexpect to make many mistakes, to seldom in an activity sii.nilar to that of many indus-produce perfect instruction. Consequently, trial designers, designing a system which isI look for ways to build on mistakes and not perfect but which can accommodate, insufficiencies because I can count on hav- through a variety of mid-course correc-ing them. At the moment, it looks as if this tions, to any problems for which it was not is a fruitful way to proceed. fully prepared. References One of our accommodation techniques, the panic button, has actually turned out to beBonar, J., Cunningham, R., Beatty, P., & quite interesting as a new source of instruc- Riggs, P.(1987).Bridge: Intelligent tion. When the tutor completely loses track Tutoring with intermediate representa- of where the trainee is in the problem space, tions. Technical Report. Pittsburgh, PA: perhaps because the set of tests the trainee University of Pittsburgh, Learning Re- performed in the exploratory environment search and Development Center. was pretty random, it tells the student this. Bonar, J. G., & Liffick, B. W. (1987). A visual Also, when the student is stumped and the programming language for novices. Tech- hints he is getting are not helpful, he can hit nical Report. Pittsburgh, PA: Univer- the panic button. In panic mode, the tutor sity of Pittsburgh, Learning Research works with the student in a top-down and Development Center. manner, discussing what the student has Gitomer, D. H., (1984). A cognitiveanalysis of done so far and what parts of the problem a complex troubleshooting task. Ph.D. Dis- space remain unsearched. It then asks the sertation. Pittsburgh, PA: University of student to proceed from the top, specifying Pittsburgh. which subspace and which subsubspace,Lesgold, A. M., Lajoie, S., Eastman, R., etc., he wishes to enter. Put another way, Eggan, G., Gitomer, D., Glaser, R., when the tutor doesn't know what to do, it Greenberg, L,, Logan, D., Magone, M., 33 24 Using Computers Intelligently Weiner, A., Wolf, R., & Yengo, L. (1986, April). Cognitive task analysis to enhance technical skills training and assessment. Technical Report. Pittsburgh, PA: University of Pittsburgh, Learning Re- search and Development Center. Scribner, S. (1984). Studying working in- telligence. In B. Rogoff & J. Lave (Eds.), Everyday Cognition: Its Development in Social Context. Cambridge: Harvard University Press. Shute, V., Glaser, R., & Raghavan, K. (in press). Discovery and inference in an exploratory laboratory. In P. L. Acker- man, It J. Sternberg, & R. Glaser (Eds.), Learning and Individual Differences. San Francisco: Freeman. Alan Lesgold graduated from Stanford University in 1971 with a PhD degree in psychology. Since then, he has been at the University of Pittsburgh's Learning Research and Development Center, where he is currently Professor of Psy- chology and Associate Center Director. With Richmond Thomason, he directs the University's doctoralprogram in intelligent systems studies, which fo- cuses on formal characteristics of intel- ligent organic and artificial systems. His laboratory has been developing prototype computer-based training systems that are based upon cognitive psychological principles and artificial intelligence techniques. These intelli- gent training systems also function as tools for instructional research. Do- mains for which systems are being developed include electronics trou- bleshooting, programming, flightengi- neer duties, electrical and logic prin- ciples, economics, and composition (writing). Lesgold has been involved in a number of efforts to plan research strategies on applications of computers to education, both in the United States and elsewhere. 34 The Physics literacy project 1987 D.E. Bailey*, P.F. Logan", P.J. Walker*, I. Walsh " School of Science and Technology, Nepean C.A.E. " Physics Department, University of Technology In 1986 the Commonwealth Tertiary Education Commission, through the Participation in Eq- student and (b) gives the student a "hands- uity Program, supported the Physics Literacy on" physical experience of controlling di- Project. This was a proposal to devise a number rectly the on-screen events. Despite diffi- of computer assisted learning packages to assist culties experienced with the language and students with little or no background in Physics, to overcome the initial hurdles to a Tertiary its implementation on our basic 512 K Macs Physics course. As the program matured this and the lack of acceptable large screen had developed into seven CAL programs using projection facilities we are still delighted the Macintosh computer and the programming with our choice of machine and method of language Forth. The project has a strong em- phasis on experiential learning of basic concepts operation. using simulations. This paper describes the progress of the project and will present a num- The Project ber of Issues relevant to the production of CAL material and particularly CAL material using simulations. The initial proposals emphasised the team design of software modeled on Bork (1984). This has had mixed success. For the team The Physics Literacy Project was reported members the process has been very re- on at CALITE - 85 (Bailey 1985) indicatingwarding as it has forced them to go back to the origins of the project from a concernbasics and to examine seriously their teach- over students who were attempting terti- ing approaches. However some designers ary Physics programs with a lack of secon- preferred to work as individuals and some dary background in Physics. This concernteams lost members due to commitments is supported year after year at many insti-outside the project. tutions by a continued attrition or high failure rate amongst these students (80- 95 It was also found that the revision and %) design process needed feedback from us- ing the machine and this was not always One approach towards assisting these stu- available. Due to the quaint way funding dents is to produce Computer Assisted works, the project has had two program- Learning (CAL) material. This project se-mers, both on one year contracts. Part of lected 7 topic areas and work has pro-their time has been spent learning the ceeded towards producing teaching mate-machine and language before producing rial for the Macintosh computer using product. Early work has concentrated on Forth as a programming language. Thethe production of graphic sequences to- justification for the computer choice waswards later incorporation into teaching based on the interaction possible via mousesequences. This has resulted in a number of between operator and machine which (a)exciting mini programs of great velz: reduces the need for keyboard skills by the towards supporting a teaching program 3a''' I 26 Using Computers Intelligently but unfortunately not specifically appropri- preparation for trialling. It is anticipated ate to the immediate goals of the project. by December that a substantially larger number will be available for demonstra- Work in progress tion, for assessment and for feedbacksug- gestions. Screen lumps of representative At the time of writing this paper eight appli-portions of each of these are included for cation packages have been completed in information. r dFile Dropping TROLLEY MASS BRICK MASS 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 ; NOIR 1;111-1111M1AZIII DM Et: ASO= 111.0111111if I Li roe VELOCITY DISPLAY INITIAL VELOCITY ( START BEFORE AFTER 1 2 3 4 5 6 TROLLEY VEL 1 4.001 A Figure 1. Dropping. Conservation of momentum is illustratedwith student control over the mass of both trolley and brick. r* File Colliding PROJECTILE MASS TARGET MASS 1 2 3 4 5 6 7 8 9 1, ICEE11141331 Ql sr 0 VELOCITY DISPLAY INITIAL VELOCITY START BEFORE AFTER 1 2 3 el76 PROJECTILE VEL 4.00-1.33 TARGET VEL 0.00 2.67 ® Elastic 0 inelastic Figure 2. Colliding. Elastic collisions are accessible fora range of masses from 1:9 to 9:1. 36 Using Computers Intelligently 27 r * Fd Sticking7-7777-77--7=- PROJECTILE MASS TARGET MASS I2 3 4 5 6 7 8 9 2 3 4 5 6 7 8 9 5141411112111g111111MIMMIIIIIIIIE S. INITIAL VELOCITY ( START ) VELOCITY DISPLAY 1 2 3 4 5 6 BEFORE AFTER M111111010111 IIk111Ii1t C »: VELOCITY I4.001 2.77 ()Elastic Inelastic Fig 3. Sticking. Inelastic collisions with variablemass ratios and total mass are avail- able in this program. * File Exploding CANNON MASS (x 1000) CANNONBALL MASS (x 10) I 2 3 4 5 6 7 8 9 1 2 3 1 1 6 7 1 9 EIMUM111 MEI :111;liffi:11:11i".?: I =II Alz VELOCITY DISPLAY MUZZLE VELOCITY (x 10) (SPIRT RECOIL VEL -0.54 2 MUZZLE VEL 30.00 Kir1 II5 6 Fig 4.1 Exploding. As with other momentum simulations this enables the studentto vary the mass of the gun and projectile over a wide range. 37 Using Computers Intelligently r 16 File Pressure/Volume Pressure so 45 40 Also& 35 ( Hide Table 3 30 25 ( Hide Graph ) 20 ( Clear Graph ) 15 10 6 5 5- 0 Temperature 0 5 10 15 20 25 30 33 40 45 50 Control 3 Volume 2 V 1 26.7 L P 127.0 Fig 5. Pressure/Volume. The properties ofa gas depend on a number of variables. Here a student can vary either the Pressure or the Temperature and observethe volume and pressure changes. r IFile Hooke's Lai Extension 13°- 0 120 - 1-0 110 - 100 - x-20 (Hide Graph 90 - -30 80 - - 40 ( Clear Graph 70 50 (Draw Line) 60 - - 60 SO - L-70 r 40 -80 30 - 90 Spring 1 I i 11,1.111 0 30 100 150 200 230 300 330 400 450 500 -100 Stiffness Mass 110 120 Fig 6. Spring. The extension of a spring of variable stiffnesscan be examined for a range of masses. 38 Using Computers Intelligently 29 r aFile Dropping in Fluid Distance Timer 100 - (o.o 90-90 (Plotl 80- Drop 70- (Hide Graph ) 60- 50- Clear Graph ) 40- ( Draw Line) 30- 20- pu to - 0 Fluid 1 lllll 1 I I 2 4 6810 121416 18 20 Viscosity Time Fig 7. Viscosity. The terminal velocity of a sphere givesan excellent ,.cperiment for ratio and proportion. r File Circuit 1 Voltage 10 - 1.6 9- (Hide Graphj 8- ( Plot Point') 7- 6- ( Clear Graph 5- Draw Line 4- Resistance 3- 2- Low 1 - Power Supply 0 Notts) V I i llllll I 00.1 02 0.30.4 05 0.60.7 0.8 0.9 1 0.160 Current G: 51 r.P.O1 Fig 8.Circuit1. The linear relationship between Voltage and Current can be seenas the student varies the control boxes and if needed the value of the resistance and the experiment repeated. 3 Using Computers Intelligently r* File Circuit 2 Voltage to Hide Graph )( Clear Graph 9 - It ( Plot Point)( Drew Lite) e- 7 - 6 - 5 - 4 - 2 - 1 - 0 0 0.05 0.10 0.13 020 0.25 0.30 0.33 Current N 41111 C..E Power Supply (Volts) Fig 9. Circuit 2. ideas for ways in which the teaching of Physics can be improved and specifically it The original seven programswere to be inhas lead to our teaching staff having the areas of: access to a number of instant experimentsm a box (the Macintosh) for illusirating and Motion sup- porting their teaching. Forces Momentum The sequences will be connected together Work and Energy and made available, on request to other Properties of Matter Fields institutions and hopefully will serveat a catalyst to foster further development and Experimental Methods. exchange of disks. One other institution has already indicated its interest in specifi- The graphic sequences chosen sometimes cally developing teachingsequences and fall into more ti-an one of the abovecatego-this could lead to further exchanges. ries and this is done deliberately to enhance learningby familiarity. Thus Experimental References Methods is concerned with ratio andpro- portion and will call on anysequence with linear characteristics as a suitable learnerBailey D.E., Logan P.F., Walker P.J., experience. The Jearner may also see the Hulme T., Drubetsky M. (1985). same graphic sequence elsewhere under Physics Literacy Project, Student Motion or Properties of Matter. Control of learning, Computers in Tertiary Education, Bowden J & Li- Conclusion chtenstein S Eds. Bork A. (1984) Computers and Informa- tion Technology as a Learning Aid. This project is far from concluded. It has Second Australian Computer Education opened up a wealth of opportunities and Conference, Sydney. 40 The Logic:hem Organic InferenceProgram Frank R. Burden, Chemistry Department Monash University, A program has been written, for the IBM/PC look at Organic chemistry is to compare it series, which deduces how a pair of organic with a game of chess r here the rules are chemicals wore Id react with one another or with common reagents. It is * forward chaining infer- definable ( at least h. some extent), the ence system which makes we of separate strategies are understandable but the game knowledge base of rules concerning organic is is complex. functional groups and multiple bonds. The knowledge may be added to or modified by the A student of organic chemistry has first to user, though the supplied set of rules is fairly comprehensive and covers the main reactions learn the language of valency and rules for which students will encounter to the middle of the connection of atoms. Then the atom:, their tertiary year. The system should be of use, are clustered together into co-called func- therefore, to chemistry undergraduates as well as be of interest to others concerned with the tional groups, which are groups of atoms representation of complex knowledge. that can be thought of as a whole and from which larger entities, molecules, can be Chemical Computing. built. There are no absolute rules as to what constitutes a fimrtional group as against Alongside the development of FORTRAN some arbitrary group but there is a general and LISP in the early sixties came theeveragreement with many exceptions. increasing use of computers in damistry. In fact, in most universities the chemistryThe chemistry of a molecule can, in part be department was one of the main users ofderived from its constituent functional the computing facilities. Of course theirgroups and the bond types (single, double work was all to do with number crunchingetc.) that connect them. This then forms a and for the most part still is. Fortran is stillgood start for thinking about the encoding the language used and will probably re- of the rules of organic chemistry. Of course main so for some time to come since itthis is not the whole story else it would all produces fast code and there is a large be too simple to bother with; however it is investment in the form of existing pro-enough for a beginning. grams. These programs are availableThe original purpose of the program was to through an organized software sharinginvestigate the extent to which organic facility that has been in existence for over 15 years. chemistry was amenable to being encoded in a logical manner. Somewhere along the way the aim was changed to producing a However chemistry as a whole is a mixturepackage that would be useful to students, of science and art in the way that goodafter the scope and limitations of the origi- cooking is, so that although recipes can benal purpose had been defined. written down, it still takes a good chemical cook to turn the recipe into a product orThe initial challenge was to encode the invent a new one. So one way that Hike torules that relate to the reactivity of the 4 Using Computers Intelligently functional groups and their associatedficulty in this area as they have become bond types. The main criteria is thata pairfamiliar with specialized chemical word of chemical compounds be typed in at the processors which enable them to place such keyboard, or retrieved from a file, anda setdiagrams in their documents. The of possible product molecules be displayedenormous difference between going from a after computation. The chemical input and string to a diagram and from a diagram to output to be as near as practicable to nor-a string is not obvious to them. mal chemical conventions and the rules to cover the main possibilities. The string then has to be parsed into func- tional groups, bonds and other chemical The language choice falls between aproce- dural language such as Pascal anda de-syntactical symbols, all of which is made clarative language such as Prolog. How-into a linked list of symbol groups. This list ever one of the constraints in the presentis then further parsed into a graph where case was that theprogram shouldbe under-each node is a record containing chemical standable and modifiable by a range ofinformation pertinent to the functional chemists and chemical undergraduates.group and pointers to other atom records. Pascal was chosen since it also hassome natural advantages for chemistry in that itsAt this point is is easier if the molecular record and pointer types match the atomicgraph is thought of as a mobile as used and bond properties of atoms in molecules.over children's cots) with the functional Hence each record can include severalgroups as the objects and the valency point- pointers to other records and so, betweeners as the strings. The object of the rules is to them, form a graph which can representamanipulate, add to and partly alter the chemical structure. mobiles. Turbo Pascal°, together withsome ex-The rules that will decide how two mole- tender packages, was chosenon the cules react with one another are really rules grounds that a large number of these com-of interaction between functionalgroups pilers were 3ready in use and thereforeand functional groups or between func- users would has the potential to changetional groups and bonds. Hence a list of all any source code pnAded. In the event s;# such appropriate pairs can be made and far only compiled code has been circulated.then compared with an external database. Pointers and Valency The encoding of rules and the Action types. The program inputs a string whose syntax is as close as possible to the acceptedway ofThe external database was referenced toa writing a chemical formula; the limitations well known organic chemistry textbook being in the super- and sub-scripting which (Morrison and Boyd, 'Organic Chemistry) is commonly used. Of course chemistsand at present contai ns about 150 rules. The prefer to write two-dimensional diatramsreaction list is first pruned of the non -rf act- to represent molecules. Parsing such la-ing pairs and the remaining rules acted grams wouldbe anextremely complex taskupon at the discretion of the operator. and well outside the scope of this project though it must be recognized asa limita-The rules and the ianctional groupsare all tion and a barrier to the acceptance of suchencoded with a two character code for programs by chemists. compactness and ease of reference. In fact it is not always recognized by 'com-One of the codes contained within each rule puter illiterate' chemists that there is a dif-is an action code. This code specifiesa 4 2 Using Computers Intelligently 33 particular Pascal procedure which willMarketing and copyright. combine and rearrange the original (reac- tant) molecules in a particular mannerA part of this years program is to be de- pertinent to the reaction under considera-voted to the marketing and copyright as- tion. The other codes contained within apects of programs written by academics. rule specify how the functional groups ofThe law, and some of the larger software the original molecules are to modifiedhouses, seems to want to treat software in a upon reaction. Finally there are also codes similar manner to books. If this is the case to specify the conditions under which thethen there is no problem as academics have reaction might take place (eg. heat,acidalways written books and usually kept the etc.). royalties, wt-,:h rarely amount to a fortune. It is important, and in keeping with the new The result of using the action procedure ispolitical philosophy, that individuals be to produce one or two new molecularmotivated to produce innovative ideas. graphs which represent the product mole-Program writing is a difficult, lengthy and cules, and can themselves be re-manipu-often tedious business for which there is the lated and even reacted further with otherneed for the prospect of a positive bonus. If molecules. the nation is to keep up with the trends in the ever changing world of computing and The program. computer aided learning then such encour- agement is needed. Of course, where grants The program as a whole has been con-are provided it up to those concerned to structed to present in a professional man- negotiate their own contracts which may or ner, in contrast to many extant chemistrymay not involve the ownership of copy- tutorial programs. An in-built editor isright. provided for typing in the molecules; help windows may be switched on and off; in-In my own case I have attempted to market formation windows pop down when somethe program myself under the name rules are fired, and the disk may beLogiChem and an organizational name of browsed for pre-recorded data. ChemSoft, which are registered business names. I took this course as there does not The program may be viewed as a chemicalseem to be a suitable avenue for distribut- calculator where molecules are input in-ing chemical software to chemists in a stead of numbers. It does not, however,commercial manner. The second advan- address the problem of synthesis which is a tage of doing this is that one is able to problem of greater complexity, though an interact with the business community on attempt at a restricted synthesis programan equal footing, which in itself promotes a may be made next year. better understanding of other peoples problems in both directions. The program should prove useful to Aus- tralian first- and second-year chemistry undergraduates and their teachers. It may Dr. Frank Burden is Senior Lecturer in also prove interesting to those who wish to Chemistry at Monash University, his brush up their organic chemistry and even research interests are in the quantum- to those who are interested in the represen- dynamics of small molecules and the tation of Complex knowledge. production of expert systems for chem- istry. 43 Grabbing knowledge graphically Kay Fielden SchOol of Information Sciences andEngineering Canberra College of Advanced Education Icardc interfaces of machines like the Macintosh neer requires a deep understanding of the provide valuable and novel ways in which knowledge acquisition may be enhanced. A system domain to select an appropriate knowledge acquisition tool has been developed representation. .to design and create problem solving Wks by simple graphic interadion.In a Ce.drolled Simulation of an object-oriented environ- domain where objects are mapped directly onto Jams for simulation, a dear mental modelcan be ment is an ideal domain for utilizing direct famed by die veer. Exploiting the conbul struc- graphic interaction. This paper will discuss anis of the iconic interface, that is, menus, but- a knowledge acquisition tool designed for tons and supimposed windows forces a dis- use within an authoring system environ- did and unambiguous mental model ofhow the simulat.:m is operated. Adoptinga struc- ment. tured sporoach in designing the interfaceen- hances knowledge acquisition. Authoring Systems Environments While acquiring knowledge fiomexpertsApplication of graphic interaction toau- has been regarded as one of the majorthoring systems environments enables bottle-necks f-,r the construction of intelli- specialist teachers to concentrateon the gent systems (Young, 1984), the underlying domain under consideration, with the problem appears robe in knowledgerepre-computer system becoming 'invisible'. sentation. Once expertise has beenac- quired by a specialist, that knowledgeAny domain which can be representedas a appears to be stored in the specialists mindstate/space relationship with constraints in 'capsule' form.Acquiring detailed on the movement and relationship of ob- knowledge therefore becomesa difficultjects to each and to the domainspace is step, as the specialist often can no longersuitable for simulation and graphic interac- articulate these details. Traditional knowl- tion. edge gathering techniques: form filling, questionnaires and interviews all requirea By adopting the same strategy described by translation process by a knowledge engi- Finzer & Gould (1984): 'only things thatcan neer in order to get the knowledge into the be seen can be manipulated', specialist system. Aiello (1987) suggests that a rich teachers can design their own simulated variety of knowl&ge representationsaretasks and implement their own creative required to allow selection of an appropri-ideas, thus eliminating the need forpro- ate representation. grammers in the design of software. Not only does task creation in such an environ- The translation process required by thement constitute a new medium for the knowledge engineer in traditional systems presentation of information and concepts, can be side-stepped by the use of simpleit also provides new ways to gather knowl- graphic interaction. The knowledge engi-edge both about the domain and the control 4 4 Using Computers Intelligently 35 structure for the domain and for the simu- However, control of the real world system lated system. is often inarticulated, details of control having become automatic for specialist Knowledge Acquisition Tools users. To the knowledge engineer, gather- ing information about control of a task is Although many researchers agree that in-difficult. The mental model is not dear. terface design is domain dependent (Hol-The problem becomes one of mapping a lan, 1984, Aiello, 1987, Norman 1987) thefuzzy model on to an explicit set of instruc- availability of a rich variety of knowledge tions in the simulated system (Figure 1(b)). yepresentations enables the selection of the most appropriate representation. WhileBy recognizing that the simulated world is presentation of the domain under consid- a different world by virtue of the control eration need not necessarily be directlymechanisms, structuring the iconic inter- related the representational system, theface to provide consistent control mecha- more direct the mapping, the dearer thenisms eases the task of learning a new mental model. The picture on the screen mental model for control. While the ideal should be close to the real world. In the situation is to make the machine 'invisible' world of shrtulatedphysicalenvironments, (Norman, 1987, Hollan 1984), there is still direct mapping from the real world is pos- control of the device. sible via the use of direct manipulation in an object-based system.Movement ofGraphic Interaction physical objects in the real world can be mapped directly on to the simulated sys-Graphic display has long been used as a tem by direct manipulation (Figure 1(a). powerful tool, particularly in simulation Domain Real World Simulation Manipulation Direct graphic of physical interaction objects Figure la Simulation Task System Fuzzy Explicit Figure lb No Direct Mapping 45 Using Computers Intelligently FileEditSearch Run Windows Introduction Task DataPath MacSoluer Help View Path Qu Step Make Change Quit Hints Make Change Quit Figure 2 (Trollop & Alessi, 1985). Graphic acquisi-Control Structures tion of knowledge by direct manipulation has only become the subject of muchre- Icon interfaces like the Macintosh offer search with the wider availability ofma- a chines capable of providing direct manipu-variety of control mechanisms- pull down lationfacilities. While it is difficult for other menus, buttons and superimposed win- than a skilled artist to draw or paint pic-dows, which can isolate foreign (to hu- tures directly on to a screen using a devicemans) control structures in a standard way, like a mouse, graphic input can be achieved separating them from domain knowledge, easily and effectively in a variety of ways. If thus differentiating the levels of activity knowledge is required about rectangularrequired to interface with the simulated regions on the screen this can be done by system. dragging a 'rubberband' rectangle. In an object-based domain, objects can be de-Domain control must be made explicit in fined in both generic form and as instantia- the simulated system. In the domain stud- tions of the generic form. Keeping a library ied, teaching problem solving tasks ina of graphic objects in both these forms al- restricted environment, the following lows the user to request the appropriate strategy has been adopted: object from this graphic library. It will beI. control within tasks is by using pull displayed as the instantiation if it is pres- down menus which are only visible ent, otherwise in the generic form. Direct when required (Figure 2); manipulation of objects on the screen is then under mouse control, allowing the2. display of a completed solution is by specialist teacher to build up steps in a using buttons displayed on the screen solution path. (Figure 3). 46 Using Computers Intelligently 37 * FileEditSearch Run WindowsIntroduction TaskPattiStep Macsolver (ClicltHer±i r. Figure 3 FileEditSearch Run WindowsIntroduction Task Path S Macsolver Start Neut Previous 8Quit Hints Have you moved a fence yet? CE) Figure 4 47 38 Using Computers Intelligently 3. superimposed windows are used forpretation, it seems likely that the acquisi- displaying text for proving guidance totion of knowledge both in quality and students attempting solution of a taskquantity, is enhanced. This knowledge (Figure 4). acquisition tool attempts to shift the inter- face to the human end of the spectrum by Control is a function which exploits leftmaking use of input techniques which brain activity and while menu selection isreduce the 'visibility' of the machine. Hol- not sequential, it does provide a linearIan (1984) maintains that the design ofan approach. Button selection gives access to interface should be considered as the con- a Sequence of steps, and keeping graphicstruction of a representational system for and text windows separate is consistentcommunication. with the approach adopted for the Macin- tosh system interface. Further exploitation of the iconic interface provided by the Macintosh should lead to There is at best a fuzzy mapping from thebetter understanding of the formation of simulated world for control functions. Thismental modelsby different classes of users. creates problems for the user who must Differentiation between control structures form a mental model of control structures.and domain activities at the Macintosh However, separating them from domaininterface provide an effective means for knowledge by a completely different formimproving formation of mental models. of interaction with the computer systemAdopting a structured approach to the use provides a dearly defined method of dif- of an iconic interface for knowledge acqui- ferentiatingbetweencontrol of the task andsition can also reduce the time taken to manipulating objects within the domain. become at ease within the authoring envi- By aligning these different levels of activity ronment. with a mapping most suited to left brain/ right brain functioning (Edwards, 1979),References the user, whether they be the specialist teacher at knowledge acquisition time,or Aiello, I.E. (1987). Why This Generation of the student at problem task execution time, Knowledge Representation Tools Will can more easily form a mental model of the Fail. In G.Salvendy (Ed.) 2nd Int. Conf. system. on Human Computer Interaction, Hawaii, Elsevier. Control features provided by the interface Alessi, S.M. dr Trollop, S.R. (1985). Com- are available only when required.Pull puter - Based Instruction. New Jersey, down menus are only visible when apar- Prentice-Hall. ticular command is required and buttonsAmarel, S (1984). Expert Behaviour and and superimposed windows are sentaway Problem Representation. In A. Elithon when they have been used. This provides 8r R. Banerji (Eds.), Ch.1, Artificial and minimum interruption to the thought pat- Human Intelligence, (pp 1-41). Nether- terns required to solve the task within the lands, North-Holland Publishers. domain. Benzon, B. (1985). The Visual Mind and the Macintosh. Byte, January 1985,113 -130 Conclusion Edwards, B. (1979). Drawing on the Right Side of the Brain.Los Angeles, J.P. If the transformation from the specialists Torcher Inc. mind to the system is aided by an interface Fielden, K. (1986). Designing an Interface for which is closer to the human specialists Knowledge Acquisition for an Intelligent view of the problem than the system inter- Tutoring System. 17th Annual Austra- 42 Using Computers Intelligently 39 lian CAE Computer Conference, Dar- win, (pp 29-54) Fielden, K. (1987). An Enhanced Interface for Improved Knowledge Acquisition.2nd International Conference on Human- Computer Interaction, Hawaii, August 10-14. Finer, W & Gould, L (1984). Program- mingbyRehearsal. Byte, June 1984, 187- 209. Hollan, J.D. (1984). Intelligent Object Based Graphical Interfaces. In G. Saluendy (Ed.) Human Computer Interaction, (pp 293-297). Netherlands, Elsevier Science Publications. Mylopoulos, J. &Leuesque,H.J. (1984). An Overview of Knowledge Representa- tion. In M.L. Brodie, M.J. Mylopoulos & J.W. Schmidt, J.W. (Eds.), On Conceptual Modelling, U.S.A., Springer-Verlag. Norman, D. (1987). User Centered System Design: Emphasis Usability and Under- standing. In Proc 2nd International Conf. on Human Computer Interaction, Hawaii. Simon, H.A. (1983). Search and Reasoning in Problem Solving. In J. Pearl (Ed.), Search and Heuristics, (pp 7-29). Nether- lands, North-Holland Publishers. Young, R.M. (1984).Human Interface Aspects of Expert Systems. In Expert Systems State of the Art Report, (pp 101- 111). Exeter Pergamon Infotech Ltd. Kay Fielden is currently lecturing in Information Systems in the School of Information Sciences and Engineering at the Canberra College of Advanced Education. Research interests in CAI are both within the School as joint co- ordinator with Jo Baskett on a pilot project examining the most effective way of introducing CAI material into first year courses and also ongoingre- search into iconic interface design con- tinuing from higher degree work al- ready completed. The interface design project is funded by a CCAE research grant. 49 What do you say after theypress "RETURN"? Intelligent feedback in CAL Geoff Foster Tertiary Education Institute University of Queensland. Much of the value of Computer Aided Learning ness in their application), but for the pur- stems from its interactive nature, its ability to poses of argument I shall take that given in elicit from the leaner a thoughtful response to a question; it is the very next step that is often Higgins and Johns (1984), as a description under-exploited. The computer's reaction to the of `drill-and-practice' programs. It will be learner's response can contribute a great deal to recognized as a paradigm followed by thelesrning that is going on. If we simply tell the many CAL lesson authors (who would learner that the answer is right or wrong (or, at most, congratulate or commiserate), then we fail often, I am sure, vigorously deny that what to take advantage of the receptive state that has they are doing is 'drilling'). Wm induced in the leaner by the concentration needed io construct that answer. The repertoire Higgins and Johns list (page 39) the follow- of skills used by skilled tutees in discussion classes has its counterpart in CAL; maximum ing components: benefit can not be got until we exploit it. This, in CAL as in a tutorial, demands both a careful 1. select task or question analysis of the student's answer and a reaction 2. display question to it which has been developed to at least the same standard as the body of the lesson. 3. accept student's answer 4. match answer against acceptable answer(s) As in so many other disciplines, we work-5. report success or failure ers in Computer Aided Learning are apt to6. adjust student's running score 7. return fall into habitual ways of thinking and to (1). doing. It is constructive to do this to a controlled extent; progress would be slow This list needs to be only slightly modified indeed if we had to start off each newto make it apply more generally. Specifi- project from scratch and were not able tocally, if we widen items (1) and (2) to read: profit from experience, if we had always to 'reinvent the wheel'. But, from time to1. select information, stimulus material time, we ought to scrutinise some ofour and/or question assumptions and practices, in case condi-2. display information, stimulus material tions have shifted without our noticing, and/or question then it will cover a making those assumptions and practices considerable number of those lessons less valid (or even completely invalid); which are commonly encountered. perhaps we might find that we no longer need that wheel, but would do better withIn this paper, I shall focus particularly on legs or caterpillar tracks.The field that isitems (4) and (5) and suggest that it is in sometimes referred to as 'Tutorial CAL' is a these that the greatest potential for im- case in point; let us use it as an example. provement resides. Sometimes, I suspect, There is a number of possible definitions ofattention has been almost entirely directed what it is (and there is considerable loose- to devising stimulus material that will be 5© Using Computers Intelligently 41 searching and challenging, and displaying questions inserted in the text (either it in captivating ways, steps (1) and (2). I do 'pre-questions' or 'post questions') not deny the value of these, merely recom- advance organizers mend that steps (4) and especially (5) behavioural objectives receive a comparable share of attention. headings, titles and emphasis (e.g. ital- ics or underlining) A mathmagenic approach instructions about note-taking (in the text, or separate). (Jonassen, 1982, vol 1, There is an extensive literature concerned pp 130-131) with learning from texts, largely initiated during the heyday of programmed learn-There is an obvious relationship between ing,but transcending it; the newcomer willsome of these text-based procedural de- find Jonassen (1982) a useful starting point.vices and the parts of a tutorial CAL lesson Among the many aspects of the field, one that were listed before, and we should cer- referred to as mathemagenics' has a par- tainly bear these in mind; but when we are ticular relevance to CAL. The term prop-talking about computer mediated presen- erly refers to those behaviours in a learnertation rather than about print there is an- (originally in one engaged in workingother very powerful mechanism at our through a programmed text) which 'givedisposal, thanks to the much more interac- birfrt to learning', but has been given an tive nature of that medium compared with extended currency as description of a 'passive' text. domain of study in its own right. We should certainly use mathemagenic The central notion of mathemagenics is thattechniques borrowed directly from the the structure and components of a text (or, world of print, but we should also take by extension, of a CAL lesson) can be usedadvantage of the receptive state of mind to stimulate particular mental behaviours, which these techniques induce in the particular ways in which the text is at- learner. At the moment when he or she has tended to and processed by the learner: just answered our carefully designed ques- tion, we can analyse that answer and then By controlling the way the learner thinksimmediately respond to it, giving feedback about textual material, we can determineto him or her which is sensitive to the how learners will convert presented infor-answer and the context in which it was mation (nominal stimuli) into encoded in-given. This will have more impact than a formation (effective stimuli).(Jonassen,stock response can ever have. 1982, vol 1, p 220) So there are at least three phases of lesson The means we have of effecting this control design involved at each point of interaction lie mainly, but not entirely, in the content, with the learner: structure and appearance of the textual material itself; Isaacs (1986) has written on the construction (both in content and the application of some of these techniques form) of the stimulus text,and of the in CAL, particularly on the use of screen task or question to be presented presentation to achieve advance organiza- the analysis of the learner's response tion and to demonstrate structure and the feedback given as a result of that emphasis. For printed text, some mathe- analysis. magenic aids which have been used or proposed include: I shall not dwell in this paper on the first of these, since there is little difficulty in seeing 42 Using Computers Intelligently the relationship between established text son depends on whether a match was techniques and their equivalents in CAL. made, and what it was. .I prefer to think However, there is not the same established more broadly, of dassification rather than body of knowledge and experience for theof matching. latter two items,which will therefore be addressed in some detail below. The central concept of matching is that of selecting a small number of predetermined Evaluating responses patterns from a large population of pos- sible ones, and subsequently winnowing A great deal of effort has been spent over out 'correct' from 'incorrect'. To be sure, the years in devising efficient and effective this is a fornt of dassification, but if we set ways of evaluating the responses that aout with the idea of classification as the learner gives during a CAL lesson, whichprimary one, it seems to me that we are at are usually (but not always) answers toliberty to choose our categories in a freer, questions. There is, as always, a trade-off;less encumbered way than mere matching here it is between forced and free re-will allow. If we start with the idea of sponses. matching, then we can implement certain restricted forms of classification; but ifwe Forced responses, such as are used in mul-start with classification, then matching is tiple choice questions or for menus, are but one of the many options available to us easy to analyse, but limit the learners' (and may sometimes, indeed, be appropri- activities to those of choice or recall from a ate). limited set of possibilities; free responses leave more to the learner, but are notori-It would be, perhaps, a little fanciful to ously more difficult to evaluate. claim that what we are attempting here amounts to introducing a higher order The usual compromise solution is to allow cognitive skill to our programs (in the same a certain degree of latitude in response, sense that we think of analysis as a higher- and accept thefact that some answersorder skill than recognition when weare which a human would judge acceptablediscussing human learning); to do this to will be rejected and some unacceptable any extent would take us into the area of ones accepted by the essentially limitedartificial intelligence, and I am not intrepid algorithms that are practicable (in termsenough an explorer to venture as far my- of programming complexity, processor self; pioneering work is proceeding else- time, memory or disk space, and so on). where, in several fascinating directions. However, even the move from matchint, But there is another underlying (and moreclassification as our basic model will, I important, to my view) aspect of the ap-think, have a significant influence on the proach to evaluation which will bear somedesign of lessons, and even ,nore impor- scrutiny.For both closed and open re-tantly on the design of audit tools sponse sets, answer evaluation in the bulk (programs or languages designed to assist of CAL lessons in service has been imple- the 'courseware author') and, in the long mented in terms of 'matching', meaningrun,on computer-based management that a learner's answer is compared with a systems. set of stored expected patterns, some cor- rect, some incorrect; this, indeed, was anAs illustration; some present authoring explicit component of the drill-and-prac- systems stipulate that a certain set of con- tice example quoted earlier. The learner's sequences must invariably follow a 'cor- score and the subsequent course of the les- rect' response, another set must follow an Using Computers Intelligently 43 'incorrect' one; no provision is made forcits in prior knowledge. There ought to be any other case. Or, in a more specific in- similar benefits in the refinement of classi- stance, some packages insist that a learnerfication that we are discussing here. be given all the marks for a right answer and none of them for a wrong one, thusOnce a approach based on dassification effectively ruling out answers of interme- has been adopted, there are several tasks to diate merit. These sorts of restrictions can be undertaken at each point in the lesson at be frustrating to someone trying to write which the learner gives a response: lessons intended to exercise learners' higher-order skills; they exert a similarly constructing a set of meaningful catego- damaging but subtle influence on the field ries and specifying program paths or of CAL as a whole. reactions to each of these Perhaps, in CAL, it would be better to re- deciding on the criteria by which re- place the right/wrong duality by a con- sponses will be sorted into those cate- tinuous spectrum which ranges from 'ex- gories cellent' down to 'awful', even if this does have to be implemented in practice as a set designing algorithms by which the cri- of disjoint categories. I still remember teria will be tested being told, thirty years ago when an engi- neering student, that in engineering there checking the algorithms against test are no right and wrong answers, just good data; responses generated either by the and bad ones; this is true of more fields than lesson author or, preferably, by a pilot engineering. group of learners who have similar characteristics to the 'target audience' We have been talking of 'tutorial' CAL; a term which, presumably, is intended to developing categories, paths, criteria suggest that there is some relationship and algorithms in an iterative way. between the activities in the CAL lesson and what normally goes on in a conven-The criteria for classification mentioned tional class conducted by a tutor. One of thehere, and the algorithms by which they are desirable characteristics of a human tutor implemented, should be developed to an (as well as those of being knowledgeableappropriate level of sophistication, where and skilful) is that he or she listens to his or 'appropriate' is the key word. It is just as her students and responds sensitively; it isineffective to waste resources and effort in this behaviour that we ought to be trying tousing refined techniques to make gross simulate here. just as I am accustomed in decisions as it is frustrating to try to make my work as a staff developer to runningsubtle discriminations with too coarse a workshops intended to enhance the listen- sieve. ing and communicating skills of tutors, I would like in this paper to advocate that we (Note also that the development process enhance the corresponding 'skills' of CAL referred to in the last item in this list is not lessons. a substitute for that which takes place (one would hope) during the production of any A tutor who is listening with sensitivity cannew learning material, but is contained do much more than determine which an- within it.) swers are right and which wrong; he or she can detect subtleties and, with experience, It was not by chance that I included two diagnose mistakes in reasoning and defi- distinct activities in the first item in the list. t-r 53 44 Using Computers Intelligently The processes of deciding on categories it (or better, paraphrasing it)so that he or and on the differential behaviour of theshe can see that 'the computer has heard program (lesson) in respect of these catego-and understood it'. After that, the learner ries are intimately linked. One mightusecould be given reinforcing or correcting here an analogy borrowed from medicine;information, which could in its most ele- a medical researcher who wishes to deter-mentary form be the words "right"or mine the aetiology of a patient's condition"wrong", but would be better if elaborated may use different tests from that of tl appropriately. This might be followed by clinician who wishes to make a diagnosisnew information, designed to consolidate for the purposes of planning treatment. Inor expand on the textual material which the same way, we must be clear to make thepreceded the question just answered. distinction between those questions which are appropriate for assessing a learner's The point at which the 'response' ends and state of knowledge and ability and those the next part of the program or lesson be- which will enable us as teachers to bestgins is somewhat undefined;we could facilitate his or her learning. Iam not sayingchoose to say that all material which is that there is never an overlap between thesesensitive to the learner's 'most recentan- two sets. swer is part of the response, but this w3uld only be the case when the subsequentroute Intelligent feedback through the lesson is fixed; k. .ally this would also depend on the learner's 'an- I have made the point inan earlier paper swering history'. In any case, it ismore (Foster, 1986) that there are occasionsonimportant to include all the features here which we may wish not to respond to a described than to attribute them to particu- learner's answer at all and otherson whichlar functions of the underlying software. a brief response is all that is warranted. I claim here, however, that asa general rule,I have said in the preceding section that it if we fail to exploit the opportunity of re- should be our aim to emulate the sensitive sponding fully, then we are missing aand knowledgeable tutor in the behaviour chance to enhance significantly the veryof our programs when they respondto a learning that we are trying to induce. Afterstudent's answer. One of the components all, the learner has just been deeply in-of sensitivity, as I have mentioned, is the volved in constructing the best answer thatability to listen; the CAL equivalent of this he or she can, and his or her attention isresides in the operation of the procedures likely to be tuned to a fine pitch. We areused for evaluating, for classifyingre- poised at the legendary 'psychological sponses into categories. The CAL analogue moment'. of knowledgeability in a tutor is embodied A well designed response,moreover, in the prior definition of the categories and should act as a powerful motivational rein- in the design of the algorithms whichper- forcer; if the learners think theyare beingform the classification (here again it is dif- 'taken seriously' they are less likely to beficult, and of little point, to distinguish turned off than it they gradually get theabsolutely between these two). feeling that, however hard they try, the computer response is going to be a laconicAnother component of sensitivity isempa- "correct" (or even worse a patronising thy; the nearest equivalent of this in a CAL "Well done, Johnny! "). system is in the extent to which the pro- gram as a whole is designed to suit the At the very least, the response shouldac- target audience and, on a more local scale, knowledge the learner's answer by echoing the extent to which it can respond to what 54 Using Computers Intelligently 45 it 'knows' about the learner; the history ofthe appropriate one of a number of re- previous answers and interactions earliersponse categories predefined by the lesson in the session and, maybe, in other sessions. author. Further; just as in an office the the- As a concrete example: if a student has saurus to be used on any given occasion, or already demonstrated his or her knowl-the dictionary used for spelling checks, are edge of a fact, we would not consider itselected from a small number of standard tactful if a tutor were to carefully explain versions, so the same generalization can be the fact to him or her at a later time; in a like applied to answer evaluation. One might manner this should be an elementary re- envisage having an answer thesaurus for quirement of all CAL software. psychology and a different one for political science; certainly it would be impracticable Remarks on implementation to construct a new one for every new course, and ludicrous to do it for each new I have recommended a move away from lesson. matching algorithms toward something more general. This was not simply an ex- The other technique that has been develop- pression of faith; there have been develop- ing quietly outside the CAL world is that of ments in computing going on outside the so-called 'natural language interfaces', CAL field, brought about by the revolution mainly for use with database packages. Just in computer use in office practice, that it as with thesauruses and spelling checkers, would be remiss of us to ignore. (As well, as extremely keen competition has stimulated I have mentioned, as some valuable re-advances in this field that are equally wor- search aimed specifically at CAL; I am surethy of consideration by CAL workers. Why we shall see evidence of this work else-should the lesson author be compelled to where in these proceedings.) construct complex logical expressions (('SPAIN' OR 'PORTUGAL') AND NOT It3 common place, now, to find on-line (ITALY' OR 'FRANCE)), while his or her spelling chc.kers and thesauruses ('the-colleague in business is benefiting from sauri'?) available as accessories for wordmuch more sophisticated and easily used processing packages used on personal techniques? computers in offices. The thesaurus, in particular, offers features very dose to When we turn to the question of providing what would be needed for intelligent clas-feedback to the learner, there is no similar sification of learner responses; and bothadvance in computational technique that these types of auxiliary programs havesprings to mind; the require! nts here are been developed to the point at which it isvery similar to those needed for presenta- possible for them to process each word as it tion of the primary information in the les- is typed inby the user in 'real time', without son. So, as in that case, we should simply serious delays (especially as learners willmake sure that we are not overlooking any be unlikely to be typing as fast as clerical ways of making the content and presenta- staff). tion of the text as effective as possible; here we are using the corpus of expertise of the What the thesaurus does now is very dosetechnology of text directly. However, it is to what would be required of answer clas- worthwhile, even at this stage, asking the sification algorithms; it is capable of takingfundamental question wh.;.:it I raised in a target word and associating it with a list ofFoster (1984); should we be using the com- synonyms, close relatives and antonyms. puter exclusively to present this informa- The same type of coding techniques can tion, or should we also rely on supplemen- equally be used to associate an answer withtal material, even on print material, using 5 3 46 Using Computers Intelligently the computer to manage this information References rather than to present it? Conclusion Foster, Geoff. '1984). TheComputer-Man- aged Book. In R. M. Russell (ed),Pro- ceedings, 2nd. Annual ComputerAided What I have attempted to do inthis paper is Learning in Tertiary Education Confer- to make two major points; the first is that ence, Brisbane, University of Queens- Computer Aided Learning hasa distinctive land. strength, compared with some other mediaFoster, Geoff. (1986). The Casefor Plain of teaching and learning, inthat it is inter- Text CAL. In G. Bishop & W. active, and therefore uniquely van Lint capable of Proceedings, 4th. Annual Computer demanding thinking and reasoningfrom Aided Learning in Tertiary the learner. I followed this with Education some sug- Conference, (CALITE 86), Adelaide,Uni- gestions as to how best to exploita particu- versity of Adelaide. lar aspect of this quality, which we couldHiggins, John and Johns, Tim.(1984). sum up in the aphorism "strike while the Computers in language Learning. iron is hot". Lon- don, Collins Educational. Isaacs, Geoff. (1986). ScreenDesign for The second point made andelaborated is Computer Assisted Learning. that we might try to emulate that In G. quality of Bishop & W. an Lint (eds),Proceedings, effectivl tutors whichwe refer to as sensi- 4th. Annual Computer Aided tivity, and add it to the Learning in equivalents of Tertiary Education Conference,(CALI7'E knowledge and skill that, presumably,we 86). haVe always tried to incorporate into ourJonassen, David H. (ed.). (1982). TheTech- CAL lessons. This might bedone, on the nology of Text, (Volumes 1 and one hand, by paying attention to 2), Engle- ways of wood Cliffs, EducationalTechnology evaluating student answers thatgo beyond Publications. mere matching or identification withpre- defined models; on the other byputting as much effort and ingenuity intoconstruct- Geoff Foster is a lecturer in the ing truly helpful feedback Tertiary to students as we Education L-Isetute (TEDI)at the Uni- have always done in devisingrich text and versity of Queensland. He searching questions. began his academic career asa lecturer in me- chanical engineering, but hisinterests Implicit in these suggestionsis a further gradually shifted away from technical one; that we should engineer- not fail to take ing and towards teaching andlearning. advantage of computationaltechniques He made the move to TEDI in developed for other 1974 after purposes; such as the fourteen years of teaching, andsince on-line spelling checker, thethesaurus and then has been engaged principally the natural language interface. in general consultation workon aspects of teaching and learning withacademic There are many excitingdevelopments staff. His present interests include which are only just CAL over the horizon of the but are not limited to thet field. CAL world; my proposals heremay be more modest in scope than some which will be considered commonplace inthe future, but I hope they have their placeand that they might offer a lasting benefit. 6 Some lazy thoughts on applying Al to CAL Marshall Harris Department of Computer Science University of Queensland The leading edge of AI research in CAL is problems, until the level of decomposition mainly concerned with problems of represent- ing knowledge, modelling studcAts' knowl- is such that each sub-problem is easy to edge, processing natural language, and intelli- handle. gent choice of teaching strategies. It will be same years before this research becomes eco- Another aspect of this technique is that it nomically and educationally viable. allows one to consider the problem at a Meanwhile, there are many aspects of CAL that high and very general level first, before can benefit from piecemeal application of AI considering the details. techniques, without attempting to resolve the above-mentioned difficult problems. The au- Putting this another way, Top-Down De- thor is currently implementing a system where composition allows one to dodge the nasty AI techniques ..a.-: applied to a conventional CAL system, to select suitable material for de- bits and concentrate on the interesting livery to the student. This was described in parts. One might, indeed, say that it en- (Harris, 1985). This paper considers another courages one to put off till tomorrow what area where such techniques could be applied, one does not feel like doing today! That is izqmely Authoring Systems. why I call it a lazy person's technique! The Lazy Person's Technique. I mention this because I believe that what- ever comprises CAL can be broken down into a bundle of sub-tasks or sub-problems, In (Harris 1985) I suggested that a consider-to each of which some AI techniques may able time would elapse before a full, "intel-he applicable and effective. ligent" CAL system would be truly user- friendly, affordable, and acceptable toBy identifying those sub-tasks where Al teachers and students as a viable teaching/ can be used, and deferring attempts to tutoring alternative to human teaching. apply AI to those tasks which are currently intractable, we are likely to make more Meanwhile, however, it is clear that tech-rapid progress in CAL than would be the niques imported from AI systems can becase if we were to try to tackle the whole applied in a somewhat piecemeal fashionproblem headon. Among the intractable to various aspects of a CAL system, withtasks I include: understanding natural lan- useful and effective results. Such systemsguage, in speech and writing; generating a are exemplified in (Webb, 1984,1985) and complete lesson from a subject domain (Harris, 1985). knowledge base; effectively representing a student's knowledge of the domain; and Students learning computer science areother difficult tasks like these. taught the technique of Top-Down Decom- position of problems. This means, simply, This is not to say that the "headon" ap- breaking a problem down into ever smaller proach should not be attempted, nor that 48 Using Computers Intelligently the difficult tasks should be neglected:I shall refer to these roles as "educator", these are precisely theareas where "author", and "student" respectively. well.funded research needs to done. But, while others tackle the hard parts, educa-The needs of the educator: teaching tors are anxious to "get on with it". So let'sstyle. help them by trying to improve the CAL techniques we now have, in thoseareas where improvement, via AI, is currentlyThe educator requires pedagogic flexibil- feasible. It was in this spirit that I proposed ity. By this, I mean that the lesson,or rather, the Authoring System that is used topro- the PSICAT system (Harris, 1985),noW under development, where Al techniquesduce the lesson, should supportmany were applied to the problem of selectingpedagogical styles and a mixture of styles. conventional CAL material appropriate to the student's currentprogress record. Styles would include "browsing"- an un- structured presentation of material giving In this paper I want to consider yet anotherthe student free choice in the selection of sub-task of CAL where I believe AI could material; structured or hierarchical organi- usefully be applied, namely, Authoring sation of material; and others. Systems. A fully intelligent system would includea Achieving flexibility inan Authoring knowledge base of criteria which would enable it to decide upon which pedagogical System. style or styles it should use fora given lesson or lesson segment presentation. In (Harris, 1983) I described the features which I thenbelieved an Authoring System In (Harris, 1985) I have describeda system should contain. That paperwas written (PSICAT since renamed "ICATS") which before I fully realised that, for CALto prog-presents a lesson composed of hierarchi- ress, Al must somehow be involved. Cer-cally organised lesson-segments, selected tainly the kind of flexibility and the facili- according to the student's performance ties I advocated then, are still highlydesir-record. I shall not discuss this particular able. However, I hope to show here that aspect of AIBAS further in this paper. In the much of that flexibilitycan be providedlight of my opening remarks, I also do not automatically by the builtin expertise ofan intend to pursue the problem of establish- AI-based Authoring System (AIBAS). Ining a system that can make itsown deci- order to show this, it is firstnecessary to sions about style. analyse and decompose thenecessary fea- tures of a good Authoring System. Rather, I would propose a system thatcan advise the educator on the appropriate What an Authoring System should be style of presentation, leaving the final like. choice to the educator. This would be, in other words, an advisory Expert System- a There are essentially three kinds of people sort of "Educator's Style Adviser". This who whose interecare affected by an puts the intelligent Authoring System well Authoring System: within the capabilities of currently avail- able Expert Systems. the teacher, in the role of educator; the teacher, in the role of lesson author;So the Educator's Style Adviser would in- the student, in the role of learner. terrogate the educator to find out what s/ he has in mind for C.e lesson, in terms of the 5 6 ar Using Computers Intelligently 49 nature and structure of the content, andwhich are appropriate are used by the provide advice accordingly. Author's Assistant. The needs of the author- hardware. The Author's Assistant would be con- cerned with such things as screen layout, The distinction between the needs of theediting of text, graphics, and sound, the author and :4 the educator is somewhat ar-control and interaction of external devices, tificial, but useful for establishing further the nature of interaction between student areas where Al can be effectively applied. and segment, and the manner in which student input should be evaluated. The Authoring System must be capable of generating different kinds of presentation-Handling text screens. text, graphics, colour, sound, simulation, and animation, and mixtures thereof; itSince the Author's Assistant is to behave as must be able to control external devicesan expert, it should interact with the author such as video disc and tape machines, tapeat a high level, rather than require the au- recorders, and slide projectors; and it mustthor to provide detailed layouts, etc. be capable of activating and interacting with other programs available on the com-For example, the Authoring System would puter, such as language compilers andrequire the author to indicate what should command shells. be shown on the screen; say: text, a ques- tion, and a response. I would not expect the Seen from this point of view, an intelligentsystem to accept natural language input Authoring System needs to have a knowl- from the author: that is one of those areas of edge base with extensive information AI which I want to place, for a while, in the about the characteristics of various hard- "too hard basket", and leave to the leading ware devices and how to use them. edge A! researchers! I would be content with a menu-based dialogue! Assume that the result of the consultation with the Educator's Style Adviser is thatThe Authoring System would then call for the author has chosen the appropriatemore information: i.e it would ask the au- style, and that the Educator's Style Adviserthor to enter the text, and automatically has used this information to select anotherinvoke a screen-based text editor (with Expert System that can deal with the par-plenty of on-screen help!) for the input. ticular choices made. I shall call this systemWhen the input is complete, the Authoring the "Author's Assistant". System knows exactly how much text there is. At this stage the Authoring System The author now has to embark upon themight ask the author to use the cursor to task of creating the actual lesson, or lessonindicate where words should be high- segments. lighted or coloured. It then calls for the question, using the same text editor, and Authoring a lesson. the style of questionsuch as multiple choice, or text string (if not already estab- The Author's Assistant would be a single, lished by the Educator's Style Adviser) and self contained Expert System whosedisplays a provisional screen layout as it knowledge components (rules, facts, etc)would be seen by the student. are marked to indicate whether they are appropriate for the choices made at theA weakness with many icon -AI Authoring Educator's Style Adviser stage. Only itemsSystems, and a difficulty for the author, is :t I ',.-;4 50 Using Computers Intelligently that the s/he has to supply pattern stringsGraphics Editor with an intelligent ability for matching against student answers. An to create graphics from the author's replies AIBAS should generate such strings auto- to questions. To avoid excessive complex- matically. The system might use a combi- ity, the Graphics Editor should developa nation of methods: "soundex", dictionary display by a process of trial and error, refin- look-up (suitable techniques might being its questions to the author and the gleaned from currently available spellinggraphics generated in response, until the checkers), transposition of characters, andauthor is satisfied with the result. so on. Handling other devices. Grammatical analysis of student input is more difficult for an AIBAS in the last analysis it is a natural language under- I do not profess to know a great deal about standing problem, to be lazily filed away in videos, tape drives, and the like, so I do not the aforementioned too-hard basket! want to comment on this aspect in any detail. However, if we wish to deal witha re- stricted subset of natural language, or with a well defined programming language, forLet it suffice to say that the principleson example, the Authoring System should which the preceding discussion is based provide facilities for syntactic definition. can be applied here too, with Editors and Expert Systems providing advice and, if Perhaps the Authoring System could pro-necessary, generating appropriate proce- dures. vide two facilities: a means of defininga "complete" definition for the language in question, and a means of defining a syntaxSome aspects of software design. associated only with a particular answer. The latter seems more attractive for the lazyThe Expert Systems proposed here would Authoring System, and is probably easierneed, in their knowledge bases, not only for the author to handle: I would expect therules and facts, but also procedures. These Authoring System to ask for lists of words, would be very high-level procedures their associated grammatical categories, which the Expert system would condition and a syntax description limited to the al- or parameterise, to produce the demon- lowable structure of the answer currently stration layouts, text screens, response being processed. This, combined with the evaluations, graphics etc. techniques for pattern matching suggested in the previous paragraph, would enableThese same procedures would, of course, the Authoring System to generate all theform part of the final AIBAS system that testing needed to evaluate textual studentinteracts with the student. responses. The needs of the student. Handling graphics It is a general requirement of CAL, whether Designing aesthetic and legible graphics is AI-based or not, that system must respond an expert job. It should therefore be done by quickly, provide help, perhaps allow user friendly consultation of databases, be gen- an Expert System. It is, perhaps, again a too hard task to develop such a system. erally easy to use and easy to see. On the other hand, the lazy AI-er shouldOther desirable qualities include some, at apply her/ himself to the design of aleast, freedom of choice (which may, per- haps, be an option specified by the author) 60 Using Computers Intelligently 51 of material, ability to terminate a session Webb,G (1985). Student control under the and resume from the last point of exit, and, feature network based courseware design especially, adaptation to the particular methodology. Proceedings, 3rd CALITE student. This latter aspect is addressed in Conference, University of Melbourne, (Harris, 1985). Melbourne. Conclusion: AIBAS is possible now. Marshall Harris has been a program- By now it must, I hope, be dear that existing mer since 1959, and a Lecturer in and proven Al techniques, especially Ex- Computer Science since 1971. Prior to pert Systems, not only can, but must be that he worked, at different times, used in CAL Authoring Systems. within the computer industry in both software design and applications pro- It should also be clear that there is an order gramming. In 1983 he convened and of priority of applica'lon of Expert Systems organised the first CALITE Confer- techniques, starting with the easiest appli- ence, at the University of Queensland. cation first. That order is, I believe: He has designed a structured CAL authoring language, which he used to 1. the Author's Assistant; produce CAL material for 1st-year Computer Science students. His cur- 2. the Graphics Editor; rent research interest is in the field of Artificial Intelligence applied to CAL, 3. the Educator's Style Adviser; with a. few sidelines in other applica- tions of Al. He is currently implement- 4. the rest! ing his Intelligent Computer-Assisted Tutorial system ( "PSICAT" - see this The above order is probably the complete paper) on a variety of machines. inverse of that which the leading edge AI researchers would assert. The point is that my suggestions are feasible and would be commercially and educationally viable within the current state of the art. References Harris,M (1983). What an Authoring System Should be Like. Proceedings, 1st CALITE Conference, University of Queensland, Brisbane. Harris,M (1985).PSICAT:aPseudo- Intelligent Computer-Assisted Tutorial System. Proceedings, 3rd CALITE Conference,University of Melbourne, Melbourne. Webb,G (1984) A Methcdology for Intermedi- ate Level Knowledge Representation in CAL. Proceedings, 2nd CALITE Con- ference, University of Queensland, Bris- bane. Teaching Introductory Programming using a language-intelligent programming environment Michael Newby, School of Computing and Quantitative Studies Curtin Unive:sity of Technology The teaching of introductory programming Various systems have been developed normally involves teaching not only an ap- proach to problem solving and the language aimed at integrating such utilities. These itself but also an editor and some operating systems include BASIC, UCSD Pascal and system commands. Learning to use the editor Turbo Pascal. All of them provide an oper- and the operating system has little to do with ating environment which is menu driven learning how to program but often causes prob- lems because of lack of keyboard skills. Lan- and allows access to the file system. In each guage-Intelligent development tools are avail- case, the editor is easy to use and switching able to aid program coding and this paper de- between the utilities causes few difficulties. saibes the author's experience using the pro- BASIC systems normally contain an inter- gramming environment ALICE to teach Pascal to first year undergraduates. preter and syntax errors are detected at run time and in the case of BASICA on an IBM- PC, the line containing the error is dis- When teaching introductory program- played allowing it to be corrected immedi- ming, no matter which language is being ately. Both UCSD Pascal and Turbo Pascal used as the vehicle, there are a number of contain a compiler and syntax errors are factors apart from language that must be detected by it. In both cases, transferring to taken into account.Firstly, there is the the editor is almost automatic and the cur- method used to describe the algorithms,sor will be on the character where the syn- secondly the computer and therefore thetax error was detected. Systems such as operating system under which programsthese have assisted learning to program, are to be run, and finally, the editor whichbut in all of them the programs have to be is to be used to create programs. The pro-entered fully every time and good key- gram development process using theboard skills are still necessary. above approach is shown in Figure 1. The programming environment ALICE (Looking Glass Software, 1986) was devel- One problem encountered by first timeoped to remove the drudgery of coding users is slow input due to lack of keyboardprograms in Pascal, by providing a lan- skills, and in addition to this there is the guage intelligent editor within a full oper- need to switch between the various utili-ating environment. First year Information ties, invoking them using operating systemProcessing students in this University were commands. introduced to programming in Pascal us- ing ALICE, and the remainder of this paper Becoming competent at usinF; these utilities evaluates its use. and gaining keyboard skills may well be valuable in themselves, but certainly haveOverview of ALICE nothing to do with developing problem The ALICE system is available for IBM PC solving and programming abilities. and compatibles, running under the oper- Using Computers Intelligently 53 attempt to type an illegal symbol in a place- holder is flagged as an error. Undeclared variables may be typed within statements, but these will then be highlighted inre- cmtualsomcsAlizortoal verse video ,and a message displayed on the top line of the screen. Theprogram is held within main memory ina tree format. COKPILII 1 IPOITTPROGIVAH 1110011All trriaRDITOR Each node of the tree consists ofa Pascal construct, with corresponding descendent nodes. In the case of a while statement, the node has two subtrees, one for the condi- tion and one for the statement. The condi- tion node must be a single Booleanexpres- sion, which is representedas a tree, 211205111 Minn AUJOR11101 111,whereas the statement subtree can consist of a list of statements so is said to bea list node. Each subtree has itsown subtrees, and so on.It is this tree structure that Figure 1 ensures that only a syntactically correct programs may be written. a ting system MS-DOS Version 2.0 or higher The second component of ALICE is the and it consists of four components: interpreter and is used to execute thepro- gram held in tree format. It is at this stage the editor that some semantic errors are detected,an the interpreter example being incorrect type expression in the debugger a condition. the help facility The third component is the debugger The editor is syntax-directed and designedwhich assists the programmer in detecting specifically to create Pascal programs. This logic errors. This is done by permitting the is done by providing templates for allpro-program to be executed one step at a time, gram constructs. For example, a while loopthe setting of 'breakpoints' to stopexecu- consists of the template tion on a particular statervmt, and the cursor to zip through the code on the top while Condition do begin half of the screen whilst the resultsappear Statement on the lower half.In addition, there is an end; immediate mode which allows values of variables to be displayed at any point whilst a var statement has the template where the program has stopped. The final component is the help facility var which like all of ALICE is menu driven and Name : Type very extensive. Not only is specific help given on aspects of ALICE and Pascal itself, Typing a reserved word, followed by spacebut it also allows such queries as 'Whatcan or tab, causes the template to be displayed I type here?' and 'What was my last error?' on the screen. The placeholders which must be replaced by the programmer are As mentioned ALICE is menu driven with underlined on a monochrome screen. Anythe main menu displayed across the top of 63 54 Using Computers Intelligently the screen, and the operations and the otherthe placeholder. Pressing theRETURN key menus are invoked by using function keys.moves the cursor to the comment line. Fl invokes the interpreter to execute theAgain typing a comment replacesthat program continuously, and F2 to single-placeholder. Thisprocess is continued by step it. Other function keys invoke differ-typing in Pascal reserved wordsto give the ent menus and a full list is given intherequired template, and then merely Appendix. replac- ing the placeholder. Shouldtyping mis- takes be made,and not discovered The file menu gives until access to the MS-DOS after the placeholder has beenreplaced, file system including DOScommands. Athe., the Undo command (CONTROL-U) request to load a file causes a list of ALICE which goes back to the previousstep, may format files to be displayed; thesefiles havebe used; there is alsoa Redo command the DOS extension .AP. No otherfile types (CONTROL-R) which is thereverse of are displayed. Selection froma menu isUndo. made either by a single letteror using the cursor keys. For the more experienced If procedures are being used, thentyping programmer, ALICE allows some func-the reserved word proceduregives the tions to be invoked with the ALTkey intemplate procedure conjunction with other keys. Proc-name (Parameter); Program development (Comment that says what theroutine does) When users first start to codea Pascal pro- Declaration gram using the ALICE environment, they begin are presented with the template fora pro- Statement gram. end; This consists of This can be completed in theusual way. Invoking the procedure by typingits name program Program-Name (input, causes thf_. parameter list to appear as place- output); holders. For example, the procedurewith (Comment that says what theroutine declaration does) Declarations procedure SUB1(var x : real; n : begin integer); Statement end. when invoked by typing SUB1 wouldpro- duce the template Cursor keys can be used tomove the cursor anywhere on the template, and the Control Subl (var x : real, n : integer); Key combined with leftor right arrow keys moves the cursor forward or backwardstoThis removes from theprogrammer the the nearest placeholder. Provided thattheneed to remember the variabletypes and algorithm is well specified andproperwhether they are called by valueor refer- preparation has been carried out before ence. starting to code, using ALICE isvery straightforward. The cursor is positionedOne further aspect is the ability of ALICEto over the 'Program-Name' placeholder and determine where certain types of stateznent the required name is typed; this will replace should go. Oftenprogrammers, particu- Using Computers Intelligently 55 larly those who are just learning, forget tocurrent beginend block. Figure 3 shows declare a variable. With ALICE, any at- section of Pascal code produced using tempt to use an undeclared variable pro-ALICE templates; (a) is correct and for duces a syntax error messagebut allows the every value of NEXTSCORE, the proce- variable to stay. At that point, typing var dure PROCESSCORE is executed, whereas causes the corresponding template to be in (b) PROCESSCORE is invoked only for displayed in the appropriate declarationthe final value of NEXTSCORE on each section and the necessary variable can be line. declared. while not eof do begin read (NEXTSCORE); On completion of coding, the interpreter is if eoln then begin invoked by pressing Function Key Fl; the readln; program runs until completion or until an end; error is detected. A single keystroke re- PROCESSCORE turns the user to the ALICE editor. end; Once a working program has been pro- (a) duced it may be saved as a text file, rather than in ALICE tree format, and compiled using a compiler system such as Turbo while not eof do begin Pascal, for greater efficiency. With ALICE, read (NEXTSCORE); the program development may be de- if eoln then begin scribed as in Figure 2. readln; Iannum, PROCESSCORE; AL0011111111 end; end; CODS PROGRAK (b) WING ALI= Figure 3 21,..JRS It is very easy to produce (b) when (a) was intended, because on completion of the block following them, the user must move Figure 2 the cursor out of the block and press return in order to generate the next placeholder. First-time users often overlook this. Difficulties encountered using ALICE The code resulting from ALICE can be dif- Given the nature of the ALICE system and ficult to read because of the large number of its reliance on templates, it is quite difficult begins and ends. The practice of append- to make changes to the program structure ing a comment to an 'end' is not available without dismantling the code to some ex- conveniently in ALICE, as comments, tent. All statements within control struc-apart from those attached in declarations, tures are nested within begin and end evenmust appear on a line by themselves. where there is a single statement following then else or do. This means the user has toChanging existing programs can cause be aware of which template generated thesome problems all of which are related to 65 411.1.1.1Mr 56 Using Computers Intelligently the free format in whichthe program isConclusions. held; two examplesare given: The most noticeable differencebetween 1. to move sections of code, itis necessarydeveloping programs in ALICEand any to: other environment is theamount of typing required by the user. With ALICEthis is select the section to bemoved, andreduced considerably because ofthe use of move it to workspace (an area ofmem- templates, and this usually leadsto faster ory: coding. It has extensive help files,is menu driven and protects theuser with such select the section again anddelete it prompts as 'Your file has not beensaved', giving the option to return to ALICEto save move the cursor to the position to whichit. Being language intelligent,syntax errors the code is to be moved, andretrieve itand type compatibilityerrors are detected from the workspace. as they occur altwing immediatecorrec- tion. This lulls the firstusers and some 2. to edit an arithmeticexpression or anexperienced ones intoa sense of compla- identifier, it is necessary to selectit andcency, to the extent where a number donot invoke the edit function. Thiscouldbother to learn Pascalsyntax rules.It is involve three key strokes. interesting to note that mostPascal pro- grammers who have learnt in amore tradi- For obvious reasons, one function thattional environment donot like ALICE at ALICE cannot perform isto create data files first. They complain it isrestrictive and directly at the keyboard. Thismeans that ifdifficult to use, although themajority even- a file processing program is to bewritten,tually accept it has much tooffer. then it is necessary to writeprograms which will create the datafiles to be proc- Its major advantage is thatit allows prob- essed. From a teaching point ofview thislem solving techniquesto be learnt inde- could be seen asan advantage, but withpendently of both the operatingsystem first time users it cancause confusion, par-and the coding process. Thelatter process ticularly as programs must alsobe writtenis relegated to its rightfulposition which is to display these files. merely as a means of getting thealgorithm into the computer. A minor problem is thata program in tree format when saved ina file cannot be dis- From the program developmentpoint of view, the most important played except by the ALICEsystem.If process is the there is a machine malfunctionit causes thedesign of the algorithm. Inthe course program to be stored incorrectly, then thetaught in this School, NassiShneiderman whole program is lost. ALICEdoes not diagrams (Nassi & Shneiderman,1973)are provide automatic backup. used, mainly to impose disciplineon algo- rithm design. From thesediagrams, the Programs may be storedas text files andrequired procedures and variables,both subsequently recovered from withinAL- local and global, may be easilyidentified ICE using the text to tree formatconversionand from that point coding usingAUCE is program called APIN. This program is relatively straightforward. invoked automatically fora program file with extension .PAS, but unfortunatelytheReferences user must remember the file nameas it will not be disilayed by any ALICEmenu. Looking Glass Software 0986). ALICE:The Personal Pascal User Gutae.Turonto: 6 Using Computers Intelligently 57 Software Channels Inc. Nassi, I. & Slineiderman, B.(1973). Flow- chart techniques for structured programming. SIGPLAN Notices, 8(8), 12-26. Appendix Key Function Run the program Single step Delete menu Insert menu Change menu Debug menu Menu of menus File system and DOS menu F9 Help menu F10 Select item SitF9 Miscellaneous Michael Newby is a lecturer in the School of Computing and Quantita- tive Studies at Curtin University. He holds a BSc(Hons) in Mathematics from the University of London and an MSc from the University of Bradford. His current research interests are into the development of software engi- neering tools and the use of such tools in teaching programming. 7 The powerful ideas ofeducational knowledge engineering Neville Stem Senior Lecturer, Information Systems GroupFaculty of Business Royal Melbourne Institute of Technology .1.1=a1IIMMAIMINI Building a knowledge base in practice in- on what models he has available. This structs the student, in more than the do- raises, recursively, the question of main-specific knowledge which may form how he learned these models. Thus its contents. Conceptual modelling is in- the "laws of learning" must be about structionally active in the culturally signifi- how intellectual structures grow out cant area of the construction of knowledge, at of one another, and about how, in the a level at once distinct from and inclusive of process, they acquire both logical and the it nsmission of contained knowledge. The emotional form. (Paper% 1980,p. vii) powerful ideas of educational knowledge- engineering thus have to do with practicalAt the risk of pointing out what has been epistemology. looked at many times, it is notable that Papert doesn't simply say "learning is Powerful ideas genetic". It is the understanding of learning that is genetic. The power of powerful ideas With the usual genuflection in Papert'shas a metacognitive foundation: they have direction (1980), powerful ideas are educa-something to do with the consciousness of tive ideas: they emerge from simple,con-a construction to acquired knowledge, and crete experience to become generalisablewith the corresponding consciousness of and applicable in contexts far removedthere being a process to that construction. from theiroriginal circumstances (Yazdani, 1984a). They are characterisedThe powerful idea by a quality of personal enablement. The power of the ideas, in cognitive terms, lies All of this is by way of introduction to what in the way they form the basis for assimila-strikes me as the "powerful idea" in knowl- tion of knowledge. Hence, in a well-known edge-engineering, a commercial and prod- passage from Minds forms: uct orientated technology of elicitation, representation and machine-acquisition of Slowly I began to formulate what I knowledge. The idea is simple: one learns still consider the fundamental fact of by developing and modelling one's con- learning: Anything is easy if you can ceptual structures in a knowledge-based assimilate it to your (mental) collec- system which performs the role of an inter- tion of models. If you can't, anything active, reconstructable medium. In effect, can be painfully difficult.... The the medium plays the student. understanding of learning must be genetic. It must refer to the genesis of Using the technology of knowledge-engi- knowledge. What an individual can neering to build knowledge-based systems learn, and how he learns it, depends is a process which can educate the bu ilders. Using Computers Intelligently 59 Feigenbaum (1984) has commented that probably go through in order to con- struct such a system. The "finished" In the end it may be irrelevant that a system may have a value in provoking computer program is available to be an other learners to reflect about their own "intelligent assistant"The gain to understandings, but that is an inciden- human knowledge by making explicit tal plus. (p.180) the heuristics of a discipline will per- haps be the most important contribu- And in the same context of primary educa- tion of the knowledge-based systems tion, Cumming and Abbott (1986) have approa-41. (p. 50) made a case for using implementations of logic programming: Knowledge-engineering has developed as ...children ...can look inside (expert a commercial application of artificial intel- systems in micro-Prolog), make addi- ligence (AI). The interesting point is that it tions and modifications and eventu- has frequently resulted in a learning benefit ally, write their own. Teachers also for those involved. An extension and should find it easy to use and modify modification of human expert knowledge, programs and write their own. So the and more efficient industrial practice are range of possible educational activities two examples of such a non-computer- is extremely broad. The suitability of based and commercially significant result. micro-Prolog as a language for use by In some cases, the knowledge-base has children and teachers, and as well or been discarded in the light of what may the implementation of educational have been thought of as a side-effect, as (expert systems) may be its single Feigenbaum (above) has suggested. greatest strength from the point of of education. (p. 133) 1(n, wledge engineering can have a gener- alised educational application, therefore,Few apart from the author and some Aus- which can be distinguished from conven-tralian colleagues (Stern & Adams, 1985; tional methods of computer-assisted in-Stern et al, 1986; Stern, 1986a, 1986b;Gild- struction. Some researchers have toucheding, Stern & Forrester, 1986) have referred on aspects of the application, mainly into the learning effects of the process as a connection with young chilaren. For ex-generalised and comprehensive approach ample, Bigum (1986) has commented: to teaching and learning, particularly at the tertiary level. Tertiary vocational and pro- ...the building of even very simple fessional education tends to be dominated knowledge-based systems by children by what can be described as skill and i.e. by having them make their knowl- knowledge transfer styles of instruction, edge and understandings explicit, and which seems to many to be suited to me- conform to the requirements of a shell, chanical tutoring methods. A knowledge- may provide their teachers as well as engineering approach, however, can be the children themselves with an oppor- easily integrated with standard and avail- tunity to expand and more importantly able educational materials and instruc- to critically question their knowledge. tional methods, including continuous and examination-based assessment. It also rep- Building a knowledge-based system resents an interesting point of departure. will not be important because of the finished product. The educational An illustration by projection worth o; the enterprise w be entirely A simple illustration of what such a depar- located in the processes children will ture may lead to can be provided in the 69 6C Using Computers Intelligently form of a projection. in conventional terti- learning by building these structuresand ary educational methods, for example,models in an external interactive andedi- students write essay s and sit examinationstable medium is an extrapolationof that to demonstrate their knowledge. Ina constructivist view. It is a synergistic activ- knowledge-engineering approachto ity of reflection, both internally andexter- teaching and learning, students willalso nally. build knowledge-bases withavailable materials. The knowledge-bases willbeInstr.tdion could be described, particularly examined and tested. The immediatepur-in the cc;ntext of mechanical tutoring pose will be to have the student-built sys- sys- tems, as the methods of buildingconcep- tem demonstrate a certain requisite knowl-tual structures in students. Itincludes edge content. strategies to bring the student's menta' model of knowledge intosome correspon- The process itself will havea broader, more dence with the machine's. The terminology important purpose. The resultant knowl-of engineering, and of knowledge-engi- edge-bases will be convincing models ofneering, is frequently associated with the human knowledge. They will be ableto beinstruction of people in thisway. The goal taken as material evidence of a human of the activity is the communication,trans- depth of understanding of theknowledge,fer and registration of well-definedquanti- originating in their creators. That under-ties of representable structures ofknowl- standing will have been gained throughtheedge. The reception of knowledgeby the process of analysis and construction, and student is acknowledged and verified by through critical examination ofthe ade- simple recall, and otherresponses to ques- quacy of both the knowledge and itsrepre- tioning requiring elementary explanation sentation in the model. The value oftheand demonstration facilities. Theparallels approach will thus inhere inprocess as wellwith data-communicationare obvious,as as product, and in personal learningasare the connections with the activity of distinct from rote memorisation andrecall. knowledge-enginr :ring. Learning as construction Reversing the direction of the effort ofreg- istration and correspondence,students The process of building knowledge-baseswho build models of theirown knowledge cou.d therefore makea good teaching and in intelligent systemsare creative instruc- learning method by placing studentsin the tors of both the machine and themselves. role of knowledge engineers. Connections This activity of instructive modellingcan can be made with important bodies ofprovide students with the role, insights and learning theory and application, inthelearning experience that normallyaccrue to broad context of using computers foredu-their teachers and other authors ofcourse cation. materials.. Learning can be viewed incog, a tive terms Broadening the focus of attention in as .undamentally- a constructionprocess,knowledge-engineering whose results and whose building blocks are conceptual structures. Intelligent sys- The focus of attention in applying these tems which incorporate machine-learning ideas educationally has to be broad enough have relied on this view, to model student to take process as well as product into behaviour during computer-assisted in- ac- count. Bigum's comments earlier empha- struction. In terms of constructive human- size the process aspect, effectively freeing directed learning processes, however, up the dependence we may feel, or want to 70 Using Computers Intelligently 61 feel, on sophisticated technology. The tech- (MYCIN etc.) gives a misleading view nology is with us, however, and there is no of what is currently being achieved in reason why the machine-based product of the field. However, the number of educational labours should be under- case histories that are available for rated. At the tertiary level in particular this general viewing is sadly limited. (p. prospect is encouraging. 10) (my emphasis) But knowledge-engineering is a relativelyA closer examination of the process, in new, commercial and product-orientated educational contexts, could lead towards a activity whose constituent processes and more general application of the technol- technology are chagrining, and are not yetogy, as well as firmer underpinnings for its well understood. The sole focus of com-use in specific areas. mercial attention and incentive to date has been the development of a finished prod-The strengths (and weaknesses) of uct. Both the processes and the productIntelligent Tutoring Systems tend to have been regarded as proprietary. The focus, purpose and the concealment from public scrutiny of the activity tend to The powerful idea of educational knowl- deflect attention from the human learningedge-engineering is an epistemological effects, and to place educational efforts to idea, having to do with the construction of use the technology in the toy category. knowledge. How does the process com- pare with tutoring systems, and their cen- A recent comment (Jones, 1987) from a tral idea? Yazdani & Lawler (1936) summa- professional knowledge-engineer in a rise the application of artificial intelligence journal under the heading "Commercialin intelligent Tutor Systems: AI: Expert Systems for Blood: Buildinga Business" nicely sums up the issues of ITS have a central objective of commu- learning benefits and proprietary silence: nicating some knowledge through computer facilities which may employ A problem for commercial clients in domain-specific expertise, error particular is that the consultant analysis, and user modelling. Their user-oriented intelligenceis con- (knowledge engineer) gets to know a trolled by instructional great deal about the customer's business. strategies This drives some organisations to at- which present (the user) with (do- tempt to grow their own in-house main-specific) problems and then test groups....The secrecy with which for understanding of the knowledge, some organisations wish to surround generally conceived of as a collection of methods for problen solving, as their expert systems activities leads well as domain-specific knowledge. to...(a) frustration. It is often impos- sible to get permission to talk about the customer and/or the application They state that the strengths of ITS derive from "good definition and... complete- to others. When embarking on an area ness" in the sense that an ITS like expert systems applications, it is pretty obvious that having good ref- erence sites gives added credibility will have a well articulated curriculum embodied in its domain expertise and both to the contractor and to the tech- an explicit theory of instruction repre- nology itself. sented by its tutoring strategies. This It is our view that endlessly talking completeness permits an ITS to pack- about the great systems of the past age existing expertise and focus on the 71 62 Using Computers Intelligently novelty, which is the use of mechani- Yazdani & Lawler (1986) take an example cally embodied sets of rules as a tool for from Brown uid Burton's work (1978) in instruction. Because an ITS can be well teaching arithmetic, which requires the ITS defined for a given curriculum, the to analyse and respond to "bugs" when a achievement of its goal, in principle, "child's logic represents a deviation from can be unambiguously evaluated. the standard addition algorithm". They go on to describe the resulting issue as follows: The weaknesses of ITS "set against these considerable strengths" are, they state: There will surely occur cases where the student uses representation be- ...inadequate complexity in what the yond the current scope of the system. user knows, how the user learns new What should the system do when it knowledge and what could be consid- determines that it does not, and even ered an appropriate instructional the- more cannot, understand what the ory for complex minds. student is thinking? The systen should have available an extensils... In terms of the knowledge-engineering repertoire of representations. No sys- approach I am advocating, such weak- tem which is too rigid to learn should nesses are not necessarily redressed by be called intelligent. increasing the internal complexity of ITS. The argument is that an appropriate in-They propose that the intelligence of the structional theory for complex minds ITS, therefore, should consist in large part needs must take a knowledge-engineeringof being able to learn by experience, rather approach into account, on the grounds ofthan by being told - a crucial step which the self-constructive character of mind.. effectively begins (but of course does not necessarily seal) the process of removal of Making ITS more complex internally such intelligence from the direct control of students: The attraction and challenge of an ITS is to get it to behave as a flexible instructor, While people learn more from experi- sensitive to the "complexity of the instruc- ence than from being told, for comput- tional situation" (Yazdani & Lawler, 1986). ers the opposite is more nearly true. The focus is therefore on the issue of de- Computers can learn from experts signing machine-learning and other expert today; for use in education, an ideal systems internal to ITS to support this in- instructional system should be able to tended role. The idea that an ITScan be learn new representations applying to improved by incorporating severalcom- its domain of primary expertise even plex knowledge-bases, and "sites of intelli- from non-experts, as when the stu- gence" (Richards 1985, 1986), whichare dents are introducing some alternative separate and distinct from the learner, and way of thinking - however non-stan- set in a particular relationship to the dard that way may be. learner's activity, has attracted consider- able research interest. One feature of this It is central to the knowledge- engineering approach, when human learning with an paradigm that what Yazdani & Lawler say ITS is discussed, is that it is predicated should, ideally, be true: that "computers upon the richness and complexity of ma- can learn from experts today". But it is how chine-learning in the student model inter- they learn which is vital. To develop an nal to the ITS. instructional theory for the knowledge- Using Computers Intelligently 63 engineering paradigm one needs to give the mind is the central need in effective students the latitude to instruct the system, education, thinking of learning as and to exploit the system's learning capa- 'adding another rule to the database' bilities in the same way that ordinarily a maybe counter productive (p. 74) (ital- teacher might a student's. If this latitude is ics my emphasis) in some way reduced, by assuming for example that it does not exist, then ofIf adding another rule to the database is course the basis for the paradigm is weak-inadequate as a view of learning, and as an ened. action has no educational significance, then the knowledge-engineering paradigm is Concern about the knowledge-engi- on shaky ground indeed. But mechanical neering approach: it's too easy reductions of the learning process however should be recognised for what they are. It is not simply the mechanical addition of Since one of the components of the knowl-another rule that constitutes learning edge engineering approach is an experien- within a educational knowledge-engineer- tial approach to learning, the real difficulty ing paradigm- It is the consistency of that of accounting for the wha t and thehow in the new rule in terms of the organisation and exponential process should not be under- knowledge-representation discipline of estimated (as Yazdani & Lawler, 1986 inthe knowledge-based system that really fact warn). Following this line of argument, matters. it can be said of many tutoring systems that it is time to be suspicious when they assert It is easy to seize upon the editability of a a narrowly specific mechanism of what and knowledge-base, and to object to it in isola- how. tion from the formal organisation, and therefore the testable behaviour of the Few would regard drill-and-practice or an whole. Logic, for example, has been from overlay approach to student modelling as the time of Aristotle a discipline of building adequate expression of how a complexconsistent and valid discourses. Adding mind learns. But there is in Yazdani & another rule to a logic program is equiva- Lawler's arguments a demonstrable biaslent to adding another statement to a logi- against leaving issues of instructional cal proof. The consequences, and the sig- complexity of a tutoring system to thenificance of such an act, are in either case human using it. What a human does to the the same. system can be, in a curious twist to the argument, derogated because a). it is too The redress for the problem of a rule added easy and b). one doesn't understand it at random lies in testing the system itself because it is too complex. As an example offor adequacy and consistency. Notwith- this bias, here is Yazdani and Lawler's criti- standing the mystery of how a student cism of learning by building a knowledge learns through experience internally, we base: will have some measure of the success of that process if that student can bring the Knowledge based systems, with rules system to the required standard. To re- stored separately from the processes phrase Yazdani and Lawler: thinking of which use them, have proven their learning in this way is productive if it can be usefulness because their very lack of seen in terms of the organisa'ion and reorgani- organisation permits the addition of sation of knowledge in the mind. In this light, new rules without reprogramming; if, some of the "considerable strengths" of ITS on the other hand, the organisation can be incorporated in the knowledge- and reorganisation of knowledge in engineering approach: there are clear proc- 7 3 64 Using Computers Intelligently esses to follow, and well-defined externally and the mental models of theuser. assessable products as a result. (p.13) Concerns about the knowledge The resulting dilemma several authors engineering approach: it doesn't have described as follows: instruct One wants to support a child but not Interest in ITS can also be fueled bycon- the development of a wrong cerns, often provoked by the experience of theory...we want to support the badly managed alternative approaches, child's commitment to his authentic point of view and its creative applica- about about guidance, control and instruc- tion. Here is the viewpoint of a critic of tion; at the same time we want to Logo usage (Cumming, 1985): modify that view. (Yazdani & Lawler, 1986) Children left to learn in an unstruc- tured, do-it-yourself hands-on way ...the central problem of humane edu- sometimes generate for themselves cation is how to instruct while respect- quite bizarre mental models. In con- ing the self-constructing character of temporary writing on education there mind. (Lawler & Papert, 1986,p. 69) is a strong theme that such laid-back, In the knowledge-engineering approach, laissez-faire approaches are to beap- plauded. Partly this is an understand- the differential between human cognitive able reaction to the worst rigidities of models of knowledge, and thoserepre- teaching machines, partly it is a ration- sentable in a machine can however be used alisation in the face of the great diffi- as a dynamic of human learning. It does not culty of writing good CAL (Computer exclude issues of human complexity,nor Assisted Learning) materials, partly it discount the useful aspects of buildinga is a healthy affirmation of choice, rele- model consistent enough to run in the vance and motivation. But we must be machine. The cause for concern arises vigilant! Encouraging children simply when one tries to bring people into line to go to it may be fashionable, but will with machines, not the other way round. for some children lead to serious mis- Papert (1980) made this point in a broader social context: conception. We must not sidestep the responsibility of curriculum designer Consider another example of how our and teacher to choose and structure images ef knowledge can subvert our and guide: helping a learner builda sense of ourselves as intellectual good mental model is too important agents. Educators sometimes hold up for this.(p.9) an ideal of knowledge as having the kind of coherence defined by formal The concern is aggravated by empirical logic. But these ideals bear little resem- results such as these (Gentner & Stevens, blance to the way in which most 1983): people experience themselves. The subjective experience of knowledge is ...all too often there is no correspon- more similar to the chaos and contro- dence among the conceptual model of versy of competing agents than to the the system that guided the designer, certitude and orderliness of p's imply- the system image that is presented to ing q's. The discrepancy betweenour the user, the material in the instruc- experience of ourselves and our ideal- tional manual that is taught to the user, izations of knowledge has an effect: It 1/ 4 Using Computers Intelligently 65 intimidates us, it lessens the sense of is related to the frequently asked question our own competence, and it leads us of: how does the system learn? The simple into counter productive strategies for answer is of course, through what the stu- learning and thinking. (p. 192) dent teaches it, and not necessarily through what it can infer from the student's behav- It is hard to see the knowledge contained iour. within an ITS as anything other than hav- ing the "kind of coherence defined by for- In terms of who or what does the shaping, mal logic", irrespective of its degree ofone should also consider the scope of the sophistication or complexity. Similarly, environment within which the student can irrespective of the complexity or sophisti-construct things and get them to "run" cation of the strategies of instruction, the satisfactorily. Here the image of knowl- central irreducible objective of tutoring edge is as a shape to an explorable space of systems remains: to transfer a quantity of "things to do". This space is a microworld: knowledge. The problem is to avoid such aa bounded environment populated with "counter productive strategy for learning computational objects, having some repre- and thinking" and yet still to guide; tosentational mapping to activities in the instruct so that the learner constructs, "real world" (Adams, 1986). rather than forcing the issue to the point of mechanical communication so that the The merits of this view include the recogni- learner remembers. tion of a hierarchical and recursive compo- sition to microworlds. The designer of such My argument is that the best resolution of a world can include in it the objects and that problem is to be found in a serious- tools to make and to contain other worlds. minded application of knowledge-engi- Hence the role of designer can be trans- neering in education. We have nonethless erred to the designed, or more precisely to to meet an important objection to com-the user within a designated space. Such a puter-based learning environments in microworld, in other words, can be views d general, that a "model-based epistemol-and presented as just another computa- ogy of instruction does not exist" (Yazdani tional object. The concept copes with (frag- & Lawler, 1986). It is argued in this view mentary) collections of microworlds, and that in a modelling system there is no set ofof computational objects within worlds. instructional techniques which derive their justification and purpose from a specific The major weakness of a simple discovery- and empirically verifiable view of the con-orientated learning strategy, as we have struction of knowledge. noted at several points in this thesis, is its openness to aimless wanderings. This ob- The question is begged if a narrow view is jection can be partially met by acknowledg- taken of instruction, the direction of com- ing that explorations of microworld spaces munication, and of the separate roles ofcan be directed by teachers, and by job or teacher and learner. If teaching can be life-related requirements to do something connected with learning, then there are specific. An example of the latter could be many strategies available to the studentthe building of industrially useful knowl- who does the instructing of the machine.edge-based systems using tools within a The knowledge-representation scheme ofknowledge-engineering environment. the system, and the means of programming in it then connect in the answer to theNonetheless the apparent lack of an innate simple question of: in what way does the instructional strategy, belonging to the student tell the system what to know? Thisrr icroworld and based on some view of 7 5 66 Using Computers Intelligently epistemology, is not fully dealt with by people: how come they can think (orsay) reference to external direction. Theargu- that? This is practical epistemology. ment is further weakened, in the knowl- edge-engineering example, by theveryEach epistemological statement has impor- generality znd scope of the working envi-tant repercussions is terms ofour under- ronment. What does one learn from a blank standing of the cultural and personal bases sheet of white paper, ora word-processorfor knowledge, and ourown views about for instance? how it is possible to alter knowledge,and hence learn. The active instructionalpur- One must therefore look for instructional pose of a model-based system therefore is strategies in what the modelling environ- simply to make it possible to learn those ment does to the student as an activeconse- important things about knowledge by quence of its design and character, notbuilding and "running" models. The point simply in the constraints,more or less, is at its sharpest in an interactive concept- placed upon him or her by externalagen-modelling system. The instructionalstrat- cies or the scope of the modeling environ- egy of such a system therefore cannot be ment. To do justice to the notion of instruc- seen in the same terms as that of a linearor tion, it must come to the student. What heorbranching or iterative drill. Its curriculum she does, whether directedor not, should is the construction of knowledge itself. give back something. The reflection in the mirror has t* teach, in other words. Conclusion Instruction through practical In summary, the powerful ideas ofconcep- epistemology tual modelling have to do with practical epistemology. Building a knowledge base A model-based system teaches something does in practice instruct the student in very concrete and specific about knowl-more than the domain-speciiic knowledge which may form its contents. Conceptual edge. Its real "instructional goal" isto communicate a number of personally and modelling is instructionally active in the socially significant epistemologicalstate-culturally significant area of the construc- ments. These are that human knowledge istion of knowledge, at a level atonce distinct at once from and inclusive of the transmission of contained knowledge. representable; I have argued that, within the knowledge- constructed; engineering approach, a putativegap be- limited and tween a student's mental model and the editable. model of knowledge internal toan ITS can be exploited as a dynamic of human learn- It is easiest to see these communicationsasing, rather than seen as a problem for the instructive when the student is givena machine to solve. Instead of bringing the knowledge -based system, for example,student's knowledge into acorrespon- which already has some significant anddence with system knowledge, and testing relevant knowledge modeled within it. Butfor that direction of correspondence, the there is also a strong basis for the argumentreverse becomes true. Both the machine- in a concept-modelling shell itself,a mes-based and the personal models willun- sage let us say, in its medium. In a sense, thedergo dynamic alteration as the one, and by ability to run a concept model isan instruc-implication the other, is brought toa new tive and live demonstration of theanswerdesignated and testable standard of knowl- to the question one frequently asks of otheredge. 7c 10111111 Using Computers intelligently 67 The role of a (human) teacher, and the have an "explicit theory of instruction rep- responsibilities to guide and instruct are resented by its tutoring strategies", except clearly not undervalued or affected by the that here the tutoring strategies are those of alternative knowledge-engineering ap-construction, and the explicit theory of proach. The communication of an existinginstruction is based on its connection with and contained body of knowledge, with construction. And finally, as expected of an which the student is to work and be ITS, the "achievement of its goal, in prin- equipped, is not necessarily impeded by ciple, can be unambiguously evaluated". doing this. It is not impossible that stu- dents' knowledge will remain unmodifiedThere is an entertaining sense, in the fore- by this process of exploration, constructiongoing, of doing things to tutoring systems and instruction of the ITS. Normal evalu- that they were supposed to do to you. More ation and assessment procedures in a terti-seriously, though, connections are possible ary level institution are not disregarded. A between the knowledge-engineering ap- constructive form of independent learningproach and the use of general computa- becomes possible, and particularly valu-tional environments which also span the able in periods of financial constraint. four standard models of computer usage in Support for the student's activity can be education. Their major characteristics can provided in a variety of forms: available be described as: an openness to user access, knowledge-based materials, such as lec-and a continuity between what the user tures, tutorials and textbooks; and content,does with them and the means of their own procedural and process adviceam hu- construction. An important example is af- man teachers. It can then be argued that a forded by I3oxet (di Sessa, 1982). student is in a better position to develop a deeper understanding of knowledge thatBibliography may otherwise have been rote-learnt. Adams, T. (1986) Towards a Theory of Mi- The approach leads me to reverse an earlier crozvorlds, in Salvas & Dowling (1986), quotation from Yazdani and Lawler (1986): 312-320. Bigum, C. (1986) Knowledge-Based Systems (ITS) user-oriented intelligence is con- in Education: TheFourthTovl? in Salvas & trolled by students' instructional Dowling (1985),176 -182. strategies which present the tutor with Cumming, G. (1985) Mental Models - the (domain- specific) problems and then world inside our heads, COM3 11(1), test for understanding of the knowl- February 1985, 7-9. edge, generally conceived of as a col- Cumming, G. & Abbott, E. (1986) Exploiting lection of methods for problem solv- Expert Systems in Education, Proceed- ing, as well as domain-specific ings of the First Australian Artificial knowledge.Emy emphasis and altera- Intelligence Congress, Melbourne, tion] November 1986, 238. di Sessa, A. (1982) A Principled Design for an The "considerable strengths" of ITS as Integrated Computational Environment, described by Yazdani and Lawler also Laboratory for Computer Science, MIT, stand up well to this re-viewing and re- Technical Report MIT /LCS /TM -223, phrasing. Once a student has finished July 1982. building a particular system that passes all Feigenbaum, E. A. (1984) Knowledge engi- the tests an assessor can devise, it then "will neering: the applied side in Hayes & Mic- have a well articulated curriculum embod- hie (1984), 37-55 ied in its domain expertise". It will also Gentner, D. & Stevens, A. L. (1983) Mental 7 7 Using Computers Intelligently Models, Erlbatun, New Jersey, 1983.. Powerful Ideas and Education, Working Gilding, A., Stern, N. R. & Forrester, C. Paper W138, Department of Computer (1986) Students as Knowledge Engineers, Science, Exeter. Proceedings of the First Australian Arti- Yazdani, M. & Lawler, R. (1986) Artificial ficial Intelligence Congress, Melbourne, Intelligence and Education:an November 1986. overview, PEGBOARD, 1(1), Exeter, Hayes, J. E. & Mid& r). (1984) (eds) Intel- Spring 1986, 68-74. ligent Systems: the unprecedented opportu- nity, Ellis Horwood, Chichester, 1984. Jones, R. (1987) Commercial AI: Expert Systems for Blood: Building a Business, Professional Computing 22, Australian Computer Society, April 1987, 9-10. Lawler, R. & Papert, S. (1986) Intelligent Videodisc Applications for Pre-readers, Journal of Educational Computing Re- search, 1(1), 1985, p. 67. Papert, S. (1980) Mindstorms: Children, Computers and Powerful Ideas, Harvester, 1980. Salvas, A. D. & Dowling, C (1986) Comput- ers in Education: On the Crest of a Wave, Proceedings of the 1986 Australian Computer Education Conference, Computer Education Group of Victoria, Melbourne August 1986. Stem, N. R. & Adams, T. (1985) Teaching Prolog to Students of Business, Proceed- ings of the First Pan-Pacific Computer Conference, Australian Computer Soci- ety, Melbourne, 1985, 1015-1030. Stern, N. R., Hinrichs, J., Lukaitis, S., & Kimber, D. (1986) New Directions in Business Technology Education, Proceed- ings of the 17th Annual Computer Conference, Australian Colleges of Advanced Education, Darwin, July 1986, 72-78. Stem, N. R. (1986a) Prolog as a Mind Tool?, Proceedings of the First Australian Arti- ficial Intelligence Congress, Melbourne, November 1986. Stem, N. R. (1986b) Prolog and Education, Proceedings of CALITE 86, Adelaide, December 1986. Yazdani, M. (1984) (ed) New Horizons in Educational Computing, Ellis Horwood, Chichester, 1984. Yazdani, M. (1984a) Artificial Intelligence, Bandwagons, hearses and other vehicles: The useful fate of Prolog in education Neville Stern Information Systems Group, Faculty of Business Royal Melbourne Institute of Technology The useful fate of Prolog in education is to open Logic) has induced a shift of interest in this and join the way to educational knowledge - direction by both its strengths and its weak- engineering, within a general framework of conceptual modelling. Together with work in nesses. wide and sometimes apparently disconnected areas, it contributes to the possibility of a new Amongst its strengths in principle: approach to teaching and learning that incorpo- rates many of the strengths and ameliorates some of the weaknesses of its precursors. it is an easily available knowledge-rep- resentation and knowledge-base build- The confluence of a number of strands ing tool, with elementary logic as a dis- cipline: of construction within a broader Building knowledge bases for educational rule-based methodology; purposes is a useful approach to teaching it has encouraged the sense of an 'or- and learning with the technology of Artifi- ganisation to knowledge' which chil- cial Intelligence. This paper briefly reviews dren can easily appropriate and master; a number of educational and knowledge- it has also facilitated, to some extent, the representation research projects which placing of novice students in the have contributed to, or can now be linked role of builder of the expert system; under the heading of, educational knowl- its major strength is its support in prin- edge-engineering. ciple for the representation of concepts and a consequent conceptual modelling Work with Logo in the development of activity. computer-based modelling environments, and with Prolog in a movement towards Amongst its weaknesses in practice: the educational construction of knowl- edge-based systems, are two important an otherwise appropriate view of logic contributory strands. Concept mapping as a discipline "in its own right' 1 !..ading and conceptual graphs are examples of to the promotion of logic as a sipgle two distinct activities, the one educational important "backbone" (Ennals, 1984) to and the other technical and theoretical, more general areas of knowledge; which can also be connected in this way. hence, conflict with the ways in which people actually 'know', and other more Prolog and the idea of educational flexible means of knowledge represen- knowledge-engineering tation; an overemphasis of the value of declara- As one of the important strands converging tive as opposed to procedural knowl- on the idea of educational knowledge- edge, which implies a static or finished engineering Prolog (Programming in state to knowledge; 79 70 Using Computers Intelligently a consequent assumption that, given alogic programming as a language for complete specification of a problem,children's learning has led to a revision of problem-solving is to be carried out byits role amongst a broad and richly varied the machine, overlooking whence theset of educational vehicles to avoid a simi- specification is to come; lar hense-like effect. inherent difficulties in declarative logic of dealing with change of state andAs a sample of aspects of the change, it is dynamic alteration of the knowledge-instructive to review Ennals (1985)com- base; ments on a report evaluating the the Logo in the controversy over procedural andexperiment at the Centre Mondial de Pru- declarative interpretations ofa logicgrammation. Ennals translates part of the program, a confusion of the procedural report as follows: interpretation with process issues to do with actually developing aprogram in ...if one looks at the computer prod- logic; ucts of the pupils...one is struck by the hence an under-valuing of the dynamic, uniformity of the results and by the modelling and prototypingprocesses of wretchedness of the programs....(An) developing a knowledge-base in de- hypothesis would be that the com- clarative logic; puter facilities...are too limited and limitations and complexities in inter- impose too much computer thinking (as face, syntax and environment militating opposed to simply thinking)on the against the construction by novices of programmers, and that the uniformity substantial and non-trivial knowledge- of the results (in Logo it is rare tosee bases; programs which escape from the and, as a database and query language, square and the house) is only a direct it has tended to sidetrack researchers reflection of the weakness of themeans into using it as an unecessary substitute at their disposal and the inadequacy of for other more general and perhaps such computational constructs (and more easily used database and query these languages) for expressing the languages. mechanisms of problem solving. (p. 59) Detail of the arguments for these assertions can be found in Stern, 1986a, 198611 and goes on to comment: Its strengths and weaknesses may besum- There had been a tragic mismatch marised as: support for knowledge-base between potential and expectation. In modelling, and the limitations ofa purely many cases...tne specification docu- logic-programming point of view. Both ments describing the problems to be have contributed to a broadeningmove- dealt with were clear and explicit ment beyond Prolog itself. (much of the text could be run directly as Prolog databases!) but the Logo Indicators of transition Turtle was hideously inadequate to deal with the diversity of problems, The Prolog education movement is in rapid particularly in the hands of unpre- transition. It could be said that it gained pared researchers. some impetus as a bandwagon from the faltering, in France in 1982, of Logoas an I have arg .d elsewhere the inadequacy of educational vehicle. Subsequent experi-argument .or Prolog as an expression of the ence following promotion of Prolog and of "mechanisms of problem solving". There a© Using Computers Intelligently 71 has been, if not a tragic mismatch of poten-tion of Prolog was concerned, has been the tial and expectation, a tension betweenextent to which other software, such as what was and could be done with the lan-simple database and query languages, guage, and what was said about it. Forcould have done what their Prolog-based example many promot^rs of problem-solv-software did (Stern, 1986a). ing with Prolog (such as Conlon, 1985; Ennals,1984, 1985) were, in the midst ofA recent PEGBOARD editorial (Autumn their assertions of the advantages of de-1986) began by asking the question: clarative interpretations of programs, as was said of Milton, of the Devil's party When does a bandwagon become a without knowing it. hearse?...There is a danger of over enthusiasm, and making claims Other researchers, such as Cumming which cannot be substantiated, or (1986) have cautiously begun to raise issues which other factors can explain. of limitations, beyond those of syntax or interface, to Prolog as a general educational and concluded with: medium. At the same time some inherent strengths to the educational use of the lan- The key concept is that of learning with guage are acknowledged: computers, where the computer is seen as a powerful tool for learning. Prolog The restrictiveness of Prolog's struc- has a triple part to play here; the teach- tures for knowledge representation ing of logic programming in order to limits its educational applications, but develop logical thinking, the writing of still allows a usefully wide range for logic ...programs to develop under- exploration. The severity and precise standing of a problem or a topic, and nature of the limitation has yet to be the use of authoring programs or shells determined. Most importantly, to achieve the same goal. The writing of Prolog and EMITSI provide an envi- Prolog programs and Prolog shells are ronment that is very friendly for learn- based upon analysis of the logical ers and can support a wide variety of structures of knowledge within subject activities, including using, modifying domains. They force the writer of a and building small expert systems. (p. program, be it a simulation, adventure 60) game or database to structure the knowledge which the program repre- Indicators of this transition towards a more sents. Prolog can have universal appli- general authoring and knowledge-base cability in the development of logically construction approach can most clearly be based thinking within the triangular seen, for example, in the work and publica- structure of subject knowledge do- tions of PEG (the Prolog Education Group mains' conceptual, procedural and at Exeter). Workers in the group produced propositional knowledge. (Nichol, a number of authoring programs written in 1986, p.5) Prolog fo use at first in the teaching of history through modelling and simulation.The concluding sentence indicates a recu- The software included simple decision-treeperation not only of Prolog's original ech*- modelling of historical choice and action,cational aims (the "teaching of logical an adventure-game building system, andthinking") but of more general educational database construction and query systemsknowledge- engineering and knowledge (Briggs, Dean & Nichol, 1985) A majorrepresentation issues (the disciplining of weakness in this work, as far as the promo-the "writer of a program...to structure the 81 72 Using Computers Intelligently knowledge which the programrepre-While modelling in Logo tends to beasso- sents"), in a context in which "proposi-ciated with the graphic presentation of tional knowledge" is onlyone component. turtle-behaviour, there is amore general In effect, it is no longer Prolog that haspoint. The behaviour of such models is controlled at the level of programstate- "universal applicability", despite Nichol'sment, a textual and linguistic level that reassertion in the context of the "develop- opens up more general possibilities of ment of logical thinking" above, but a us- moreing a list-structured language for symbolic general educational approach usingrepresentation. It is at this level, that the broader knowledge-representation andmost important link with Prolog exists, knowledge-base building tools. Thatcon-since developing a knowledge-base model struction environment, which ofcoursein Prolog is essentially a must include a rule-based and declarative process of testing and extending the meaning ofstatements logic methodology, need not be restrictedin a concept-description language. to it, as a review of commercial knowledge- engineering systems and architectures will The assertions that a Prologexpert-system confirm (for example: Friedland, 1985;is a learning microworld and Fikes & Kehler, 1985; Hayes-Roth, 1985; an object-to- think-with, and that Prolog is a computer- Genesereth & Ginsberg, 1985). based learning environment for creating Examples of this kind of integrated andsuch microworlds, can be confirmedby tool-kit providing environment aboundinreference to current work in microworld industry, and to certain extent in education theory (Adams, 1986) and bya reading of for limited purposes such as the study ofPapert's Mindstorms (1980) from thepar- artificial intelligence or cognitive mecha- ticular point of view of modelling with nisms (for example POPLOG: Sloman eta',concepts. 1983; Gray, 1984). Most examplesdo not focus on the educational effect of buildingA learning microworld (Adams, 1986)is knowledge bases themselves. Boxer (diboth an environment for the construction Sessa, 1982) is worthy of mention in this of models, and a working model itself- a context, because as Nichol reports (1986) it recursive, hierarchical composition. Every is beginning to attractsome interest inconstructed model possible withina given Europe as an environment for learning. As environment is defined by the (essentially an "authoring environment" it is describedlinguistic) set of instructions which consti- by PEG workers as a kind of nextstep, tute its program. Expert-systems no matter something which "combines the learninghow smaller prototypicalcan be viewed as features of Logo, Prolog, LISP and Small-microworlds -as- models, with defined behav- Talk in one environment" (Nichol, 1985, iour and scope of activity, created withinan p.6). microworld -as- environment. Logo Modelling and Concept Concept models in Prolog display key fea- tures of the modelling process. One is able Modelling to examine the model's behaviour, circum- scribed by the program that defines it, and Logo hat: done more than provide a spring- to change the scope of its possibilities of board for Prolog. There is an importartbehaviour by concrete means- that is, connection between modelling a set ofsimply editing the program and re-running concepts in a language like Prolog, and the it. The 'closed world' of any Prologpro- body of modelling theory that hasgrowngram, within an environment that allows up around Logo. this world to be changed, matches thecon- Using Computers Intelligently 73 cept of a learning microworld very closely. The first, concept mapping (Novak & The important point is that one can alwaysCowin, 1984) is an educational teclinique edit that dosed world within the envonn-which has so far been done without com- inent supplied by the Prolog progran ngputers. It derives from studies in cognition system. (Ausubel, Novak & Hanesian, 1980), and connects with knowledge representation An insistence on a declarative world intechniques for computers, in particular which nothing ever "happens" makes a semantic networks and more generally a problem out of this creative editing pr ocessformulation of concepts and their linkages where none really exists. Furthermore, in terms of connected graphs. Kay (1986) simply adding procedural capabilities tohas described the use of computer -based Prolog as a programming language, orconcept mapping in an educational con- turning to procedural interpretations of a text, but restricted her interests to a means logic program, does not represent a fullof representing knowledge in a form ',)y response to the issue of building concep- which an intelligent tutor system could tual models in and through time. For ex- confirm that a certain body of knowledge ample, extending Prolog to include Turtlehad been successfully communicated to a Graphics (Kowalski, 1984) does not extendstudent. the notion of a learning microworld in terms of the significance of the editing andThe second is in fact a significant generali- modelling process. sation of the first, but one in mhich all educational implications have so far been The body of microworld theory that hasignored. Computable conceptual graphs, grown up around Logo expresses the ideaand hence the idea of a conceptual graph of the computer as ultimately a tool for processor which could include all the fea- thinking about thinking - an extension oftures of logic programming, are the work the more particular instances of the com-initially of Sowa (1984). This work is almost puter as means of thinking about some-exactly contemporary with thrf of Novak thing. Building knowledge-bases ex-and other researchers in educate. nal con- presses this idea perhaps even more neatly cept mapping. Neither Sowa n or Novak than building abstract mathematical struc- acknowledge the existence of each other in tures in turtle geometry. A small Prolog their work, nor so far have th.'olio iers. prototype expert-system is explicitly a microworld for thinking about thinking. There is an obvious and missing connec- tion to be made. The link is of course the Concept Mapping and Conceptual educational knowledge-engineering ap- Graphs proache, provisioned with an easy- to-use and yet universal means of vnowledge The confluence of the tributary strands representation - a conceptual g a ph proces- discussed above, with the experience insor with the technology and the interface industry of the human - learning effect of that would make it umble by novice stu- engineering a knowledge-based product,dents. At present theae provisions by the can form the basis fr,. a generalised and technology are pumly speculative. inclusive notion of conceptual modelling as both a learning activity, and a knowl-Conceptual modelling as practical edge-base construction activity. There are epistemology paths to this idea, from two distinct and yet-to-be connected areas of re;carch inThe confluence of the four str?'s above knowledge-representation. with the educational effects enc. -ed in 83 74 Using Computers Intelligently commercial knowledge-engineering leadsBibliography to a more general notion of conceptual modelling to describe the process. Adams, T. (1986) Towards a Theory of Mi- Central to this idea is a view of the leaner croworlds, in Salvas & Dowling (1986), as a constructor of his/her own knowl- 312-320. Ausubel, D. P Novak, J. D. & Hanesian, H. edge, as proposed by Papert (1980). Itcan be summed up in the proposition of learner (1980) Educational Psychology: a cognitive as epistemologist. It captures an important view, 2nd ed., Rinehart and Winston, abstraction from learning, which broadens New York, 1980. and distinguishes it from a matter ofcom- Briggs, J., Dean, J. SC Nichol, J. (1985) Toolkits munication and of memorization. Concep- for Naive Users of Prolog, Proceedings of the AISB Artificial Intelligence and tual modelling in an interactive andrec.:al- structable medium is a concrete expression Education conference, Exter, 1985. of the proposition. That is to say,concep-Conlon, T. (1985) Start Problem-solving with tual modelling directed by the learner is Prolog, Addison-Wesley, 1985. practical epistemology. Cumming, G. & Abbott, E. (1986) Prolog and Expert Systems in Education, PEG- This view helps to meet an ol-,!ection,put BOARD 1 (2), Winter 1986, and in Pro- forward by lazdani and Lawler (1986) that ceedings of the First Australian Artifi- a "model-based epistemology of instruc- cial Intelligence Congress, Melbourne, tion does not exist". Concept model-based November 1986. systems teach very concrete and specific di Sessa, A. (1982)A Principled Design foran things about knowledge. Its real "instruc- Integrated Computational Environment, tional strategy" is to demonstrate, through Laboratory for Computer Science, MIT, a student's modelling activity, a number of Technical Report MIT/LCS/TM-223, personally and socially significant episte- July 1982. mological statements. The ability torun a Ennals, J. R. (1984) Teaching Logic asa Com- concept model is an instructive and live puter Language in Schools, in Yazdani (1984), 164. demon.-tration of the answer to theques- tion one feequently asl.f °ther people: Ennals. R. J. (1985) Artificial Intelligence: how come they can thinkr say) that? This applications to logical reasoning and his- is practical epktemology. torical research, Ellis Horwood, Chiches- ter, 1985. Conclusion Fikes, R. & Kehler, T. (1985) The Role of Frame-basedRepresentationin Reasoning, CACM 28(9), 904-920. The useful fate of Prolog in education is to Friedland, P. (1985) introduction to Spc-vial open and join the way to educational Section on Architectures for Knowl- knowledge-engineering, within a general edge-based Systems, CACM 28(9), 902- framework of conceptual modelling. To- 903. gether with work in wide and sometimes Genesereth, M. R. & Ginsberg, M. L. (1985) apparently disconnected areas, it contrib- Logic Programming, CACM 28(9), 933- utes to the possibility of a new approach to 941. teaching and learning that incorporates many of the, strengths and amelioratesGray, M. (1984) POP-11 for Everyone, in Yazdani (1984), 252-271. some of the weaknesses of its precursors. Hayes-Roth, F. (1985) Rule-Based Systems, CACM 28(9), 921-932. Kay, J. (1986) Interactive Student Modelling 84 using Concept Mapping, Proceedings of Using Computers Intelligently 75 the First Australian Artificial Intelli- gence Congress, Melbourne, November 1986. Kowals1-1, R. (1984) Logic as a Computer Language for Children, in Yazdani (1984), 121-144. Nichol, J. (1986) "When does a bandwagon become a hearse...", PEGBOARD 1(2), Winter 1986, 5-6. Novak, J. D. & Goin, D. B. (1984) Learning How to Learn, Cambridge University Press, 1984. Papert, S. (1980) Mindstorms: Children, Computers and Powerful Ideas, Harvester, 1980. Salvas, A. D. & Dowling, C. (1986) Comput- ers in Education: On the Crest of a Wave, Proceedings of the 1986 Australian Computer Education Conference, Computer Education Group of Victoria, Melbourne August 1986. Sloman, A., Hardy, S. & Gibson, J. (1983) POPLOG: A Multilanguage Program Development Environment, Informa- tion Technology:Researchand Development, 2,109 -122. Sowa, J. F. (1984) Conceptual Graphs: infor- mation processing in mind and machine, Addison-Wesley, 1984. Stern, N. R. (1986a) Prolog as a Mind Tool?, Proceedings of the First Australian Arti- ficial Intelligence Congress, Melbourne, November 1986. Stern, N. R. (1986b) Prolog and Education, Proceedings of CALITE 86, Adelaide, December 1986. Yazdani, M. (1984) (ed) New Horizons in Educational Computing, Ellis Horwood, Chichester, 1984. Yazdani, M. & Lawler, R. (1986) Artificial Intelligence and Education: an overview, PEGBOARD, 1(1), Exeter, Spring 1986, 68-74. 85 Increasing active learningin a CAL environment William A. Stewart and Jack Flanagan, Nepean College of AdvancedEducation, Computer are claimed to have a powerful role to play in tertiary education, yet mostprograms Problems for business education available in the business education field have software not displayed a quality which makes the claims for computers convincing. Beyond specific-pur- pose computational aids, progress has been There are substantial differences between slow. At the heart of this slow program is the business software and educational soft- demand on the computer- learning innovator to ware on business topics. The focus for cover three disparate fields: the academic skills business software is on input-output rela- for her/his discipline, the educational skills for effective communication, and the systems and tionships where the outputs from thesys- programming skills to achieve an understand- tem are the major benefit. Most business ing of possibilities for CAL. software on any specific topic willuse a This paper uses one element of a managerial "black box" approach, biding theprocess- finance course to explore these issues as they ing aspects from the user. were experienced by the presenter and his co- author. By demonstration the paper traces the Educational software on business topics origin of a program, its development in early has a different set of needs. While theout- use, and program modifications as the experi- ence of four semesters' use demonstrated that puts, the results of data manipulationor learning experiences fell short of the authors' computation, are useful, the educational goals. Modifications were continually at- aspects of most interest are usually the tempted to Improve the active learning aspect. relationships, and the processes them- The paper presents a preliminaly evaluation of the learning outcomes and generalises theseto. selves. The assumptions that have been similar tasks in the business education environ- made about the environmentare also of ment. importance. All of these aspects thecom- puter must make available to the student. A feature of accounting and financial soft- For such reasons many business educa- ware is that the system controls of account- tional institutions have takenup the ing programs are often a direct hindrancespreadsheet as a contribution to business to the new user. The penalties for wrong education. entries are usually re- entry of data insepa-Of course the spreadsheet is alsoa widely rate runs, enshrining all mistakes as a per- used tool in business, and knowledge of manent reminder of misdemeanour. Asuch a tool must enhance employability, potential benefit of computer learning has but the educational aspect is that the stu- always been the non-threatening nature ofdent must develop the model her or him- mistake correction. Mistakes are electroni- self, and thus focus on the process, not cally erased, and only finishedsuccessessolely the outputs. show: the accounting security and audit trails are thus in direct conflict with educa- For the one-page process such as a budget tional goals. or cost analysis the learning time for the SC Using Computers Intelligently 77 spreadsheet is reasonable and the ability to computer-specific applications of educa- manipulate data on a "what-if" basis can tional ideas. (For brevity of this paper, we provide powerful insights. Our purpose inhave chosen to provide discusion on these this project, one in a series of attempts to points by way of a handout to workshop use the computer in business education,participants.) was to provide some of these aspects, but as well to develop something more elaborate What emerged from this study was a set of as a teaching tool. general guidelines. We did not have a position on whther our students were ho- Problems from developing your own listic or serialistic learners. It was in keep- software ing with the method of teaching that the process of the progrma followed a serialis- As a general comment, much of he soft-tic pattern, bue important that the student ware received in business education to date should have the widest possible control is poor, particularly by the standards ofover the sequencing of access to aspects of commercial products. An exception to thatthe program. assertion is the range of accounting sys- tems, where the available products in bothUser-friendliness was a vogue term not business and education vie with one an-given much "operability" in the literature other in inscrutability, unfriendliness andwe read, but in our efforts was to be repre- just plain "orneriness". sented by careful control of inputs to avoid program crashes, context-sensitive help, There are reasons why rod educationand simple recovery processes from errors. software for business courses is slow toThe setting in which the user operated was arrive, for three skills, not the usual twothe crowded typical college/university (knowledge of problem/knowledge aboutcomputer room, and use of bell or beep, a systems-programming), are demanded.humiliating announcement to the room The software developer must have a good that the user had "goofed" again, was to be technical understanding of the area, mustavoided. have strong educational motivations, goals and insights, and lastly must acquire someWhere possible, we would try to have system and programming skills to execute educational software mimic the better ef- the technical and educational tasks to meetforts of commercial software to provide a the objectives of the system. We might add familiar interface, and one that operated also that there are few if any financial re- bett_ r than the restrictive and censorious wards for such developments. software we were currently using. Beyond that, the prescriptions for develop- As proceeding in such poorly charted terri- ing appropriate software at the tertiary tory represented such a threat, it seemed level are difficult to find. appropriate to tackle small tasks with ap- parently simple requirements. Specific In a previous working paper (Flanagan/ feedback on success or failure could be Stewart 1985) we explored ideas from theascertained more readily if the task were education area which might guide the small and specific. preparation of business education soft- ware. Of particular interest were the works Our purpose in this workshop is therefore of Gagne(1965) and Ausubel(1968) on the to explore our experiences so far, and by cognitive processes of education. In addi- such exposure help others attempting to tion, Rushby(1975) had written on the cover the same territory. We hope to have 87 78 Using Computers Intelligently commentary giving other explanations asOperationally, the program would require: to why/where we went wrong, or where no instruction beyond disk loading and we should have gone instead. program name crash-proof data entry This particular task non-threatening error correction no consumables beyond hard copy The subject chosen for explorationwas the minimized likelihood of assistancere- financial evaluation for the comparison of quests from data processing personnel leasing with other financing options,some- outputs Ma familiar or expected format. times shortened to "lease versus borrow- and-buy". As an educational task it in-Program failure and the first revision volves a series of computations, present value analysis, and the understanding ofa The first implementation of theprogram set of environmental relationships which ran in compiled Microsoft Basic on DEC affect the financial outcome. A simple 5- Rainbows. It had been tested on a student year example involves about 70 sequential who had an unreasoning fear of anything computations. involving mathematics. As an assignment, she was asked to use the program to hf-'? Because of the computation volume, stu-her with the task, report on difficultiesor dents, in our teaching experience, are loathweaknesses, and suggest improvements. ro do much exploring of possible combina-She was tests.: for competence at thecon- tions, and gain insights to the interactions clusion of her assignment. and influences which determine the deci- sion outcome. Lecturers are none too keenAs a result of her suggestions, beforesome either to create and mark examples, and the processes could be accessed, a multiple- texts cart usually afford one simple-to- choice question which addressed the major medium complexi ty illustration. Rarelyare purpose of the process was installed to act there enough illustrations for practice,noras a "sentry" to each process. Before a proc- sensitivity-tested parallel examples. Illus- ess could be used, the question had to be trations in class are excessively time-con- answen...1 correctly. A short tutorialon suming unless trivial examples are chosen what was wrong with an answer followed and demonstrated at speed. n incorrect response. The program goals Confident in the usefulness of the en- hanced program, the program was used for The program goals were to :- several classes the following semester. make available a demonstration pro-Apart from some minor problems with gram which solved the lease versus data entry, sit -tents succeeded in using the borrow-and-buy analysis program for a conventional assignment. permit the student to generate prob- Subsequent testing of their ability to do the lems for self study with answers to same type of problem manually showed no permit self-assessment sign that the program had assioted them to tutor and explain the problem and pros gain the intended skills or insights. Post- ess sufficiently to permit students tomortem discussions with students showed have their questions answered withoutthat few had been curious enough to use lecturer assistance the help screens or the tutorial, and had also provide references for further reading failed to absorb the understanding of the provide a browsing facility to stimulatebroad processes guarded by the multiple curiousity. choice questions. Using Computers Intelligently 79 The second revision suggested further computations and pres- entations which would have made data For the following semester, a more lengthy available in forms that matched their way assignment required the students to do a of thinking about the problems. While this variety of problems using the program; a represented progress of the sort we had series calculating tasks which would have hoped for, there was still a disappointing been too onerous manually. It was thought lack of competence in performing the ac- that more continuous exposure to the pro-tual task of an analysis under exam condi- gram might have stimulated experimenta- tions. As a generalisation, it could be said tion and enquiry with the use of the tutorial that their interest and understanding had and help screens. been increased, but that the performance of particular skills had not improved to the Tutorial discussions, class testing and oral degree hoped for. A series of intensive quizzes revealed that more use had been workshops would, on past experience, made of the program features, but that too have brought the students to a higher skill large a proportion had asked other stu-level. dents for the answers or rote- learned the required responses and processed the in- The present program formation mechanically without learning taking place. As the inability to display the technical skills of the analysis seemed the problem, It is reasonable at this point to divert to the present revision requires the student to question whether we had arrived at anycomplete one iteration of each of the key more than a reminder of the adage : "youanalyses. The program makes the remain- can lead a horse to water but you can'tder of the analysis available after an at- make it drink" and therefore that you can- tempt. Wrong answers are flagged, but the not force learning to take place. There is analysis is always provided. truth in that, but it must be understood also that the student is under competing pres- After two successful attempts at each of the sures to acquire all sorts of skills andtasks, the student is freed from further knowledge. Knowing when one has the demands by the program, and may make required knowledge and skills so that it is any number of further changes to inputs safe to switch priorities, is one of the diffi- without entering one line of the iteration. cu).fies of a content-overladen curriculum. Class testing indimtes that this procedure We decided therefore that the signals on has had some success in improving this when it was safe to stop needed to be level of skills, though at the time of writing clearer to the student. The structured set of(July 1987) there has not been a testing problems was extended in the followingunder final examination conditions. semester. These required resetting of the problems, particularly the environmentalSummary features, and needed closer study of the computations if the questions were to Our purpose in this workshop has been to be answered. discuss the progress of an attempt at the development of a computer-assisted learn- The results of this stage showed that stu-ing process. The goals of this process can be dents had made more use of the help facili- seen in hindsight as naive in that the pro- ties and had used the options in the pro-gram provided an opportunity for experi- gram more than previously. Some studentsmentation and serendipitous discovery. I., 89 80 Using Computers Intelligently the task- intensive environment of a busi- ness school the stresses on the student do not provide the atmosphere for such a learning process for most students. Some level of success has been reached by a more exacting setting of the problems which the student has been required to explore or solve. However, the program became again a tool to assist lea ning rather than a self-contained environment of learn- ing, which had been the unrealistic goal. Because many business problems require both an understanding of the setting and the problems within that setting, with also a set of exacting technical skills for prob- lem solving, others may be forced to the stratagems we have employed. The power- ful computational processes which the computer can carry out and display may pass before the eyes of the passive student without any 1;osponse. Forcing the student to participate in an iteration of the solution, as well as problem solving for answers on the influence of environmental factors, seem to have been respomible for the improvements we have noted. Bibliography Ausubel, D.P., (1968), Educational Psychol- ogy: A Cognitive View. New York: Holt, Rinehart and Winston, Inc. Gagne, Robert M., (1965), The Conditions of Learning, New York: Holt, Rinehart and Winston. Inc. t,CI V r Applying intelligent CAL principles to help subsystems for interactive software David A. Waugh, Department of Computer Science Kevin P. Stark, Department of Civil and Systems Engineering James Cook University A multi-level dynamic Help subsystem for in- 1. Hierarchical Help Systems teractive software is described. By combining 2. Computer Coaching information about the user's history and pres- ent situation a system can offer help that is 3. User Modelling directly related to a user's needs. A number of help techniques are discussed including dy- Hierarchical Help Systems namic example generation. Most software today comes with some There are many similarities between offer- form of help system. These can be on-line or ing intelligent help to a user of an interac-simply a set of manuals or some combina- tive software package and intelligent tutor- tion of both. Of the on-line help systems ing in computer assisted learning (CAL) available, there iS a wide range of assistance systems. In both environments, an accurate offekei, from simple electronic copies of model of the student, or user, must be documentation to complete elaborate tuto- maintained if appropriate assistance is to rial systems (Houghton, 1984). An example be offered. A help system, however, is not of the elaborate form of help is the Do What concerned with teaching concepts in- I Mean (DWIM) facility in the Interlisp volved with learning how to solve prob- environment (Teitleman & Masinter, 1981). lems with a computer, or presenting les- This facility will attempt to diagnose and sons to the user to be answered and evalu-correct incorrect input while leaving it up ated. Intelligent help system,. need to offerto the user to accept or reject the correction. more than static script based assistanceObvious corrections are made without driven by some form of lookup mecha-consulting the user. Other sysems like the nism. To be truly useful, help must be per-SIGMA message processing system have sonalized. Personalized advice implieson-line tutorial assistarce with built in les- keeping a model of the user and analyzingsons and exercises (Rothenberg, 1979). this model to determine the appropriate form of coaching to provide. In this paper,Another consideration for help systems is we investigate the potential for intelligent whether an active or passive role should be help and present an implementation of ataken. Fischer et al (Fischer et al, 1985) form of user modelling for interactive soft- advocate passive help in a system for the ware with an intelligent help system. subset of commands a user knows a little about and uses occasionally and for the set Discussion of commands the user thinks exist in the system. However, active help is required to There are three main elements to consider introduce the user to the set of concepts and when looking at providing an intelligent commands that exist in the system but the dynamic help subsystem. These are: user has no idea about. 91 82 Using Computers Intelligently Passive help systems wait to be requested analyzing and attempting to correct their by the user. Typically, passive help is avail-own errors. A tutor must know when to able in the form of a menu drivenor key-interrupt a student and what tosay when word system. Menu driven systems havethis interruption occurs (Burton & Brown, the advantage over keyword systems in1982). These principles apply to designing that they are hierarchical in nature andan effective help system. In a help system display a set of keywords on which help isthe advice giving element is basicallya available. Natural language help systems computer coach. To offer the most effective exist but are still not sophisticated enc-:gh advice, a tutor must know something about to provide adequate levels of help, priraar- how students learn and when theyare the ily because of the problems inherent witha most receptive to such advice. The same natural language interface. In general,applies to help systems, except thata user passive help systems do not take intoac- often initiates the interaction. Witha dy- count the user's present or past situationnamic help system the 'coach' takeson the (Fischer et al, 1985). role of tutor by not waiting for theuser to ask for help. But, the scope of assistance Active help systems are characterized by that a help system provides is limited to the their ability to interrupt a user to provideconcepts and commands of the interactive some form of advice or coaching. Interrup- software that is being used. A tutor,on the tions are not neccessarily restricted tocor-other hand, offers advice on all aspect of recting errors made by a user, but also offerthe problem at hand, not just the toolcur- a more efficient method of accomplishing arently being used. This last pointmeans goal than the current method being used. that an intelligent dynamic help system An example of this latter form is the Unixneeds to embody much less information in advisor (Kay, 1986). The important aspectits knowledge base. This also implies that of active help systems is that they monitor less knowledge needs to be stored when the user's behaviour and some attempt iscompiling a user model. made to reason about their goals (Fischeret al, 1985). One major concern with active User Modelling help is when and where to interrupta user. A help system that continually interferesA student model has been defined by becomes a nuisance whereas an active help Clancey, (Clancey, 1986), as a set of records system that never offers help is simplya that describe a student's knowledge about passive help system. Some form of judge-what is being taught so that theprogram, ment needs to be used to adequatelyserv-with the interaction of the teacher, ifneces- ice the user's needs. Support shouldstopsary, can adapt its presentations to the when a command has been successfullystudent's needs. The classification of stu- mastered by the user. One method is to give dent or user models can be achieved using the user control over the level of help inter- the following major characteristics (Rich, face they are using. 1979): Computer Coaching the model is of a canonical user or the models are of individual users Personal tutoring is a very effective method construction is explicit by the user or of teaching, especially when coupled with abstracted by the system from observ- additional formal instruction either indi- ing users' behaviour vidually or in a group. It is important thata the model embodies short-term tutor does not interrupt too frequently as specific information or long-term this can lead students not to discover from general information. Using Computers Intelligently 83 For a user model to be useful it must con- The kind of knowledge we need to tinually evolve to accurately reflect the model. change in a user's knowledge. This can be The method of implementing this achieved by maintaining a record of a model. user's correct and incorrect usage of a sys- The application of this knowledge by tem. According to the classification above, the help system. the model described in this paper acquires its knowledge implicitly to create an indi- The kind of knowledge we need to model. vidual model of a user. The kind of knc edge that is embodied is primarily short-The kind of knowledge we need to model term knowledge about specific constructs includes the number of times a user has and concepts. However, as a user builds correctly and incorrectly used each of the knowledge over time, a long-term model is commands and/or concepts available in a available by continuous analysis of the system. Incorrect usages need to be catego- acquired data. At some point, a user could rized in terms of the determined severity of be determined to have mastered a system the error, ranging from a typographical and the form of advice offered would auto- error or misspelling to a complete lack of matically change to a passive role. knowledge of the correct form of a com- mand or concept. Each command may There are various methods of implement-have a list of possible errors and the num- ing a model. Some of these include overlays ber of occurrences of each kind can be using genetic graphs (Goldstein, 1981),compiled as in fig 1. Also, to aid the help stereotypes (Rich, 1979), scripts (Schank, system in determining its effectiveness, for 1977), frames (Minsky, 1975), schemaseach command a record of the number of (Bobrow and Norman, 1975) and specifi- times help has been offered and what strat- cally for CAI using a procedural student egy was used should be kept. A later analy- model (Self, 1974). The model discussed in sis may show some forms of help to be less this paper uses a form of frame-based rep- effective than others for some commands/ resentation. concepts. A further analysis of many user models could indicate some common areas The User Model of trouble or misunderstanding. These last two points are very attractive to educators The only way to build a help system that using software with such a help system is not continually interrupting is to build built in. and maintain a model of the current user. This model would represent a profile of a particular user's history of usage with a particular interactive software package. This model would need to embody the various attempts (correct and incorrect) made by a user when interacting with a system. The obtrusiveness of the help system can then be controlled by scan- ning the user's model to determine if help should be offered for the current situ- ation. -...., To employ an effective user mode:, we need to consider the following: 84 Using Computers Intelligently The method of implementing this model Users Knowledge The model is implemented in a hierarchy of # of users stored frames. The top level frame represents the list of 10 most common errors knowledge about all users of the system, collection of individual models (see fig. 2). It includes the number ofusers for which a model is kept, a current list of El CD El ED Q D the ten most common errors in the system DEIDEEID and one model for each individualuser. The model for each individualuser, (see fig. 3), includes some form of user identifica- tion, a distinction between the number of Figure 2 spelling errors and constructerrors and a collection of usage records for eachcom- mand/concept that user has used. This Individual User Model collection of usage records growsas the user idenaeon user attempts new commands/concepts in 5 Mg .:iff0f5 the system. The omission ofa record for a ' construct errors particular construct implies theuser either usage records does not know of its existence or simply has not had a need for its use yet. Each usage EJ NI record,( see fig. 4), includes whichcon- 0 struct/concept is being recorded, the III p MI p III number of correct usages, the number of incorrect usages, the number of times help has been offered, the number of timesa user has asked for help for this construct, Figure 3 which help strategies have been used anda compilation of the kinds of errorsencoun- tered as in fig. 1, and a link to the set of static library examples provided for thiscon- struct by the system. Figure 4 The application of this knowledge by Usage Record the help system command/concept To apply the knowledge embodied # correct usage in the model, the help system evalu- *incorrect usage ates the user model to determine # times help offered both the appropriate level of help * visas user asked for help and the specific example or sugges- strategy used to help tion to provide to the user. Forex- error statistics ample, if the user has used acom- error record I mand incorrectly the help system will check first to see the number of canned help times a user has correctly used this Ltbrary command. If no record exists, then Examples this is the initial usage by the user and a dynamic example could be j4 Using Computers Intelligently 85 presented to the user. A dynamic exampleTo determine adequate levels of help and is a nowerful way of illustrating to the user appropriate examples requires a substan- the correct form of their specific input tial amount of time and effort. This is analo- through the use of an example. If a record gous to a tutor who must determine the does exist, the confidence factor is calcu-correct form of help to provide to a student. lated. The confidence factor for a commend No two students are the same, hence an or concept is the ratio of correct usages to effort must be made to personalize help incorrect usages determined for the typesand tuition. Of course, when a student of errors a record is kept. If the confidence masters a system, help becomes passive factor is high, say at least 3:1 and the kind of and inexpensive to provide. error is a spelling error, or minor omission, then the user is determined to have mas-Domain of Applicat;on tered this command and the form of help maybe to automatically correct the errone- Novice programming environments (NPE) ous input while asking for the user's ap-secm to fall roughly half way between a proval. If the form of the error is deter- software engineering tool and a CAL sys- mined to be major, then a library example tem for teaching programming. Users are would probably be an appropriate form of trying to develop software to solve some help. specified problem but the reason for solv- ing this problem using a computer and a One problem arises when a user is deter- mined to have mastered a command orprogramming language is to learn how to program in that language (in addition to concept and an erroneous usage occurs. If learning other problem solving concepts). this error is merely a spelling error, the help A useful programming environment (PE) system treats this as noise in the model. must thus provide the user with some tools However, the help system has no method to enhance the efficiency of the program- of handling the situation where a friend of the user for whom the model has beenming task. To deal effectively with novice created uses the system. Suddenly, we users, some consideration must be given to accommodate the fact that the user is learn- have a completely different pattern of us- ing by doing rather than an expert purely age evolving that contradicts any assump- hying to write a program. This considera- tions made using the existing model. The only way to handle this currently is totion can be in the form of a useful help system that changes dynamically with the strongly discourage the above practice and user's anticipated needs. One method of if this situation is detected by the help sys- accomplishing this is to model the i:ser's tem by encountering many errors in a pre- history of the various language constructs viously mastered command, the system and concepts to use as a reference when can ask the user if they are really who they offering help. should be. One possible honest cause of this situation would be a user who has mas- tered the system then not used the system A NPE for Pascal programmers using for a prolonged period of time and re-Macintosh computers is currently under turned to use it again. One could argue that development at James Cook Ui 'y, the returning user really is not the mod- (see Waugh, 1987). The current version of elled user from before. To compensate for the system uses a static help system that is this, the model could include time stamp- obtrusive and predictable. A user builds a ing and erroneous usage could be checked Pascal program statement by statement. against the previous date to detect anyWhen an error is encountered by the sys- 'staleness' about the user and this com- tem, an attempt is made to generate a sug- mand. gestion which is then offered to the user to 41r.' Q.,t.) 86 Using Computers Intelligently User Help Subsystem- Interface Second First Software 411---4 Level 4*Z-Level Package User Model ___ 1 Figure 5 accept or reject. If a suggestion cannot bewhich level of help they would like, but generated, the user is told the current state- many novice users would usually choose ment is unrecognized by the system. An the interface that would be most beneficial indication as to where the error occurred to them. This allows an experienced user to together with an explanation of the type of avoid the overhead in the system of build- error is provided (as per typical compiler ing and maintaining the user model as well output). A static heir system is alwaysas avoiding the dialogue and interruptions available in the form of a menu-driveninherent in the intelligent help level. The keyword based system. At any time a userfirst level interface maintains information may browse the help system for an expla-about the user by modifying the particular nation and/or example of any concept or user model (hence the one way connection construct for which help is available. in fig. 5) in case the user reverts to the second level interface during a lat,r ses- This form of help can be improved bypro- sion. viding a multi-level dynamic help subsys- tem as described in this paper. The fullThe overall implementation of such a help potential of these ideasrealized by giving system incorporates the lecturer, tutors the user some control over the level ofand students for a particular course and the interface they are inter acting with. The helphelp subsystem fo' a particular package, system is organized into two levels (see fig. introductory Pascal programming and a 5), the first being for experienced users Pascal PE in this case. The interrelation- while tha second is for novice and inexpe- ships of these various elements are indi- rienced users. The first level provides acated in figure 6. Thus if a particular stu- script based system coupled with the abil-dent continues to make a particular error, ity to generate a dynamic example usingthe level of HELP provided is increased the user's erroneous input should the userand if a class of students persistently make re"uest it. The second level provides thethe same error, the HELP facility detects a same functionality as the first only the user teaching problem and advises the LEC- model is consulted on every erroneousTURER to check and modify, if appropri- input to determine if advice should beate, the tutoring approach for that class, given. All u sers have the option of choosingboth human and HELP, as indicated by the 5 6 Using Computers Intelligently 87 connection from HELP to LECTURER la- The development of this student model belled 'Data from Runs' and the corre-therefore involves elements of artificial sponding links be' omen the LECTURERintelligence with cybernetic feedback loops and HELP and TUTORS. In each session, a and a heuristic learning facility combined student enters INPUT to the system, (pri-with elements of database management marily Pascal statements), which are incre-with accumulated statistics for individuals mentally compiled by the Pascal environ- and classes of users. After a number of ases ment. Upon detecting an error, the HELP of the HELP facility, assessment o', each system is invoked which offers advice to student's ability in Pascal can be deter- the student depending on the evaluation of mined in structural terms both semanti- the student model. If no error is detected, cally and syntactically and by aggregation the student is allowed to continue uninter-the whole class can be rated. The HELP rupted. system has been developed heuristically Figure 6 Data from Runs Student/Class Profiles LECTURER Help System Changes Advise Problems Detected TUTOR [ HELP Modify More Advice Advice Automatic No Correction STUDENT Made st Pascal Input Input Environment OK?/ Yes Program Output 9 7 88 Using Computers Intelligently with an interactive learning component- beginning to be attempted to be automated thus along with other logicerror detection prob- lems. A potential problem with intelligent the first HELP procedures are deter-help systems is where the machine incor- mined using our experience as a lec-rectly perceives the user's ability forcing turer of Pascal and introductory pro- the user to do something they did not in- gramming over the past 3 years. tend to do. This can be solved by giving the user control at all times to accept or reject a record of the last 10 errors of each type any action taken by the system. Also, the is maintained continually for access byhelp system should be as transparent as the teacher. This allows modification of possible to the user so that responses are HELP procedures, introduction of newbrief and succinct so as to minimize the prompts and the hierarchical ordering delay and disruption in a user's thinking of the search for errors used by HELP. and acting cycle. It is important to realize that the intelli-This form of help also provides the useful gence level of the student as a Pascal userfeature of individual as well as group feed- will normally increase at a rapid rate andback to the lecturer which promotes the the need for the HELP facility should corre- continual growth and modification of the spondingly quickly decline. The better theteaching style of the lecturer and tutoring HELP system is, the lower the number ofstyles of the tutors and help system. Such a intrusions required by the HELP facilitydynamic situation can only enhance the will be. efficiency and quality of the teaching proc- ess thereby providing room for enriched Conclusions. learning environments. In this paper we have shown the parallelReferences. between tutoring and on-line help systems and presented a method for tailoring theBobrow, D. G. & Norman, D. A. (1975). help to a specific user with the inclusion of Some principles of memory schemata. a user model. Script based CAI models only In D.G. Bobrow & A. Collins (eds), Rep- capture the statistics of the number of in- resentation and Understanding. New correct matches of the user's answer York: Academic Press. against the set of expected answers.The Burton, R. R., Brown, J. S. (1982). An inves- method presented overcomes this restric- tigation of computer coaching for infor- tion of script based CAI models by the mal learning activities. In (Sleeman & inherent embodyment of the system expert Brown, 1982), Ch.4. in the help system and subsequently theClancey, W.J. (1986). Qualitative Student user's model. Models, in Traub, J.F.,Grosz, B.j., Lampson, B.W. and Nilsson, N.J. (Eds) It should be noted that we do not advocate Annual Review of Computer Science, Vol. the use of such intelligent heip systems to 1. Palo Alto, California: Annual Re- replace the role of human tutors, but rather views Inc. to enhance the environment in which stu- Fischer, G., Lemke, A. and Schwab, T. dents and tutors interact. Such a system (1985).Knowledge-basedHelp would take care of many minor problems, Systems. Proceedings of the ACM CHI leaving tutors with more time to handle the Conference. substantial problems like the classic 'whereGoldstein, I.P. (1981). The genetic graph: a do I start?' situation which is only just representation for the evolution of pro- 9 6 Using Computers Intelligently 89 cedural knowledge. In (Sleeman & Brown, 1982), Ch.3. David Waugh holds a BSc (Hons) Houghton, P.C. Jr.(1984). Online Help Computer Science, 1985. The Univer- Systen,.A Conspectus. Communications sity of Western Ontario, London, On- cf the ACM, 27(2). tario. MSc in Computer Science, James Kay, J. (198?). Interactive student modelling Cook University, Townsville. Current using concept mapping. Position is a Lecturer, Department of Minsky, M. (1975). A framework for repre- Computer Science, James Cook Uni- senting knowledge. In P.H. Winston versity. His research interests: Novice (ed.), Psychology of Computer Vision. Programming Environments, Soft- New York: McGraw -Hill. ware Engineering Tools, User Model- Rich, E. (1979). User Modelling via ling, User Interfaces Stereotypes. Cognitive Science 3. Rothenberg, J. (1979). On-line tutorials and documentation for the SIGMA message service. In Proceedings of AFIPS Na- tional Computer Conference, New York, vol. 48. AFIPS Press, Arlington, Va. Schank, R. C. & Abelson, R. P. (1977). Goals, Plans, Scripts and Understanding: An Enquiry into Human Knowledge Struc- tures. Lawrence Erlbaum Associates. Self, J. A. (1974). Student Models in Com- puter-aided Instruction. International Journal of Man-Machine Studies 6. Sleeman, D., Brown, J.S. (eds) (1982). Intel- ligent Tutoring Systems. London, Aca- demic Press. Teitleman, W. & Masinter, L. (1 :1). The Interlisp programming envii onment. Computer 14(4). Waugh, D. A. (1987). MacTutor: An Intel- ligent Novice Programming Environ- ment and Tutor. Proceedings of the Fifth Annual Conference on Computer-aided Learning in Tertiary Education, Univer- sity of Sydney, Sydney. 4-,..., 9 Mac Tutor: An intelligent novice programming environment and tutor David A. Waugh Department of Computer Science James Cook University ImMi111111. The design and implementation of a novice the assistance provided by programming programming environment with an intelligent tutor is described. The implementation pre- environments to code a solution often sented is for use on Macintosh microcomputers imposes a significant learning overhead by first year undergraduate computer science just to master the operation of the tool. This students learning Pascal. Mac Tutor provides an incremental compilation environment that results in a vicious circle that often frus- detects syntactic and semantic errors on entry. trates novice users enough to give up. A suggestion generator provides an 'intelligent' reconstruction of an erroneous statement and An ideal environment for novice users presents this to the user for acceptance. The environment includes full run-time simulation would provide useful feedback in terms of abilities for source code debugging with fea- a creative suggestion as to why the error tures such as single stepping, breakpoints and occurred and even perhaps a suggested variable traces. correction. This kind of assistance can be developed for all of syntactic, semantic and Introduction run-time errors. As much effort as possible should be spent in reducing the overhead Programmers face the challenging task of and effort in learning to use both the PE producing correct, efficient and reliableand the programming language. Another software. This process has been simplifiedpossible area of assistance available to the by many programmer's development aids novice user is a help subsystem within the such as syntax directed editors, (Donzeau-environment. Gouge et al, 1984; Morris & Schwartz, 1981; Marlin, 1986; Moretti & Lyons, 1986; Welsh Such a programming environment is being et al, 1986), and by self contained program-developed and is presented here. The prin- ming environments (PE), examples includeciples for this system are investigated and INTERLISP (Teitelman & Masinter, 1981), presented followed by the presentation of the Cornell Program Synthesizer (Teitel- an environment for novice users learning baum & Reps, 1981), and Smalltalk-80the programming language Pascal on (Goldberg, 1983). However, there exists a Macintosh computers. This environment is subclass of programmers who dependbeing developed for specific usage by a greatly on human assistance to complete first year computer science class to solve their task, these being novice program-programming assignments related to the mers. The task of a novice programmer iscourse. twofold. One goal is to learn a program- ming language. The other is to produce Some of the features to be provided by the coded solutions to various tasks set out byenvironment include suggested correc- their instructor. Existing environments do tions, syntactic and semantic error detec- not provide assistance in both areas. In fact,tion as each Pascal statement is entered, the 100 Using Computers intelligently 91 ability to delay subprogram definitions programming language LISP. INTERLISP, until later, the ability to test and save sub-(Teitdrnan & Masinter, 1981), is perhaps programs individually and the automaticthe most well known LISP environment insertion of variables for declarations.that includes such features as automatic Some of the run-time features availableerror correction, an integrated structure include single stepping through the sourcebased editor, sophisticated debugging, a code, breakpoints, dynamic variable trac-compiler, and a file system. This environ- ing, variable watchpoints that indicate ment progressed from work by Teitelman, each modification of a particular variable, (Teitelman, 1969), where early ideas on plus the indication of any error that is about programming environments were formed to occur like an array index out of bounds, from observing experimental program- or pointer referencing to nil. Static help is ming using LISP. Another significant con- always available to tne user in addition to tribution to the area of programming envi- the obtrusive help a at the beginning. The ronments is the work done on the Cornell system reduces the problem of responseProgram Synthesizer, (Teitelbaum & Reps, time degradation by using an incremental 1981). The Synthesizer views programs as a compiler. collection of computational structures and editing is restricted to structures defined by The major advantage of this system is athe underlying programming language great reduction of time spent by a novicegrammar. The grammar is organized by user editing, compiling, executing, re-edit- templates for the utructures and creation of ing, recompiling etc. The learning over-a program is accomplished by selection head has been minimized by making fulland completion of these templates. The use of the menu and icon features available Synthesizer also provides a run-time envi- in the Macintosh toolbox. The appearance ronment with features such as single-step- of the environment to the user was influ-ping and tracing. Another PE based on a enced in part by the environment pre-structured view of a program is the MEN- sented by MacPascal, (Apple, 1985). TOR-PASCAL environment, (Donzeau- Gouge et al, 1984). Editing in MENTOR - Discussion PASCAL is restricted to using a tree ma- nipulation language, called MENTOL, to Programming Environments edit an abstract syntax tree. Limitations on the rigid nature of structure based editors The term 'programming environment' can are discussed by Welsh et al, (Welsh et al, include systems that manage the complete1986), who advocate that editing should software development process or an inte-conform to the user's conceptual form of a grated tool that supports programming. Inprogram and not impose what may be an this paper the following definition for a unnatural representation on the user. Edit- Programming Environment given by ing should be accomplished using what- Ardis et al, (Ardis et al, 1985), is adopted. ever conceptual level a user is comfortable This being: an integrated software systemwith which may vary at different stages of that directly supports the activity of pro- the program's development. gramming. The systems that support the overall programming process are better Incremental Compiling labelled as software engineering environ- ments. A major problem with programming envi- ronments is the degradation in response A significant amount of work has beentime due to repeated compilation of the done developing environments for the complete program for each modification. 1 01 92 Using Computers Intelligently Techniques to reduce this overhead in-cant progress made towards predictive clude incremental compiling. The earliesttheories about the programming process work on incremental compiling was by and that much of the data collected about Lock, (Lock,1965), who describes a method novice programmers varies significantly of editing Algol 60 programs at the state- depending on the language used and the ment level. This allows for syntactic error environment in which testing was carried detection and correction as each statement our. They conclude with: is entered. Semantic checking is left until run-time. The interactive Algol system Bearing in mind these reservations, it COMA, (Atkinson & McGregor, 1978), was is clear that all aspects of program- developed for use in an introductory pro- ming cause novices difficulty, includ- gramming course and involved complete ing typing, planning, coding, testing recompilation of the source code. The au- and debugging. Many novices are thors reported that the resulting overhead frightened away right at the start by imposed was quite acceptable to the novice the amount of arbitrary detailed infor- users but not for sophisticated users. In an mation that they have to remember, as attempt to reduce the overhead, further well as the problems of typing accu- work was undertaken to develop a form of rately and quickly.Languages in incremental compiling using context sensi- which the novice can very quickly start tive editing with a structure editor, (Atkin- writing and running interesting pro- son et al, 1981). A more flexible approach to grams (e.g. LOGO) have a definite incremental parsing allowing for modifica- advantage in that early success breeds tions of any form is presented by Ghezzi confidence. (du l3oulay & O'Shea, and Mandrioli, (Ghezzi & Mandrioli, 1979). 1980). However, the flexibility gained in the edit- ing unit is at the expense of the storage andThese principles can be applied to many maintenance of significant amounts of in-other categories of users, not just novices. formation. To avoid recompilation, the scope of each statement and variable must Suggestion Generation be maintained to determine the range of effect a modification will have. Work on suggestion generation began with early work on error-correcting compilers, Teaching novices programming (Aho & Peterson, 1972; Graham & ithodes,1975 ; Kantorowitz & Laor, 1986; An excellent source of information con-Moura, 1986; Feycock & Lazarus, 1976; cerning the various aspects involved with James & Partridge, 1973). The techniques teaching novices programming can bedescribed in this paper most closely re- found in the report by du Boulay and semble ideas from James & Partridge. They O'Shea, (du Boulay & O'Shea, 1980). Therepresent a method of representing the report is a comprehensive literature reviewterminal symbols of a grammar as a tree on the topic up to 1980. Findings presentedand perform an attempted match of the include ideas on system use like: not to input statement's tree to the correct form's underestimate the amcunt of effort a nov- tree. One interesting technique also pre- ice user may require t) master a system,sented is the confidence jump that allows and statistics concernir g the major sources the parser to 'jump to a conclusion' based of errors in a programming language, e.g. on a partial match to a certain degree. This the most error-prone construct in Pascal could lead to an erroneous intention diag- was the semi-colon. Some reservations are nosis and if returned to the user as a sugges- expressed concerning the lack of signifi- tion, it could be misleading. However, Using Computers Intelligently 93 erroneous intention diagnosis and result-from prototyping and successive versions. ing suggestion can often indicate the source At all times we maintain a system that runs of the error to the user. and is therefore testable while only offering a limited subset of the projected capabili- Requirements Specification ties.This makes the project continually exciting to the developers because we al- The following requirements were identi- ways have something real to work with, fied and specified before a design was instead of a simulation of 'what it is going initiated. to be like'. Successive versions have the a. Syntactic and semantic check as theadded advantage of continually testing the code is entered. existing software and design principles b. Suggestions provided when the system under new circumstances. This helps to can anticipate the user's intent. remove many otherwise unobtrusive er- c. Force corrections to be made as the er-rors. Also, we managea to discover how rors occur. the use, interface principles and assump- d. A file system to save work in progress astions worked out over a long period of time well as finished programs. (1 yr.). An idea for a response mechanism, e. The ability to save procedures and func-like mouse dick, may sound like a good tions separately to build a library ofidea at the start, but repeated use can illu- routines for later use. minate the frustration that can occur. f.To allow the delay of procedure and function declarations. The basic system was broken down into the g. Offer dynamic help with examples andfollowing modules. (see fig.1), User Inter- rules. face, Editor, Command Processor, Run- h. Menu driven user interface. time Controller, Lex/Parse and Encoder/ Decoder. This breakdown minimized the i.Use independent windows for display of input/output. area of effect of the Macintosh specific fea- j.The ability to trace code and variablestures, the toolbox routines in particular during 'simulated' executionusingThese machine dependent peculiarities breakpoints and watchpoints. were confined to the Editor and User Inter- k. Automatic insertion of variable declara- face modules, a technique known as infor- tions. mation hiding, (Pamas, 1972). Information hiding permits the independent develop- 1. Automatic indentation at various block levels. ment of modules, a useful software engi- m. The ability to test a procedure or func- neering technique, while also encapsulat- tion independently. ing details in one area thereby minimizing n. Detect run-time errors during simu-the effects of changes in other modules. lated execution. Using this technique and porting the soft- ware to another machine implies that The user is assumed to be a novice pro-modifications will be restricted to the grammer learning the programming lan- module containing the machine dependent guage Pascal. No previous programmingdetails. experience is assumed. It would help if the user has had a single tutorial on using theThe Editor and User Interface modules Macintosh prior to their first session with were developed first resulting in a working the system. editor. We then used this editor in our first year class as a simple editor for entering Design Principles programs into MacPascal. This gave us on- The system was designed using top-down si te testing by the actual users of the system principles combined with the advantagesenabling us to compile feedback on design 103 94 Using Computers Intelligently Figure 1 issues much earlier than if we had waited to matical category at a time. This allows for release the final version of the system. Thecontinual testing and the maintaining of a next module under development is Lex/ working environment at all times. Work Parse. This involves the lexical analysis and on the Suggestion Generator module is parsing of the Pascal source statements en- proceeding in parallel with the Lex/Parse tered by the user. The lexical analyserwasmodule. This is facilitated by the strong originally developed using Lex, (Lesk &relationship between the original parse Schmidt, 1975), under Unix by specifyinga and the method of generating a sugges- set of regular expressions to describe thetion as discussed later. syntactical elements of Pascal. The output from Lex was a working lexical analyzerDescription of the major features of which was then modified to work on thethe environment Macintosh under Lightspeed C and our environment. This method of development proved to be useful due to the increased Immediate detection of syntactic and seman- tic errors: speed and versatility of Unix on a Microvax II with spare capacity, however, a unified development environment has some obvi- A user enters a series of Pascal source ous advantages. All development wouldstatements into the system to create a program. As each statement is entered, it have taken place on the Macintosh if Unix is parsed by the incremental compiler to had been available. determine if the statement is constructed in a valid form. This will detect any syn- The development of the Lex/Parse module is incremental, implementing one gram- tactic, and some semantic errors present in the statement. If any errors are detected, L14 Using Computers Intelligently 95 the user is informed and must correct thempersonal conceptual form. The last feature before being allowed to continue.Thisis the ability to save routines in a library for ensures that the code entered so far, is errorlater re-use. This feature should be brought free as far as the parser is concerned. If the into context though, in that Pascal routines system can anticipate the user's intent fromare very difficult to use in two different an erroneous statement, a suggestion isenvironments due to the strong typing in generated and presented to the user for the language but this does not make re-use acceptance or rejection. If the user deletesimpossible. any code from the program so far, the sys- tem will check this deletion for any errors Simulating execution: introduced intothe remaining code. Again, the user is notified and suggestionsAn extremely effective method for a user to are generated if possible. discover errors is to actually run their pro- gram.The problem with conventional Procedure /function testing and saving: compiler environments is that a run is ac- complished independently from the edit- Students taking an introductory course ining environment. Any errors that occur computer science are often taught to de-must be corrected by leaving the run time velop their sof tware in a top-down manner. environment and entering the editing envi- This involves the continual refinement ofronment. The cycle of edit - compile - run is modules and routines. The Pascal pro-repeated until a correct program is gramming language is often used to teachachieved. One advantage of a self con- introductory programming skills, how-tained development environment is the ever, Pascal requires that all routines beability to switch from one function to an- declared before their usage. This requires other immediately. Another advantage is bottom up implementation for a piece ofto be able to control the execution of the software that was usually conceived in acode segment and stop and examine vari- top down fashion. This anomaly is oftenable contents at the user's discretion. the source of confusion in many noviceMacTutor includes the ability for a user to programmers. They are required to worrydefine where they would like a simulated about all tl e! detail s before getfing the over- execution to suspend and wait for further all process correct. One alternative is to use commands through the use of breakpoints stubs, merely a place holder, for routines atand watchpoints. A breakpoint is associ- a lower level, returning to define them in ated with a source code statement. When detail later. The MacTutor environment the run time controller encounters this provides another alternative, this being the statement, execution is suspended until the ability to use a routine call without defining user chooses to continue. A watchpoint the body of the routine first. A list ofcauses execution to pause when a certain routines that need to be declared before avariable is modified. Another function is program is complete is kept. The user may the ability to trace certain variables test routines individually as well. A testthroughout the program execution. The shell is provided to specify the parameterscontents of the specified variables are dis- required for a routine and then simulate theplayed in a window at all times with the execution of the routine. This feature al-appropriate modifications being made as lows the user to test using a bottom up, top they occur. The user also has the ability to down approach or both, depending on thestep the run time controller through the user's preference. This provides a means to source code one statement at a time using allow the user to develop their programs in the single stepping function. These tools a form that directly corresponds to theirgive the user a very effective method to 105 96 Using Computers Intelligently trace the program's execution under vari- ronment, the help system can take more of ous conditions and inputs. Program de-a back seat and wait to be asked for assis- bugging becomes an enjoyable and infor- tance. This would provide a model closer mative task while also being efficient. to the environment existing between a Using these techniques, a user can discovernovice user and a human tutor. Of course many logic and semantic errors that wouldthe ultimate machine based environment otherwise have taken many runs witha would incorporate a complete history of substantial amount of 'guessing' as to the the user and use this to generate intelligent appropriate correction to make. prompting, correcting and encouragement by anticipating the errors and pitfalls of the Suggestion generation and help: users. The substantial overhead imposed by such an undertaking does not imply When a user makes an error, the systemthat steps towards this goal should not be attempts to determine the original intenttaken. A useful environment can be cre- and suggest a correction. The user is givenated using a modified history of each user the opportunity to accept or reject thisto generate assistance in the form of sug- suggestion. A user's statement is parsedgested corrections and examples using the and a partial parse tree is constructed. Thisuser's incorrect and correct statements. is analyzed to detect any errors. Ifan error(for a more detailed description of the is detected a nearest match is attempted application of an intelligent help system, with a correct partial parse tree segment see Waugh & Stark, 1987.) stored in the system. The closest match 'esembling a correct structure is used to try Editing capabilities: to reconstruct the user's partial parse tree segment. This is then traversed to generateThe editor allows a user to enter Pascal the suggestion. The details of this method programs which can consist of Pascal state- are fairly technical and are thus beyond the ments and comments in the form of English scope of this paper. It is not possible to text. The editor provides the useful fea- always generate a suggestion as the omis- tures of automatic indentation in block sions from the user's entry makes identifi- levels resulting in a pleasing visual presen- cation of the intent a guess at best. In this tation of the code. Other features include case the system will identify the error to the the ability to cut and paste, use the clip- user as per a typical compiler, requestingboard to make copies of code for duplica- the user to correct before continuing. tion purposes or moving text around and scrolling both up and down the page as Static help systems fail to take into account well as left and right as the page size is 120 the previous history of the particularusers. columns wide. Other useful functions in- Users must understand a substantial clude saving of files, resuming a previously amount of the problem area to ask for the saved session and undoing previous com- correct kind of help. A profile could be mands. maintained for each user that would record the usage of both the features of the envi- The user interface: ronment used and of the syntactic con- structs mastered in the language. UsingThe user interface is menu driven with this profile, help can be offered in an incre- independent windows, (see fig. 2), for code, mental way. To the complete novice anprogram output, variable tracing and obtrusive help system is active makingcommunications between the user and the suggestions at each fault of the user. As the system, (known as dialogue boxes on the user accomplishes tasks within the envi- Macintosh). The interface was designed to 106 Using Computers Intelligently 97 Mac Tutor Windows File Edit Run Trace Help Code Window Output Window Variable Window Dialogue Box 1 Figure 2 take advantage of as many features of the shell that is independent of any particular Macintosh toolbox as possible. This allowsprogramming language. Implementations for a clean, simple and efficient interface would include a language dependent All interactions involve some form of indi- module being 'plugged' into the shell re- cation using the mouse or typing using the sulting in a novice Programming Environ- keyboard or menu selection. The amountmentfor any language that a language of typing required by the user is thus mini-module exists. This is a rather ambitious mized. Many users enjoy the novelty and undertaking and is the direction of future ease of use of a Macintosh, especially when research. Other extensions are to include compared to the alternative systems of- dynamic data flow analysis. Basically, data fered by typical operating systems. flow analysis provides a trace of the use and abuse of variables, an example being Conclusions re-initialising an unreferenced variable, allowing the detection of anomalies in In this paper a truly useful Programming variable usage which can be corrected to Environment for novice programmingprovide a more efficient and less error students has been described. The methodprone program. Dynamic data flow analy- of implementation has also been presented sis is analysis performed while the pro- with the advantages and lessons learned gram is being created and compiled. This along the way. The major outcome of thelast feature can be accomplished with whole undertaking has been the idea to minimal modifications to the existing envi- generate a Programming Environment ronment. 107 38 Using Computers Intelligently Finally, the environment and its toolscan Donzeau-Gouge, V., Huet, G., Kahn, G. & be extended to the professional develop- Lang, B. (1984). Programming Environ- ment arena. This would require the optimi- ments Based on Structured Editors: The zation of every facet of the environmentas MENTOR Experience. In Barstow et al, professional software development is not (1984). concerned with teaching programming, Feycock, S. & Lazarus, P. (1976). Syntax- only the development of software.pro- directed Correction of Syntax Errors. Soft- gramming environments, and automated ware-Practice and Experience, Vol 6. software engineering tools in general,areGhezzi, C. & Mandrioli, D. (1979). Incre- gaining much enthusiasm as software con- mental Parsing. ACDA Transactions on tinues to become larger and more complex. Programming Languages and Systems, Vol. 1, No. 1. A dosing thought: Goldberg, A. (1983). The Influence of an Object-Oriented Language on thePro 'Least is most' gramming Environment. In Barstow et al, Mies van der Rohe. (1984). Graham, S. L. & Rhodes, S. P. (1975). Practi- (especially when applied to the amount of cal Syntactic Error Recovery. Communi- information a programmer needs to keep cations of the ACM, Vol. 18, No. 11. in their head and the amount of effort to James, E. B. & Partridge, D. P. (1973). Adap- achieve the end result.) tive Correction of Program Statements. Communications of the ACM Vol. 16, References No. 1. Kantorowitz, E. & Laor, H. (1986). Auto- Aho, A. V. & Peterson, T. G. (1972). A Mini- matic Generation of Useful Syntax Error mum Distance Error-Correcting Parser For Messages. Software-Practice and Experi- Context -Free Languages. SIAM Journal of ence, Vol. 16(7). Computing, Vol. 1, No. 4. Lesk, M. E. & Schmidt, E. (1975). Lex- A Apple Computer Company, (1985). Macin- Lexical Analyzer Generator. From Unix tosh Pascal. Programmer's Manual. Ardis, M. A., Gehling, J., Gill, .D., Lock, K. (1965). Structuring Programs For Glushko, R. Sr Vishniac, E. (1985). An Multiprogram Time-Sharing On-Line Ap- Assessment of Eight Programming Envi- plications. Proceedings AFIPS FJCC. ronments. Technical Report TR-85-10, Marlin, C. D. (1986). Language-Specific Edi- Wang Institute of Graduate Studies. tors for Block-Structure Programming Lan- Atkinson, L. V. & McGregor, J. J. (1978). guages . The Australian Computer Jour- CONA A Conversational Algol System. nal, Vol. 18, No. 2. Software-Practice and Experience, Vol. Moretti, G. S. & Lyons, P. J. (1986). An 8. Overview of GED, A Language-Independ- Atkinson, L. V., McGregor, J. J. & North, S. ent Syntax-Directed Editor. The Austra- D. (1981). Context sensitive editing asan lian Computer Journal, Vol. 18, No. 2. approach to incremental compilation. The Morris, J. M. & Schwartz, M. D. (1981). The Computer Journal, Vol. 24, No. 3. Design of a language- Directed Editor for Barstow, D. R., Shrobe, H. E., Sandewall, Block-Structured Languages. ACM (eds)(1984). Interactive Programming SIGPLAN Notices 16(6). Environments. U. S. A., McGraw-Hill. Moura, A. V. (1986). Early error detection in du Boulay, B. & O'Shea, T. (1980). Teaching syntax-driven parsers. IBM Journal of Novices Programming. DAI Research Research and Development, Vol. 30, Paper 132, University of Edinburgh. No. 6. 106 Using Computers Intelligently 99 Parnas, D. L. (1972). On the Criteria To Be Used in Decomposing Systems into Mod- ules. Communication.: of the ACM Vol. 15,12. Teitelbaum, T. & Reps, T. (1981).7 he Cornell Program Synthesizer: A Syntro-Directed Programming Environment. in Barstow et al, (1984). Teitelman, W. (1969). Toward a program- ming laboratory. International Joint Con- ference on Artificial Intelligence, Wash- ington, May. Teitelman, W. & Masinter, L (1981). The Inferlisp Programming Environment. In Barstow et al, (1984). Waugh, D. A. & Stark, K. P. (1987). Applying Intelligent CAL Principles to Help Subsys- tems for In teradive Software. Proceedings of the Fifth Annual Conference on Computer-aided Learning in Tertiary Education, University of Sydney, Syd- ney. Welsh, J., Rose, G. A. & Lloyd, M. (1986). An Adaptive Program Editor. The Australian Computer Journal, Vol 18, No. 2. David Waugh holds a BSc (Hons) Computer Science, 1985. The Univer- sity of Western Ontario, London, On- tario. MSc in Computer Science, James Cook University, Townsville. Current Position is a Lecturer, Department of Computer. cience, JamesCook Uni- versity. His research interests: Novice Programming Environments, Soft- ware Engineering Tools, User Model- ling, User Interfaces 1 G 9 Generative CAL and courseware abstraction Geoffrey I. Webb Computing and Informatiorl Studies Griffith University Courseware abstraction is an approach to CAL with the capacity to generate, solve and whereby the lesson author creates a general critique students' solutions to novel prob- parameterised CAL lesson that is then applied to many concrete examples. This approach has lems.Once one leaves the domain of the following advantages mathematics, relatively difficult AI based it provides a powerful framework within techniques are required to creatc genera- which to adapt tuition to a student's tive CAL. (See, for example, Sleeman and knowledge and aptitude; Brown, 1982). it encourages the development of detailed treatments of the subject matter; it reduces the cost of lesson development Generative CAL offers several as a ratio to student lesson time; and benefits it enables large numbers of examples to be made available for individual students. Generative CAL is an example of courseware By selecting problems with appropnate abstraction. It is argued that the advantages of attributes it is readily possible to adapt generative CAL do not arise directly from the generation of the examples to be examined but to variations in student's initial compe- rather can be directly attributed to the use of tence and learning rate within a do- courseware abstraction. main. Generative computer-aided learning is a Because one general treatment of a CAL technique that was developed in the domain is defined that is then applied to late sixties (Uttal, Pasich, Rogers & Hiero- many instances from that domain, the nymous, 1969; Uhr, 1%9). A generative lesson author is encouraged to make the CAL system generates unique problems treatment more detailed than if a sepa- and presents these to the student for solu- rate treatment has to be written for each tion. The system must be able to evaluate example to be examined as is the case and critique the student's solution to the with normal programmed learning. problem. This implies that the problems to be generated belong to a class of problems Because only the one general treatment for t 'hich the system has the capacity to has to be written, if the system is able to generate, evaluate and critique novel in- generate many examples from the stances. domain and consequently the one gen- eral treatment is presented to the stu- Most generative CAL systems have been in dent a large number of times then the the domain of -.ematics. This is not a relative cost in authoring time to stu- matter of coin _Ince. Mathematics is one dent terminal time is likely to be very of the few domains for which it is relatively low. This can make the approach very simple for the computer to be provided cost effective. Using Computers Intelligently 101 The student is provided with the oppor- They must be able to generate meaning- tunity to examine as many problems as ful problems. may prove desirable, all within the one consistent framework. They must be able to solve the problems generated. However, a careful examination of these benefits will show that none of them result They must be able to monitor, remedi- directly from the fact that the computer ate and assist the students' attempts to generates the problems to be examined. solve the problems generated. Rather, they result from the fact that one general treatment of a domain is applied to many specific instances from that domain.As mentioned above, although this is rea- In short, the benefits of generative CAL all sonably tractable for arithmetic domains, arise from the fact that it is a technique thatonce one leaves the field of mathematics employs an approach to creating CAL thatvery sophisticated artificial intelligence will be referred to as courseware abstraction. based systems are required. The develop- ment and operating overheads of such systems are prohibitive for most CAL ap- Courseware abstraction is a method of plications. creating computer-based courseware that is both time saving for the lesson author and encourages thorough treatment of the However, courseware abstraction does not subject matter. The author creates once require that the system should generate the only a detailed abstract CAL treatment of problems that the student is to examine. the subject matter. This treatment is par-Rather, the course author can create a body ameterised. That is, those aspcct of the of examples that the system can then use, subject matter that change from example to thus avoiding the problem of generating example are identified, and a variable is novel problems. This approach has been provided for each.This parameterised used with success by a number of systems. abstract treatment is then applied to a body It is a key factor in the power of the ECCLES of specific examples from the subject area. (Richards dr Webb, 1985) and DABIS Associated with each example are the cor- (Webb, 1986) systems. Most Al based CAL rect values for that example for each pa- systems u tili se it. Despite being a key fac tor rameter in the abstract treatment. The re-in many systems, this approach to CAL has sult is a detailed treatment of the particularnot previously been identified and the role example for the student. The one lesson canthat it plays has not previously been recog- then be presented to the student many nised. times, each time using a different example and providing a relevant analysis for thatAs the generative approach has been well example. documented, the rest of this paper will address non-generative approaches to courseware abstraction. To provide a con- Generative CAL is one approach to CAL text for this discussion the next section that employs courseware abstraction. The examines a typical example of an applica- vast majority of generative CAL systems tion for courseware abstractionCAL have been for mathematics.As statedtreatments of the French passe compose. above, this is not a matter of coincidence. The remaining sections provide a general discussion of the advantages and applica- Generative CAL systems must meet thebility of the approach and means of imple- following requirements. menting it in existing CAL languages. 102 Using Computers Intelligently A computer-based lesson on the If a main verb takes the auxiliary verb French passe compose avoir and there is a direct object in the phrase that precedes the main verb then the main verb must agree with it in both The passe compoce the past tense verb number and gender. If there is no direct agreement systemis an extremely diffi- object, or if the direct object follows the cult aspect of French grammar. The past verb then the main verb must be un- tense is marled by the presence in a phrase marked for both number and gender. of one of the auxiliary verbs etre and avoir. Each verb can normally only appear with one of these auxiliary verbs. This aspect ofTo further complicate matters, the auxil- French grammar has many irregularities. iary verbs etre and avoir take many differ- That is, for many French verbs it is notent forms depending upon their context, obvious which auxiliary verb the mainmaking it difficult for the student to iden- verb takes. This is simply a matter of ac-tify whether they are present. quired knowledge. From the complexity of the above descrip- The agreement rules of the passe composetion it should come as no surprise that can be summarised as follows students of French grammar typically ex- perience great difficulty in mastering this aspect of the grammar.It is dearly of If there is a reflexive pronoun in thebenefit to students to provide them with phrase as either the direct or indirectpractical experience in identifying whether subject of the main verb then particular French phrases do conform to these rules and, if not, how exactly they If the reflexive pronoun is the direct conflict.In short, this aspect of French object of the main verb then the verb and grammar is a prime target for CAL. direct object must agree in both number and gender. If the reflz)dve pronoun isThe only practical manner in which to the indirect object of a pronominal mainprovide this practice is to create a large verb then the verb must usually be series of exercises, El, E2,...En,each of which unmarked for both number and gender. examines a separate French phrase, Pi, such This means that it takes the masculine that E1 examines Pi. singular form. (The reflexive pronoun is treated here as the direct object of the verb in phrases such as Elle s'est lay& The standard test and branch approach to but not in phrases such as Elle s'est lave the authoring of such a lesson is to create a les mains.) list of example phrases to be examined and for each of these examples write a CAL segment that provides a detailed exercise Irrespective of which auxiliary verb theexamining that example. This will be called main verb normally takes, the phrase the traditional approach. This approach can must contain the auxiliary verb etre. be characterised as the creation of CAL segments si, s2, ... s, such that si generates If there is no reflexive pronoun in athe interactions with the student required phrase and the main verb takes theby Er As a result, the provision of n exer- auxiliary verb etre then it must alwayscises will require the creation of n CAL agree with the subject of the phrase insegments. both number and gender. Using Computers Intelligently 103 It wouid be necessary in such a lesson to If the student believes that the example is examine the following questions, among ungrammatical and it is not, each of the others, with relation to the examples reasons why it could be ungrammatical should be examined and the student Is the phrase grammatical? should be shown how they do not apply to What is its main verb? the example. Conversely-, if the student believes that the phrase is grammatical and What is its auxiliary verb? it is ungrammatical, then s/he should be Does the main verb agree with its sub- shown exactly how it is ungrammatical. ject? Figure1 shows a fragment of such an Does the main verb agree with its directexample exercise written in Common Pilot. object? This fragment tests if the student can iden- What is the main verb's number? tify the direct object in the phrase Jeanne je What is the main verb's gender? l'aivuehier. Common PILOT is described in Does the main verb have a direct or Khieraty & Gerhold (1980). indirect object? Does the direct object precede or follow R: Examine whether there is a direct the main verb? object in the phrase. T: Is there a direct object in Does the phrase contain a reflexive "Jeanne je l'ai vue hier"? pronoun? U: ANSYES TY: Yes, well done! However, not all of these questions would TN: Wrong, "L'" is the direct object apply to all of the examples. For example, in this phrase. it would only be important to determine JN: @P the relative location of the direct object and T: main verb of a phrase if the main verb took T: What is the direct object in avoir. "Jeanne je l'ai vue hier"? A: Each exercise would have to examine only M:%L'% those questions that are relevant to the TY: Very good! example on which it is based. Each relevant JY: AGREE question would be asked of the student. M:%JE% The student's answers would be evaluated and appropriate feedback provided. TY: No, "JE" is the subject of this phrase. TY: The direct object is "L'"! The structure that each exercise might take TN: No, the direct object in this would be to first enquire of the student phrase is "L" whether s/he believes the example to be *AGREE grammatical. If the student answers cor- rectly then it can be assumed that s/he can ANSYES is a subroutine that sets the Y conditioner correctly analyse the example and another if the student gives an affirmative response and sets example can be selected and examined. If the N conditioner for a negative response. this assumption is incorrect and s/he has only guessed the correct answer, this FIGURE 1: A fragment of a standard test would be picked up at a later stage when s/ and branch lesson on the French passé he again has to analyse a similar phrase. compose 113 104 Using Computers Infetligently Let us assume that we are going to provide or approximately 165 hours. All of this for one hundred exercises for the student. This1 hour and 40 minutes of student lesson is far fewer than would be desirable, but it time! This amount of time might be feasible will do to demonstrate the point. for a software house, but it certainly is not for a teacher wanting to provide some For each exercise there will be a complexremedial exercises for a few students! schedule of This would be a prime target for generative frames of text to be displayed to the CAL if only it were possible for the com- student; puter to generate novel examples and to questions to be asked of the student; determine the correct analyses for those examples. The creation of such a system answer evaluation procedures to applymay be currently feasible using state of the to the student's responses; and art artificial intelligence programming. O conditional branching dependent uponHowever, the developmental cost involved the student's responses. would almost certainly far exceed that of the traditional approach. There will be some similarities in this schedule from exercise to exercise, in that The alternative approach that is being pro- the text frames, questions and answerposed is to write once only some general- evaluation procedures will be drawn fromised courseware which can then be used to one pool, but the exact set used will vary examine any of a large number of concrete greatly from exercise to exercise. examples from the domain in question. This can be described as the creation of a If the traditional approach were used eachsingle CAL segment S(0, which takes as a of these exercises would have to be au- parameter a description of the item to be thored as separate CAL segments. Cer-examined, and generates the appropriate tainly, there would be some savings in exercise for the student, E,. In the case of the authoring timedue to being able to copyFrench passe compose this means writing a frames, answer evaluation procedures, etc lesson that can examine any combination of from one exercise to the nextbut eachverb type and agreement (or lack of agree- exercise would still have to be separatelyment) for any phrase. Such a lesson has all manually constructed by the author. the advantages of a generative system with the added bonus that it is feasible to imple- It is probable that each CAL segment ment! would take at least 100 minutes to author. Assuming that each exercise would occupy Figure 2 contains a portion of an example approximately 1 minute of the student'slesson which utilises this approach, again time, as each exercise is embodied in awritten in Common Pilot. The code in separate CAL segment, this results in an Figure 2 is a subroutine which examines authoring ratio of 100:1. This estimate iswhether the student can identify the direct probably quite conservative. A ratio of 200 object in a phrase. This lesson fragment is authoring hours to one student hour is notfunctionally identical to that in Figure 1. uncommon for this authoring methodol- That is, given that the fragment in Figure 2 ogy. is being applied to the phrase examined in Figure 1, both respond identically to the student. Despite the fact that the abstracted On these figures, for 100 exercises the au- code can be used in examining any phrase, thor could expect to spend 10,000 minutes, it is less than twice the size of the traditional 1 i 4 Computers Intelligently 105 code which can only be used to examine Clearly the abstracted courseware requires one particular phrase. The greater utility of some means of determining the details of the abstracted code is gained at very little the concrete example under consideration. cost. Otherwise it cannot determine what the correct answers are for any particular ex- *EXAMDO ample phrase. The lesson from which the R: fragment in Figure 2 is taken reads the R: Examine whether there is a direct phrase and various parameters about it in object in the phrase. from a file. Figure 3 shows the parameters T: Is there a direct object in "$A$ that are read in for exercises on the phrases Jeanne je l'ai vue hier and Nous nous sommes U: ANSYES arrates devant la bibliotheque. J(D$ "") : NODO *DO TY: Yes, well done! Phrase Jeanne je l'ai TN: Wrong, "$D$ " is the direct Nous nous sommes arrgtes object in this phrase. vue hier devant la bthliothique EN: NEXTPH T: Grammaticality GRAMMATICAL CROMMECAL Subject je mxm T: What is the direct object in "$A$ Direct Object mxm indirect Object la A: bthliotheque T(%B D$): Very good! Reflexive Pronoun now Verb vu arrites EC: Auxiliary Verb noir etre T(%B E$ & E$ <> ""): No, "$E$ " is Location of direct object PRECEDING REM% the indirect object. Number SINGULAR PLURAL TC: "$D$ " is the direct object! Gender FEMININE WC LI .1E EC: T(%B C$): No, "$C$ " is the subject FIGURE 3: Parameters for two examples of this phrase. from the abstracted passe compose TC: The direct object is "$D$ "! lesson. EC: T: No, the direct object in this phrase is "$D$ " Using this authoring method, if we assume E: that it will take the author approximately *NODO 40 hours to specify thr initial lesson and TN: Excellent! then approximately 1 minute to specify TY: Wrong. There is no direct object each example phrase and its parameters, for this phrase. then we have an authoring time of 2500 EY: NEXTPH minutes for the lesson with 100 examples. E: This is an authoring ratio of 25:1, a fourfold improvement over the traditional ap- ANSYE.S is a subroutine that sets the Y conditioner proach. It is not really important whether if the student gives an affirmative response and sets the (probably already excessive) figure of the N conditioner for a negative response. 40 hours for the specification of the ab- stracted lesson is not accepted. No matter C is the subject of the phrase. how exorbitantly this figure is inflated, the D is the direct object of the phrase. E is the indirect object of the phrase. ratio will still come out in favour of course- ware abstraction in the long run. This is FIGURE 2: A fragment of an abstracted because, having written the abstracted les- lesson on the French passé compose son and being in the position that each new 115 106 Using Computers Intelligently phrase only takes a minute to specify, theGREATERP system involves the applica- author is unlikely to stop at one hundredtion of a general lesson to a sequence of examples. The more examples that areconcrete examples. Thus, it is also an ab- specified, the better the authoring ratio will stracted lesson. However, it clearly does be. This contrasts sharply with the tradi-not use a programmed learning based tional approach where the authoring ratioapproach to CAL. Rather, it utilises ex- remains constant no matter how many trzraely sophisticated Al based CAL. examples are provided. The advantages of courseware abstrac- Several points should be noted about the tion lessons of which portions appear in Figures 1 and 2. First, as already meLtioned, for the phrases covered by the traditional lesson, An improved ratio of authoring to student both lessons are functionally identical. The time is just one of the advantages of course- traditional lesson only covers three ware abstraction. phrases. However, with just these three phrases the author has already written lessLesson validation is far easier with ab- code (albeit more complex) for the ab-stracted courseware. With the traditional stracted lesson than for the traditional one.approach, the author will never be able to Further, the smaller abstracted lesson canthoroughly determine that no errors lie already handle far more than just the three hidden in some section of code in some examples that the traditional lessonexercise or other. With courseware abstrac- handles. Indeed, the abstracted lesson can tion, any error in the abstracted courseware handle any grammatical phrase with a re- will quickly come to light, as the same code flexive pronoun as the direct object or withis used for every exercise. Once the gener- a main verb that takes avoir. All that is now alised lesson has been debugged, the entire needed to make the abstracted lessoncourse can easily be validated simply by complete are short treatments of phraseschecking that the parameters passed into with reflexive pronouns as indirect objectsthe system are correct. If the parameters and of verbs that take etre and a treatment have been given sensible names then this is of ungrammatical sentences. The latterstraight forward. The author need only would be approximately the same sizescan through a list like that in Figure3 in again as the existing lesson code. By com- which any errors will be transparently parison, the task of the author using theobvious. traditional methodology has barely begun with the three phrase example which has Another advantage of the methodology is been written. that it encourages the author to provide an exhaustive treatment of the topic. With the The final point that should be made abouttraditional approach the author knows that the example lessons is that it is not impor-for every aspect that is built into the lesson tant whether the reader believes that theythere is going to have to be separate code provide a good or a bad treatment of the written for each example to which it ap- subject matter. They are provided only to plies. This can result in the author having to serve as a concrete example of the differ- choose between providing a thorough ence between abstracted and non-ab- treatment of a smaller than desirable set of stracted courseware. Courseware abstrac- examples, or a less thorough than desirable tion is not tied to this programmed learning treatment of a larger body of examples. based style of lesson.For example, theThis is not true with abstracted courseware 116 Using Computers Intelligently 107 authoring because the author only has toBut ...ifficulty level is not the only or the provide the one thorough treatment of themost powerful distinguishing feature that domain. This is then applied to all ex-could be taken into account. Consider the amples examined thus ensuring that everyabstracted passe compose lesson.If the example receives the same level ocanalysis. student is not making errors for examples that take the auxiliary verb avoir but is for Abstracted courseware is also easier to examples that take the auxiliary verb etre, update. If the author decides that some then it will be quite trivial for the system to aspect of the original treatment of the identify this fact and to concentrate on the domain under examination should beexamples for which the student needs fur- changed, such a change need only occur inther tuitionthose which take the auxil- one place in the generalised lesson. Under iary verb etre. All of the information that it the traditional approach, any such changeneeds in order to determine thisthat is, would need to be made separately for each the auxiliary verb for each exampleis exercise to which it applied. This would already explicitly stated and thus readily cause two problemsa) determining accessible. which exercises the change applies to; and b)changing the lesson code in each exer- A final advantage of abstracted courseware cise. is that it greatly facilitates the analysis of a student's performance during the lesson. If Another advantage of courseware abstrac-a simple record is maintained of the ques- tion is that it provides a framework in tions at which the student makes mistakes, which it is readily possible to adapt instruc- then it will be readily possible to determine tion to each student's current level of exper- what difficulties the student is experienc- tise in a domain. If the CAL system is able ing. For instance, if the student often speci- to identify relevant features of the ex- fies the wrong word as being the main verb amples that it is examining, such as theirin the phrase then it is clear that they are relative difficulty, and is able to judge from having difficulty identifying the main verb. the student's performance how well theyThe standard test and branch approach perform in relation to these features, then it makes this information far more difficult to is relatively straight forward to select ex- obtain as there is no clearly defined struc- amples for examination that are appropri-ture by which to assign identity to similar ate for the student's current needs. Forquestions in different exercises. At best, instance, if the examples are graded with this becomes another chore for the author regard to difficulty, and the system can to look after. At worst, this information is determine that the student makes no errors simply not collected. when examining examples at difficulty level one, makes errors for 50% of the ex- The applicability of courseware amples at difficulty level two and 90% of the examples at difficulty level three then itabstraction will be able to determine that it should concentrate on examples at difficulty level Courseware abstraction is clearly not a two. All that it need do is start at the easiestmethodology that can advantageously be difficulty.level and work its way down toapplied to all domains for which a com- the more difficult levels, remaining at eachputer-based lesson may be written. Rather, difficulty level until the student achieves a courseware abstraction only comes into its set success rate. own when large numbers of concrete ex- amples are to be examined in terms of a general theoretical framework. L17 108 Using Computers Intelligently I An obvious area in which this applies is ent values for different examples; and for it grammar. Typically languages have very to allow conditional branching dependent complex grammatical systems for which itupon the value of those variables. is advantageous for the student to examine large numbers of specific examples withinProbably the simplest way to implement a general framework. The passé compose this in most existing CAL languages would lesson is an example of just one such lesson. be the approach taken in the lesson frag- ment in Figure 2. That is, to define a set of However, languages are far from the onlyvariables which are then read in from a file. domains to which such lessons may beFor each example that is presented a new applied. A lesson has been developedset of values is read into the variables from under the DABIS system (Webb, 1986) that the file. Once a lesson has been written, a examines the leading theories in the mind/ text editor can then be used to create and body debate within philosophy. The lesson update the input file. identifies a number of central questions within this debate and examines how each However, for courseware abstraction to be key position in the debate has a different set possible, it is not necessary for a CAL lan- of answer3 to these questions. guage to support files.If the language supports variables then it is possible sim- Another possible use for the methodology ply to assign values to them within the is in the control of practical classes where program. Thus, each time that an exercise the students must follow a strict order ofbegins a new set of values are assigned to enquiry to reach some conclusion.For the variables. instance, the students could be given a chemical solution and then the program Although courseware abstraction is pos- could guide them through the step by step sible using many existing CAL languages, process of identifying its chemical compo- specialised systems, such as ECCLES sition. The same lesson could inanage the(Richards & Webb, 1985) and DABIS examination of any of a large number of chemical solutions. (Webb, 1986), that take advantage of the special features of the methodology in- crease its benefits to the author. Courseware abstraction is a methodology of wide but not unlimited application. Conclusion Courseware abstraction in existing Courseware abstraction provides a meth- CAL systems odology for the authoring of computer- based courseware which can create de- As is demonstrated with the Common Pilottailed lessons of a type not feasible through lesson from which the fragment in Figure 2 other approaches due to their huge cost in is taken, courseware abstraction is possible authoring time and difficulties of mainte- using at least some existing CAL lan- nance. guages. The benefits that courseware abstraction All that is required for a CAL authoringhave to offer have already been demon- system to allow courseware abstraction isstrated by the success of generative CAL. for it to support some method of specifying Among the attractive features of course- variables and then assigning them differ-ware abstraction are reduced authoring 118 Using Computers Intelligently 109 time; the provision of a convenient frame- In Proceedings of the 1985 Western Educa- work for adapting tuition to the student's tional Computing Conference, Oakland, level of expertise; improved lesson verifi- C.A., pp. 151-157. cation; improved lesson modifiability; andSleeman, D. & Brown, J. S. (Eds). (1982). improved student analysis. Intelligent Tutoring Systems.London: Academic Press. Although courseware abstraction has beenUhr, L.(1969).Teaching machine pro- used in many previous CAL systems, there grams that genera te problems as a func- does not appear to have been any apprecia- tion of interaction with students.In tion of the fact that it was courseware ab- Proceedings of the 24th National ACM straction to which many of the desirable Conference, pp. 125-434. features of these systems could be attrib- Uttal, W. R., Pasich, T., Rogers, M. & Hiero- uted. This is particularly notable in the case nymus, R. (1969). Generative Computer of generative CAL where the benefits of the Assisted Instruction.Communication approach have been attributed to the gen- 243, Mental Health Research Institute, eration of the problems to be examined University of Michigan. whereas they can actually be attributed to Webb, G. I. (1986). Knowledge Representa- the use of courseware abstraction. tion in Computer-Aided Learning: The Theory and Practice of Knowledge-Based It is to be hoped that a better understanding I Student Evaluation and Flow of Control. of factors underlying the success of these PhD thesis, La Trobe University School approaches will in turn lead to the develop- of Mathematical and Information Sci- ment of yet better approaches. ences. Acknowledgements The research presented in this paper grew out of the ECCLES project which was founded and led by Tom Richards. Thus, the ideas presented herein have directly evolved from Tom's original inspiration for ECCLES. I am also deeply indebted to Bob Hooke and Roly Sussex who advised me on the finer points of French grammar during the preparation of this paper. The lesson fragments in Figures 1 and 2 are based on a lesson written for the ECCLES system by Bob Hooke. References Khieriaty, L. & Gerhold, G. (1980). COM- MON PILOT LanguageReference Manual. Bellington Wf : Western Washington University. Richards, T. J. & Webb, G. I. (1985). ECCLES: an "expert system" for CAL. 119 Using Computers Intelligently 111 2 Text-dominated instructional material Such expectations were never realised and usually defines and prescribes the it is unlikely that they will be. Videotex is a learning experience and seldom draws simple technology with a pleasing mixture on the experience or potential of theof colour, graphics and text. Its menu ap- learner. proach, numerical branching technigue, 3 There is an assumption that self-moti-and split rate access (sending at75bps and receiving at 1200 bps) make it slow and vation of the remote learner will help cumbersome when compared with com- him or her overcome the lack of learn-mand-driven systems. It is therefore not ing resources. effective or efficient for large databases, 4 There is not enough understanding ormail and sophisticated information re- appreciation that a remote learner maytrieval; but is a promotional medium more not have participated in a course for the suitable for business or commercial appli- sake of merely getting a qualification. cations. A videotex-based AESNET Most institutions and home-learners who rushed to subscribe to Viatel soon discov- King and Fosteis chapter is in a book on ered its educational limitation and as a study centres (Castro, Livingston and result, any it I euphoria and enthusiasm Northcott1985)published in August1985, towards a national network based on Viatel the year that saw the launching ofhave vanished. There are also different Australia's videotex system (Viatel), firstenvironmental factors in Australia and the satellite (Aussa 0, two public electronic United Kingdom to account for the failure messaging systems (Minerva and Tele-of educational videotex in the former and memo, now merged to form Key link), and the relative success in the latter (Castro the year that the International Council of 1986). Soone could say that the notion of a Distance Education held its Thirteenthvideotex-based AESNET was virtually World Conference at Melbourne. The Hon. shelved earlier this year when the Govern- Susan Ryan, then Minister of Education, in ment finally withdrew its limited support her opening address to the 600 delegatesin early1987to the Australian Caption from around the world, announced theCentre, a nen profit-making organisation Government's "very important" commit-appointed to develop Edutel. Edutel has ment to Edutel, an educational network on since been sold to a commercial videotex Viatel. The vague expectations of the Gov- bureau. Recent reports, however, ini 'cate ernment and the external studies fraternity that the notion is back in vogue with the were that Edutel would develop into a Commonwealth Tertiary Education Com- national network like Prestel Education in mission and some selected Technical and the United Kingdom, and that it wouldFurther Education sites will be asked to have a comprehensive national course di- develop and manage such a network for rectory which could be continuously up-use by all external students and institu- dated by all participating institut. ons. In tions. other words, there would be an Australian external studies network (abbreviated in It would be useful, therefore, to pause at this paper as AESNET) built on a publicly this point and ask what can such a network available and simple to use videotex sys- hope to achieve. Is it to replicate the current tem. Study centre personnel, by signing onconventional support services which, in to Viatel, would be able to show students the view of King and Foster, are more di- the courses on offer, and perhaps even offer rected to meeting just the needs of the insti- online counselling. tution? Or should it try to provide 0 raring and more efficient environment which 121 112 Using Computers Intelligently some would regard as beneficial for indi- cally competent, can take on the role of vidual learning? a mentor and a catalyst, and who are personally committed about their These are difficult questions which do not facilitative and gatekeeping responsi- seem to have been widely or seriously bilities towards the student users of contemplated or addressed by people who such a network. have expert knowledge about the strengths and weaknesses of educational technolo-3 That approaches be made to established gies and, more importantly, an apprecia- and operational national, state and tion of the crucial issues of the need for commercial networks (e.g. ACSNET, technology management and user educa- VICNET, CS!RONET and KEYLINK) tion. While I certainly do not pretend to to request for interconnection to institu- have all the answers and can only speak tionally-based networks to form a na- from limited, personal experience as an tional AESNET at some stage in the instructional designer, an on-line network future. manager, and a researcher of information4 That an institution start training both technology, I do not subscribe to the view: staff and students to help them become effective users of the network. That an AESNE1 has to be based en- tirely on computer technology or even All four tasks have to proceed concurrently on a particular kind of system; and but criss-crossing in a timeline that may That a nationally directed and managedstretch to three, four or even five years. network will attract a workable user base from within and outside the terti-For the first task, I shall now quote some ary institutions. examples of using computer and telecom- munications technologies at Deakin Uni- The philosophy of a home-based versi ty. The purpose is not to publicise how network for tertiary distance and open advanced Deakin University is but to draw education attention to then fact that they are the sort of resources or embryonic services already A different philosophy, and in my view a existing in many tertiary ir.stitutions and more realistic approach, would have been which in time will be used by an increasing to visualize the student support system as number of staff and students at a rate beginning at home - the home of the stu-commensurate with how they are intro- dent and the home of the institution. Ifduced or promoted. institutional decision is that a technology- based support system should complementThe following is a summary of the comput- the current conventional one, then I woulding resources at Deakin; machines and propose a serious consideration of the fol-connections which are irrelevant to the lowing four tasks: current discussion have been omitted for the sake of clarity. 1 That there should be a thorough review of current computer resources within 1 Machines: each institution to see if they could be integrated to provide a reasonable elec- Charlie (Gould PN6080, UNIX central tronic infra-structure of student sup- machine for academics and students) Alice (Data General Eclipse S/160 for port. Library) 2 That this structure be managed by a Microcomputers (app. 300 units in mi- small team of people who are techni- crolabs and staff offices, mainly IBM- 1 2 2 Using Comuters Intelligently 113 compatibte3 for Schools of Sciences. 3 Read and contribute to Deakin news. Management, Architecture and Li- brary; and Macintoshes for Schools of 4 Read and contribute to Australian Education, Social Sciences, and Hu- news (free of charge) and overseas manities) ones (at reasonable rate) via ACSNET. Terminals (app. 300 units) 5 Access thelibrary catalogue and 2 Connections search for books, then phone to ar- range for a loan, or send electronic The bigger machines and terminals are mail to a librarian to ask for books to linked by cables to Micom and Ethernet be posted to me to save time and a trip ports. to the Library. Microcomputers are nearly always used as stand-alones and currently few 6 Send a request to a reference librarian of them have modems or communica- to ask him/her to do a database search tions software or are in a local area net- for me and forward the results or work. Only the Library microlab have abstracts of relevant titles to my ad- Appletalk (for Macintoshes) and dress on Charlie, which I theriprocess EasyLAN (for Olivettis, IBM-compat- and refile into a personal database. ibles) for linking to printers. Staff and students can dial in from their 7 Run statistical programs or CAL pro- offices or homes Charlie using STD and grams without having to buy them Austpac but permission must be previ- pirate them at the risk of breaking ously granted and payment guaran- copyright laws. teed. The Austpac PAD has eight lines, one for dialling out, the rest mainly for 8 Have a word-processed file printed dialling in by external students on the on one of the expensive laser printers Graduate Diploma of Computing attached to the big machines. course. 9 Build a bibliographical database with Staff members and students must register search software and invite some cot- on Charlie first before being give a disk or leagnies or students to use it or to add directory quota and permission to use the to it. centrally installed programs and access to certain database, mail, news and printing 10 Use the Austpac line to dial out to facilities. Once that is done, I can do any of external systems on whicn I am a the following things by using the terminal permitted user (e.g. Keylink, in my office or the microcomputer in my etc.) office or at home with a modem, a commu- nications program and a telephone. 11 Ask my students what they think of a course through personalised mail or a 1 Exchange mail with my colleagues questionnaire. and students. 12 Conduct computer conferences to simulate tutorials, either in real time Exchange mail with people in Austra- 2 or in a store-and-forward messaging lian institutions (free of charge) mode. with people in overseas computer networks (at reasonable rate) via 13 Get students to produce an on-line ACSNET, of which Deakin is a mem- journal to exchange their writings and ber. thoughts on their course. 123 114 Using Computers Intelligently All theme things have not yet been per- The role of an external studies ceived in Deakin as a kind of package serv- netwo, k ices which we can offer to our external students for every course or program, but These big nd small networks all share the one could see the potentials for home ac- following jectives and features: cess by a remote student if arrangement can be r,...1de to allow him or her to dial in using 1 Institutionally-based and home-acces- a modem linked to a microcomputer. Now sible. if one compares these services with the kind which has been suggested should or2 A more direct communication channel could be mounted on a videotex AESNET, between the student and the teacher one could immediately see the advantages which is not constrained by time, dis- tance and place. of having a home-grown and home-based network which is capable of being custom- 3 Promoting peer group interation to ised to suit the personal needs of the indi- remove the feeling of isolation. vidual student who does not want to travel to a local study centre (usually after work,4 Making available resources and oppor- at night or during weekends) which can be tunities which would have been beyond 6 or 60 kilometres away. Direct and almost the reach of the home-learner. instantaneous communications can take place between an instituti, a teacher and It is a little early to know it ,ney will help a student without the constraints of time, remove all or most of the disjunctions as place and distance. observed by King and Foster, IAA they are commendable first steps to take to cultivate a caring environment. And if members are The Deakin scenario reflects the potential properly led and encouraged, they will offerings of a well-endowed institution, find imaginative and innovative ways of which may even be able to bear the tele- using even the most modestly-resourced communications costs of the students. Fornetworks, as it has been found in some institutions which prefer a low-cost front-research projects (Castro 1987). end setup, more modest designs may be used. For example, Murdoch University Mere is now a small but increasingly ar- has a small bulletin-board type networkticulated view that computer-mediated built on an IBM-compatible microcom-communication systems should be puter (Atkinson, 1986, 1987). Brisbane Col- adopted in external studies. Bates (1986, lege of Advanced Education, on the other p42) explains why he sees it as more valu- hand, favours a different approach alto-able than the promised wonders of com- puter-assisted learning: gether by registering their students and staff on a network built on Keylink. North The structuring of the teaching is not con- American and European distance-teaching tained in or restricted by the architecture of institutions have gt, ne a slightly different the computer, but developed and negoti- route again by purchasing off-the-shelf ated between learners and teachers. computer conferencing systems because they are considered to be easier to handleTo this, it could be added that learning is by staff and students who may not be too enriched and multiplied by the sharing of computer-sophisticated. (Kaye 1985, 1987; wisdom and experience among the stu- Harasim 1986) dents themselves which is far more impor- 2 Using Computers Intelligently 115 tant than making the institutional grade The SUN software (now being upgraded to (Castro 1987).The facilitating of good include X.400 capability) is distribu ted fora teaching and learning must remain the nominal charge to academic and research single most important role of any external institutions. There is no central administra- studies network. tion, each host is usually managed by a small team of experts who also educate or The "messages" of ACSNET help the users if difficulties are encoun- tered. The original developers and Robert Elz from Melbourne University act as na- I feel that it would be discourteous of me, tional co-ordinators. There is no govern- delivering a paper with a title like AESNET, ment funding; each host pays for its own not to pay a tribute to the great network mentioned in passing earlier which started link. here in the University of Sydney. The addressing format of ACSNET is simple to remember and use: user - ACSNET (Australian Computer Science name ©host- name.oz. Oz stands for Network), cited by Quarterman and ACSNET, which could be regarded as a Hoskins (1986) in their classic paper as a "cooperative network", is based on thesubdomain of AU, the domain name for Sydney UNIX Network (SUN) softwareAustralia. Messages are passed along on ACSNET, from one host to another, possi- primarily developed by Piers Dick- Lauder bly utilising intermediate hosts, in a store- and Robert Kummerfield of the Basser and-forward manner. Messages sent Department of Computing Science. The within Australia is free, overseas messages first two machines connected were from are charged at the rate of between two and the Universities of Sydney and New South Wales in 1979. ACSNET currently spans three cents per 64 bytes. the continent with approximately 350 hosts at universities, institutes of technology,The history and the operational mode of CSIRO divisions, Telecom Research labo- the ACSNET have some important lessons ratories, and a number of private and de- for our discussion of AESNET. It shows velopmental organisations; and operates that a network stands a much better chance over fixed lines, leased lines, Ethernet,of success it it starts small, is designed and CSIRONET and Austpac. managed by people who are competent and committed about a set of clearly de- fined and workable objectives, and who The purpose of ACSNET is to support mail take the effort to persuade others to join traffic, file transfer, print jobs, or anything them to create a critical mass of users. The that can be transferred in a string of bytes. ACSNET's decentralised model also True ACSNET hosts are UNIX machines, means that there is no complex, overbear- but there are gateway programs to make it ing top management to stifle local initia- also possible to communicate with nontives and experimentation, and that the UNIX systems. There are several UUCP naive and new user will have the additional (UNIX to UNIX COPY) gateways to North advantage of having to learn only the host America and Europe irom Melbourne. The system and be able to correspond with underlying transport protocols are also others on remote systems without hassle, used to support the USENET news trans-thus reducing the learning curves and mitted from America to Australia via the operating costs but gaining in self-confi- Midas X.25 service each evening and paid dence and efficiency. for by the Division of Information Technol- ogy at the CSIRO (about $22,000 a year). 116 Using Computers Intelligently References tional Communications, 16(2). Kaye,T. (1985) Computer-mediated communi- Atkinson,R.J. (1986) Computer bulletin cations systems for distance education. boards for distance education students. Report of a study visit to North Amer- Proceedings Ed Tech 86, International ica, September/October 1985. Project Educational Technology Conference Report, CCET/2. Milton Keynes, Insti- and Exhibition, Perth, Western Austra- tute of educational Technology, The lia, December 1986. Open University. Atkinson,R.J. (1987) Development priori- Quarterman,J.S. & Hoskins,J.C. (1986) ties for computer communications in Naotable computer networks. Commu- distance education. Paper for ASPESA nications of the ACM, 29(10), 932-971. 8th Biennial Forum, University of New England, Armidale, NSW, 27-31 July 1987. Bates,T. (1986) Computer assisted learning or communications: which way for in- formation technology in distance education? Journal of Distance Education, 1(1), 41-58. Castro,A.S.(1986) Videotex in tertiary edu- cation: the missing links. A keynote address delivered on 1 July 1986 at the National Forum during the 2nd Asian- Pacific Videotex Conference and Exhi- bieon, Sydney. Castro,A.S. (1987) Teaching and learning as communication: the potentials and current applications of computer-me- diated communication systems for higher distance education. Paper pre- pared for the UNESCO Higher Dis- tance Education Meeting, 6-11 Septem- ber 1987, at Deakin University, Gee- long, Victoria 3217, Australia. King,B. & Foster,A. (1985) A South Austra- lian perspective on student support. In A.S.Castro, K.T.Livingston & P.H.Northcott (eds.) An Australian case- book of study centres in distance education. Victoria, Distance Education Unit, Deakin University, p99-112. Harasim,L. (1986) Computer learning net- works: educational applications for computer conferencing. Journal of Dis- tance education, 1 (1), 59-70. Kaye,T. (1987) Introducing computer- mediated communication into a dis- tance education system. To be pub- lished in the Canadian Journal of Educa- An automated library and borrowing system (ALABS): Implications for CAL R.J. Clarke and L. Athanasiadis Microcomputer Laboratories University of Wollongong ALABS was designed as an operational control The second major set of problems ad- mechanism for the loan of floppy-based soft- dressed in this paper is that of appropriate ware, hardware, and audio-visual materials to mass distribution (delivery) and manage- students and staff, and the reservation of micro- computer resources in a spatially distributed ment of CAL and other resources after they microcomputer centre. This paper describes the have been acquired or created. Potential experience of the Microcomputer Laboratories problems include the structuring of labora- Unive.sity of Wollongong, which consists of 5 laboratories of 20 microcomputers each, plus tory time, managing loads through labora- another 20 scattered over the campus. ALABS is tories, determining if items on loan are designed to provide software utilization data overdue or held by staff, finding where a by student (for example CAL courseware), particular person is in the centre, managing laboratory loading statistics, activity rates, as well as maintaining an accurate inventory of machine bookings, and the allocation of audio-visual, software and hardware re- resources to these activities. In order to sources. ALABS is suitable for computing facili- solve them in the general case the Micro- ties where the diversity of machines and appli- computer Laboratories restructured its cations means that a single network structure is neither desirable nor feasible. The impact of methods of material's distribution and ALABS on the provision and management of management, producing a computer as- large scale CAL facilities is discussed. sisted solution called ALABS (Automated Library And Borrowing System). Two sets of problems beset individuals and organizations attempting to provide CALFacility Description and Physical facilities in the tertiary education-envi- Layout ronment. The first set of problems is the determination of the type and variety ofThe Microcomputer Laboratories provides CAL products to be supported (software,teaching support for all faculties at the hardware, audio-visual aids and printed University of Wollongong, including materials) by academic and computer sup-Commerce, Education, Science, Engineer- port staff. Attendant problems includeing, Humanities and Arts. Subjects sup- accessing the need for new software, deter-ported range from machine languagt pro- mining the sources and alternatives avail-gramming to creative arts. The facility able for software acquisition, making theundertakes consultancies (internal and choice between the internal developmentexternal to the University), provides com- or purchase of software, installation andputing expertise for academic and research implementation issues, evaluation of theactivities and runs courses for university software, the rating of candidate systemsstaff and the general community. and the ensuing costs and contractual obli- gations of software acquisition and mainte-The Microcomputer Laboratories consists nance (Borovits 1984, 66-9). of 5 teaching laboratories with 20 micro- 127 r 118 Using Computers Intelligently computers each, an Operations Room, a mercially available CAL, computing facili- Data Preparation Room, Workshop and a ties may opt for internal production. The Systems Resources Room. Three of thetools required to internally produce CAL laboratories contain Sperry PC Model 25s, products (bulk software, high grade docu- with 512-640K RAM, dual 5.25" floppy diskmentation etc.) are also expensive. In the drives and Colour Graphics Adaptors. tertiary environment demand for CAL re- Two of the Sperry laboratories are config- sources always outstrips the available ured using 2 Sperry USERNET 286 LANs funding. each with a Sperry PC IT as a file server with 30 Mbytes and 44 Mbytes of secon-Pressures exist to maximize usage of the dary storage apiece. The remaining twoCAL resource yet minimize damage and laboratories consist of Apple 2e machinesloss. These requirements are in conflict. with 128 Kbytes RAM, dual 5.25" floppy disk drives. One of these laboratories hasOrganizational history and activity colour displays and the other mono-profile chrome. Computing facilities in the tertiary envi- The laboratories are spatially distributed ronment evolve and can change function (separated by large distances). There are no and scale in unexpected ways. For instance, direct lines of sight into any of the laborato- the facility from which the Microcomputer ries from the Operations Room. Six staff Laboratories evolved served as the admin- members are employed, only a maximum of three are available to periodically check istrative data processing facility for the the laboratories. then College of Advanced Education. Fa- cilities consisted of a Perkin-Elmer 7/32 minicomputer and nine Apple II+ units CAL Resource Management in a Tertiary Environment (used for teacher training and to offload word processing users from the minicom- There are three vexing and interrelated puter). Upon federation of the College and issues in CAL Resource Management the University, the facility possessed 60 within a tertiary education environment. Apple II+ and IIe microcomputers and had divested itself of the 7/32 minicomputer. They are the nature of the CAL resource An emphasis on computer awareness re- itself, the organizational history and activ- ity profile of the facility, and the lack ofplaced the administrative role, and the internal support function was replaced by general support tools (network and stand- student training. alone) for CAL resource management. Equipment funding is often sporadic and CAL resources have the following linked to factors outside the facilities' con- attributes trol (for example, resource statements asso- ciated with new course proposals). For- CAL resources are often multi-media in ward planning becomes difficult. nature. Video disks and tape, kits including workbooks, slides, and manuals as well asIn the case of microcomputer systems the the more conventional software based tu- market is so volatile that both hardware torials, drills and simulation can constitute and software may become obsolete (this is CAL resources. a consideration in Information Systems area). The systems may not be supported CAL resources are expensive. In an attempt by the vendor, or in the worst case, the to reduce the long term expense of com- vendor may have ceased trading. l. 2 8 Using Computers Intelligently 119 The diversity of activities and course re-instance, the Operations literature (Fried quirements conflict with CAL Resource 1979, 222-260), including Information Management priorities. For example, Edu- Centres (Perry 1987, 127-147), Service cation users of our facility would like to Centre Management in general, and the PC train teachers on as many microcomputer Resource Centres in particular (Jackson systems (those on Government contract) as 1985, 151-154). One package recently re- are likely to be encountered in schools. viewed (Syk,1987), is called Micro Re- Business Systems users would like to keep source Management (Atrium Information up with the corporate state-of-the-art. Group). Again, business resource manage- Users in general want to increase system ment priorities are not those of an educa- diversity. Because centre management are tional CAL provider. This system places charged with the support and maintenance emphasis on keeping track of machine lo- of existing CAL resources, if not the pro- cation, current configuration, and financial duction of them, time and effort is reducedmanagement of microcomputers within a if systems are as homogeneous as possible. business (Sykes, 1987). Lack of CAL resource management tools Whereas the business microcomputer sec- tor has arguably been of greatest benefit in Networking existing machines and relying the production of relatively inexpensive on the network management system IS and fully functioned software applicable as NOT a solution to CAL resource manage- educational tools (Sperry 1985a, 1985b, ment. 1985c), the means of managing and distrib- uting these resources in a tertiary educa- Network hardware and softwa. e are verytion environment are not available off the expensive, especially with the number ofshelf. This necessitates the creation of in- workstations required in the tertiary envi-house tools tailored for this purpose. With ronment. Coupled with the diversity ofdramatic increases of student numbers and hardware generally found in tertiary com- transactions (see Table 1), new clients re- puting facilities, this means that it is un- sulting from the federation of the Institute likely that all machines will be networked. and the University, and the increasing timetabling demands of existing depart- Network management tools assist withments, it was necessary to provid'a new network software and global resourcessolutions to cope with the increased only. The network tools required in a terti-throughput. This necessitated the restruc- ary environment are different from busi-turing of laboratory time, borrowing privi- ness priorities. leges, and development of a computer system (ALABS). Business networking systems are created fora small and static number of users. Restructuring of laboratory time Network management tools reflect this. For example, USERNET 286 network con- The restructuring of the timetable con- tains a function to allow users to be addedsisted of dividing time into three catego- to the system, but only one at a time! ries. Free time or 'slack', is time not directly utilised for university courses. It is used for There are apparently no commerciallyspreading abnormal student loadings, available stand-alone microcomputerunscheduled classes, external short based systems suitable for CAL manage-courses, and hardware maintenance. Free ment tasks. The closest commercial a:la- time is not allocated to a given subject. logues follow methods proposed in, for 12 9 120 Using Computers Intelligently The regular usage of laboratory time by centre (software, hardware, and audio-vis- students is divided into two forms. Super- ual). This barcode format has a line of text vised laboratory time occurs when a tutor under the code itself. The coding used or lecturer has face-to-face contact withenables the item to be identified directly students in a class room role. The second from this text. The three types of barcodes form of borrowing is by a student for indi-are delineated from each other by the value vidual study purposes. This is called unsu-of the first character. A Sperry PC and an pervised laboratory time. Time slots areEpson RX-80 printer are used with the assigned to each class generally in a ratio of Barcodes on Demand package (Barcode one hour unsupervised to every two hoursTechnologies) to print item barcodes onto supervised. Unsupervised lab times areadhesive address labels. reserved for a given course. If over the period of several weeks these machines Software Barcodes are affixed to the disk- remain idle, students not in the course are ette and covered with contact to prevent allowed to use machines. If this is the situ- removal and deterioration. Surprisingly, ation then lab staff allocates the time as the use of a contact wand does not appear to equally as possible. Special notice is given lead to mechanical damage of the disk to the distribution of unsupervised eve- media. A back-up barcode is kept in a soft- ning access times (5:30pm - 9:30pm). Care is ware barcode fol der. The folder is useful for taken to ensure that evening students have quick entering of a class loan. Audio-visual equal access to the facilities. items (books, journals, videotapes and slides) and hardware items have barcodes Allocation of the time resource in a spa-attached to them in an appropriate format. tially distributed microcomputer centre is analogous to other timetabling and opti- Machine resources are barcoded as well. In mum packing problems (Ending an opti-order for a machine to be used, a student mum allocation for limited resources). Notmust present their student card to the only can there be a feasible solution, but a Operations Room and receive a lab pass best arrangement. A collected slackalong with any software they may need. method, where Free Time is gathered to-Each laboratory has a set of colour coded gether (Brown 1972, 3), appears to be appli- cards, one per machine. The machines are cable to this centre. Although the timetable numbered with barcodes. This allows is currently pi:Tared manually prior to theclients to be assigned to a particular ma- start of each session, the availability of thischine for the duration of their visit and method bodes well for the future automa-ensures against multiple bookings of ma- tion of this task. chines. A machine is available only when its card is held by the lab staff and when Unique identification of items and us-that laboratory does not have a tutorial ers: Barcoding class currently assigned to it. Similarities between the control measuresA High Density version of Code 39 Stan- employed in libraries and supermarketsdard is used by University Administration led to the adoption of the machine readable on the student identification cards. The barcode as the method of uniquely identi-barcode light wand used for ALABS (MR fying all items. Barcoding is an ideal 300), can read both high and medium den- mechanism for keeping track of numerous sity barcode formats without a configura- items required of ALABS. tion change. The barcode on the student identification card is used to record trans- The Medium Density Code 39 Barcodeactions for the student, and the 'labpass' Standard is used on all types of items in the provides their location. Medium Density 130 4 Using Computers Intelligently 121 Code 39 barcodes are generated as needed rently on loan by students, staff, tutors and for staff and tutors who use the centre. long term staff loans. Included is a count of the active student loans. Students and tu- ALABS System tors may have extra items appended to current loans. ALABS was targeted for an Apple He sys- tem, running dBase II under the CP/MThe system regularly and automatically operating system, as most of the requireddetermines if any items on loan are late. hardware was already available at theOverdue student and staff loans may be Laboratories. The initial targeting of anchecked. Offending students who do not Apple He gave the advantage of a lowerreturn items within the required period are entry cot (see Diagram 1 for ALABS hard- automatically placed on an offence data- ware). Mindful of an eventual target to base. Depending on circumstances their another machine, the barcode wand pur- account with the centre may be frozen for a chased was a stand-alone unit. given period, so that they cannot use the facilities outside of supervised time slots. AL ABS was designed as a system to control and monitor the borrowing of lendable Machines in laboratories may be booked in items (software, audio-visual items, andadvance according to loading demands hardware) for the Microcomputer Labora- (due assignments!). Weekly timetables for tories in the absence of both a single net- all laboratories can be initialised. Functions work structure and network usage control are available to assign a laboratory time slot facilities. ALABS currently consists of 178 as free, supervised, or unsupervised. Given modules totalling approximately 582 other constraints, individuals may book Kbytes of dBase H, with some Z-80 and 6502 machines no later than a day in advance. machine code routines. The system utilizes Terminal booking sheets can be printed. 38 permanent databases and several tem- Booked machines are released if the stu- porary databases. dent has not arrived within the first 15 minutes of their time slot. In operation this Two main modes are provided in ALABS. section is reminiscent of a spreadsheet. The first mode is an operational mode; the second is a supervisor mode. Super-visorAn Auxiliary Commands section allows mode groups together all powerful or ses- some of the reporting functions of the sional management functions. These in- supervisor mode to be performed in the clude erasing the student and tutor data-operational mode. A hardware mainte- bases (occurs at the end of a session), delet- nance and repair system and a function to ing staff members, deletion of the student change a computer's status (operation or sessional borrowing file, viewing deletedunder repair) is provided. Also provided is library records or archived repair records, a Batch Command Processor which allows checking and re-indexing databases, batch long reports to be printed overnight. printing of student sessional borrowings, printing staff and tutor databases, initialis-Complete library functions are provided ing and altering laboratory bookings or for item databases. These can be quickly adding/deleting machine bookings. edited to reflect changes in stock. Software records consist of a barcode, name, version, In the operational mode, loans and returns publisher, commissioning date, license may be made (see Usage Scenarios section number, machine usage and disk media for more detailed descriptions). Interactive variety. Audio-visual records are defined functions include determining items cur- according to type of item. Journal articles 131 122 Using Computers Intelligently and books are recorded with author, date(student card, 'labpass' and at least one etc. in the manner of a library. If the item is item). A keyboard entry mode is also pro- a videotape, then the record consists of the vided in the case of any adventitious failure title, presenter, duration, and category. If of the barcode wand. the item is a set of slides, then the number of slides are also recorded. Hardware records Usage Scenarios include barcode, item, model, supplier, serial number, machine number, commis- Unsupervised Student Borrowing sioning date, location, cumulative down- time, and the last date down. When borrowing an item the student number from the student card is read using Extensive database back-up facilities are the barcode wand. If the student is not on provided. These include all Library data- the sessional database, the operator is bases (Software, Hardware, Audio-Vis- alerted and the student manually provides ual), Deleted Library database, Staff, Stu- full details on a Sessional Student Borrow- dents, Tutors, Loans (Staff Loans, Staffing Form. The details are entered into Permanent Loans, Overdue Student ALABS by the operator, and loans can be Loans), Sessional Student Borrowings, made. The student card is retained by Offence, Daily Machine Software Usage, centre staff member for the duration of the Hourly Loans Counts, Hourly Borrowing loan. A machine/lab pass is read and given Rate, Borrowing Duration, Down Time,to the student. Requested items are found Down Equipment, Repaired Equipment,and each barcode is read. Machine alloca- Faculty Usage, and New Students data-tion and loan duration is currently two bases. consecutive hours. Students may re-bor- row at the end of each two hourly interval The system may be temporarily closed if demand for the item is low. Borrowing of down for forms printing or maintenance. A audio-visual media (example training complete close-down is issued at the end of packages) and manuals, is also subject to daily transactions. After back-up of trans- similar procedures. Hardware is currently actions onto floppy disk, ALABS can be left only lent to academic and general staff unattended. At this time daily reports andmembers. housekeeping functions are performed. Functions performed include the printing Returns are performed in a similar fashion, of a transaction file, updating sessional with the labpass' and items retained by the borrowing file, appending New Studentscentre and the identification card returned to the student database, the calculation ofto the student. hourly loan rates, faculty usage figures, the printing of machine usage figures, andStaff and Tutor Borrowings processing overdue student and staff loans. Also, expired offences are cancelled,New tutors and staff members are intro- and a report on which files to back-upduced to the centre by a representative before starting operations the next day is from their department. At this time, all produced. The machine will power itselfrelevant details are entered concerning the down after close-down procedures haveperson. This information is manually re- been performed. corded on a Staff Borrowing Form, then entered into the system. Each tutor is iden- Most of the data entry for ALABS consists tified by a unique two digit barcode which of single keystroke menu navigation fol- is stored in a folder for quick access. In lowed by three or more barcode readsaddition, staff numbers, unique three digit numbers, are stored in the same folder. Using Computers Intelligently 123 Staff members may also be tutors, that is an provides the minimum cost consistent with individual may have two borrowing num- the ability to meet the evolving needs of bers. A staff member will borrow items tertiary educational requirements. useful to their research work, whereas a tutor will borrow teaching sets of software. ALABS is currently being ported to a PC Delineating loans in this way enables dif-environment where it will run as Clipper ferent loan durations to be exercised.compiled dBase III+ application. Sections ALABS prints out a form let ter overnight to of the system have been redesigned to take e staff members with overdue loans. advantage of the more sophisticated lan- guage. The change in language has already Class Loans resulted in simpler code and fewer data- bases. Increased speed will result fa the use of a faster host machine. cyst ..1 per- For a class, tutors will generally borrow one formance trials have indicated a constirva- or more class sets of disks. Once the trans- five 2.2 times speed increase when this port actions are recorded using ALABS, it is is completed. then the tutor or lecturers' responsibility to log the student using a particular software Some operational performance bo t tlenecks item against the item number on a sheet exist. Different types of users (tutor class provided (Class Loan/Borrowing Sheet).loan/returns, terminal bookings and nor- At the end of the class, the items are mal student loans), can lead to servicing checked against the name of the student so delays during peak activity. To correct this, that an audit trail is maintained. the new version of ALABS will be multiuser, and operate from PCs con- Implications and Future Extensions nected to the LAN. During peak loads two operators might work on the system servic- Organisational arrangement is as impor- ing different user types. tant as a technical solution to the problem of CAL and general course based resource Table 1: Sessional Unsupervised Transac- distribution and management. The combi- tions (1985-1987) nation of an appropriate organisational structure and a computer system ALABS Year Session TransactionsComments has proved to be a reliable solution. 1985 1 1,591 Manually recorded Resource distribution and management in 2 2,277 Manually the CAL field, need be no d i fferent to that of recorded providing resources for more general computing courses. An appropriate design 1986 1 3,164 ALABS 2 5,389 ALABS can satisfy both. 1987 1 10,318 ALABS ALABS has proved to be a successful mechanism for the distribution and man- agement of CAL and related items in a Ackowledgements tertiary education computing service centre, Successful software for the manage- The authors wish to thank Cathy Nicastri ment and distribution of CAL and related for her efforts in testing and managing this items imbeds the operating procedures of system, and Helen Sigalas for her contribu- the organisation within the system and tion to the implementation of ALABS. 124 Using Computers Intelligently Bibliography R.J. Clarke. Previous positions in- Borovits,I. (1984). Management of Computer clude Computer Operator at the Wol- Operations. Prentice-Hall longong Institute of Education Com- Brown,A.R. (1971). Optimum Packing and puting Centre, and Tutor for School of Depletion: The Computer in Space- and Industrial and Administrative Stud- Resource- Usage Problems. London: ies, University of Wollongong teach- MacDonald ing Microsoft COBOL and BASIC and Fried,L. (1979). Practical Data Processing application packages including Lotus Management. Prentice-Hall 1-2-3, and Microsoft WORD. Rod de- Jackson,I.F. (1986).Corporate Information signed that schools first External Short Management. Prentice-Hall Course curriculum, and training notes Perry,W.E. (1987). The Information Center. for SIAS commercial programming Prentice-Hall and computer applications courses. Sperry. (1985a). Sperry Personal Computer He has been Operations Supervisor of Microsoftware. Sperry PC Pilot Support the Microcomputer Laboratories since Sperry. (1985b). Business Packages for Class- 1985. Current research interests in- room use. Sperry PC Pilot Support chide Systems Analysis, and Digital Sperry. (1985c). General Business Programs Image Processing. His CAL interests which operate on the Sperry PC andare relate to resource distribution and useful for classroom instruction. Sperry management and PILOT software PC Pilot Support development and production. Sykes,J. (1987). PCs Present and Accounted For. PC World April, 202-208 L. Athanasiadis. Support Programmer for the Microcomputer Laboratories since 1985. Apart from ALABS, Louis has written software for a number of educational systems at the Laborato- ries, including the Educational Testing System for optical mark scanner as- sessment of secondary school student results. Bringing the distance student on campus the Riverina way Brian Cornish, Mark Johnston and Michciel Whitelaw School of Information Studies Riverina-Murray Institute of Higher Education The study of computing at tertiary institutions considered for an applicant who has dem- normally involves a substantial component of onstrated professional development by practical work. This usually takes the form of prognraiming or use of applications software. one or both of the following: Traditional approaches to the teaching of com- puting in the external mode generally use the Membership of a professional associa- residential school as the means by which the tion which requires academic experi- practical component'; satkeecL ence based qualifications; This paper describes the use of dial-up facilities as the primary mode of communication and Professional seniority or status. teaching in a post graduate tours,: for students interested in computer applications. The devel- Students will not be expected to have any opment of software to enhance various aspects of communication is documented. Results of academic knowledge of computing at time surveys of student attitudes are provided, to- of entry. Students who have completed gether with statistic of usage and achievement. any degree in computing, or an academic major in computing as part of a degree, will not be eligible to enrol in this course. The School of Information Studies at River- ina-Murray Institute of Higher Education This course is offered in the external mode; commenced the design and development during the development of the course, the of a new Graduate Diploma in ComputerSchool considered various ways of provid- Applications in 1983. ing computing power for students study- ing the course without requiring them to As described in the course proposal docu-attend a large number of residential mentation (Graduate Diploma in Com- schools. puter Applications Stage II/III Proposal, Riverina-Murray Institute of Higher Edu- The following options were contemplated: cation, May 1985): All the students being required to buy This course is aimed at providing the one (1) of a small number of particular information management expertise brands of microcomputer. for use re- that is necessary for significant com- motely; or puting activity in various professional All the students being required to buy groups. one (1) of a small number of particular brands of computer and to use direct The general minimum requirement for dial to connect to the Institute's com- admission to the Course will be an Under puter; or Graduate Degree from a college of ad- The students being allowed to buy any vanced education, university or equivalent suitable terminal and use the AUST. institution. However, admission may be PAC system. 126 Using Computers Intelligently The decision was made to use the AUST-at the end. At the residential school at the PAC facility, and as stated in the coursebeginning of the course, the students are documentation ,"the high level of machineintroduced, amongst other things, to the excess would normally prove a problem to AUSTPAC system and taught how to log a course being offered in the external mode, on to the Institute's VAX system from but as the Institute has an AUSTPAC mode, AUSTPAC. as well as telephone dial-in facilities, stu- dents fro m a nywhere in Australia will haveThe first intake into this course was in the access to the Institute equipment, at a cost Spring Semester 1985, and some of these of something less than $8.00 per hour.students recently graduated. Students are required toprovide them- selves with terminal equipment which will Communication with Distant connect to the AUSTPAC facility." Students. This procedure has proved to be highly AUSTPAC usage satisfactory.Students have used a full range of microcomputers, and some stu- The Institute rents AUSTPAC from Tele dents have used main-frames to connect to corn and for the first eighteen months of the the Institute's VAX system. The Institutecourse, bore all costs.These averaged leaves the responsibility of selecting suit- approximately $825 per month, being able equipment for connection to the comprised of monthly rental, a fixed charge AUSTPAC system to the student. per call and a variable charge based on volume of traffic and connect time. It then The students attend two (2) residential became necessary for the Institute to charge schools during the period of this course, students for use of AUSTPAC over a basic one at the beginning of the course, the other allowance for each subject. end of session ordinary user Figure 1: AUSTPAC Accounting system 136 "ST1 Using Computers Intelligently 127 As Telecom was unable to charge individ-lecturer's mail concerning this course are ual users, it was left to the Institute to devisealso included. appropriate accounting procedures. VMS identifies users by a terminal nuinber and Initially, electronic mail was used to handle type, with AUSTPAC users being denoted all of the above transactions.The mail by an NVA prefix. It thus proved to be a facility in VMS version 3 had no sub-direc- simple matter to identify NVA users at tory or folder capabili ty, with the result that login time and call a program which checksmail files were large and difficult to organ- a file of valid users. The file contains the ize. One can easily appreciate that witl 7-7 balance of connect time available for usemultiple assignments and 30-50 students, which is displayed on the user's screen at good organization and frequent access to a login time. This is automatically updatedcomputer terminal is essential.It also daily by the VMS accounting facility. Usersquickly became obvious that users should can thus monitor their usage. Post gradu- keep copies of all important mail sent in so ate students receive an automatic alloca-that problems or disputes be quickly re- tion of five hours of connect time per sub- solved. VMS mail has provision for the ject.Further time is paid in advanceusemame to which the message is being through the School office where a program directed and a subject heading. When re- updates the balance in the valid user file.ceiving :nail, the username originating the Any user not contained in the valid user tile message and the subject appear when a or whose balance is exhausted, receives thedirectory of mail is obtained.Ordinary appropriate message and is automaticallyusers of mail tend to be rather vague and logged out. The overall system is depicted casual in the details entered in the subject in Figure 1. headir6. Typically, they deal with small numbers of messages and may delete them Use of electronic mail quickly. When lecturing in a course such as this, a much more efficient use of mail is The course is structured so that students required since one has to glean information attend a residential school at the com- as quickly as possible without laboriously mencement and at the end of their study.reading every message. When sending a At other times the main vehicle for commu- message a copy is kept for oneself. Since nication is electronic mail via the Institute'smail shows the originating username, one VAX computer.Communication, how- soon finds nu- :serous messages with one's ever, ercompasses many different aspects own username and no way of distinguish- of study. Common categories are listed ining them short of reading each one.It Table 1 below.Proportions of one quickly became apparer.t that the subject Table 1 Common categories of communication Direction of Nature % of total mail communication over 12 months student -> lecturer query re coursework 32 lecturer -> student response to query 29 student -> lecturer problem report 5 lecturer > student assignment handout 2 student -> lecturer assignment handin lecturer -> lecturer problem report/response 3 lecturer -> student assignment feedback 28 lecturer -> student acknowledgement of receipt N.A. 137 128 Using Computers Intelligently heading should be used to record the tain if a particular student has completed student's username and nature of transac-certain work or responded to a request. tion so that directories of mail could be quickly scanned for information.e.g. The use of mail can be a time consuming Subject DPG1111 Tute7marks activity, especially on a multi-user com- puter with numerous other demands. A VMS version 4 offered a considerablyparticularly noticeable issue was that of the improved mail facility with provision for handling of student assignments. Initially, separate folders o r sub-directories to whichthese were sent to the lecturer concernedor mail could be directed. Three foldersareto a class account via mail. The lecturer normally in operation : NEWMAIL, MAILwould then extract each message/assign- and WASTEBASKET. New unread mail ment, taking care to place it in an appropri- appears in the NEWMAIL folder; once ate sub-directory using the student's user- read, mail is transferred to the MAIL folder name for later marking.One method and once deleted, is transferred to the which was tried initially was to have stu- WASTEBASKET from which it can be re- dents send their assignments to a 'clearing covered before the end of that mail session. house' account, from where a notification Other folders can be created to contain the mailing was done. The assignments were various categories as listed in Table 1. Itthen mailed to other marking accounts. then becomes a simple matter to scan the This resulted in considerable douole han- directory of a small folder in order to ascer-dling and inefficient use of time. An on-line Figure 2: Common communicatioi: transactions 13S Using Computers Intelligently /29 menu-driven system of automaticSome subjects, such as the introductory handin.was therefore developed, so as to programming subject, enable students to allow the student to select the subject and do almost all their work on their own the exercise for which he/she wished tomachine. Others, which require special submit a file.The file was then copied software, make more demands on connect directly to the appropriate sub-directory oftime to the VAX. Some statistics of student a 'secret' account for later marking. At the usage are presented in Table 2. i itese were same time, a notification of receipt mail averaged over the months February to June message was automatically sent to the stu- 1987. dent; thus considerable unnecessary han- dling of mail was eliminated. As can beFrom these data it may be reasonably con- seen from Table 1, most mail is now con-cluded that students are making efficient fined to student queries and responses. use of connect time. This may also account for the fact that up to 50 postgraduate stu- Some exercises, which are multiple choice dents have been able to coexist using only and/or short answer in nature, can be auto-five AUSTPAC lines and three local dial-up matically marked in the sub-directory tolines to the VAX. It can also be seen that which they are sent. Result files can then be most usage is concentrated in the off peak 6 automatically mailed to students and re-pm - 8 am period. sult file directories can be automatically scanned to feed results into a spreadsheet Student Profile mark book. The automatic handin pro- gram has also been adapted to offer anSome of the characteristics of the first co- interactive on-line test in some subjects. hort of students were measured. Perhaps the most important characteristic is that of A flow diagram representing these transac- success. Two measures of success are used. tions is shown in Figure 2. The first is quantitative and is the propor- tiLn of the course completed in minimum Student usage time. The course consists of 6 equally weighted subjects and one double subject. As connect time is limited and since work-This gives a measure with a scale 0 to 8. ing on-line at 300 baud (the most common rate) is relatively slow, it is important to useThe other measure of success is qualitative; time efficiently. Text or data entry is bestit is the current status of the student. The done off-line on the student's own micro-student may be a graduand , i.e., have computer and connect time to the VAX completed the course. The student may be used for uploading, handin or testing.continuing. Such a student has taken Average connect time/session/user 12 minutes Average connect time/month/user 3 hours Maximum connect time/month/user 10 hours 30 minutes Average number of connections/month/user 15 Maximum number of connections/month/user 45 Average total connect time/month 80 hours Average connect time/month (off-peak cost) 54 hours Average connect time/month (peak cost) 26 hours Table 2 Statistics of student usage 139 130 Using Computers Intelligently Excluded Continuing GraduandTotal 8 11 11 Proportion 7 1 1 6 of 5 4 1 2 3 Course 3 2 1 Completed 1 6 6 0 5 5 Total 13 3 11 27 Table 3 - Successfulness longer than the minimum time to complete subjects studied. Indeed 85% of the stu- the course but still, at the time of writing,dents now excluded had been excluded by may be ultimately successful.Finally a the end of the first programming subject. student may be excluded. This makes early performance a good pre- dictor of ultimate success. Exclusion is not to be interpreted pejora- tively. The Institute e..cludes any studentsSome other predictors that could be used who do not maintain an adequate rate ofare verbal ability, spatial ability and atti- progress. The student' may indeed be at-tude. These predictors have the advantage tempting subjects and failing them throughof requiring only a short interview to be inadequacy; however, many of the stu- assessed, so they are a tempting screen for dents who decide to withdraw fail to notifyadministrators of a popular course who the Institute of their intentions. It is only the have to decide who is to be admitted to a zero rate of progress that enables the Insti- course. These predictors were measured at tute to recognize the withdrawal and re-the start of the course and compared move the student from the course. Unfor- against final success. As will be shown tunately, the number of students who with-below, none were found to be as satisfac- draw and are polite enough to inform thetory a measure as success in the first sub- Institute are sufficiently small that it has jects. not been felt necessary to create a special category. The test used for verbal and spatial ability was Heim's AH5 Test (Heim 1968) of group Those first cohort students who completed intelligence. This test is a group intelli- the course in minimum time are now grad-gence test designed to assess people al- uands. Table 3 shows the distribution ofready known to be towards the top end of the students between the two measures ofother intelligence scales. Results have been success. related to University and Teachers' College students. The test is a relative test,in- Current graduation rate is 41% and could tended to disperse the group across the rise to 52%. Most students who do not scale rather than measure ability in abso- succeed, fail in one or both of the first No lote terms. Using Computers Intelligently 131 8 gggggg gg g g g Proportion 7 c 6 of 5 4 c ec Course 3 2 e Completed 1 e e ee ee e e g 1011 12 13 14 15 16 17 18 19 20 21 22 2324 E I D I C I B Verbal ability Table 4 - Successfulness versus Verbal Ability (one student did not do the test) For ease of cc mprehension, scores are con- It is clear from Table 4 that success does not verted to graaes such that the first 10% aredepend on verbal ability.It must be re- given an A, the next 20% are given a B, themembered that the population is already middle 40% a C, the following 20% a D and pre-selected so such a statement may not be the lowest 10% an E. Analysis of the testtrue of the full range of verbal ability. was carried out by Reid-Smith(1985). The second component of the AH5 test The first part measures verbal ability. The used was that which measured spatial abil- students scored between 10 and 24. This ity.Again, it is a test intended to show spreads the students between E and B at differences in a narrow range of abilities University level; or E and A at Teachers' rather than measuring absolute abilities. College level. Table4 provides the distribu-The results are in Table 5. tion of verbal ability versus achievement level. A student is recorded as an "e" if heThere is no apparent relationship between is excluded, a "c" if continuing and a "g" if spatial skills and success shown in Table 5. a graduand. 8g g ggg ggggg g Proportion 7 c 6 of 5 4 c ce Course 3 2 e Completed 1 e e e e ee 0 e e e e 1011121314151617 1819202122232425 E I D I C I B Spatial ability Table 5 - Successfulness versus Spatial Ability (one student did not do the test) 141 132 Using Computers Intelligently 8 g gg ggg g gggg Proportion 7 c 6 of 5 4 c c e Course 3 2 e Completed 1 e ee ee e 0 e e 6 27 28 29 30 31 32 33 34 35 3637 38 39 40 41 Attitude Table 6 - Success versus Attitude. In fact, there may be some relationship to communicate with their lecturers and to between very poor spatial ability and fail- use the Institute's VAX computers. ure, as the student who performed badly on the spatial test but graduated was sick From the results of the four measurements on the day of the test. Her subjective im- used with the first cohort of students in the pression was that her performance wasGraduate Diploma, it would appear that significantly affected. the probability of success is enhanced by a good attitude towards computers, is not Another characteristic measured was thatlimited by spatial abilities, is even less of attitude on entry. The instrument used toaffected by verbal abilities,and is best measure attitude was a subset of the ques-predicted by the results in the initial sub- tionnaire used by Wilson and Braun to jects. distinguish between groups of students (Wilson and Braun,1985). The subset were References those questions in the original question- naire found to show significant differences Reid-Smith, E. R. (1985). Some Attitudes in attitude towards computers. On analy- and Skills of Computing Students: An sis, Reid-Smith showed this subset to be Analysis of a Survey undertaken by M 88% perfect in being uni-dimensional with Whitelaw at RMIHE, Internal report. a neutral point at 33 (Reid-Smith, op city. Heim, A.W. (1968). AH5 Group Test of The attitude of students on entry is meas- High-Grade Intelligence: manual. ured against success in Table 6. While it is Slough (England): National Founda- clear that a neutral to positive attitude tion for Education Research, r. e. helps, it is still possible for a student to [Restricted Test] graduate in minimum time even if a rather antagonistic attitude towards computers isWilson, J. D., Braun, G. F. (1985). Psycho- held on entry to the course. logical Differences in University Com- puter Student Populations. ACM Conclusions SIGCSE Bulletin, .5. 166-177 AUSTPAC has proved to be a very satisfac- tory method for allowing Distant Students X42 Using Computers Intelligently 133 The three authors are members of the School of Information Studies, River- ina-Murray Institute of Higher Educa- tion. Brian Cornish is the Dean of the School and as such, is responsible for course development.He has had many years experience in computing, commencing on Silliac in 1960. His research interests include effective distance education teaching tech- niques. Mark Johnston is a lecturer in the School and his recent teaching experience includes introductory pro- gramming subjects in the Post Gradu- ate Diploma, and robotics and expert systems in the degree course. Current research interests include robotics, expert systems and the applications of computers in astronomy. Michael White law is also a lecturer in the School and his recent teaching experi- ence includes algorithms and lan- guages, business programming and data structures subjects at both gradu- ate and undergraduate level. His re- search interests include automated warehousing and effective teaching techniques. 143 Computerised formative assessment A powerful new tool Peter Fleming Director of Educational Development Family Medicine Programme The Royal Australian College of General Practi- It is now possible to provide computerised tioners has some 2000 young doctors undertak- ing post graduate training in the family medi- formative assessment to students, in the cine programme. They are scattered allover area of knowledge and problem solving, Australia in supervised practical experience using software written for this purpose and supplying them with high quality educa- called FACTS (Formative Assessment tional materials is a major problem. An interac- tive computerised, formative assessment sys- Computerised Testing Service). FACTS is tem has been established on Viatel called written in Fortran and was developed by CHECKUP (Computerised Home Evaluation of the RACGP's Family Medicine Programme Clinical Knowledge, Understanding and Prob- for their 2,000 or so young doctors in an lem solving) .It consists of a bank ofover 8,000 TF questions and a series of diagnostic and apprenticeshipbased, four year post management problem solving exercises. The intern programme in hospitals and prac- CHFCKUP software runs on a VAX and is avail- tices throughout Australia. able for CAEs, universities and continuing education programmes for almost any subject area. The RACGP/Family Medicine Pro- gramme call their medical version of Most teachers are concerned about makingFACTS CHECKUP (Computerised Home Evaluation of Clinical Knowledge, Under- their teaching more effective and efficient. standing and Problem solving). Delivery The best way to do this is for your students' learning to be more effective and efficientof CHECKUP is through Telecom's Viatel and the best way to do that is to ensure high system which provides ready access to the levels of motivation to learn and to providecomputer from anywhere in Australia via feedback for them on their learning. the public telephone network at a reason- able price (9 cents a minute in business hours and 6 cents a minute outside). There are many ways that feedbackcan be provided to students. One such way is to The Family Medicine Programme cur- give them access to assessmentprocesses during their learning that reveal whererently has 50 Viatel terminals circulating among its trainees, so that each trainee they are going right and where theyare should have access to CHECKUP at least going wrong, and why. This type ofassess- ment is called formative assessment and once a year. Advertising on CHECKUP aims to improve performance by providing supports most of the operating costs. feedback. The more common- summative FACTS consists of two main parts: - assessment usually takes place at the end of learning, is designed to grade students, a bank of true/false questions designed and generally provides little feedbackon to assess factual knowledge their specific strengths and weaknesses. a bank of diagnostic problems designed to assess problem solving ability. Using Computers Intelligently 135 Question bank those topics requiring further attention. This is provided (if required) as a printout The question bank is designed to be ac-to both the FMP doctor and his/her train- cessed in two different ways. A menuing adviser for discussion at their next structure allows the user to focus on an areameeting. No score is calcnlated, as the of interest or the computer will select ques-emphasis is on remediation rather than tions at random from a "pool" of questions. ranking. For example, the CHECKUP system is ar-The system also allows users to send mes- ranged to allow access via: sages to FMP at the end of a session, e.g., Diseases and Conditions, complaining about the accuracy of a ques- Random, tion or the ambiguity or relevance of a Signs, Symptoms and Abnormal Tests, question. This is an important feedback Special Subjects, and mechanism for maintaining the bank in a Systems, healthy state. where Random and Systems are pools. Questions are entered in several parts of the Statistics are collected on the use of the bank; for example, all questions go into thesystem which will allow comparisons to be random pool, while cardiac questions willmade between performance of different also be in the Circulatory System pool as groups of doctors and it also identifies well as in part of Diseases and Conditionsquestions with high failure rates. These and perhaps in Signs and Symptoms ormay indicate poor questions or significant Special Subjects.A summary of the areas of deficiency in doctors' knowledge CHECKUP menus is attached. which would then need to be addressed through educational channels. The True/False questions are arranged in sets of four parallel questions a,b,c,d. If theProblem bank user answers la correctly, 2a is presented, but if it is answered incorrectly, lb is pre- The problem bank is designed to give feed- sented followed by lc and ld. This is to test back on the learner's problem solving whether a small slip has been made or a strategies.It does this by presenting the significant lack of knowledge has been initial data on the problem and then allow- identified. If these additional questions are ing the learner to ask questions in a number all answered correctly, 2a is then presentedof categories. The format is open and the learner can return to any category as often and no "error" is recorded.If, however, one or more of them is answered incor-as he likes and ask as man, questions as rectly, the computer records this topic as needed. When the solution has been en- tered, feedback is provided. one needing attention. Likewise if three or more of the questions in one topic are an-The CHECKUP problems provide an open- swered incorrectly, but all the b, c, and ding description (the patient's presenting questions following are answered cor- symptoms) followed by a pathway frame rectly, this is also considered an area re- which offers alternative types of action quiring attention. Arranging the questions which could be taken (history, examina- in sets of 4 also has the advantage of being tion, investigations, emergency treatment, able to present a fresh set of questions each etc.). This pathway frame can be returned time the user reenters a topic (up to 4 to at any time to select a different type of times). action. At the end of a session, a list of topics Once a particular type of action is selected, undertaken is presented, along with a list of the user specifies what he wants by typing 136 Using Computers Intelligently in the first three letters of the specific action their students by establishing FACTS on or information required (e.g. ,DIE for Diet if their computers and encouraging depart- you want to know about the patient's diet ments and schools to enter their own ques- or SMO for Smoking History or CAT for tions and problem banks in the system. CAT scan). The computer then provides aThis could lead to a substantial increase in menu of all items in that section startingthe effectiveness and efficiency of their with three letters, or, if there is only one teaching programmes, at least in the area of item, it supplies the response. knowledge and problem solving.The pooling of questions and problems by all The user works his way through the prob- institutions around Australia which con- lem until he thinks he has solved it, when duct courses in the same area, e.g., nursing, he enters his solution. The computer thenmedicine, chemistry or botany, would cre- gives the expected solution plus a descrip-ate a magnificent data base for each course tion of how the problem could have been to use. solved. This is followed by feedback to the user in the form of a list of items chosenOther developments which would have been useful in solving the problem, marked + ve, but also includes The RACGP Family Medicine Programme those which would have been counterpro- is also providing a management problem ductive (in CHECKUP this includes unnec-on Viatel; here the diagnosis is given, ques- essarily invasive, harmful or costly investi-tions are asked about management, and gations) marked ve. These enable the feedback is given on the choices made. This user to compare his list with a similar one is a branching programme originally de- for each of two competent professionalssigned by Monash University's Depart- who did the same problem. It is illuminat-ment of Community Medicine and based ing to see which questions they asked and on the Modified Essay Question (MEQ) to realise that even experts make mistakesformat, which does not require interactive occasionally! capability and so is provided on normal Viatel frames. The problems can be scored; in fact this system was used by the RACGP candidates A Special Interest Group (S.I.G.) has been sitting for the Patient Management Prob- set up on Viatel called General Practitioner lem component of their Fellowship exami- Network. This is in effect an electronic nation this year in all the capital cities in scribble board for GPs on which they can Australia. put notices about clinical matters, confer- ences, interesting articles, useful resources, The exam problem choices are printed out etc. Notices drop off automatically after a in chronological order. This feature can benominated time (14 days in this case) or can used to follow the thinking of the person be removed by their owner. This S.I.G. will solving the problem. This process is helped help to break down the professional isola- if the users are asked to write down their tion of many GPs. hypotheses and indicate which hypothesisA read only bulletinboard has also been they are testing and whether the responseestablished with dates and times of meet- confirms or refutes the hypothesis. ings, conferences, etc, and important infor- mation. Facts in tertiary education The future Universities and CAEs could provide aAn update service will be added to valuable and powerful learning tool toCHECKUP which will allow the doctor to 146 Using Computers Intelligently 137 read the latest developments on, say, AIDS. Acknowledgements This should prove a very valuable service, but requires substantial work to establishThe FMP development team consists of: initially and to maintain at a high standard. Ms. Jude Biackwood, Computer Services Co-ordinator, Dr. Wes Fabb, Director of A community Health Information Service Education, Dr. Peter Fleming, Director of is proposed which would be open to the Educational Development, Dr. Jack general public and enable them to access Marshall, Director of Educational Research health information and possibly pay for theand Resource Development. time on the system through coin in the slot terminals in public places such as pharmacies, hospitals, libraries and schools.If successful, this would allow most Australians to access high quality health information at a reasonable cost. r: Access will probably be ,ia menus of dis- eases and conditions or peoblems which list infonnatic,a under headings such as: what causes it? what are the symptoms? how can I prevent it? what treatments are available? can it be cured? Itis expected that other professional groups will establish data bases on Viatel similar to CHECKUP to provide continu- ing education for their members in the field. For further information, about the education aspects, contact: CHECKUP COORDINATOR Family Medicine Programme, 70 Jolimont Street, JOI.IMONT, 3002. about the software, contact: Merlin Communications, 147 Eastern Road, SOUTH MELBOURNE, 3205. Computer enhanced distance learning S. Latham, W. Moore, G. Ritchie, B. Rothwell and L. Wilde Mitchell College This paper presents preliminary results and and applied science. A number of units suggestions on the use of micro-computer/ studied in the courses provided by Mitchell mainframe linkages in the teaching of under- graduate courses. The findings presented are College require the use of some form of the resultof a co-operative effort between computing. This usage is split into two Mitchell College of Advanced Education and areas at Mitchell College;that of units IBM Australia. Concrete data is Oven on the use 'about' computers and units 'using' com- of microcomputers in the study of a computing subject and a social science subject for first year puters. The latter units typically would use undergraduates at Mitchell College. Various a computer to run statistical and word preliminary findings based on computer usage processing packages and the former units data and information from questionnaires dis- are those using the computer to compile tributed to the participants is presented and appropriate conclusions drawn. and execute programs in computing courses. Mitchell College of Advanced Education isMuch effort is expended at Mitchell Col- 1. a College that has the majority of its stu-lege by lecturers and external studies staff dents studying in the external mode. Thein preparing the study guides used by the external mode of teaching means, in students. The staff at Mitchell are con- Mitchell's case, an extension of the Newstantly seeking new methods to enhance England model as distinct from the corre-distance learning. One of the major en- spondence model. The seudent is requiredhancements addressed in this project is to to attend a four or five day residentialincrease the lecturer/student, student/ school at the College in the middle of each student and student/institution interac- semester as well as optionally attendingtion. some weekend schools presented at vari- ous centres. The student receives a set ofMicrocomputer Assisted Distance study materials containing assignmentsLearning and unit information, a study guide and a booklet containing selected readings. The main driving force in improving dis- Other material such as cassette recordingstance education has been the development and videotapes may also be made avail-of new technology. The application of new able. technology to distance learning has re- ceived a great deal of attention and is re- The students studying in this mode areported in the literature (Bates, 1984). distributed over a wide geographical base. The units in various courses studied byThe use of microcomputers as a delivery external students range from business andsystem for material, linked to computer(s) law through to computing, social scienceat a central location has been discussed but 14 8 Using Computers Intelligently 139 not as yet implemented (Clark, 1986). At-mainframe studies in distance education. tempts have been made, for example, by Most information in this area is of an anec- Capricornia Institute of Advanced Educa- dotal nature. It was decided at Mitchell to tion, to use computers in distance educa-embark on a pilot study to collect and tion. This attempt ,however, was to use publish some information in this area. microcomputers at study centres and not in individual students' houses (Cook, 1985). Project Hardware Some tertiary institutions, Mitchell College amongst them have encouraged studentsThe facilities needed to implement the who have microcomputers at home to use a project have been provided by a co- opera- communication program and modem totive venture between IBM Australia and dial into the institution's mainframe and to Mitchell College.IBM Australia lent a use the facilities to complete assignments mainframe and associated equipment and and use any other facilities that may beMitchell College supplied microcom- available.This method of use has been puting facilities to students participating in encouraged but not investigated along any the project. cohesive development plan. The mainframe equipment used was an A pilot project using a microcomputer/ IBM 4361, a console, 4 colour terminals, a local area network/mainframe linkage line printer, a tape drive and 1.2 Gbytes of was initiated at Mitchell College in the lat- disk storage. The ,ralue of this equipment is ter half of 1986, with the co-operation of estimated by IBM to be of the order of $750 IBM Australia. 000. The operating system used for the project was MUSIC running under the IBM Project Aim VM operating system. IBM also provided Pascal, COBOL and FORTRAN compilers The aims of this project fell into three main as well as a graphics package, a communi- categories: technical, student background cations utility called PCWS and a package and educational enhancement.Specifi- for the implementation of computer aided cally the aims were:- instruction material. 1. to investigate whether the use of a mi-Mitchell College supplied each of the 30 cro- computer at home would enhance students participating in the project with the students' learning process; an IBM JX personal computer, colour monitor, two floppy disk drives and a 2. to investigate the technical and organ-Netcomm 1200A Smartmodem. Each stu- isational aspects of establishing a reli-dent was supplied with a copy of a able communication network between wordprocessing package, the MS-DOS Mitchell College and external studentsoperating system and utilities and a copy of for use in distance education; and PCWS. PCWS enabled communication using IBM 3270 emulation between the 3. to investigate the suitability of different microcomputers at home and the IBM 4361 types of unit presentation and content mainframe at Mitchell College. The use of a coupled with the students' background synchronous protocol emulation at home is in using computing facilities. a significant feature of this project. AUST- PAC communication was installed to the One of the main incentives for these objec- ethernet Local Area Network (LAN) that is tives was the lack of published experimen-used at Mitchell College. The students dial tal information of any microcomputer/ into AUSTPAC from home and are con- 149 ,. 140 Using Computers Intelligently nected to the Cot`- LAN. Once con-These students were asked to attend a day nected to the LAN, provision for further long briefing session at the beginning of connection to any of the host systems can beautumn semester 1987. During this session established using a simple LAN menu fa-they were to receive their IBM JX micro- cility. The IBM system is one of the host computers and also learn how to use the systems on the LAN. following:- Project Software 1. PCWS (Communications program) The key piece of software for the project2. Modem connection was a microcomputer program running under the MS-DOS operating system called 3. The IBM MUSIC editor and environ- PCWS. ment The PCWS program allows the student to4. PCWrite (Word processing program). use the IBM JX at home as if using one of the colour terminals connected to the main-The students were issued with a manual frame at College. The PCWS program also detailing function key sequences, Austpac allows the student to transfer programs/connections, a tutorial (brief) on using PC- assignments from/to the mainframe in a Write and other technical information very convenient fashion. This is achieved needed to use the complete system. The due to the integration of PCWS and the two groups of students were briefed on two MUSIC operating system supplied by IBM. different occasions at Mitchell College. This operating system was developed at Magill University in Canada andwas de-The students were also given a network signed to be simple to use and yet provideuser identification (NUI). This is used for a powerful working environment.A accounting with the Austpac system. The wordprocessing package, PC Write, from students were trusted not to abuse the use the public domain was also supplied to of the NUI. The College paid the bills that students allowing them to wrndprocessaccrued through the use of the two NUIs assignments on the IBM JX machines for that were given out one to each group of later transfer to Mitchell College. students. The students only had to pay for a local c«11 when using the Austpac NUI. Project Organisation and Implementation This is not neceabarily a definit: ye arrange- ment and will change according to circum- Thirty students were selected for participa- stances. tion in the project. These students were drawn from two units of work; Compuf....;Sway r,.)1,1, I and Methods and Statistics in Social Sci- ence. The first unit is from the IndustrialThe two sets of students had different Mathematics and Computing degree study needs from the project. Tne social course at Mitchell and the second is .7., unit it science students were studying a unit that the Social Science degree course at Mitch-was split into two distinct sections over the ell. Both units are first semester units. Thesemester. The first section required them to project had two groups of 15 students, one learn some basic statistics for later applica- studying ;he social sciences and the other tion to social science problems. The later studying the introductory computing sub-section consisted of essay type assignments ject that required them to learn Pascal pro- on various social science topics.These gramming techniques. students were able to use an elementary 150 Using Computers Intelligently 141 CAL program on the mainframe to reviseing topics that applied only to these stu- and learn some elementary mathematics dents. work as well as participating in a regular conferencing session every 2-3 weeks. Prelimir Iry results from this project can be broken to two groups:technical and At the beginning of these sessions, the lec- educational. turer would pose a controversial topic and invite discussion on the topic using elec- Computer Usage and Empiricai Results tronic memo techniques from the partici- pating students. Both of these uses of the The project had as cne of its aims the inves- mainframe were voluntary and were not tigation of the technical feasibility of estab- part of the required coursework of the unit. lishing a micro/mainframe link and to Students were also encouraged to :23e the gather some facts on the establishment and PC-Write package at home to wordprocess running of such a link. After an initial time assignments used in the latter part of the in early March 1987, most of the students unit and then to electronically transmit had mastered the technical problems of them to College for marking. establishing a usable linkage from their homes to the mainframe at Mitchell Col- The computing students on the other hand lege. From answers in the questionnaire, were required as part of their unit to com- the average time to become familiar with plete 8 assignments tlt require?: the appli- the equipment was 2-3 weeks. cation of the Pascal progran.ing lan- guage. The first assignment was a writtenFrom the 16th March 1987, records were one, but the other 7 assignments needed to kept of machine usage on the IBM. The be run and developed or. a cmputer. Foraccounting files were processed using these students an IBM Pascal compiler was some application programs and relational available on the IBM 4361 as well as thedatabase techniques to produce a file with required data sets to be used in each of theonly student logon information. This con- assignments. These students were to use sisted of a relational table with the attrib- the computer as an integral part of their utes of logon identification, type of student, assessab'e work in the computing unit. connect time, connect port, time of connec- tion, date of connection and an IBM IPL- All students were able to make use of the identification value. From this table three electronic mai'_ facilities to leave or read figures of interest arise. memos for lecturer, or other students. Another facility that was available was the Figure 1 shows the hourly usage of the 4361 access to library facilities using theduring the day. The results show what CLANN library system. This system al-would normally be expecte., that is, most lows access to a database of library titles students use the machine from early after- and topics from various participating li- noon through to approximately 2300 braries. Mitchell students are able to re-hours.It should be remarked that this quest these books ...Is required. figure includes weekends, which partly explains the low usage of the computer link Results during working hours. At the end of the autumn semester 1987, a Figure 2 shows the usage by day of the questionnaire was sent to all participathig week. Monday is taken as day 1 and Sun- students. An add. tional questionnaire wasday as day 7. Again the usage is as ex- sent to tlu., social science students concern- pected, Saturday, Sunday, Monday and 142 Using Computers Intelligently HOURLY USAGE of REMOTE CONNECTION MINIM SIMMS - 3OCIAL $C . 1911IL . . ... ,..._1 , . i .,. e-...'.1 ..,,,S. 1.0 0.0 L 0 12.0 15.0 10.0 WA 24.0 HOUR OF 0 RYt24haur iimel Figure 1: A graph of the hourly usage of the IBM 4361 connected to the LANat Mitch- ell College Tuesday are the heavy usage days. The lastdevelopment work done on the microcom- figure shows the connect time and the day puter at the students' homes. The total in semester. From the graph it can be seenusage of the system is given in Table 1. that the usage by the social science students tailors off as the semester progresses. This Table 1 Total Average / may reflect the approach of exams and the Hours logon day lack of any real need for the computer ex- cept for transmission of asisgnment mate-Computing stud.304.62 4.35 rial. Social Science 134.82 2.32 All Students 439.44 6.01 It is interesting to note that usage of the machine was minimal over the Easter breal (days 20,21,22 on the graph). TheColumn two of Table 1 shows the average usage of the machine by the computingcontpudng time per day for the indicated students is again what might have beenstudents. anticipated: sharp peaks of usage when assignments are due! All the figures pro-Educational Results vide some empirical data on micro/main- frame usage. The data does not includeThe results presented here are a result of access to the CLANN library system oranalysing the questionnaire and other re- 1 5 2 Using Computers Intelligently 143 HOURS LOGGED ON tMONDAT_SUNDAT) 1 I TITX CON/17111 ...mac III ...... I --I N, 1I e.../ N, /1 ....%i, LI Lt 1.1 3.11 i.1 LI LO 7.1 $.1 REEK OAT Figure 2: A graph showing the usage of the IBM 4361 connected to the LAN at Mitch- ell College for each day of the week. DRILY USAGE THROUGHOUT THE SEMESTER etaronieMOWS SCJPII 10.1 21.1 31.1 41.0 $0.1 70.1 400 DAT In SEMESTER tiros ISth MARCH) Figure 3: Graph showing the usage of the IBM 4361 connected to the LAN at Mitchell College for each day of the Autumn Semester starting atAth1)3 March. 144 Using Computers Intelligently sponses. A selection of questionnaire re- sults are presented in brief form in Figure 4. QUESTION yes no unsure Students were given a scale of 1-10 from 1 Computer link reliable 113 4 which they could reply to questions on the 2 Computer link was easy to use134 1 questionnaire form. Students answering in 3 Confident using the link 132 3 the range 4-6 were put into the unsure 4 Computer at home was helpful category in Figure 4. in studies 152 1 5 Linkage useful in contacting lecturer 123 3 From the results given in Figure 4 it is easy 6 Definite advantage over other to see that the students generally wel- students 106 2 comed the use of the microcomputer in 7I was adequately prepared to use the equipment 49 5 their studies at Mitchell. Some responses 8More time was needed to reflected distinctive biases from each of the repare me 17 1 selected student groups. This difference is 9 I will obtain the use oft micro in the future 152 1 shown quite clearly in the response given 10 Assignments were done more in question 11 (interaction with other stu- efficiently 133 2 dents). In the social science group, 6 of the 11 Benefitted using elect. memo 9 responses agreed that they benefitted with other students 8 8 2 12 Electronic memo was easy quite markedly from interacting via elec- to use 143 1 tronic memo with other students. In con- trast, the computing students stated thatFigure 4:Summary of some questions they did not use the opportunity to send/ asked of the project students at the end of get memos from other students very much the semester. and it was not all that helpful. At this stage we can only conjecture at the reasons for this response. It is worth noting, in connec- tion with the results from question 11, that acquainted to using the computing link. all the students indicated that the electronic This area of the project can be improved memo facilities were relatively easy to use using past experiences and having learnt and no clear bias was seen from eitherwhat are the common mistakes and needs group on this question. of the beginning student. Improvements have been made as the project progresses. Another topic that showed a clear biasOne such improvement was the writing of between the two groups was that of t,enefit a simple logon procedure that made dial- of the facilities in doing assignments. Eight ling up and logging on much easier. of the nine responses from computing stu- dents agreed that they had a definite ad-The social science students responded that vantage over other students in their unit bythey spent less than 12% of their time using using the microcomputer/mainframe Brix.the remedial mathematics program The explanation for this seems to be that (MATHE) that was set up using a CAL doing computer assignments interactively package supplied by IBM. This was quite a is far superior than doing them via Austra-surprising result. lia Post. It is suprising that the social science students did not think that they had anThe conferencing sessions was another advantage over other students in the unit. area where the usage by the students was very disappointing ( 15% of their time). Questionnaire Responses The use of the wordprocessing package All the students agreed that more timewas rated very highly by the social science should ha-- been allowed in getting them students. Most of the time was spent in 154 Using Computers Intelligently 145 sending assignments to College ( J 20% of ing and incorporating the use of microcom- their time) and in sending/reading elec-puters in the teaching of some external tronic mail ( J 30% of their time) units. The final grades obtained from exams by The preliminary results of this pilot project students in both groups was of a high stan-indicate that the students do benefit from dard (a number of distinctions and A the use of a microcomputer at home linked grades), but no significant results can be up to a mainframe at College, this linkage reported due to the project selection criteriabeing an integral part of a students' studies. not being completely random and the smallThe use of electronic mail as part of a unit of sample size. However, the general level ofwork has proved to be of benefit to students grades gave an indication that the use of the and staff. However, no conclusions can be micro/mainframe link was beneficial anddrawn from this pilot study about the the use of a more rigorous experiment ramifications of such computer technology might perhaps be able to determine on the final exam results. whether students' grades are influenced by having access to a computer linkage asThe second goal of the project, that of inves- described in this paper. tigating the feasibilty of establishing a reli- able computer linkage, has shown very Another interesting, although not unantici- promising results. The establishment and pated, result emerged from the computingmaintenance of a reliable, remote comput- students. This was the abandoning of theing environment for students throughout use of the IBM Pascal compiler to use the the semester, has proved to be one of the Borland product, Turbo Pascal. This wassuccesses of this pilot project. It has been because of the ease of use of this productdemonstrated that a section of studies at compared to debugging a program over a Mitchell College can be enhanced and have phone line. incorporated successfully into it a micro- computer/mainframe linkage that pro- There was a general feeling amongst all ofvides students with a better means of the students to have a small printer avail-communication and study than that which able with the microcomputer. presently exists. The majority of students ( 95%) stated quite The project has given some preliminary strongly that the use of electronic mail in facts on which to base any further work. contacting lecturers and other studentsThe project has established an organisa- was their greatest benefit. This contact withtional and educational framework for fu- the College and having almost immediate ture developments in the use of computer response made the students feel "as if weenhanced distance learning. It remains to are part of the stu dent body'. The use ofbe seen how other units of study can be electronic mail seemed to reduce the sense adapted to exploit this particular technol- of isolation for students studying in theogy. external mode and made some of them feel that they were not being doubly disadvan-References taged by not hiaving access to large city libraries and postal services. Bates, A. W. (ed) (1984) The Role of Technol- ogy in Distance Education, New York, St. Conclusion Martins Trees. Clarke, J. A. (1986) Some Reflections on The project has provided Mitchell College CAL at the Tertiary Level in Australia, with some valuable experiences in organis- Proceedings of the Fourth Annual Com- 146 Using Computers Intelligently puter Assisted Learning in Tertiary Educa- tion Conference, Adelaide 1986, 64-71. Cook, L. G. (1985) Considerations From an Attempt at Implementing CAL in Dis- tance Education, Proceedings of the Third Annual Computer Assisted Learning in Tertiary Education Conference, Mel- bourne. Contact Author: Dr. W. E. Moore School of Applied Science Mitchell College BATHURST 2795 Dr. W Moore and Dr. G Ritchie are lecturers in computing and social sci- ence respectively and have been re- sponsible for developing the unit con- tent and evaluation of the IBM project at Mitchell College. Dr. B Rothwell is the Deputy Principal at Mitchell and is the director of the IBM project. Techni- cal and administrative expertise has been supplied by Mrs. S Latham and Mr L Wilde. Mr L Wilde has been the operator/systems programmer for the whole project. 1 5 6 Quality distance education = computer based feedback + electronic mail Donald Munro School of Information Studies Riverina-Murray Institute of Higher Education The Riverina-Murray Institute of Higher Educa- Students were to purchase and install ter- tion offers a Post-graduate Diploma in Com- minal equipment in their place of study. puter Applications. The course is offered only externally and utilizes AUSTPAC as the prime Now, two and a half years later, several instructional and communication medium. The salient features of this style of education paper describes the types of facilities which have have emerged which are worthy of men- been investigated and developed, in an attempt tion. to provide an enhanced learning environment for the external student. Two of the facilities discussed are the use of ELECTRONIC MAIL, The Institute is no stranger to the task of and COMPUTER BASED FEEDBACK. The use teaching computing. In 1979 an Associate of Electronic Mail has meant that the external Diploma in Computing was introduced, in student can have the same types of regular communication with lecturers which an on- the full-time, part-time and external stud- campus student takes for granted. Computer ies modes. This course has always filled its Based Feedback has provided a periodic assess- quota of student places, and it was the ment mechanism, controlled by the student, persistent demand in the external mode which highlights areas of learning difficulty by directing the student to material in the pre- from applicants already possessing an scribed texts and notes. The combination of these academic qualification which led to the facilities has proved highly effective. Not only development of the post-graduate offering. has the marking load on staff resources dropped, In the last five years, pressure from external but students are able to ask questions relating to their particular difficulties, and receive direct students possessing their own micro-com- advice on overcoming those problems. puters encouraged the Institute to install three DIAL-UP lines to the Academic VAX A lot has been said and written about the11/780 and 8200 computers. This facility benefits which could be achieved in dis- allowed the preparation of programs and tance education by the utilization of mod- data off-line, followed by concentrated em communications technology. At the periods of transmission during the STD off- Riverina-Murray Institute of Higher Edu- peak periods. Students utilizing these fa- cation in Wagga Wagga, the approachcilities made generally better use of the adopted by the School of Information Stud- time available to them during residential ies has been typically careful and low key,schools, learned more and achieved better with the Institute proposing a Post-gradu-academic results. In short, a higher quality ate Diploma in Computer Applications in of distance education was being achieved early 1985;. the Diploma commenced withthrough little more than providing appro- an initial intake of 28 students in June of priate hardware. that year. The Course was distinctively unique in that it was offered only externally The introduction by Telecom of the AUST- and utilized AUSTPAC as the prime in-PAC Communication facility presented a structional and communication medium.challenge to the computing lecturers: 157 148 Using Computers Intelligently With the removal of punitive STD charges, tion of the VAX computing facilities, so that to what extent could extensive access to students have an idea of the location of the existing and newly developed softwaremachinery with which they will be inter- packages, support and enhance the ;earn-acting.This is an important concept to ing opportunities for the External Student?establish, as most of the students' home terminals will be micro-computers with The Post-graduate Diploma in Computertheir own processing and storage facilities. Applications concentrates on six majorWe wish to distinguish between "local" aspects of computing. The student is pro-and "host" storage and processing, and the vided with: a comprehensive program-various communications between them. ming background; a solid understanding of computer management; a good knowl-Student equipment edge of fourth generation facilities; the fundamentals of data base design and dataDuring the design of the course, the Insti- communications, and experience usingtute lecturers looked at the advantages Structured Analysis Techniques. A finalwhich would accrue from definitively individual case study major project drawsspecifying the equipment which the stu- together the student's newly acquireddent had to acquire.Standardization knowledge and experiences. would allow the preparation of programs which could be "down-loaded" and RUN The course, which consists of two yearsin LOCAL mode, and would allow the part-time external study, commences inexploitation of graphics and colour.In June with a five day residential school.addition, the equipment could be com- Unlike residential schools in the Associatepletely checked out and the support effort Diploma, where the emphasis is centredwould be reduced. On the other hand, around practical tuition and computermany potential students already had com- access, in this residential school the Post- puter equipment which would be suitable graduate Diploma students are oriented to for textual transmission, and regardless of the Institute and its expectations of them,which equipment was standardized upon, the equipment which they will be accessing it would be obsolete in a relatively short from their remote locations and the lectur-period of time. The decision to remain ers they will be interacting with (sight centred on the Institute VAXs was made on unseen) for the following two years. Ex- the basis of: amples of appropriate terminal equipment and the use of AUSTPAC are also covered, the relatively small amount of graphics together with a general introduction to and colour related instruction LOGGING-ON, and using the HELP and electronic MAIL facilities. insufficient staff to develop a new set of programs centred around a particular The distinguishing feature of this course is micro-computer the use of the AUSTPAC communication facility as the primary communications the existing set of programs used in the link between the student and the lecturers Associate Diploma and based on the involved. VAX being seen as providing a good base of appropriate material for the As these students have Lo extensive back- Post-graduate Diploma students. ground knowledge of computers (there is no pre-requisite computing knowledge),The introductory programming subject (a the residential school also allows inspec-full year subject) is taught using the BASIC 0 8 Using Computers Intelligently 149 language. Other languages were consid- ment. This submission file is then MAILED ered. However, as most of the micro-com-for marking. As well as each lecturer hav- puters have BASIC built-in, and as the ing a personal account on the VAX, there is Associate Diploma course already used a class account for each subject. The student BASIC in its introductory subjects, anyactivates a "HANDIN" procedure which move to another language without signifi-automatically transfers the submission file cant advantages would have produced an to the class account and an ACKNOW- unacceptable strain on staff resources. The LEDGEMENT OF RECEIPT is sent back to pedagogical advantages of being able tothe student. test and run small programs in LOCAL mode was seen to be a great advantage, Students may also MAIL questions directly particularly when the same program could to the lecturers' personal accounts. In the be uploaded and then run on the VAX. case of a lecturer being absent from duty, that message will remain unanswered for This approach works well because the ini- that period of time. Of course if the lecturer tial programming instruction concentratesis simply off-campus (at home, at a confer- on a STANDARD SUBSET of BASIC state- ence), then by using portable terminals the ments which are compatible across a widecommunication with the students can be range of equipment, rather then exploringmaintained regardless of physical situ- the richness of various BASIC dialects. Ination. I have even used a portable terminal most cases initial programs ran in LOCAL from a telephone box in Melbourne while at and HOST without change, and more ad-a conference in that city, and have been vanced programs required only a few successful in establishing communications global changes using the EDITOR (whicheven through switchboards with STD bars had been taught by that stage). Anotherin larger organizations. advantage of this approach is that the stu- dent develops a respect for the power andBy retaining a separate electronic mail flexibility of the micro being used; the FOLDER for each student in the class ac- "even my own machine can do it" syn- counts, the lecturers have access to the total drome. In fact some students took great history of submissions, acknowledge- delight (and spent many hours) convertingments, questions and answers which have advanced examples provided on the VAX passed between the Institute and a student. into versions which would run on their own machines. The value of this to theAnother feature of electronic mail is the educational process is self-evident. BROADCAST facility which allows a lec- turer to send a message to all usernames on Electronic Mail a distribution list. This allows for easy and efficient communication of general notes, explanations, and words of encourage- Most exercises and assignments are sub- ment, and because each student must mitted on the VAX by electronic mail. Using this approach, the student tests and communicate with the system twice a week, the lecturers can assume that such runs programs on the VAX until ready for messages are received. The number of submission. In this process a sample execu- LOGINS and duration of connection are tion of the program may be captured into a stored in the VAX Operating System LOG log file which will show all of the interac- and the statistics can be distributed to both tion. The student then concatenates the lecturers and students. Input and Output files, the program listing and any captured sessions into a singleElectronic mail is not restricted to commu- "submission file" using the COPY state- nication between the Institute and the stu- 150 150 Using Computers Intelligently dent but also facilitates STUDENT TOsubmission. Once the submission arrived STUDENT communication. A number of at the class account, the file was read into a individuals make regular use of mail tomarking program which tallied the score share difficulties and approaches to solu-and MAILED the mark, grade and correct tions, and there was a 'TUTORIAL" hour answers to problem questions back to the established by some students for the ex- student. These marks were then automati- change of pressing problems. cally transferred to a spreadsheet for as- sessment determination by the lecturer and The use of CAI techniques for other statistical operations (e.g. correla- tion between student responses). With such a brilliant communication facil- ity it was natural that approaches would beComputer based feedback made to include elements of CAI into the main stream educational material. WithinThe second program to be developed actu- the Associate Diploma there are alreadyally asked the student the questions and three packages, provided by Digital Equip-accepted the answers interactively. The ment Corporation, which are used in a tuto-program allowed the student to move for- rial instruction mode. These are not avail- ward or backward through the questions, able to the majority of Post-graduate Di-or to jump into a particular question. There ploma students because they require was also the facility for asking for help with VT100 compatible display terminals andthe question where more details of what single keystroke responses rather than linewas required would be given. When the oriented responses (this is a costing restric-answering of questions was completed, the tion rather than an AUSTPAC restriction). program would then tell the student the mark which had been achieved (without The CAI "TUTORIAL" method, while of-indicating the questions which were an- fering an enormous potential for individu- swered incorrectly) and it would list the alized instruction, also places a significantpage numbers of the text and the notes load on staff time in the development phasecontaining the information relevant to the of the tutorial construction. Various incorrectly answered questions. The stu- sources have quoted a ratio of 100:1 (100 dent then had the option to submit the hours of staff time required to produce 1 current attempt, or to re-read the references hour of tutorial). Again, even though thereand then answer the questions again. The will be savings in staff time once the tutorialstudent decided what level of mark (and programs are in place, the shortage of staff effort) was acceptable; there was no restric- time for the development cycle renders thetion on the number of attempts by the stu- tutorial approach practically unviable at dent.If an attempt was to be submitted present. then the program would MAIL the answers and the marks to the class account for proc- The first local CAI developments have fo-essing as earlier described, except that in cussed on reducing the amount of hackthis case the student already knew the marking by lecturers, and improving feed- mark for the submission. back of results to students. This second program is ased in an Intro- Marking programs ductory Systems Analysis subject which is based around a fairly large and detailed The first of these required the students to text. The students find reading the text create a file of answers to multiple choice quite a chore, and often skip or skim read questions and then to MAIL the file as a sections. Traditional methods of motivat- Using Computers Intelligently 151 ing students to do this reading thoroughly used the same text and sat the same exam would require the setting of various discus- (but who did not have access to the Com- sion questions or short answer tests. To be puter Generated Feedback), we found that of best use, these in turn would have to be while the distribution of marks was similar, quickly marked and have comments at-the post-graduate students' results were tached, before being returned to the stu-centred around a higher mean. In the fol- dent. There are insufficient staff resources lowing semester we made the CBF system to allow this approach to be effective. available to the next group of on-campus students and experienced a similar upward The Computer Based Feedback approach shift in the mean mark for that group. A breaks the 100:1 ratio of the tutorial ap-dramatic upward shift in mean for a cohort proach (VAX based and PC based pro-of Associate Diploma in Computing stu- grams in use have typically a 10:1 ratio),dents, who were also allowed to use the because most of the required questionsCBF system, occurred in the same semester. being presented in the exercises are avail- able as True/False or Multiple Choice The quite dramatic increase in perform- questions to be found in most instructor'sance, as measured by the examinations manuals. All that remains is the task of (consisting of 100 short answer questions), locating a page reference which will assisthas also been noticed at QIT following the the student should the incorrect answer beintroduction of Computer Based Training given. By allowing multiple attempts at theprograms by the Educational Research and set of questions, most students do not justDevelopment Unit (ERDU), and reported accept the grade and go on to the next task,atCalite '86 (1). but rather review the references and try again. This can be contrasted with the tradi- Reasons proposed for this success have tional testing methods where students willbeen: often not review the results of a test and, in some cases, not even read the comments that a lot of students no longer study appended to their submissions. effectively; be this due to lack of knowl- edge about how to study, or lack of In fact, the motivational aspect of using this motivation (2) delivery method became apparent when we included a counter of the number of that with this method the student is attempts which individual students were provided with only positive feedback, making at the tests. These ranged from 4 or and references to the correct answers, 5 up to 20 to 30. b an attempt to reduce the and so the student does not "learn the pressure on terminal lines we distributed distractors" (3) printed copies of the questions, so that students could note down answers as they that using the computer to answer the read the text. This decreased the duration questions sustains an interest in the of each test (as the student did not have to subject material which not only extends carefully read the question as presented), the time a student spends on the mate- but had only minimal impact on the num- rial, but also makes that time more effec- ber of attempts. tive (4). Exam results Conclusion When we compared the final exam resultsWhatever the reason, Computer Based of the post-graduate students with a group Feedback seems to: of on-campus degree students who also 161 152 Using Computers Intelligently 1. WORK, in the sense that students achieve a better learning program re- Don Munro entered the computing gardless of their distance from the Insti- industry in 1967. He has worked 11 tute. years in commercial computing envi- ronments and 9 years in academic 2. WORK, in the sense that students get computing environments. He is cur- appropriate feedback at the time of rently a senior lecturer in computing at submitting their answers. the Riverina-Murray Institute of Higher Education, and Course Co- 3. WORK, in the sense that lecturers are ordinator of the Post-graduate Di- freed from repetitious marking and ploma in Computer Applications. His record keeping, and can devote time to efforts oz.. c'omputer Based Feedback addressing areas of difficulty identified programs are a practical response to by the incorrect answers to particular both the difficulties experienced by the questions. external student, and a desire to see "cost-effective CAI" being incorpo- When Computer Based Feedback is com- rated into the mainstream teaching bined with a fast response electronic mail activities of educational institutions. system, and is provided over a low cost Much of his approach to the design and communication facility, the effects of the implementation of CBF programs was physical isolation of the external student investigated during the Autumn se- are significantly reduced. mester 1987 when he was granted studyleave to visit one commercial and References four educational institutions in Austra- lia, and to review their application of Ellis, H.D. (1986) Computer based educa- computing in their teaching programs. tion at Queensland Instituteof Technol- ogy. Proceedings of the Fourth Annual CAL1TE Conference. Adelaide 1986 (pp 92-96). Ellis, H.D. (May 1987) Private Communica- tion. Whitelaw, M. L. (July 1987) Private Com- munication. Gregor, S. (June 1987) Private Communica- tion. 162 A case study in computer education by external study John A Palmer, Ken Peirce, and Graham Pervon Curtin University of Technology This paper covers the introduction of a Busi- In order to overcome these inherent prob- ness Systems unit to the Bachelor of Business lems Curtin offers Undergraduate and degree at Curtin University of Technology and the changes in teaching approaches and tech- Postgraduate programmes for external niques which became necessary when it was students in the areas of Business and decided to make the unit available to external Administration, Arts, Education, Social students. EDP Systems 200 was developed in Sciences, Engineering and Science, Health 1984 with the objective of providing an intro- duction to Data Processing and Programming Science and Mining. A total of thirteen for second year Amounting students. It became courses are available entirely by external a required unit for their studies and was taken study and a further twenty seven are par- as an elective by students elsewhere on the tially available. The majority of the stu- campus. dents enrolled in these courses are located In 1986 it was decided that the unit shoild be over widespread areas of Western Austra- made available to external students through- lia, and in addition there are significant out Western Australia because many Account- ing students were held up in their studies numbers of students who live in other through not being able to study EDP Systems States and overseas. 200 which was a pre-requisite for other units. This unit was designed in 1986 and imple- External Studies in Business mented for the first time during the Autumn Semester 1987. In addition to a general Graduate Diploma This paper describes the design, implementa- in Business, the Division of Business and tion and problems associated with bringing a unit with many different requirements to a Administration offers three of the Majors relalively small number of students spread from its undergrauuate Bachelor of Busi- over a vast area. ness degree; these are Accounting, Finan- cial Management and Economics, and External Studies at Curtin Marketing. In addition, some Graduate Diploma courses are also offered through From its inception in 1968 the Western external study. All five Schools within the Australian Institute of Technology (nowDivision are involved to some extent, due Curtin University of Technology) has rec- to the fact that the degree includes a com- ognised that it has a responsibility to pro- mon core of units from each School. Three vide tertiary education to WAIT students of the Schools which also offer a major ha ve regardless of their location, age, status and greater involvement. The amount of in- domestic or work responsibilities. These volvement in external studies for any par- problems frequently mean that students ticular unit depends on the degree of diffi- are unable to complete studies in the nor-culty in conducting that unit over large mal internal mode. distances. I 6 3 154 Using Computers Intelligently The School of Computing and Quantitative into further computer studies. In addition Studies services two of the common coresome faculties other than Business have units and one of the specialist units in the made the unit part of their curriculum. It is Accounting Major, EDP Systems 200. now usual for over 400 students to enrol in EDP Systems 200 each year. EDP Systems 200 at Curtin The unit requires students to attend lec- Historical background tures in each of the three segments, work- shops for the programming sections and In 1984, at the last review of the Bachelor of tutorials for the data processing compo- Business Degree, it was agreed that all sec- nent. ond year Accounting students should study a unit which would give them someThe student evaluation consists of continu- knowledge of both data processing andous assessment for the D.P. area; this is programming, including hands on experi-made up of case study work plus tests and ence with microcomputers and an IBMan examination. The programming sec- type mainframe computer. tions are assessed from hands -on assign- %rents carried out in the workshops. A course was prepared for on-campus stu- dents. It was a five hour per week unit In 1986 it was agreed that EDP Systems 200 made up of studies in the following areas: should be made available to eternal stu- dents spread over Western Australia, but Data Processing not open by external study to students in Introduction to dBASE II the Metropolitan area of Perth. This deci- Introduction to COBOL programming sion was made to avoid the overscheduling of available laboratory facilities (already After two semesters it was decided that highly utilised) for internal student COBOL did not satisfy the objectives of the unit and another mainframe programmingDeveloping the External Unit language was selected. The program cho- sen was SAS, which was easier for non-The idea of using computers or learning specialists of computing to learn on thesome aspect of computing by external Amdahl computer because it removed thestudy is not new. The Open University has need for the students to overcome thebeen doing this since the early seventies complexities of Job Control Language (Bramer, 1980). In the case of EDP Systems (JCL) and concentrate on quickly develop- 200, there were specific problems which ing applications. In addition, it provided aappeared different from other attempts. sound preparation for the use of the lan-Other hands-on computing units have of- guage later in the study of Auditing. ten involved the study of the BASIC lan- guage (for example, Jennings and Atkin- In 1986. dBASE III was introduced to re-son,1982). As EDP Systems 200 involved place the earlier dBASE II. This has been some IBM mainframe requirements as well retained for 1987, but will be upgraded to as microcomputers and a mix of applied dBASE III Plus for both internal and exter-systems software, these past experiences nal students next year. were of little use in developing this unit for external study. The unit has proved satisfactory as an in- troduction to these key areas and many There were three main problems to be dealt students are using it as a stepping stonew0 due to this overall decision. The first (. 4 Using Computers Intelligently 155 one was to convert the Data Processingconduct a concentrated course on SAS at segment from a lecture and tutorial mode Curtin University during the extended to a format compatible to distance learning Easter break. concepts. The second problem was to ie- velop a microcomputer nands-on" seg- Once this information was at hand it was ment for dBASE III which could be handled possible to commence preparation of the by students in remote areas with very little course material. In the areas of Data Proc- reference to Curtin University. The third essing and dBASE III, a seriesof topics were goal was to create a teaching programme selected and readings were chosen to sup- for SAS on suitable mainframe computers plement the topic presentations. A variety in remote areas.The authors of this paper of exercises, both assessable and for prac- were given the responsibility of producingtice, were included in the topic material, so this programme by the end of October 1986 as to provide a comprehensive set of study notes for the students. For SAS a set of Before any of these activities could begin, it introductory notes were prepared, so as to was necessary to ascertain the interest and give the students a basic idea of the lan- likely enrolment of eligible students andguage prior to attending the session on also to find out the possible availability of campus. both microcomputers and IBM type main- frame computers in the country centres. A Once enrolments were completed in Febru- circular was sent to all prospective students ary 1987, all students were telephoned to explaining the format of the total unit, determine the type of PC which was acces- advising of the availability of the unit in sible and whether all students would be 1987 and requesting a response regarding able to attend the on campus session at the student access to both IBM compatible Curtin for the SAS. Answers to these ques- PCs and mainframes. In addition, teaching .ions indicated that the students knew the institutions in country areas were con-requirements before hand and they all tacted to see if ar.y of the facilities needed appeared able to conform to the pro- were available at these centres and if so, gramme. Questions from students were what conditions would apply to the use of dealt with and they were left with a clear their equipment and teaching staff. understanding of the objectives and format of the unit. The overall response to these letters was good. There were over thirty replies and This initial contact proved very helpful, in from the information given it was apparent that the students commenced their studies that the bulk of the students would be ablewith very few problems and misconcep- to gain access to microcomputers. Many tions about the unit. They knew what to do had PCs at home, others had them at their and they already had received personal work place and other individuals were able counselling from their contact in the unit. to use equipment at local High Schools.The Outcomes of the First Run This meant that there was no need to use the institutions which had been contacted,Following the unit development as de- although there had been good supportscribed in section 2.2 above and after the from those bodies. communication with the students and any necessary counselling, it was agreed that The response regarding the availability of the unit was ready for a trial and it was run mainframes indicated that it would not bein the first semester of 1987. The activities feasible to handle this segment of theinvolved in each section of the course and course off campus and plans were made to the overall outcomes are set out 1.below. 1 A '7) 156 Using Computers Intelligently General D.P. Section files, backing up data, and appreciating the role of the operating system. Secondly, we The data processing section was generallywanted to expose the students to concepts suitable for changing from internal mode to of data storage, fields, records and files, one satisfactory for use with external stu- and how data in files can be accessed. The dents. The general principle was to use the vehicle chosen for the achievement of these text book (Principles of Data Processing,objectives was dBASE III, although origi- Ralph J. Stair) as a basis for learning about nally the course had been based on dBASE hardware and software. Readings and II. other references were used to supplement any deficiencies in the chosen text book. It The largest single task in preparing the is usual that one book will rarely cover all dBASE III part of the course was writing the the topics in a set curriculum. The first taskexternal notes for students. Fortunately we was to develop individual modules for thehad a good basis for starting these notes. unit, each of these to cover a specific sec-For several years we had been teaching tion, such as Computer Input. If the text dBASE II and dBASE III to the public and book adequately covered the area theproviding detailed notes as well as sample mxlules were not large, but theywere data. These notes provided the basis for the e..1 tended if it u!:.-.0 considered that the text external course notes. co:^Prage was insufficient or if the particu- lar topic was not dealt with at all. There A complete set of internal course notes does were exercises within the text book which not suffice for an external course, how .-er, the students were expected to complete.and much of the material had to bere- These were used for practice only andwere written. Unlike teaching in a classroom or not assessable. Assessable exercises wereeven a laboratory, in distance learning the provided for areas not covered by the text notes are the students' only source of infor- book and were also use to test whether themation. It was necessary, therefore, to in- students were able to use the knowledgeclude in the notes sample screen layouts gained in pr'ctical circumstances. Anex- and detailed explanations which were not ample of this wa3 a case study dealing with part of the original notes. We knew that the an organisation which was having prob- first set of notes would be experimental and lems with its data processing andwasthat a significant re-write would be re- considering introducing a new computer quired for the second offering of the course. system. Other exercises dealt with areas As this w as a significant part of the teaching ,uch as Data Flow Diagrams and Cost/in th^ undergraduate Accounting degree- Benefit Analysis which were not coveted to in fact the students' only exposure to mi- any degree in the text. crocomputers - we knew we had to put a lot of continuing effort into the development The final assessment was an end of semes-and refinement of the notes. ter examination which mainly covered material from the relevant parts of the text Although we were licensed to copy and book. distribute a tutorial version of dBASE III, we could not provide our external students dBASE III Section with any version of MS-DOS. Our internal students were provided with the software The objectives of the dBASE III sectionlegitimately but the external students had were several. One was to give the studentsto find it themselves. Many of them did not adequate exposure to microcomputers,knov, what an operating system was, let handling floppy disks, formatting, copyingi o8e how to acquire MS-DOS! Using Computers Intelligently 157 Fortunately the solution to that problemarose when we tried to test the student's was also part of the solution to the problem knowledge of interactive commands, the of the acquisition of hardware. Once theexecution of which we had no way of wit- students had access to an IBM PC or com- nessing. It became necessary to teach the patible, it was a simple matter to ensureconcept of stored commands; in effect we that the student had a copy of DOS. The funhad to teach the students how to create began when we started to get phone callsprograms. As it turned out, the students from students asking whether their par-coped very well with the concept of com- ticular micro was an IBM compatible. mand files as well as being able to sequence There were machines we had never heard and extract data from files and store these of before- neither had our computer centre. conunands in a program. In time, we had all of the students equi pped SAS Section with PCs or compatibles as well as MS-DOS or PC-DOS. They needed compatible hard- The on-campus session used SAS on an ware because we were supplying dBASE Amdahl (a plug compatible IBM main- III and several databases on 5 1/4 inch frame computer) and proved to be success- floppy disks formatted by MS-DOS. These ful. At the start of the semester, the students disks did not contain any means with were supplied with notes on SAS and its which to boot the system. That had to be implementation on the Amdahl in a done by the students' version of MS-DOS workpack. A "short answer" test/assign- prior to loading our copy of dBASE III.ment was include with this material. This gthough we had no responsibility to pro-simply tested whether the students had vide the operating system, we neverthelessread the material. It was required that this took the responsibility of teaching the stu-work should be submitted prior to the on- dents how to use the micro, how to boot thecampus session. This material gave an in- system, copy files, rename files, look at the troduction to the types of commands, directory, and so on. A simple set of exer-screens and keyboard which the students cises had to be devised to enable the stu- would be using at Curtin. A concentrated dents to gain confidence in manipulating three days made up of lectures (about one files under MS-DOS. This had not been day) and laboratory work (about two days) done for the earlier course in dBASE IIIenabled them to learn enough to write a offered to the public, so we had to under-reasonably complex application and then take extensive writing of notes. fully test and debug the program. All the students successfully completed this pro- The final, and somewhat late, aspect of the gramming assignment and commented teaching of dBASE III externally was devel-favourably on how much had been oping an assignment for assessment. By theachieved in such a short time. end of the dBASE section of the course the students had created their own databases,The approach taken was similar to that sequenced them in difference ways, ex-previously used by the authors in the exe- tracted data according to quite complexcution of professional short courses for conditions, and reported on the data ex- people learning the basics of SAS on an IBM tracted. It was not difficult to set an exercise mainframe. The methods used had proved to test a student's ability to create a report successful in the professional courses and on a particular database. This was in factwere no less so with these external stu- done by using the CREATE REPORT com- dents. In addition the students expressed mand. Similarly we could test the student'sappreciation for being given the opportu- ability to create a database. The problem nity for face to face contact with their lectur- 167 158 Using Computers Intelligently ers which is not usual in the isolation of tended for the next offering in firstsemes- normal external studies. ter, 1988. Student performance was above expectation and questionnaire responses Student Performance indicate general satisfaction with the course. As expected, the general data proc- On the completion of the semester, 20stu-essing component ran smoothly, probably dents out of the 22 original enrolments hadbecause the requirements were no different completed all the requirement's of the unitfrom a normal distance learning course, and all of these had achieved good results.and there was plenty of experience to call The average result percentagewas 72%, upon in that area. The SAS component also, which is better than the average for other as expected, ran successfully, given the units. In addition the retention ratewas input of previous experience to its develop- very satisfactory. ment. The greatest uncertainty lay in the success of the dBASE III section, but the Student response care taken (and previous short course expe- rience) led to a successful outcome. Once the semester had ended and there- sults had been distributed,a simple ques-In addition, early planning and contact tionnaire was sent to all studentsto deter- with intending students well prior to the mine their reaction to the unit and alsosemester starting meant that there were asking for any recommendations in regard very few administrative problems. The to changes to the unit. The areas which only changes contemplated for 1988 will be were covered in the questionnaire are set upgrading to dBASE III Plus and reviewing out as follows: the text books to ensure that those usedare up to date. Administration of the unit Organisation of the unit Bibliography Academic content: Data processing Ashton Tate, (1984). dBASE III, California: dBASE III Ashton Tate. SAS Bramer, M., (1980). Using Computers in Distance Education: the First Ten Years General comments and of the British Open University. Comput- recommendations. ers and Education, 4, 293-301. Fillery, P & Palmer, J.A. (1985). Introduction Due to the smallness of the group itwas not to Personal Computers, Peiih: Curtin feasible to make valid statistical tests of the University of Technology. answers to the questionnaire, but the over-Jennings, P & Atkinson, R.J. (1982). Learn- all response indicated that the students did ing Computer Programming ata not believe that there was any need for Distance, Distance Education, 3(1), 157- major changes to the current format. 169. Shelly, G & Cashman, T. (1986). Learning to Conclusion Use dBASE III, Boston: Boyd & Fraser. Stair, R., (1984). Principles of Data Processing, Illinois: Irwin. In summary, the first run of thecourse has gone surprisingly well. No major problems have arisen in any of the three parts of the course and so no major revisions are in- 168 Using Computers Intelligently 159 John Palmer, Ken Peirce and Graham Pervan all teach commercial comput- ing for the School of Computing and Quantitative Studies and have done so since the School was formed in 1975. All the authors have been involved in the development of several public courses offered to the Perth business community in areas related to the selec- tion of microcomputer hardware and software, and programming in various microcomputer-based and mainframe- based software packages. 1 6 9 CAI A necessity not a choice Douglas L. Carthew School of Small Business and Computing Adelaide College of TAFE This paper sets out to explain some of the reasons Computer Aided Instruction is not being used by Lecturers to prepare their own material for Aspirations for C.A.I. presentation of self directed learning. The aim- plidty of text based authoring to a large extent has not been adopted in the classroom. The Dedicated educators gain reward from writer has been involved in the development of student learning. The Richmond College in a text based system suited to authoring by staff Victoria had pioneered the application of with minimal computer familiarity. The student the package "Author". The package re- has control at all limes, and provision has been made for branching facilities based on response quired an IBM with Colour for graphics and time. Emulation of multi-field screens al- presentation. Skills required were minimal lows for business systems training, and vide- but :;uitable hardware had not been pur- odisc access is also part of the program facilities. chaf ed by TAFE in Adelaide. Widely available for evaluation in TAFE the program requires little more than word process- ing skills for preparation of text. Few h...turers Many lecturers inspired to use Computer have elected to use the package. Assisted Instruction lacked the determina- tion and zeal to continue application of the The integrated authoring facility devel- most simple software. Graphics C.A.I. was oped by pragmatic response to TAFE lec-left to the few with the necessary zeal. turers requirements is the "Q" instruction package. This paper addresses the issue of Pioneering S.A. TAFE efforts in C.A.I. introducing C.A.I. into traditional class- rooms where computers are unfamiliar. The testbed for the emerging methodology will undoubtedly be the new college being Classrooms are designed for presentationdeveloped at Tea Tree Gully, currently of a familiar lecture. C.A.I. caters for theunder construction. individual learner learning at their own rate, often in their own time. A pioneering site for the new technology using the "Q" package was Croydon Park. ProfessionalEducators in TAFE (S.A.)The College was an ideal environment in were seeking alternative teaching methodswhich to develop the package from a Uni- and looking for a C.A.I. package whichversity Module to a System for wide appli- could be used on already purchased NEC cation in TAFE. APCIII computers.This required the simple skills of a Wordprocessing Typist The following screen illustrates the system and the minimum management skills ofoptions available within the package. lecturing staff. 170 Using Computers Intelligently 161 direction and implement the technology. MAIN MENU Without the support of lecturers staid in traditional teaching methods administra- Catalogue/EDIT existing TEXT tors are doomed to failure. Create new TEXT CONVERT session TEXT for use by Q UN-CONVERT session files to TEXT Off campus Tutorial Test-run Q Print a copy of CONVERTED TEXT Student learning is leaving the classroom Make an authoring work disk and taking place at home. Adelaide Col- Create/Edit a USER authority list lege of TAFE is one of the leaders of dis- QUIT tance education in this country. Staff Development took place in conjunc- More and more pressure is being applied tion with the trials at Croydon Park andby our society, by students who recognise "Q" found application in Matriculation the need for further education, and those studies and Adult Literacy. seeking advancement in present or future employment. Lecturer Hopes Many are buying computers to enable them Student Directed Learning to both study and prepare assignments. They can learn any subject with their com- With packages written today using C.A.I.,puter using C.A.I, a t home and be assessed lecturers know the their needs and theusing the same package in the college. students. The packages are available but when the time comes to convert preparedThe following Illustrates the system op- text for presentation on computer, thetions available to the administrator/man- problems appear greater than the desire toager/lecturer. improve methodology. MAIN MENU Curriculum Question Bank Modify session delivery options Recover data from a user disk Education administrators are applying Edit/read a QRESULT.TNT file pressure to meet rising costs and increase Test-run Q productivity.They are recognising the Print a copy of session text need to have lecturers more efficient but are Make a new user (student) cask not always committed to the application of Create/Edit a USER authority list computer technology and the costs in- QUIT volved ir. implementing the new method- ology. Misunderstanding Administrators know that the curriculumMost new methodology comes under criti- can have assessment on computer. Thecism with a resistance to change. C.A.I. is "Q" package is capable of holding a ques- no exception. Lecturers resist, administra- tion bank with all the questions for all thetion fails to see value, and funding institu- curriculum in 1 Ar'E on one large database, tions are loath to spend the money. and when required, extracting questions Administrative Direction for a lecturer from the database. The pro- gram exists but directors responsible forGood managers are often misunderstood implementation need to be willing to giveby educators who feel their creative teach- 171 162 Using Computers Intelligently ing efforts are unrewarded. Of all thefunc-The difficulty is not the student bi.e, the tions in our society today, educationneedslecturer feeling threatened by computers. sound direction. If senior managementare The role for the lecturer changes with C.A.I. unwilling to support change, then theen- to one of providing support for the learning thusiastic Lecturerwill introduce new environment. methodology "doing their own thing" with unnecessary duplication of effort while Lecturers looking for leadership. Some lecturers have adopted Computer Coordination needed Assisted Learning, but the majority lose Innovation needs direction, and without it,enthusiasm when the full realization of the an otherwise successful project may be-commitment becomes known. Few realize come a failure. Creative staff are difficult to that they already have much of the basic manage and inventors are noted for work-material prepared in lesson plan form and ing outside ni sidelines which shouldbehand-outs. imposed. The ievelopment of C.A.I.re- quires a team erort, the support ofa direc-Success tor, a willing typist and a facilitator who will participate in the developmentof theSuccess in alternative teaching methods is new methodology in the classroom. En-symbolized by the use of overhead projec- couragement .s needed for the student tors in the modern classroom. participating in the use of thenew method- ology. The recognition of student satisfactionin completion of a "Q" session need-, to be Work involved experienced. Once experienced byan open Computer Assisted Instruction requiresminded lecturer, that experience will be- both dedication to the machinesuse and a come motivation for more material prepa- conviction that focus should Jron theration. learner not the lecturer. The Lecturermust be persuaded that C.A.I. will aid inpresen- Australia's software crossroads tation of material in the sameway that the overhead projector is used in the classroomA well known phenomenon is the loss of or a slide-tape session may be included inqualified staff to overseas countries. The the methodology. emergence of significant user friendly Computer Assisted Instruction by Austra- Enthusiasm lian authors may lure Australians seeking recognition to countries willing topay for Students greet Computer Assisted Instruc- their expertise. tion with a zest born of a fascination with the new technology and the desireto Package success achieve. Using C.A.I.in the classroom daily has convinced me of its value inmost The "Q" instruction package developed to subjects. Students pair and answerques- meet the needs of TAFE has suffered the tions with both active participation andrelative ignomony reserved for poets in discussion. their own land. British Publishing Houses have taken interest in this software pro- Students duced in Adelaide, and are seeking to All students having overcome fear of tech-market the product in the large U.S.mar- nology value simple self directed learning.ket. 172 Using Computers Intelligently 163 The modem Personal Computer and soft- ware packages have matured.Mature Packages which are simple to learn will succeed. "Q" is one such package. It stands well without graphics on a monochrome display, it performs better when enhanced with full colour graphics, accepting colour frames. "Q" has the ability to interact with other computer packages providing tutoring capability with direct links to another computer program . It is suitable for use in the computer arena while compatibility with Videodisk rates the package with todays leaders providing the very best Video quality graphics with simple Word- star generated text. Management solution for friendly software Likeall successful computing installa- tions, Computer Assisted Instruction Management Assessment is simply an- other one which will stand or fail on the support it receives from senior staff. Conclusion Educators today, have may choose to be part of the change, but tomorrow there will not be a choice. Students directing their own learning will demand packages they can take home, packages which they can use to tutor their children, and packages whichcanbeeasilymodified. Wordprocessing, the tool we use daily, will be the prime tool for preparation of tutorial material.The "Q" Instruction Package succeeds in making minimal assumption o f computer familiarity.Its success will be recognised in Australia or overseas. It is to be hoped that the phrase "A man is not without honor but within lus own country" will not come to apply to the developers of the "Q" instruction package. 173 1 Interactive videodisc in the teaching of Orthopaedics in Phriotherapy. Allan Christie School of Physiotherapy 11South Australian Institute of Technology rather than a passive, process (Copeland, The use of an innovative educational me- 1981; Young, 1984; Pemberton et 1985; dium such as the interactive videodisc was Clark & Sandford, 1986). The interactive seen as a way of maximizing student learn- videodisc (IVD) system (combining the ing by using its attributes of excellent visual microcomputer with a videodisc player) images, interactivity (active student learn- was seen as an ideal medium as it presents ing), feedback, quick response time, and a flexible image-rich learning environment the potential for student-directed and stu- dent-paced learning. A videodisc contain- which has the potential to be both student- ing 1981 slides and approximately 34 min- paced and student-directed. Second year utes of video was produced in PAL format students could use the system to learn new and the authoring language PC/Pilot was material and at the same time it would be used for programming. Evaluation of the available to fourth year students to supple- effectiveness of interactive videodisc as a ment their practical experience. learning medium will be conducted using the illuminative method rather than the In accord with Baker & Ziviani (1986) the more traditional outcome-oriented or use of an information transfer system such comparative model. as ND was also considered beneficial for the following reasons; The teaching of the subject "acute ortho- paedics" is performed over two years of the having subjects available for student undergraduate physiotherapy course at observation is desirable though not SALT. Firstly, there is the expansive but always possible. necessary theory input in second year with even when they can be observed, re- little practical supplement and, secondly, peated performance for the benefit of there is the relatively large practical input student analysis cannot be guaranteed. in fourth year with little chance to revise the the financial and time costs of organiz- orthopaedic theory except from notes ing such sessions are becoming increas- taken in second year. ingly prohibitive. the instruction is consistently reproduc- Student-paced and student-directed learn- ible. ing was seen as a way of addressing these problems, however, a conventional form ofLearning Theories self-paced study using student guides/ notes was not considered suitable as muchLearning theories have their basic in two of the information to be learnt was of a principal schools of psychology Lugnitive visual nature. Also feedback and interac- and behaviourist. Bruner (1966) and Sten- tivity were considered to be important ashouse (1975) are leaders in the field of cog- there appears to be general acceptance in nitive psychology which emphasizes the the literature that learning is an active,importance of the process of learning and 17,y Using Computers Intelligently 165 learning by discovery. Behaviourist psy-and what are its salient and critical features chology was popularized by Americanwhich make its acceptability and success psychologists, Tyler (1949) and Skinner more or less likely?". In this way one would (1958). The hallmarks of this approach arebe looking at both content and quality of behavioural/performance objectives, out-the learning experience as well as the con- comes of learning and the stimulus-re-text in which learning occurs. Methods sponse model. Not surprisingly, the be-used to obtain information using the illu- haviourist approach to learning formed theminative approach include; basis of early programmed instruction methods and subsequent computer as- observation sisted learning (CAL) initiatives. Until the questionnaire (both open and closed mid-1970s much of the evaluation of CAL questions) was also based on this behaviourist ap- interview proach with the traditional hypothetico- written appraisal deductive model (in the guise of compara- tive studies) being used. There was, how-Interactive Videodisc ever, growing dissatisfaction with this approach to evaluation which sought to According to Sweeney (1985,p. 90) the fol- analyse a complex situation such as thelowing goals should be set and fulfilled learning milieu in terms of outcome only. when developing a videodisc; A few of the pitfalls of this approach in- clude; 1. utilize all optical videodisc functions in a variety of ways. a. not recognizing features salient to the2. cohesive topical program even though success or acceptability of the project it contains individual modules. because the information to look for has3. human quality demonstrated as well as already been decided at the beginning archival information storage. of the project in a way which excludes 4. high repeatability value - one or two "unsought" information. viewings should not exhaust informa- b. the assumption that if an "experimen- tion available in a sequence. tal" model is adopted that salient vari- 5. should be amenable for setting up pro- ables can be identified and controlled grams for learn-rs interested in varied for; an assumption unlikely to be met in (but related) topics and provide accessi- a real and complex situation. bility to different levels of information or expertise. Par lett and Hamilton (1977) highlighted these and other shortcomings of the "objec- Interestingly, the author independently tives" approach to evaluation in a book developed the Orthopaedic interactive "Beyond the numbers game" and pro-videodisc with similar goals in mind. Put posed a more comprehensive and descrip-simply, the aim is to use the pedagogical tive form of evaluation which they called potential of interactive videodisc to its full- fa:: illuminative method (as light was est. thrown onto the topic under considera- tion).Following on from this approachIVD Workstation Jones and O'Shea (1981) indicated that the traditional question always posed when The IVD workstation used at SAIT consists considering CAL; "How effective?", of a microcomputer, mouse, videodisc should be replaced by "How is the projectplayer, and monitor.The instructional perceived from the client's point of view design of the videodisc material required a I.75 166 Using Computers Intelligently single screen presentation so that theEconomic. Evaluation broadcast quality video pictures couid be overlaid with computer generated text and Materials colour graphics. Two different single screen configurations dubbing high band tape to have been used by the author; 1" C format tape and to U-matic low band time code 1. Computer video mixing- video mixing tapes (for off-line editing) card (Eg.,Videologic MIC-2000I + EGA) (8 hours © $160/hour) A$1,280 combined with a fast-blanking monitor use of 1" editing facility (Eg., Sony 10(-14CP1). (31 hours @ $150/hour) 4,650 1" premaster tape 144.60 Pros slide production 908.64 MIC-2000 card initializes the vide- high band tapes (17 @ $25 odisc player and so the same set of Sony KCS-20BRK) 425 commands controls all of the popu- broadcast video unit hire lar videodisc players thus simplify- (5 days © $30/day) 150 ing programming. royalties for music 80 MIC-2000 card can perform soft- single-sided videodisc editing, i.e., real-time editing of the pressing charges (Sony) 4,840 video, audio and computer signals. 10 videodiscs @ $57.85 578.50 the single card satisfies both major colour television standards; PAL A$13,056.74 and NTSC. Cons Filming and off-line editing were per- expensive - MIC -20001 + EGA formed by the audio-visual services unit of the Institute. (Quadram Quad+ rec.) A$3400 recommended monitors such as the Sony KX-14CP1 do not supportMan-hours EGA. scripting 50 hours editing videotape 60 hours 2. Monitor video mixing - e.g., Mitsubishi layout of material on disc 40 hours monitor EUM-1471A combined with cataloguing slides 60 hours EGA card. programming 200 hours Pros (approx) 410 hours monitor supports CGA/EGA/PGC. PAL video can be overlaid withThis time (approx. 410 hours) was spread computer text and graphics via the over a period of 12 months but certainly RGB analogue unit. these figures indicate that a videodisc must reasonable cost - monitor A$1600 + be produced by a team (ideally consisting EGA. of a content expert, instructional designer, Cons programmer and video producer). programming is more difficult and so there is a time/productivity cost.Hardware costs continuei decrease but videodiscs and associated software are The CAL lessons were developed using the labour-intensive products, so there is not PC/Pilot authoring language (see Appen- likely to be a decrease iu their cost in pro- dix A). portion to the hardware. 176 Using Computers Intelligently 16i Pedagogical Evaluation fo:0,"still 5"!!cr$ r: videodisccommand r: turn video on As indicated in the introduction, innova- fo:0,"video on"!!cr$ tive educational technologies have been r: videodisccommand r: turn audio on inappropriately and inadequately evalu- fo:0,"audio on"!'cr$ ated in the past due to predominance of r: videodisccommand -- "scientific" comparative studies. Conse- r: play videodisc playerfrom frame 5 quently, the author plans to evaluate this to frame 1200 fo:0,"play ,1200"!!cr$ interactive videodisc using, what could be r: videodisccommand termed a holistic approach, and producing r: wait until end of play a multifaceted description of its use in the fo:0,"wait endplay"!!cr$ learning milieu. r: videodisccommand r:-- turn video off fo:0,"video off"!!cr$ Evaluation tools will include: r: main orthopaedic menu 1 Content written appraisals j:@p *menu physiotherapists ts:m0;e7;v1,38,1,23;e0;f6 orthopaedic surgeons t: pretest/posttestsecond year MAIN MENU 1 undergraduates ts:f14 Process attitudes/opinions t: questionnaire ORTHOPAEDICS interview ts:f4 t: observation 1. Introduction computer program in/out time 2. Fractures audit trail through program 3. Fixation Appendix A 4. Main Menu 2 5. Main Menu 3 An example of rile ease of programming using the PC/Pilot authoring language is 6. Exit the following segment of code used to ts:f6 control a videodisc player connected to a t: microcomputer with a MIC-2000 card and PRESS #7.>#6 TO SELECT then displPy a computer generated mem% --ESS $7SPACE BAR#6 TO CHOOSE ANOTHER e: p: r: open MIC device drivel umenu fx:mic ts:f2;g10,7;a1. Introduction r: carriage returnis used with each *nol c:x=key(0) commandtovideodisc ts(x=13'x=49):g3,21;f7;ALoading is defined as a string r:-- player so it 1(x=13!x=49):intro for ease of programming j(x>49&x<54):number d:cr$(1) j(x=54):guit c:crS= chr(13) j(x<>32):nol videodisccommand r: ts:f4;g10,7;a1. Introduction videodisc player r: initialise ts:f2;g10,9;a2. Fractures fo:0,"init"'!cr$ *no2 c:x=key(0) command r: videodisc ts(x=13!x=50):g3,21;f7;ALoading r: search for frame 5 and stop videodisc 1(x=13!x=50):frac player 177 168 Using Computers Intelligently j(x>48&x<54):number ts:f2;g10,13;a4. Main Menu 2 j(x=54):quit j:no4 j(x<>32):no2 *no5a ts:f4;g10,9;a2. Fractures u:menu ts:f2;g10,11;a3. Fixation ts:f4;g10,13;a4. Main Menu 2 *no3 c:x=key(0) ts:f2;g10,15;a5. Main Menu 3 ts(x=13lx=51):g3,21;f7;ALoading j:no5 1(x=13!.:=51):fixat *no6a j(x>48&x<54):number u:menu j(x=54):quit ts:f4;g10,15;a5. Main Menu 3 j(x<>32):no3 ts:f2;g10,17;aE. Exit ts:f4;g10,11;a3. Fixation j:no6 ts:f2;g10,13;a4. Main Menu 2 *no4 c:x=key(0) ts (x=13 !x=52) :g3, 21; f7;ALoading References 1 (x=13!x=52) :menu2 j(x>486x<54):number j(x=54):quit Baker,J. & Ziviani,J. (1986). Interactive j(x<>32):no4 videodisc: two bites at the cherry. Pro- ts:f4;g10,13;a4. Main Menu 2 ceedings of 4th Annual Computer As- ts:f2;g10,15;a5. Main Menu 3 sistedLearning in Tertiary Education *no5 c:x=key(0) ts(x=13,!x=53):g3,21;f7;ALoading Conference, 1.delaide, pp.38-45. 1(x=13!x=53):menu3 Bruner,J.S. (1966). Toward a theory of in- j(x>48&x<54):number struction. Cambridge: Harvard Univer- j(x=54):quit sity Press. j(x<>32):no5 ts:f4;g10,15;a5. Main Menu 3 Clark,D.R. & Sandford,N. (1986). Semantic ts:f2;g10,17;aE. Exit descriptors and maps of meaning for *nob c:x=key(0) videodisc images. Programmed Learn- j(x>486x<54):number j(x=13!x=54):quit ing & Educational Technology, 23(1), 84- j(x<>32):no6 90. Copeland,P. (1981). The educational sig- *quit nificance of electronic media. In ts(x=13!x=54):f7;b0;m2 ec: F.Percival & H.Ellington (Eds.), Aspects ts:f4;g10,17;aE. Exit of educational technology, vol XV. Lon- ts:f2;g10,7;a1. Introduction don, Kogan Page, pp 231-237. j:nol Jones,A.C. & O'Shea,T. (1981). Evaluation *number methods in distance computer-assisted ts(x>486x<54):g3,21;f7;aLoading learning. In F.Percival & H.Ellington 1(x=49):intro (Eds.), Aspects of educational technol- 1(x=50):frac ogy, vol XV. London: Kogan Page, (pp 1(x=51):fixat 1(x=52):menu2 176-182). 1(x=53):menu3 Parlett,M. & Hamilton,D. (1977).Evalu- ation as illumination: a new approach to *no2a u:menu the study of innovatory programmes. In ts:f4;g10,7;a1. introduction D.Hamilton,D.Jenkins,C.King, ts:f2;g10,9;a2. Fractures B.MacDonald & M.Par1,2tt tads.), Beyond j:no2 the numbers game. Lyndon, Macmillan, *no3a pp 6-22. u:menu ts:f4;g10,9;a2. Fractures Pemberton,P., Ta) lor,J. & Toleman,M. 's:f2;g10,11;a3. Fixation (1985). Interactive videodisc and student j:no3 control of learning.Proceedings of the *no4a u:menu 3rd Annual Computer Assisted Learning ts:f4;g10,11;a3. Fixation in Tertiary Education Conference, Mel- bourne, 69-78. 178 Using Computers Intelligently 169 Skinner,B.F. (1958). Teaching machines. Science, 128, 269-277. Stenhouse,L. (1975). An introduction to curriculum research and development. London: Heinemann. Sweeney,M.A. (1985). The nurse's guide to computers. New York: Macmillan. Tyler,R.W. (1949). Basic principles of cur- riculum and instruction. Chicago: Uni- versity of Chicago Press. Young,J.I. (1984). Videodisc simulation: tomorrow's technology today. Comput- ers in the Schools, 1(2), 49-57. Allan Christie is a Lecturer in Ortho- paedics in the School of Physiotherapy, Set.ikh Australian Institute ol Technol- ogy.Using the PC/Pilot authoring language, he h 's developed several computer packages which are used by undergraduate students. He has also produced an interactive videodisc to assist student learning in the field of orthopaedics. Current research inter- ests relate to the use of the interactive videodisc and videotape systems and how to best evaluate these educational technologies in the learning milieu. Using computers in education inteigently: What happens when computers are removed from the classroom? Kelvin W. Duncan Department of Zoology University of Canterbury, New Zealand This is a case stuay of a first-year university These costs in time and money are recog- course in whit. computers w^re used ex- nized well enough and are a reason often tensively in the teaching and administra- advanced as a reason for not using CAL tion of the course until, following a change in staff a few years ago, the computing techniques in the class room. But what is component was removed from most of the not generally appreciated is that rejecting course work. This paper describes what CAL techniques also involves a cost. While happened following this change to more academics are accustomed to change in traditional ways of teaching. their research fields, it is surprisingly diffi- cult to convince them that a static approach The intelligent use of computers in class- to teaching is not a no-cost option. Lost rooms is possible provided the following opportunities for change in search of excel- desiderata are observed: They should belence and more effective teaching can be used only when appropriate. They shou' dam - ging and thus much more costly in be cost-effective, time-effective and peda the longer term than if CAL programmes gogically effective. The teaching programshad been instituted. should be comprehensible, relevant and non-trivial with a good student-machine Rather than consider the difficulties of es- interface. Costly establishment overheads tablishing computer -based teaching meth- should be minimized. Future hardware ods in a course, I wish to examine the rather incompatibilities should be avoided. novel question of what happens if the new technology is abandoned and traditional There is a general recognition that the techniques reverted to as a matter of delib- computer is here to stay in tertiary educa erate policy. Consideration of the effects of tion and that it can make a great contribu- this may give insights into such questions tion to teaching effectiveness and attrac- as the following: Is it always intelligent not tiveness.It can facilitate interaction with to use the new technology? Should we the subject matter, generating a real in-always rely on tried- and-true traditional volvement and deeper understanding. Butmethods? Is there a cost if we do take the to a teacher considering introducing CAL conservative approach? Is use of only a techniques the difficulties in establishing apart of the new technology worse than not CAL programme are very large, as are the using it at all? In short, is it intelligent not costs. Teaching with computers requiresto use the new technology? special skills, and the time required to develop and trial CAL modules can be very It is often not appreciated that the nature of great. Costs, for equipment software andtertiary teaching has changed fundamen- assistance, can be so high that they may tally in recent times, so that cow which pose another large hurdle to the develop- rely solely on traditional teaching tech- ment of CAL courseware. niques can fall into the following traps even ISO Using Computers Intelligently 171 though they may have been successful in This is especially important now that the pant: expenditure on tertiary education is being closely scrutinized. a. The course content (subject matter) may appear old-fashioned and therefore g.Skill development and learning of con- non-relevant. This is especially so if no tent may be less, often much less, than modern techniques or aids are em- that attainable in a CAL course. ployed at all. To a generation of stu- dents brought up in an environment ath. In courses taught in a traditional man- home and at school or university in ner there may develop a strong ten- which the new technology - in the form dency to nonparticipatory, passive of television, audio, computers, arcade learning leading to a tendency toward games and the like - makes up a vital surface, or shallow learning. On the part of daily life, chalk and talk can other hand, CAL can encourage partici- seem very dull. Those courses which patory learning with the development reject the new technology and employ of deep understanding. Furthermore, only traditional 'spray and pray' teach- learning in such courses can be more ing techniques may loose elective en- thorough, faster, more enjoyable, and rollments with serious consequences with more opportunities for remedial for funding if this is based on en- instruction than in courses taught in rollments. traditional ways. b. The teachers in courses relying on tradi-Let us now examine the fate of one course tional teaching techniques may appearwhich fell into these pitfalls because of a 'old fashioned' in their approach and reluctance of the teaching staff to become out of touch, leading to an impression of involved with computers and computing. mediocrity in their discipline. BIOL1OX A subject which went c. Course revisions and updates can beaway from CAL more difficult to do. On the other hand, computer-based courseware is easilyIn the early 1970's elementary biology edited and updated so the days of tattyteaching at Canterbury University, Christ- notes resulting from 'cut-and-paste'church, New Zealand, was rationalized editing can be left behind. and reorganized into more logical units. d. The teachers using traditional methodsThus BIOLIOX, amongst other biology in a course may find it difficult to takecourses, was begun.It was to present advantage of new developments ormaterial which had been taught reasonably teaching aids because they are locked in successfully earlier in a different course to a conservative form of teaching. Over format. However, the reorganization the years, they might find that the even-caused the staff to think about their effec- tual expenditure of time and effort intiveness, and so new computer-based attempts to improve and update theirmethods were tried. courses may be far greater than if CAL courseware had been developed in the Early on in the history of the course various first place. computer simulations were developed. These were extended and improved as f.Demands on teaching assistants (mark- experience accumulated. Later, the learn- ers, tutors and demonstrators) maying was modularised, and computer-based become too costly in traditional courses. tests and practice programs introduced. 181 172 Using Computers Intelligently The students were allowed to progress All the ordinary administrative tasks more-or-less at their ow pace. When they were performed by the computers. felt competent, they sat a computerized test of the material covered in that module. The This was all before micros made it much computer program devised the questions easier to carry out such work- these appli- from an approved pool devised by the cations were done in the 1970's on main- teachers; it marked the answers, kept theframes, and were largely carried out in student records and made recommenda-batch mode to minimize cost. tions to each student regarding remedial work for areas of weakness. Using a mas- The teachers of BIOLIOX were enthusiastic tery model of teaching, failing students had and worked well as a team. New ap- opportunity to resit until they achieved a proaches were encouraged and no compro- sufficient standard to advance to the next mise was made to dilute the subject matter. module. Difficult topics were attempt( -I, and usu- ally conquered. The students were encour- Thus at the end of 1978 the course madeaged to go beyond the confines of their heavy use of computers for the following course work and to work on their own to purposes: greater breadth and depth. 1. Simulation The course was very popular as was shown a. of biological phenomena impossibleby replies to independent questionnaires to demonstrate otherwise; administered by the student body. These b. of practical work done by students. surveys indicated that most students found The computer simulations acted as the course to be very worthwhile and inter- checks on the students' progress and esting, with a well-structured and logical comprehension of students; theory and practical component. 2. Aralysis The popularity of the course was also indi- rthe results of student experiments cated by the greater proportion of the class using statistical methods. taking the cour se on a voluntary basis than in any other biology course. Enrollments in 3. Practice some other biology courses were nearly as Practice programs and quizzes werehigh, or even higher, but their numbers available for non-compulsory review. were boosted by prerequisites making enrollments compulsory. The high num- 4. Testing ber of elective enrollments in BIOLIOX Using marksense cards, a system of were a matter of pride to the teaching staff. quizzes, which terminated and tested a sequence of instructional modules, was A change of scene instituted. At the end of each unit of work the student sat the test. Retests Departmental policy was for staff to be were available. Marking and compre-rotated around courses at intervals of some hensive analysis was done by com-years to "keep them from going state". In puter; the machine producing sugges- due course the staff of BIOLIOX were re- tions for remedial work for each studentplaced, in spite of their considerable oppo- if necessary. sition and reluctance. 5. Administration The new staff were far less keen on comput- Intermarker performance was moni-ers. A few simulations were retained, but tored and compensated for if necessary. much less emph.sis was given to the heavy 182 Using Computers Intelligently 173 involvement of computing in the courseThe third proposition, that the new teach- work. So it proved impossible to maintain ers were not as good as the original team, is 'mastery' type learning where each indi- debatable since the previous teachers were, vidual maintained his/her own rate ofin the opinions of their colleagues, rather progress, and their attainment and prog-bad! One was notorious for reading his ress was closely monitored and remedial lectures from a disjointed series of texts- he work set in place immediately. never wrote lecture notes and merely pre- sented a series of quotations which were The long-termed consequences for thisoften nit in a very logical order. The stu- course were disastrous. From a vibrant, dents, too, were often critical in their ques- well-regarded course it became voted "the tionnaire replies, of the lecturing style of worse course at the University of Canter-the original teachers, even when they sang bury" within a few years of the changethe praises of the course as a whole. over.This had a terrible effect on staff morale. Now the course is notable for the The most probable explanation is that the lack of cooperation between the staff, theremoval of computers caused a variety of lack of staff enthusiasm and involvement. things to happen which resulted in the Students are bored and merely do thedecline in student enrolment. Firstly, much course because they are forced to because of the old teaching programme had to be of prerequisite demands.Elective en- discarded or drastically simplified to fit the rollments have disappeared. constraints of a lecture-based course. This resulted in boredom and a lack of challenge Table 1, which gives the class sizes inamongst both students and teachers. BIOL1OX since establishment, shows dra-Learning became passive and non-partici- matically the decline in numbers of electivepatory which increased student dissatis- students in the years following removal of faction. The previous close monitoring of computers.There are, of course, other student progress and provision of feedback explanations for this decline; of whichhad to be given up, which resulted in in- three spring to mind: (1) the subject mattercreased student anxiety about their prog- may have become less popular or unfash-ress and a focussing on _narks rather than ionable; (2) there may have been changes in mastery of the subject. The students no the structure of university enrollments longer felt as though they could conquer such that the size of the elective pool be- the course incrementally, but had to master came progressively smaller; or (3) the new it all in one step for finals examination. team of teachers were not as good as theThis, too, increased student disquiet. And old. they no "longer felt that the course had a discovery learning component.Instead, There is no evidence for any change in thethey had to work at a pedestrian pace in popularity of the course.Very similarlarge classes with minimal staff help or courses at other universities have shown enthusiasm on material which seemed dull steady growth - not the dramatic decline and devoid of challenge. The staff also felt experienced at Canterbury. It is significant the changes. From a closely planned course that CAL plays an ever-increasing role in it became a set of individual teaching ef- the similar courses at other universities. forts, uncoordinated to the whole. The 'my patch' syndrome developed, and a lack of Nor is there any solid evidence for the cooperation affected every teacher. second explanation - if anything, students can take more elective courses now than in Further evidence as to what went wrong, the late 1970's since many prerequisiteshere are a selection of quotations from the have been relaxed. course reviews conducted by the students 183 174 Using Computers Intelligently and based on the replies to questionnaires. TABLE 1. Class sizes in BIOL1OX in differentyears. 1978 (the first year of these independent At the end of the 1978 academic year computers were removed from the classrooms. The column headed course assessments) "An interesting enforced enrolment gives the estimated number of course. Well structured and logical prac- students who had to take the course for prerequisite ticals and lectures." reasons. 1980 "Depressing. Course made up of Year No of Enforced Non-Compulsory dregs of all other Bio. courses. Poorly StudentsEnrolmentEnrolment (%) structured. Labs confusing. Pathetic." 1975 245 98 60.0 1976 252 101 1982 "This course should be avoided at 59.9 1977 260 103 all costs.It is incredibly boring. Labs 60.4 1978 248 badly organized, dull and irrelevant. 100 59.6 Dedication is required to complete. Only (Computers were removed at the end of 1978) bright sr t is that assessment load is light, th . is if you can fight back the 1979 270 111 52 8 boredom long enough to do some work." 1980 232 113 51.3 1981 227 126 44.5 1984 "This is the worst course at the 1982 218 115 47.2 University of Canterbury. Avoid at all 1983 176 109 38.1 costs." 1984 179 110 38.5 1985 210 135 35.7 The course enrolment is protected from 1986 125 125 0.0 becoming zero by enforced enrolment on 1987 139 139* 0.0* account of the prerequisite requirements of advanced second and third year courses. If* Estimated a student wishes to advance to these second and third year courses he must take were adopted in other courses. Following BIOL10X. We can estimate this enforcedthe changes, the flagship function disap- enrolment by adding the number of stu-peared. Finally, the enthusiasm of the old, dents in advanced courses with BIOLIOXand now disbanded team for CAL tech- as a prerequisite plus first year failures andniques was dissipated.They had been drop-outs. The figures for this are given in taken against their will away from a course the second last column of Table 1.By in which they had close personal involve- calculating the percentage of elective en-ment, and placed in new courses with new rollments (given in the last column of Tablecolleagues indifferent to, or even positively 1) it can be seen the voluntary enrolment inhostileto, computer-based teaching. the course fell rapidly as the course becameSomewhat naturally, the old enthusiasts unpopular until today there is no evidenceran out of steam. Faced with new, hostile for any elective (voluntary) enrolment atenvironments it proved difficult to main- all. tain enthusiasm for the new style of teach- ing with its great establishment overheads. The change had even more seriousconse-As a consequence, we now lag behind in quences for other biology courses.the CAL area, not merely in the particular BIOLIOX was a flagship course wheresubject matter taught in BIOLIOX, but in expe.imental methods of instruction andmost other aspects of biology as well. teaching were welcome and enthusiasti- cally trialled. After a successful triai in this It is trite to say that computLrs are here to course many very valuable innovations stay and that they have a vital and neces- 184 Using Computers Intelligently 175 sary role to play in teaching. But the con- verse of this is hot trite, and is often not recognized:to not use computers may seem like the easy option, but it is not a "no cost" one. Kelvin Duncan is Senior Lecturer in Zoology, he has been using comput- ers in teaching since 1967. He has worked with a variety of .UK and USA groups interested in CAL, in- cluding Computers in the Under- graduate Curriculum, University of London, University of Lancaster, and others. Main interest area lies in the use of computers in simulation and analysis in biology practical classes, but he has also worked with primary and mentally handicapped children. Beside major professional research work on ecological physi- ology of terrestrial invertebrates, he has been working on ways of mak- ing the computer more accessible to intellectually handicapped children in the belief that computers have a major role to play in their education. To this end he works with a number of teacher groups on a voluntary basis to develop courseware for sell(Is. Software Simulation of Microprocessor Hardware David T Edwards and Robert G Craig, School of Engineering Darling Downs Institute of Advanced Education later but near identical version of the D3 This paper reports on a TurboPascal lan- board. guage simulation program running on personal computers being developed to One of the authors (Edwards) has devel- aid the teaching of microprocessor funda- oped a software simulator for this Motorola mentals to both internal and external un- D3 (and D5) kit. This program, written in dergraduate Electrical Engineering stu- the TurboPascal language, is intended for dents of the Darling Downs Institute of Advanced Education. Traditional teach- running on an IBM-PC or equivalent. ing has used a Motorola MC6802 based Access to such desk top computers is made single board development system (the D3 available to students in the D,D.I.A.E. ex- or D5i, where program debugging is ternal study centres located throughout achieved by halting the program and se- Queensland. quentially examining the CPU registers. The enhanced display capabilities of this Compared to the original hardware the simulation package allow students to si- simulator has a number of significant en- multaneously view memory and register hancements designed to expedite the contents as a machine language program teaching/learning process. An early ver- executes. Preliminary trials with internal sion of the simulator was used on a trial students indicate a significant speed up factor in the learning process. basis in first semester 1987 with an internal associate diploma class. In second semes- ter 1987 the simulator is being used with an external associate diploma group and an The D.D.I.A.E. School of Engineering has internal degree group. for a number of years been teaching micro- processor fundamentals to students en- This paper will briefly describe the existing rolled externally in the Associate Diploma hardware system. The simulator software in Electrical Engineering.To developand the results of the first semester trials hands on skills, these students have at- will then be discussed. tended an on campus residential schoo.. They have then carried out a programmingObjectives of developing the 6802 assignment using a microprocessor devel-Simulator opment "kit in a suitcase" L the D3 kit based cr. the Motorola MC6802 microprocessor. The kit was collected at residential school The principal objectives of introducing the and "passed around" among a group of6802 simulator were twofold. students to complete their assignment(s). Internal associate diploma and degree stu-1. To provide the same learning opportu dents use a similar microprocessor devel- nities for external students as exist for opment system, the D5 board, which is a internal students. 18G Using Computers Intelligently 177 As internal students have class access to the A ROM monitor which performs house- hardware for approximately 22 hours, and keeping and provides the user with a a large amount of "out-of-hours" access, number of utility routines. spread over the whole semester, external 256 bytes of RAM for user program and students should have a similar amount of data storage. "hands-on" access. The present "pass-the- A keypad for inputing either hexadeci- kit" situation for external students limits mal code or invoking some of the moni- their hardware access to about 10 hours tor routines. "driving instructions" at residential school An array of 7-segment digits to display plus one intense burst when they have a kit both user and monitor information. at home. To provide each external student A tape input/output port for magnetic with a kit of Their own is prohibitively storage. expensive at a component cost of approxi- An RS232 interface for serial communi- mately $1000 per kit plus many hours of cations ( ACIA ). technician time both to build up the :.its A programmable interface for parallel into a self-contained entity and to service communications ( PIA ). them when faults occur. A reset button which clears the CPU registers and returns control to the sys- 2. To improve the laboratory support for tem monitor. the teaching of microprocessor funda- mentals. The above configuration permits the stu- dents to progress from the design of simple Students find difficulty in coming to gripsdata manipulation programs to the design with many of the essentials of microproces-and implementation of more complex real- sor programming. Many find conceptstime algorithms such as digital filters and such as flags, program counter, stack interrupt driven feedback control. Unblie . pointer and indexing quite difficult to fered access is available to the system ad- grasp. Displaying these and other features dre-s, control and data buses enabling such simultaneously should enhance the stu- operations as DMA, or the accessing of the dents' learning. digital output of an A/D converter directly onto the system bus, as alternative input/ The existing hardware has no convenientlyoutput mechanisms. accessible off-line program storage so stu- dents must either key in their programsSpecification of the D3 simulator each time by hand or down-load from one of the Institute's main-frame computers1. The simulator should have the same (an option not available to external stu- display as the actual D3 kit. dents' Easy access to off-line program stor- age is a needed feature. This is a 6 digit 7-segment hexadecimal display with the first 4 digits for display of Description of existing hardware the address of a memory location and the last two digits for display of the contents of The D3 kit i, a combination of hardware that memory location. (see figure 1). and firmware designed to assist with the understanding of the Motorola MC68022. The keyboard input to the simulator microprocessor and commonly associated should mimic the keyboard input to the components. The system as configured by D3 kit.The D3 keypad has 25 keys the School of Engineering has the following arranged in a 5 by 5 matrix (see figure 2). facilities :- 187 178 Using Computers Intelligently ADDRESS DATA ous memory location when editing EX invoke 'prompt' without resetting the CPU registers; or cease the exe- cution of a running program RD CPU register display Figure 1: Data display GOrun a program; or allow access to 1.=1,, next memory location when editing RS FS F C 3. All needed Motorola 6802 opcodes should be simulated. 7 8 9 M ( The only opcodes not implemented are those that refer to interrupts :- RTI, WAI , SWI. ) 4 5 6 iEX .111111* 4. Sufficient of the Motorola D3 monitor `41 routines as is necessary for students to 2 RD complete their assignments should be implemented. D GO Many of the monitor routines are effec- tively implemented through the keyboard keys. In addition the following monitor Figure 2: D3 Keypad routines were required: The 16 keys for inputting hexadecimal GET waits for a key to be pressed digits (0...9,A...F), for address or memory and then loads that key's contents should be implemented by the "value" into accumulator A. corresponding keys on the simulator mi- DYSCOD takes hexadecimal values crocomputer keyboard. The remaining 9 stored in a buffer (HEXBUF) keys perform housekeeping by accessing and produces a 7-segment the Motorola monitor. Since these keys display code for each digit in have a multi-character label, a suitable the buffer ( DISBUF). single keyboard character would have to be PUT displays on the output dis- found to be used on the simulator. The D3 play the 7-segment displays function keys have the following actions: stored in (DISBUF). PROMPT returns to the prompt and al- KeyFunction lows further monitor instruc- tions. (The last action at the RS reset the CPU registers and invoke end of a machine language the monitor 'prompt' routine program). FS select alternative key function FC clear alternative key function 5. Compared to the D3 hardware the P/Ltape store/load simulator should have enhanced dis- T/Bstep a machine language program play characteristics. The standard dis- one instruction; or insert a break- play of numeric information should be point in hexadecimal. The display on the M allow editing of tne memory loca- simulator screen should include, in tion the address of which has been addition to the standard display of the entered; or allow access to the previ- hardware: 8 8 Using Computers Intelligently 179 The contents of all registers. The con- TurboPascal compilers are readily tents of the condition code (flag) register available for a number of different desk- should be displayed both in hexadeci- top computers. mal and in binary. The binary display makes it easier to see which of the 6 flags This means the --)urce code can be ported have been set. from one n47 to another using a file The current position of the program transfer prograw like "Kermit" and then counter (PC) and the stack pointer (SP). recompiled on the new target machine. The contents of the page (256 bytes) of (The original simulator program was de- user memory. This display should dis- veloped on an NEC-APC 1 running CP/M- tinguish between values that have been 86 and then ported to the MS-DOS IBM-PC set (or defined) by the user as program compatibles). or data and the "random" values that the memory locations have either from The Electrical Engineering laboratories at previous use or switch-on. ( This option the D.D.I.A.E. have a mixture of NEC-APC was chosen to unclutter Cie display. ) l's and CCS IBM-PC compatibles. The ex- Any changes to SP, PC, register orternal study centres have a mixture of memory contents caused by an instruc-Sperry, Hitachi and CCS IBM-PC compat- tion being executed should be high-ible computers. For the wide range of tar- lighted. get computers, there was a need to make the system as machine independent as 6. There should be a capability to load possible. All screen displays were imple- machine language programs (memorymented using only standard ASCII charac- contents) from disk and also to save to ters for compatibility reasons. The only disk. "graphics" capability used is screen high- lighting to improve the readability of the 7. On-line help should be available. display. For external students all the informationThe program consists of almost 3000 lines required to use the simulator should bzof Pascal source code divided into almost available on the single floppy disk. 100 procedures and functions. Half of the procedures were required to implement 8. All functions of the simulator should bethe 6802 opcodes. An "Include" file is used "menu-driven" with single key stroketo handle the different program require- choices at each stage. ments of the CP/M and MS-DOS ma- chines. (This only occurs to handle differ- The simulator program should be "crash-ing screen highlighting requirements im- proof" with the user protected agaimiposed by different "IBM-PC" graphics making mistakes that could cause prob-screen/board combinations.) lems. An "Include" file is also used to handle the Description of the simulator different requirements of the D3 kit as used program by external students and the D5 kit used by internal students. The differences are :- the The program has been developed using the page of memory available ($00 on D3, $E3 TurboPascal language. This was used for a on D5) and the addresses of the monitor number of reasons including: routines and buffers. The friendly development environ- To meet the design criteria the following ment. principles were adopted: 1 8 8 180 Using Computers Intelligently 1. A portion of the lower left-hand cornerb. Provision of a number of sample pro- of the simulator display mimics the 6 grams (exercises) together with an ex- digit display window of the D3 kit. planation of their operation. 2. Most of the 25 keys of the D3 kit keypad All help files are plain text files that can be have been implemented. modified using any standard editor with- out the need to recompile the simulator SimulatorD3 key Function program. key 4. Display enhancements. 0...9,A...F"same" hexadecimal digits Screen highlighting is used to make more RS reset readable and obvious any changes to the T/B single step display caused by the execution of an in- M memory struction. Because different hardware/ GO go screen combinations give different effects EX return to promptwhen highlighting is used, the user is of- F/S - T/Bset /clear fered a choice of three different highlight- breakpoint ing options:-higher intensity, inverse video, or none. As well for the simulator keyboard there is 5. Loading and Saving. a "hidden" menu of simulator only keys. When loading or saving a file the user is L load allows a machine language offered a choice of file names from a menu. program to be loaded from This limited choice of file names avoids the disk problem of users forgetting their file names H help displays help screens or choosing file names that dash with S save saves a machine language simulator system files. Enhancements of program to disk the simulator over the "real thing" include Z zero dears the simulator display ability to save a program to disk, on-line Q quit quit the simulator help available and much more informative P peek allows the contents of a singledisplay capabilities (see figure 2). memory location (or register) to be displayed in binary as memory crintents of a full page well as the standard hexadeci- simultaneous display of all registers mal disassembly of the last instruction exe- 0 observedisplays the contents of the cuted more commonly used moni- dynamic display of program counter tor buffers and stack pointer positions Y display a trace of previously highlighting of changes caused by last executed machine language instruction instructions (up to 20 since ability to examine memory contents in last reset) binary rather than hexadecimal ability to see trace of last 20 instructions 3. Help is available in two forms. distinction on screen between defined and un-defined memory contents a.On-line Help screens to explain specific simultaneous display of position of all feat Tres e: the simulator. breakpoints Using Computers Intelligently 181 Wiressiog UN to Off I Limitations of the simulator. Pap tIM t350 0: . . . 34 II: 14. 3104 U* El X .. 1. Only one page (256 bytes) of user pc 4313 memory available. The D3 kit as used by external students has only the single " " N 14, zero page available. A multipage D5 74: MI version of the simulator was tested but If: 68: . sp was felt to be confusing with the need to ,,, " " statooLI switch between pages of display as the 111-3YL areas of memory being accessed "' ..< 110110 :. . CII 2 3 4 34.I 11111CIEF changed. ------*ma lito 2. I/O is not simulated. Programs cannot interact with the real world. I Comrloit icw, s ..,c* 4.4.1.I.E. ihae 3. Interrupts are not simulated. Figure 4 shows the screen after 'he first instruction has been carried out. The disas- 4. Only a limited number of the monitor sembled instruction (LDAA) displayed in routines are simulated. top right-hand corner and screen changes shown highlighted (underlined in figure). 5. On start up the program counter and stack pointer are initialised to the top and bottom of memory respectively rather than being unspecified. Pogo U Wend,* I CIO to 4333 3* UN Operation of the simulator Figures 3 to 7 show the computer screen as MI " " " 4"b 31 a stored machine language program is run. IT .00« Figure 3 (below) shows the screen with the : tht. program loaded from disk and the starting address ($E310) keyed in. 3 3 4 3 ma scot7 01rtss Pato Pogo 0 Mitoostal4344 to MI 1 ! ! o. . M El 3* 14 32 is c34 II X Figure 5 shows screen after jump to subrou- .. pc 304 34, E.1 0 34 tine (JSR) with stack loaded (bottom right- *II II 00 44: hand corner). KO 00 07s: . . ores** Figure 6 (on the next page) shows screen at end of program (HALT used to end siml ila- s' " " .7:3 tor programs). . 01114C . 1103040 I I 2 3 4 3 I 11ICIEF 11111111 191 182 Using Computers Intelligently hp EI kitten* 1 UN to Ur Mll years who did not use the simulator, led to 111 M 1:3M f1E3 33 31 )3C the conclusion that the simulator users had " " " " " " " "II picked up the basics of microprocessor Ni a operation in about half the usual time. mments from students who had previ- N ously studied this or the similar degree level imit were usually that the enhanced display capabilities were of significantas- sistance. Small group interviews ofran- ...... 11its 11111111 domly selected students elicited 3 a number Alms iota of possible improvements for the simula- tor. These included: 1. Providing a selection of simple (sample) programs that demonstrate the usage of Figure 7 shows the trace option with all various microprocessor instructions program steps being displayed. (especially the monitor read and dis- play routines). This has now been incor- Intractias Neemelc EA s 1 cc sa pc porated in the latest version. address 2. Tying in a 6802 cross-assembler w,th the ENO Lltti E331 1000 14 M I EVE E313 E313 EMI E332 0110 14 31 ENT ON capability to download to the D3/D5 1:316 411 ENO NCO 14 37 EMI UM E331 MN 010 11 37 11 N 09 EMI kit. This assembler is now available for E331 VANUN 0010 El 31 N I MI U34 use by semester 2 students. BM ITS UN 400 El 31 N N ENT EMI 1311 IV 0000 El 31 N I EMT MA DIA 1Nll 1101 El 31 N N ENT EMS Conclusion The simulator differ3from those normally Rest.its of student trials found in microprocessor development systems in that it was primarilydesigned as The simulator was tialed during first a teaching aid and not as a software devel- semester 1987 with students taking the unitopment tool. The consequence cf this de- 77201 - Engineering Computers. This unit, sign philosophy is the great emphasis that taught by one of the authors (Craig), intro- has been placed on such dynamic features duces students to computer numbersys- as the stack pointer, condition code and tems; computer architecture; machine andmost importantly the program counter assembly language programming; input/(PC). Being able to watch the PC move output methods and interfacing tech- through a program (with the registers and niques. memory contents changing as appropriate) has been of great benefit; a conclusion In 1987 the students received their intro- supported by the very favourable reaction duction to the 6802 microprocessoron thefrom the student users. simulator. Sole instruction on the simula- tor was given for about 6 weeks and then the students were introduced to the actualAcknowledgements D5 kit so they could begin using the serial and parallel I/O ports. Tilt: authors would like to acknowledge the assistance of fellow engineering staff Comparison of the progress of this group of memb,-_,rs who provided valuable input in students with students from previous the initial stages of the simulator design. 192 Using Computers Intelligently 183 David Edwards is currently Lecturer in School of Engineerir g DDIAE. Teaching interests include classical control engineering and structured programming. Interested in the devel- opment of teaching aids for students (both internal and off-campus) using video and print media as well as CML and CAL. Robert Craig, has been a Lecturer in School of Engineering DDIAE since 1985. Previously involved with mili- tary electronics in UK. Main fields of interest are digital design with CAE and microprocessor based equipment for the mentally and physically handi- capped. Also interested in the devel- opment of CAL packages for distance education. 1 3 3 Innovative Applications of Computer- Assisted Learning. Margaret Farrow School of Occupational Therapy South Australian Institute of Technology being dependent upon teachers acquiring This paper challenges a number of controversial knowledge, experience and competency in epistemological issues inherent in the common model of Computer-Assisted Learning (CAL) in authoring skills and that 'armed with these tertiary education in which individuals or small skills, the author has the opportunity to be groups of students sit at a terminal and learn creative in design by using the instruc- from material which presented to them, that is, tional versatility of the technology' (Sims, the computer(in essence) being used in the traditional 'tutor' role. The central concern is 1986). He specifically mentions the need that this model 'dramatizes the absurdity of for imagination in utilizing the interactive continuing to think of computers,or about capabilities of the learning medium. Inter- education itself in traditional ways' (Papert, active qualities are frequently discarded in 1979). Two innovative applications of CAL which address these theoretical considerations the application of the computer as a tutor, of education will be outlined. whether the 'mode' be Tutorial, Drill and Practice, Simulation or Testing techniques. The development of a framework based on an unsophisticated model which is being used at earlier letels of education enables the use of the It seems to me that the dilemma fz,ed in computer as a 'tutee', and as a means of aiding striving to achieve the goal of high quality and stimulating tertiary students' thinking. courseware comes largely from the notion The subsequent application of the 'tutee ' frame- work to teacher education is proposed as a that the starting point for the development means of encouraging teachers to develop an of all courseware requires the application interest in using the computer as tutee, rather of either an authoring language or an au- than just as a tutor and so become a more useful thoring system. The starting point, I con- educational tool. The two applications, devel- oped through co-operation between two educa- tend, is the nature of the knowledge area in tors in different vocational disciplines demon- which the educator identifies a need to strates the value of the CAL educator's ability to facilitate or enhancestudents' learning apply imagination and innovation to instruc- ability .A statement of the educational tional design. objectives underlying the need to learn the particular subject material is essential as a Replication of traditional instructional means of defining the nature of the infc r- models rather than innovative varia tions of mation or knowledge. It is necessary to go these techniques has been suggested (Pa-back as far as defining the very beginning pert, 1979; Sims, 1986) as a reason behindeducational concepts, such as, why it is the failure of computers to achieve their necessary for students to acquire this infor- potential as an instructional resource. Sims mation. Having locked ourselves into the (1986, p. 262) states that 'The versatility of notion that authoring languages and sys- the computer has been neglected through tems are the cornerstone for computer -as- it's use as a surrogate instructor'. Rightly,sisted learning it appears that we may be on he argues the need for the training of teach-a misguided mission as we attempt to find ers in the use of computers as instructionalsubject materials which we can apply to tools. Courseware quality, he describes as these authoring tools. 11) 4 Using Computers Intelligently 185 As an educator concerned about factors the information to be learned (Steen, 1971; which affect the incentives, efficiency and Young & Austin, 1979). effectiveness of studits' learning, I feel a responsibility toecognize the areas ofWhen considering the most appropriate learning which students find particularly CAL mode for this educational problem, it difficult due either to monotony, complex- was felt that whilst the preferred choice, a ity or unfamiliarity with the material ordrill and practice programme would pro- concepts. Having recognized the educa- vide students with the opportunity to prac- tional problem, my responsibility extends tice learning and using medical terminol- to exploration of the most appropriate ogy, three major problems occur red in the means to alleviate the difficulties. In the application of this learning mode. Firstly, production of the interactive video lessons the number of medical terms and their required to provide exposure to real client abbreviations are in the region of thou- examples for Health Science students, the sands and vary both nationally and inter- importance of an authoring language (PC nationally. Secondly, there are multiple Pilot) was indeed recognized(Farrow, examples for the prefixes and suffixes used 1986). In this instance, the identified educa- to make up medical terminology and tional need was for students to transfer coupled with both ordinary speech words static information in the form of facts, dia-and medical words, the inclusion of exten- grams and illustrations into a dynamic, sive examples would necessitate the link- motio.1 based concept as a means of inte- ing of unrealistically large data files to a grating principles, and facts into clinical drill and practice lesson. To give examples practice. to these problems:- Recently, a need has been identified to in- 1. With respect to medical abbreviations; crease motivation within Health Science in different contexts, 'LE' may mean left students to learn medical vocabulary and eye, lower extremity or lupus erythe- the relevant abbreviations which com- -itosus. monly appear in client files, medical re-2. With respect to medical terminology ports and student lecture notes. The educa- the prefix 'ab' means away from and is tional problem is that the content is boring used in the words abductor ( leading and requires rote learning, not unlike learn- away from), aboral ( away from mouth) ing to spell in primary school. Although and many more. some aspects of the terms do not refer spe-3. A drill and practice lesson is usually cifically to medical or scientific terminol- undertaken individually by students, ogy (Young & Austin, 1979, p. 8) as they are an aspect which as an educator, I believe a part of words in ordinary speech, unfor- CAL has the potential to overuse with tunately, until these terms are learned in a detrimental effects. medical context as well as learning the additional specific medical terms, the stu- An exciting program, Shrink 'n' Stretch dent is unable to even begin to understand (WESOFT,1983) is designed on the 'tutee' terms used in subjects such as basic anat- (Taylor, 1980, p. 4) in that the computer omy, physiology and the aetiology ofpresents a contraction such as 'wasn't', to medical conditions. There exists an abun-be matched by the expanded form, in this dance of texts which contain the facts to be case 'was not'. The student is then asked to learned, the educational need was simplytest the computer on a contraction known to find a motivating and interesting way ofby the student. Through this form of scor- practising the use of the information and ing game, in which the student scores if the not one of collecting and collating together computer either cannot match the contras- , 5 186 Using Computers Intelligently tion presented by the student, or the the data file increases similarly. It could be computer's presentation is matched cor-argued that the 'framework' is in itself an rectly, the student is contributing, initiat-authoring language, however, the innova- ing (Hancock, 1985, p. 21) and drawing ontive element of the program is the capacity knowledge already learned rather than for the student to teach the computer, thus simply following instructions. The pro-eliminating the need for the computer to gram has extensive interactive aspects, an hold an exhaustive store of correct facts element frequently discarded from other- which must be 'judged" against student wise excellent courseware (Sims, 1986). Itresponses. appeared that, if used by small groups of students, there existed in this type of pro-The program has no facility to detect an gram, the potential for each student toincorrect answer which has not been in- learn something about how his or her owncluded as one of the initial pertinent ex- thinking works' (Taylor, 1980, p. 4) com- amples in the data file except by the educa- pared with other students at the same aca- tor regularly checking the data file as it demic level. Having observed the enthusi-builds up with further student examples. asm with which numerous young adultsTo add an exhaustive data file of correct (the age applicable to the majority ofanswers against which students could today's tertiary students) not only enjoycheck their own answers would defeat the arcade games but will work apparentlyinnovative and exciting aspect of the pro- tirelessly to beat their previous scores, an gram and once again place the computer in attempt to design an adult version of a the tutor mode. 'Skrink 'n' Stretch' seemed worthwhile. Most CAL materials produced are based on Discussion about the feasibility of writingthe five commonly recognized modes and the proposed program led to an arrange-wl-ether developed using authoring lan- ment whereby the writing of a 'framework' guages or authoring systems continue to for the program was undertaken by anperpetuate the tendency for the CAL au- educator in the field of teacher education.thor to place the computer in the 'tutor' The basis for this arrangement was that therole, the traditional model of the teacher in 'framework' of the program could then be the educational process. Imagination and available for trainee teachers to use toinnovation have been applied to the in- develop confidence in using computers instructional design of a program designed their regular teaching (Anderson 1984, p.to overcome the motivational difficulties 88). By teachers developing and evaluat-experienced by tertiary students in learn- ing i n the classroom programs based on the ing basic medical terminology and the rele- 'tutee' model for topics such as capitalvant abbreviations. Through it's develop- cities matched to states or countries there isment in co-operation with an educator a potential for teachers to become clearer asinvolved in training teachers and the sub- to why they would introduce the computer sequent application of the program 'frdme- to the classroom as an educational tool work' as a means to encourage teachers to (Anderson, 1985). explore potential uses of computers in the classroom, the applications of this tutee Written in basic for the Apple GS, the pro- 'framework' challenges a number of con- gram is short and simple and links to a data troversial epistemological issues inherent file into which the educator inserts a lim- in the common CALITE model. ited number of clear or pertinent examples (between 20a i 100) of subject content. As the student(s) add more known examplq Using Computers Intelligently 187 References Margaret Farrow has been a lecturer in Anderson, J. (1984) C. mputing in Australian the School of Occupational Therapy, schools: An Australian perspective. S.A. Institute of Technology for past 9 Australian Education Review No. 21. years in areas of physical dysfunction Hawthorn, Victoria: Australian Coun- and therapeutic applications. In 1985, cil for Educational Research. developed an interest in CAL through Anderson, J. (1985) Computers in the lan- a research project aimed at increasing guage classroom. Perth: Australian Read- both self instructional learning oppor- ing Association. tunities and exposure to actual clinical Farrow, M. (1986) Interactive video as a problems (via videotape) for Occupa- specific training medium for health sci- tional Therapy students. With author- ence students. In Bishop, G. a van Lint, ing assistance from Rod Sims, she has W.,Proceedings of the Fourth Annual produced an Interactive video lesson Computer Assisted Learning Conference in (using PC Pilot) which was Walled Tertiary Educar.Conference, Adelaide: with students in 1986, the findings ASCILITE. being presented at CALITE the same Hancock, J. ( 1985) Computers for Learn- year. Currently a member of the S.A. ing. In J. Anderson (ed.) Computers in the Institute of technology's Computer- language cla,sroom. Perth: Australian Based Education Subsommittee of Reading Association. Academic Computing Bard and is Papert, S. (1979) Computers and learning. working towards the establishment of In Dertouzos, M. and Moses, J. (Eds.) a CAL Unit. Continuing to upgrade The computer age: A twenty year view. and produce Interactive Video lessons Massachusetts: MIT Press. for Health Science students at the Insti- Steen, E. (1971) Dictionary of abbreviations in tute. In early November, 1987 partici- medicine and the related sciences. London: pating in the ADCIS Conference in Bailliere, Tindall and Cassell. Oakland, California, U.S.A. Sims, R., (1986) On the Development of high quality courseware. In Bishop, G. & van Lint, W., Proceedings of the fourth annual computer assisted learning confer- ence intertiary education conference, Adelaide: ASCILITE. WESOFT (1983) Skrink 'n' Stretch in punctuation. Perth: Schools Computer Software, Education Department of Western Australia. Young, C & Austin, M. (1979) Learning medical terminology, step by step.St Louis: Mosby Co. 1 9 7 Computer managed learning in Physiology laboratory classes R.E. Kemm Physiology Department, University of Melbourne. acquired than with more expc Isive mod- The Physiology Departmor: has been in- em conventional physiological apparatus. vestigating the enhancement of its under- The greater versatility of computer-based graduate laboratory teaching by the use of equipment would allow new experiments computer based apparatus over a number to be performed with a more stimulating of years (Delbridge & Kemm, 1985) and has variety of investigations of physiological now drawn up specifications of its require- phenomena in a standard class arrange- ments for such a 'Laboratory Machine' fol- ment. lowing extensive testing of hardware and software. This paper discusses the results of these investigations using peripheral This paper is intended to discuss the real interfaces connected to the Apple Be and problems in the upgrading of the labora- IBM-PC computers as well as the final tory equipment used in the teaching of specifications for the 'Laboratory Machine' undergraduate students in a l'hysiology for which funding is now being sought. department. We are faced with the com- mon problem in many Universities of The overall requirement for the machine is upgrading our laboratory equipment for versatility, so that it can act as an intelligent undergraduate students in the fact of a instrument (incorporating chart recorder declining funding and decreasing purchas- and oscilloscope functions) which can ing power fur imported apparatus. Costing measure a wide range of physiological parameters, as a analyzer of the data gath- of the replacement of our 20-year old obso- ered, as a simulator of experiments which lete and potentially unsafe equipment by are too difficult to mount in an undergradu- conventional apparatus puts it well be- ate laboratory, and for computer assisted yond our means, even if we phase in the learning functions to relate and test the purchases over many years. students knowledge of lecture material. A fully-integrated computer managed learn- We were led to investigate the use of com- ing package is required to allow the stu- puter based laboratory apparatus by the dents to swap between any of these modes discovery of inexpensive digital oscillo- of operation, without requiring significant scopes based on the Apple He microcom- computer experienre. puter, which we could see as the central data gathering device for most of our labo- ratory experiments.The installation of Enhanced student teaching in and ergradu-such equipment has been carefully consid- ate laboratories is the aim oft epla :ement of ered in a staged programme over a number obsolete practical class equipment by com-of years. We thus purchased In Apple He pute; -based apparatus, referrz.d to as theComputerscope for evalu;ion and and Laboratory Machine'.The 'Laboratory then ran a pilot project to investigate its use Machine' would permit a more extensive in a typical class experiment using four range of scientific laboratory skills to besuchinstruments(Delbridge & 198 Using Computers Intelligently 189 Kemm,1985). We were able to compare on compromises between cost and facilities costs of conventional and computer based offered as well as the ability of the machine apparatus, putting us in a good position toto operate a significant range of commer- argue their relative merits when seekingcially available simulation software and funding and also to justify the develop-other packages. Similarly, the choice of a ment and evaluation of a prototype whichnetwork and file serving computer to dis- would allow the development of specifica- tribute software and data should also be tions for the final 'Laboratory Machine'. delayed. Connected to the computer, we The testing phase using the IBM PC as a prefer a free standing unit which houses all reference machine is virtually complete. Itthe electronic couplers and amplifiers, so is felt that it would be most cost effective to that the computer can select the appropri- have a single company tender for the deliv- ate devices required in particular experi- ery of a whole system to our specifications, ments, without the time consuming task of so that all components of the 'Laboratory the technical staff having to setup the appa- Machine' are integrated into a coherentratus for particular experiments. This is package and so that a commercially viable especially important when the apparatus completed package is available. will be in much more intensive use than many separate pieces of conventional ap- The Department of Physiology and the paratus. Medical Faculty of the University of Mel- bourne have provided approximately Overall specifications $50,000 over the last four years for software and hardware. Two academic and oneThe 'Laboratory Machi7-.Y comprises a technical staff member have also had acomputer with its operating and applica- significant involvement in the project in tions software, together with amplifiers this period. IBM have also been of consid- and electronic coupling and the devices erable assistance by providing up to four which measure or control the physiological computers for development and evalu- parameters being examined. The 'Labora- ation. tory Machine' is required to operated in the wides, range of physiological studies, us- Specifications of the laboratory ing an integrated software package to use machine the full facilities of the machine. This computer managed learning package We have drawn up a detailed specification would have four major areas of operation of the minimum performance required of the 'Laboratory Machine', although only in the laboratory class. These are: the general specifications are included1. As an instrument for data acquisition here. It describes the requirements of the and control in physiological experi- transducers, amplifiers and couplers, elec- ments, operating as an oscilloscope, trical stimulators, data sampling rates and chart recorder and stimulator.The precision, data logging and review, display chart recorder function 1,uld inuuue a requirements, data storage and data re- rolling screen display and a slow data- view facilities. Some of these features are logging mode direct to an ink-jet covered in Appendix A. printer. Data would be directly stored to disk for later review or hardcopy. So Although the IBM PC was used as a proto- that this one instrument can replace a type, the computer to be used is not sped- variety of standard physiological appa- fled since changes in such hardware are ratus, it must be versatile and easy to very rapid and the final choice will depend setup and use in a wide range of labora- 19 9 190 Using Computers Intelligently tory situations. A range of electronicbly mouse-driven, to give optimumuser couplers and transducers would beinteraction.Changes in parameters for required. The computer should be ableprinting, display, data acquisition rates, to control the sensitivity of the couplersstimulus parameters are to be be easily or amplifiers. made using on screen displays. A colour monitor is required, particularly for dis- 2. Analysis of experimental data using plays of multiple graphs a nd in simulations statistical tests, curve fitting and com-involving complex diagrams.Data ac- parisons with class or published data. quired from experiment or simulationmay In many experiments data gatheringrequire further analysis, which is not pro- would proceed slowly in the back-vided in the package supplied. The data ground over a long penod, allowingshould be able to be exported to standard ccncurrent operation of a progressive"user-friendly" packages. review and manipulation of previously acquired data Development of the specifications of the laboratory machine 3. Computer-assisted learning asan aid to directing students towards various expenmental investigations and to re- There have been two major phases ',ri the development of our detailed specifications. lating material presented in the lecture course. These were the pilot project involving the Apple He in a typical physiology class and 4. Simulation of experiments where the then using the IBM PC as a testbed for new procedures are too difficult or expen-software and hardware which would be used in the wide range of physiological sive for an -_ndergraduate class. This is seen as an adjunct to real laboratorymeasurements. experience and not a replacement. The Pilot project using the Apple lie computer-managed learning package must be able to swap easily between the Aims four areas, with data gathered during the experiment able to be examinedor operated upon in any other area. The specific questions we sought to answer in the pilot project were:Could the com- Operation of this management packageputer-based apparatus perform as well or must be easy for lecturers and tutors, who better than conventional apparatus as a will be required to link various routinesdata acquisition instrument? Was the within a computer-assisted lessor. These apparatus acceptable and easy to operate lessons would include gathering of databy undergraduate students? Could com- from experiments or simulations, student puter managed learning improve student analysis of data, and questioning the stu-understanding of the experiments and dent before and after the experiments.their relation to lecture material? Were These academics can be expected to besimulated experiments a useful supple- familiar with using computers with stan- ment? dard software packages, but should not beResults expected to have any professionalpro- gramming skills. We found reasonable answers to the above questions and indications of future direc- Management software should be based ontior.s which might be appropriate for fully a Window and Menuing system, prefera- developing computer based apparatus. 200 Using Computers Intelligently 191 As an data acquisition instrument, the software for user customisation and lack of Apple Computerscope performed betterlocal support from suppliers. It was also than an affordable digital oscilloscope. The difficult to modify the software to present provision of a hard-copy of the resultssimple and relevant screen labeling for during experiments was the highlight of its student use. operation for many students. The pilot project only looked at one type of The majority of the students were confident experiment which is normally performed in the operation of the instrument after ain undergraduate physiology classes. brief familiarisation lesson on the com-There are a wide range of parameters puter, although some students still had an which are normally measured it, animal aversion for computers. (Its operation was and human experimental physiology and simplified from the commercially available these are indicated by the descriptions of package by using a specially encoded key- the current experiments (Appendix B) and pad which allowed access to a limitedby the range of transducers and couplers range of the functions available to an expe-which would be required in the future rienced user of oscilloscopes. The software (Appendix A). was also configured so that it started up in the appropriate mode for particular experi-It was concluded that the Apple He was not ments.) a suitable machine for a major re-invest- ment because of declining support for data The students did show a better apprecia-acquisition modules and because better tion of the experiments they were perform- software and hardware was becoming ing and their relationship to lecture mate- available for the IBM PC and Apple Macin- although the it was not possible to tosh. It was decided to proceed with the distinguish between dependence of this investigation on the IBM PC because of its improvement on the computer managed open-architecture, colour facilities and the learning component or greater interest in interest in local software groups in devel- the project produced by its novelty. oping both the data acquisition and com- puter aided-learning facilities. The simulated experiments were both popular and usenil, since they bridged the A prototype system based on the IBM PC gap between the lecture material which was developed to demonstrate the remain- was more closely related to the data ana- ing facility which was not available ( a 4- lysed in the simulation and the less directchannel chart recorder) and to test the measurements on the phenomena whichtransducers and couplers which would be could be investigated with class equip-required for a full range of class experi- ment. ments. Conclusions from pilot project Development and performance of a prototype laboratory machine (based on The major criticism of the apparatus used the IBM PC) in the pilot project was that the software available on the Apple Ile for the individual The prototype was designed to answer two components of computer-managed learn- broad questions: ing exercise (data acquisition, computer- 1. Could we source the hardware and assisted learning and simulation) could not software, other than the computer, be integrated on the Apple He because of within Australia? incompatible languages, unsuitabilty of2. Could the prototype machine provide 192 Using Computers Intelligently satisfactory performance so thatwe unsuitable on price as well as complexity of could confidently demonstrate to fund- operation. ing authorities that therewas a better alternative to conventional laboratory Software performance apparatus. Laboratory Software Associates (LSA of The specific tests we had to performto Melbourne) have produceda pre-release satisfy these requirements were: package which shows the feasibility ofa 4- channel vertical chart-recorderon an IBM- 1. to evaluate the existing hardware andPC. We tested this extensively in parallel software which would permit dataac- with a conventional Grass 4-channelre- quisition and the development ofcom-corder in a research laboratory which puter aided learning of the typewe had measures many of the parameters likely to investigated on the Apple Ile andto see be encountered in undergraduate classes. whether there would beany problems A number of suggestions have beenmade in integrating their operation. to improve the package and LSA have 2. to seek and test suitable amplifiers and provided an updated version for evalu- couplers from local manufacturers. ation. A slow chart-record usingan ink jet 3. to have a local softwaregroup develop printer needs to be developed. 4-channel chart recorder software. Their general purpose data acquisition Performance of prototype and compari- package (DAOS) already has most of the sons with commercially available facilities required for the oscilloscope pack- apparatus (for student use) age. This package was originally devel- oped for the DEC PDP-11 computers and There are many authoring systems and received worldwide acceptance. DAOSis many data acquisition systems available, now available on the IBM PC, although but there is no integrated approach whichsome work would be required to provide combines oscilloscope and chart recorder the facilities in the specifications of the functions with computer assisted learning 'Laboratory Machine'. The cost of the full and simulations all within theone ma- DAOS package is too high for generalstu- chine. A number of Physiology Depart- dent use ($3,100 per copy withup to 30% ments in the U.K. and elsewhere havequantity discount) Many of its features published reports on theiruse of micro-would notbe required for teaching,so a less computers as either data gatheringma-expensive run-time system would be used chines in undergraduate laboratory classeson student machines and the full version or simulations or for computer-aided test- for program development. ing instruction. Many of the smallcomput- ers (e.g. BBC and Apple He) in use are notMicrocraft (Melbourne) have produced suitable for a fully integrated approach. aad authoring system ('Author') which would be satisfactory for computeras- The current state of development of thesisted learning.Preliminary discussions hardware and software which might bebetween Microcraft and LSA have estab- suitable for use in the 'Laboratory Machine' lished the feasibility of linking their soft- and other sources are described in the fol- ware packages. 'Author' also has the abil- lowing section, together with commentsity to be linked to simulation software and about a few commercial data acquisitionwould probably form the foundation for systems.It should be stressed thatwethe overall computermanaged learning require an inexpensive system,so that package. The price of this package (approx. many of the research-oriented devices are$600 for quantity) is reasonable. 2o2 Using Computers Intelligently 193 There are a number of simulation packages some modification before it is used in con- available for a variety of PC's, some of junction with the computer. They also have which would need to be re-written for usea stimulator under development. on the computer chosen. This department has designed and built, The major limitation in software is the over- for the Apple-He pilot project, general- all management package which is not cur- purpose pre-amplifiers for recording ex- rently being considered by any of the soft-tracellular potentials in man and animal. ware groups which have been consulted. ItThese have been designed with optically would not be a very difficult task to adapt isolated inputs so they could satisfy safety the Authoring and data acquisition pack-standards for recording from patients. J- ages to operate within such a management RAK have built a similar unit based on this package. design, which we will evaluate. We have examined many alternative sys- When using equipment in large under- tems including MacLab System (Otago,graduate classes, it was found unaccept- N.Z.) based on the Apple Macintosh Thisable to have ambiguity in the control of has some features not yet available in the modules by either the computer or by a LSA software for the IBM. The latest infor- switch on a panel of a coupler. The conflict mation we have shows that they have anbetween controls on the front panel and excellent software package available for any computer controlled gain would have data-gathering with oscilloscope func- to be resolved by covering the front panels tions, and a chart recorder function hasor by modifying the J-RAK modules. been released. There is no computer as- sisted learning option. The Otago (N.Z.) couplers for the Macin- tosh consists only of 4 general purpose Coulbourn Instruments (U.S.A.) Vide- electrophysiological amplifiers, with none oGraph software operates their standardof the other couplers we require being physiological couplers via an IBM-PC, but mentioned as under development. Their this appears not to have all the featuresamplifiers all have complete computer required for our measurements and has no control of gain. They are working on a computer assisted learning facilities. stimulator module. Their apparatus cur- rently performs some of the laboratory There are many other packages for data functions specified. It could be enhanced to acquisition, but their suitability would satisfy our requirements, perhaps with the depend on price and whether or not they assistance of J-RAK, although colour op- could be integrated into a management tions on the Macintosh II are very expen- package. sive and the small screen monochrome Macintosh limits its appeal for complex Hardware performance diagrams. J-RAK Instruments (Melbourne) produce conventional couplers for physiologicalCoulbourn Instruments (U.S.A) have an measurements using standard chart re- incomplete range of physiological couplers corders and these have been purchased forand transducers available, but again they evaluation with the chart-recording soft-are principally designed for operation in- ware on the IBM-PC. So far we have tested dependent of the computer and would the bridge couplers and universal couplers require modification to provide an inte- successfully. A prototype Cardiac Output grated operation with the computer in device has been briefly tested and will need control. 194 Using Computers Intelligently Nucleus (Sydney) produce a satisfactory but conventional, laboratory apparatus IBM-PC computer interface board to con- and costed its replacement at current nect to the couplers and this is supportedprices. The apparatus included 4-channel by the LSA software.All other interfacechart recorders, 2-channel storage oscillo- boards (which allow timing and conver-scopes, stimulators, spirometers, amplifi- sion of the incoming signals to digital val- ers, transducers and couplers. The esti- ues for the computer) are imported, atmated 1987 replacement cost would be greater cost. over $27,000 per student workplace (as- suming work in pairs iit most experiments There are many special purpose data-log- and in fours for the remaining experi- gers and oscilloscope systems based on the ments). This compares with our current IBM-PC, but most are expensive and de- estimates of approximately $13,000 for signed for research purposes. No commer- computer based equipment. The price of cial systems are yet available with a satis- computer based equipment is likely to fall, factory range of couplers which are inte-or its performance increase,while the grated with the computer system. conventional equipment price could rise further. Also the major cost of the com- Comparisons of costs: puter system, couplers and amplifiers, Conventional apparatus versus would be significantly reduced by larger laboratory machine orders. Implementing the replacement of One of the strong arguments for the use of computer based equipment is the potentialexisting apparatus Future cost savings.The individual items ofdirections and financial equipment are less expensive, and theconsiderations apparatus is more versatile than conven- tional equipment, so we will need fewerFurther progress with the 'Laboratory pieces of apparatus. The versatility of the Machine' is now dependent on a major apparatus means that it will be in more financial commitment by the Physiology constant use in laboratory classes, and can Department or others to purchase a com- also be used for student revision outsidemercial number of machines. We are cur- normal class times. rently seeking funds with some urgency, as our obsolete equipment is becoming un- Our Physiology Department requires 42 serviceable. Whether the final computer is 'Laboratory Machines' for its normal com-based on an IBM, Macintosh or other com- mitment for undergraduate laboratory puter, no system has reached the specifica- classes for 180 Medical, 50 Dental and 200 tions required to perform the wide range of Science students. These machines could physiological measurements which would also satisfy the requirements of the Phar- be undertaken (Appendix B), let alone the macology department classes, givenintegrated approach of combining data timetabling adjustments or a slight in-gathering, analysis, simulations and com- crease in numbers of machines. E:adentsputer aided learning within one computer would work in pairs in most experimentspackage. and in fours when the experimental proto- col required group irvestigations. A lie individual components required have been demonstrated as being viable, but We examined another University Physiol- their final development requires that there ogy Department's inventory of modern,be a large enough order to finance the rela- 2 ri 4 Using Computers Intelligently 195 tively simple task of providing the manage- FSD 40 mm lig 1 mV (± 20 nun Hg) ment software and production runs of 0-50 Hz couplers. This will require that a depart- ment takes the first step by itself, or a group 1 Muscle Force Strain gauge Bridge & of Physiology and Biological Science De- (a) FSD 1Kg D.C.amp partments form a consortium of buyers to (b) FSD lOg draw up common specifications and call 1 Muscle Length Optical ? Special for tenders. The latter approach has many FSD 1 an Coupler advantages, particularly in the pooling of 1 Core Temp. Thermistor Bridge & resources in the development of applica- 35-39 °C D.C.amp tions software for simulations and for de- 0-0.1 Hz veloping experiments which use the spe- cial expertise of each department. 1 Resp. Flow & Turbine etc Special Volumes Coupler 0-40 Hz We believe that the 'Laboratory Machine' can be produced for much less than a simi- 2Extracellular Specific Optical- larly equipped 4-channel pen recorder and Potentials (FSD's) Ag ElectrodesIsolated ECG .03-150Hz, 10mV Pre- a digital storage oscilloscope, but with all EEG .05-50Hz, lrAV amplifier the advantages of computer based appara- (NG 10-6KHz, 10mV tus. There seems to be a very large market NERVE .03Hz-10KHz,10mil in teaching and research for the 'Machine' 1 PCG 5Hz -2KHz Microphone A.C. described here and the consortium could Amplifier attract government interest in the potential for export. 1 Cardiac Output Monitoring Spedal (Imp. Method) Electrodes Interface Reference 2 Auxiliary Inputs Devices withDifferential (general use) own bridge etc. D C. Amp. Frog Skin 200mVwith Offset, Amplifi( Delbridge,L & Kemm,R.E. (1985). Com- tion, puter managed learning in an under- Attenuation graduate laboratory. In J.A.Bowden & Filtering. S.Lichtenstein (Eds.) Student Control of Learning: Computers in Tertiary Edu- 1 Double Pulse Stimulus cation, Melbourne.. CSHE, University Stimulator Isolator of Melbourne, pp . 5-97. Appendix B Appendix A Outline of requirements of current Indicates the requirements for transduc- laboratory experiments and the ers and couplers in the Department of 'laboratory machine' in first and Physiology. second year Physiology classes. #Measured Transducer Signal parameter Coupler The variety and technology required to perform experimental physiology is not 1 Arterial B.P. Strain gauge Bridge & FSD 300 m Hg 2-8 mV D.Camp widely appreciated, so an outline of the 0-50 Hz experiments currently performed is given below. As well as describing current 1 Venous B.P. Strain gauge Bridge & experimentsjorf 2rtd Year physiology D.Camp k.1 J 196 Using Computers Intelligently classes, indications are given of the easy tors. New transducers and muscle bau.,, upgrading of these experiments. How- are required, particularly if anothtr muscle ever, new experiments will also need to (e.g.sartorious) is used to allow direct be devised, or adapted from other institu- stimulation of the muscle, rather than via tions, to further modernize the teaching the fatigue prone neuromuscular junction programme. Final year experimental used in sciatic/gastrocnemius.It is pos- physiology experiments are not included, sible to introduce human muscle studies since they are usually performed in small using EMG and force measurements (if groups and are more easily adapted to additional heavy duty force transducers change in equipment or emphasis by the are purchased). This may be an optional lecturing staff. We would expect to be facility for students proceeding to exercise able to present students with challenges physiology. much closer to the experimental physiol- ogy they might encounter in their future Experiment 3:Action potentials in periph- workplace. eral nerves One of the significant advantages of the The current experiment examines the char- new equipment is that more experiments acteristics of Action Potentials initiated by will be able to be performed on human electrical stimulation in the sciatic nerve of subjects with the greater data gathering the toad. This requires fast data acquisition and analytical power of the computer in oscilloscope mode. It could be extended based apparatus. to study the transmission of activity across the neuromuscular junction by simultane- Experiment 1:Permeability of Cell Mem- ously recording the electrical activity in the branes attached gastrocnemius muscle. The Apple He pilot study introduced recording Demonstrates the factors influencing the of action potentials in the human ulnar transport of materials across membranes nerve for the medical students. The previ- using erythrocyte and yeast cell.It requires ous apparatus was not safe for human re- no computerized data acquisition or anycordings and computer averaging tech- particular laboratory niques are needed for recording activity from sensory nerve fibres (electroneuro- Computer Simulations could considerablygra ms ENG).Measurement of propagation assist the students understanding of diffu- in central pathways may also be possible in sion, osmosis, isotonicity and active trans- the human, using evoked cortical poten- port mechanisms. Computer assisted tials. learning would also be important for this conceptually difficult area of physiology. Simulations are important in the irea of membrane potentials and excitability Experiment 2:Characteristics of Skeletal where students have to deal with many muscle difficult concepts. Examines the factors which influence the Experiment 4:Properties of smooth muscle development of force following electrical stimulation of the isolated toad muscle. Examines the control of the rhythmical con- Fast data acquisition, in oscilloscope mode, tractions and overall tone of visceral of muscle force and length is required, with smooth muscle of the rabbit in an organ the possible addition of recording the elec- bath. The importance of local environ- trical activity of muscle (EMG) and recep- ment, temperature and humoral agents are 20( Using Computers Intelligently 197 examined. Slow data acquisition using thebenefit many other experiments, e.g. exer- chart recorder mode is required to measure cise tests.Access to stored ECG's of show- the rhythmical contractions. A sensitive ing normal population variations, such as isotonic length transducer is needed. The with orientation of the heart, would assist computer could calculate the frequency of the medical student's appreciation of the the underlying rhythm while recording the ECG. Some abnormal ECG recordings cyclical length changes. night also be valuable. Experiment 5:Haematology Experiment 8:Mammalian blood pressure Involves the analysis of the students own This experiment investigates the control of blood. No on-line data acquisition is re- blood pressure in the anaesthetized rabbit. quired although a central computer would This experiment would be much more assist in the collation of class results. Re- valuable if there were more than the cur- sults from som of the more modem ana- rent single recording of arterial blood pres- lytical machines could be directly interface sure supplemented by calculations of heart to the central machine. These could now rate.Simultaneous recording of venous include blood gas analysis. blood pressure would allow students to better examine compensatory mecha- Experiment 6:Cardiac muscle nisms. The ECG could also be measured. The experiment would need to be re-con- Investigates the properties of cardiac sidered to take full advantage of the new muscle in the toad, relating the sequence ofapparatus which' could also include the events and the coordination of contractionmeasurement of cardiac output by ther- of the heart chambers. The experimentsmodilution techniques. could be supplemented with the new Experiment 9:Haemorrhage equipment to include a pseudo-intracellu- lar recording of ventricular action poten-This experiment investigates the cardio- tials (using a suction electrode) so permit- vascular compensatory responses to blood ting a better understanding of the electrical loss in the rabbit.This is an important events associated with contraction. experiment in the understanding of cardio- vascular control and would be improved Experiment 7:E.C.G. and heart sounds by the better monitoring of responses as outlined in Experiment 8. Investigates the ECG (electrocardiogram) and PCG (phonocardiogram) in the hu-Experiment 10: Human blood pressure man. Experiments are also performed withand heart rate the exposed heart of the toad, so that the basis of the ECG recordings can be appreci-Currently involves the measurement of ated. These experiments would be consid-human blood pressure and its variation erably improved when two channels ofwith posture and exercise. The computer recording can be made. The ECG and PCG would be of little use in the measurement of can be simultaneously recorded withoutblood pressure with cuff and stethoscope, the confusion of superimposing their elec-but there is an important missing piece of trical signals in a single record.Vector information in this experiment, the vari- cardiograms can also be generated fromation of Cardiac Output. The computer two simultaneous ECG recordings. The will permit the calculation of cardiac out- ability to record the ECG with a computer put in all these situations so that a better calculation of instantaneous heart rate will understanding of cardiovascular compen- .2 0 7 198 Using Computers Intelligently sation is possible. As well, during exercise, occurs at the end of each practical class. It the beat to beat measurement of heart rate could also be useful for some more complex from a simultaneous recording of the ECG calculations. will permit a better understanding of re- covery mechanisms. Simulations ofExperiment 13:Gastric secretion cardiovascular function are particularly important, because of conceptual difficul- This experiment examines the modifica- ties the students have with many of thetion of gastric secretions under various interdependent variables influencing theconditions. Some accumulation of class circulatory system. results with a single computer could be helpful. Experiment 11:Respiration Experiment 14:Body temperature control Part 1: Measurement of respiratory gases, breath-holding etc.There is no require-This is a very important practical class for ment for on-line data acquisition. the students because of the significance of understanding temperature stresses and Part 2: Lung volumes and metabolic rate.because it re-examines a wide range of Students need to monitor the flow of gas inphysiological mechanisms involved in the and out of their lungs and collect samples integrated response to these stresses. The for gas analysis. There are an inadequate experiment would be considerably im- number of conventional spirometers for proved if cardiac output could be moni- this task at present, and the use of flowme- tored, as well a having on-line recordings ters connected to tht computer wouldof ECG, respiratory flow, as well as the provide a more cost effective replacement. current recording of core temperature from the tympanic membrane via a thermistor Part 3: Exercise responses. Currently, stu- probe. dents work in large groups collecting a variety of data manually supplemmted by Experiment 15:Sensory and motor func- an ECG recording to calculate heart rate. It tions in the C.N.S. will be possible to have more stu,:.Its per- forming the experiments if the chart re- Current experiments use a limited number corder directly monitors the respiratoryand range of apparatus to investigate sen- and cardiovascular changes. Mos', impor- sory mechanisms and motor functions in tantly, the measurement of cardiac output the central nervous system. The use of changes will allow a better understanding computer based apparatus al the standard of the cardiovascular and respiratory re- configuration with 2-4 equipped with sig- sponses to a graded exercise challenge. nal generators to act as audiometers would considerably expand the range of experi- Experiment 12:Renal Physiology ments which could be offered. Examines the human renal responses to a Experiment 16:Structure and function in variety of stresses such as water and salt the C.N.S. loading. There is no requirement of on-line data acquisition, but a central computer This is an orientation class for science stu- which could collate the students' extensivedents to provide them with an elementary results within the various subgroupsunderstanding of the relationships of would considerably assist the staff in pre-human brain structures during demonstra- paring for the lecture-discussion which tions and investigation of prepared mate- 208 Using Computers Intelligently 199 rial. No computer use is envisaged at pres- ent, although pre-recorded computer im- ages demonstrating brain scan results could be useful together with 3-d rotatable representations of the location various brain ssuctures. Experiment 17: Endocrinology At present, the only endocrinology practi- cal class is an anatomical investigation of relevant structures in the rat.The new apparatus would permit students to inves- tigate the action of anti-diuretic hormone on ion transport in the frog skin as a model for their understanding of kidney function. Experiment 18:Salivary secretion This experiment is for the dental students and examines the control of salivary secre- tion. Some accumulation of class data on a single computer is all that would be envis- aged. 209 Interactive laser disc in Community Medicine Trevor Lord Department of Community Medicine Monash University Teaching Community Medidne requires a Medicine Faculties to increase the commu- great deal of small grqup work. In order to nity medicine components of their courses provide time for this and still expose students to the necessary range of information a new have added to this. There is also an increas- approach was required.The Department ing need to teach consulting and communi- developed a learning laboratory where the cation skills a role that has fallen to depart- students could spend time working on learn- ing packages. The packages have taken the ments of community medicine. Teaching form of video, computer problem solving, car- in this area requires high staff to student omed, audio tapes and printed material ratios yet the more basic content of general Interactive laser disc is the latest addition to practice still has to be covered. For these the learning laboratory. Some 3000 slides including all the departments slides have been reasons our department has been develop- written onto a 30an laser disc. Using an intel- ing learning packages for some years. The ligent laser player and a touch screen monitor students in our department spend almost packages have been written using the laser 50% of their time in the department work- disc as the material source. The department intends converting the software to EPROMS ing in a learning laboratory going through allowing free standing players and monitors the packages at they own pace. The enthu- to be set up without the need to have a com- siasm for this kind of learning has encour- puter attached. aged us to continue in producing more packages and move into computer based Much of medical teaching is based around packages. clinical situations requiring live patients, clinical slides or video. Using computers toOur Apple network based CALS have been mimic the clinical situations has been lim-in use for three years and include an assess- ited by this inability to provide the illustra-ment CAL that marks each student per- tions. Graphics have not solved this prob-forming it and produces the scores. These lem. Another problem is the large collec-clinical cases have always been hampered tions of clinical slides retained within theby the inability to illustrate them. In paral- clinical departments. lel with this has been caromed based clini- cal slides and video tapes of consultations Laser disc offers the opportunity to have awith work through programs in booklet source of carefully selected clinical slides or form. video footage that the computer programs can then use to illustrate the clinical prob-The principle behind all the programs is lems. Large numbers of slides can be stored that they aie interactive with the emphasis on the one disc. on problem solving and not on presenting The Department of Community Medicinelarge quantities of factual information. The Monash University, has like many otherprograms are kept to around 20 of 30 min- departments, had its teaching resourcesutes completion time. stretched.The increasing pressure for 210 I..- Using Computers Intelligently 201 The laser disc technology offered us an Ilvas when combined with an intelligent iozal opportunity to combine all these into laser player allows the program to be this exciting and fun to use medium. Our dropped into the player from the computer enthusiasm to use the medium has encour- or from an EPROM in a plug in cartridge. aged us to go ahead with out waiting to see The player can then be driven by a simple which direction the technology is going to infra red hand held keyboard or a touch develop. screen monitor. We were very impressed with the ease of use of the touch screen We have chosen the thirty centimetre laser monitor which particularly suited our discs because of the volume of material style of packages. they can hold. Approximately 30,000 clini- cal slides can be stored on the disc or about The hardware available at that stage was 30 minutes of video footage. Any combina- the Philips VP835/E Laser player and the tion of slides or video can be used. The Philips VP120 Touch)screen monitor. The material for our first disc was prepared in cost of this approach was about $500 more South Australia and was sent overseas for than using the computer based card sys- production. We have taken a 10% share of tem with a plain monitor and simple laser this disc. This allowed us to load our de- player. partments collection of some 1,000 slides. The Ilvas software runs on a IBM compat- We also obtained a first class selection of dermatology slides from a teaching Der- ible machine using two floppy disc drives. matologist, a collection of slides from It has its own operating system and is not currently available in a hard disc version. Melbourne's Ear nose and Throat hospital The second drive contains the data disk. It and a collection of ophthalmology slides. A generates text in videotext format suitable total of 3,000 slides. for the monitors videotext decoder and has commands that allow full control of the Our choice of hardware was based on the player. The language is Al lva which al- direction that we wanted the packages to lows high level. commands. take. We wanted the packages to be simple to use ideally free of computers and key- A supervisor allows control of the different boards. We wanted good interactive pack- modes of the program. The programs are ages that could be easily altered or up- made up in small simple fragments which graded. We wanted the students to be able are then linked. The text frames are gener- to select from a range of short interactive ated separately and called up by the pro- packages and easily start the package with- grams. The programs are compiled by out supervision. Ilvas and an then be dumped to the 64k ram of the player. Once the program is Two directions were open. The first was to complete and has been fully tested Ilvas use a simple laser player connected to a manager can then be used to send the computer. Install a card in the computer program to an EPROM burner for produc- that allowed the computer to drive the laser tion of a chip which is then placed in a player to obtain the illustrations and pro- simple cartridge. The cartridge can then be duce the text for the interactive programs. labeled and available to the students in a This system required that a computer was fairly robust form. constantly attached and that the student would need to drive the computer.The sec- Writing instructions for the touch screen is ond option was to use the laservision tech- relatively simple. Screen coordinates are nology and the Philips Ilvas authoring sys- given for the positioning of the touch sen- tem. sitive areas. Corresponding coloured 211 202 Using Computers Intelligently boxes are then produced in the text frames.system is a simple stand)alone player and The boxes can be made to change colourtouch monitor. The student selects thecar- when touched. Information as to choicestridge containing the program, plugs it in made on the touch screen can be retained and switches on the player. The player for scoring or later use in the program. boots from the cartridge and presents the information through the touch screen Control of the player from within the pro-monitor. The learner touches the appropri- gram is also powerful. Any frame can beate boxes to select choices. Clinical slides called up from the disc within 0.5 secs. Aare taken from the disc to illustrate the series of frames e.g. video can be playedproblems. The presentation is rapid and forward, backward, slow or fast. Sectionscolourful. The packages have a great deal can be repeatedly replayed. Writin pro-of impact are fun to do and a very useful grams to utilize slides or video is relativelyway of repeatedly presenting useful pack- simple. ages. Integrating the videotext output and the laser disc output is also easy. Text can beOur first packages were on ECG programs. over-layed over slide frames taken off theThey were surprisingly easy to produce. disk or over running video frames. TheLessons and programming ideas devel- video text can be altered independently ofoped on one package can easily be copied the laser disc material so that a piece of and used in other packages. The on-screen video can be run for a while before super- editing of packages can be done on com- imposing videotext. Videotext can be dis-puter monitor or with the TV monitor al- played by itself. The text material can belowing quick assessment of the presenta- held on screen for a period of time movingtion colours and methods. on automatically or sit awaiting a promptA number of limitations are present. The from the touch screen. Text can be ina inability to import files from a standard range of colours with varying back- word processor. The screen editor is rea- grounds, limited graphics are available.sonably good but production of text could Text can be made to flash or change colour. be enhanced if it could be performed on a The size of programs is limited by the 64kwordprocessor and then moved into Ilvas. RAM of the player. We prefer short 20)30Another problem is not being able to use minute programs and this limitation has Ilvas from the hard disc. Reading and writ- not been a major handicap except when weing fragments to the floppy is slow and also have used a lot of text material. Programshampers production. that source most of the illustrations from the disc are not affected by this. Larger Overall we are very pleased with the choice programs could be produced and written we have made. Given the goals that we had on the disc as part of disc production how- for the package design and presentation we ever this limits the flexibility of being ablefeel the Philips system has met our needs to change or upgrade the programs whichvery well. We are keen to start working with video footage. At present the Royal is a serious limitation in the medicalarea. Ideally any large volume text material College of General Practitioners (U.K) are should be written on the disc as long as onemastering a dedicated 30 cm disc for gen- is confident that it will stay in date. Theeral practice. This will include video of advent of read)write laserdisc technologyconsultations as well as slides. We are will resolve this problem. looking forward to obtaining this as a source disc and developing our own pro- From the learners view the face of these grams. packages is very attractive. The complete 212 Using Computers Intelligently 203 Monash is considering producing 30cm laser discs. Currently we are negotiating Trevor Lord, a medical graduate, funding to produce a disc on Back-pain, spent four years in general practice communication skills and on AIDS. Once developing educational interests the discs are produced one can build an through the Family Medicine Pro- almost infinite number of programs using gramme. Computing interests de- the source material. Communication skills veloped through the Royal Austra- will be the most challenging to produce in lian College of General Practitioners a learning package. However the ability to initially in the development of video a good counseling type consultation, on)line computer assisted medical run it slowly, freeze it at critical points, records.Secretary to the Colleges superimpose points and questions offers a National Computer Committee, unique opportunity to provide this type of Victorian Computers in General learning. Practice committee and co-ordinator of the RACGP Na- The laser disc system will solve a number of tional Computer Assisted Practice problems in presenting our learning pack- Project.Currently working in the ages. The simplicity reduces the need for Department of Community Medi- any supervision. The presentation ensures cine developing the departments the students interest. The technology al- computing role including produc- lows imaginative programs to be produced tion of patient management CALS, in areas that traditionally have required medical computing courses, medi- considerable staff input. cal computing research and assist- ing in the development of commer- cial medical record, prevention and health promotion software. 2 1 3 Lateral teaching software:A programmer's perspective. Ralph S Sutherland & BenjaminK Se linger. Chemistry Department Australian National University The microcomputer has opened up a new In a typical Macintosh application,each era in teaching tools and the time has come individual keystroke or to assess its use.A project to develop mouse button Mac Fourier gave one of us (RSS),a com- push is intercepted individually bythe puter programmer and a student, theop- main loop and passed to the appropriate portunity to explore the situation froma handling routine, so that atany given variety of angles. This paper willcover moment the program is ready to respond some of the major design parameters that to the user. In MacFourier a user can have went into the development process. Along just transformed a function, spottedan the way, specific examples of the principles error in the transformed display and al- involved will be illustrated. most immediately proceed to edit the original function. This has the obvious advantages. It encourages semi-random User initiation and authorization. exploration of the program's abilities with- out fear of losing everything because ofan One of the most important parameterswas error.It means the program will notgo to keep the user in control. Theuser de- through a procedure unless and until the cides what process occurs next, and when user is both ready to, and has authorized a chain of processes can be stopped. This the operation. Thus routines thatare not 'user driven mode' is especially evident in wanted for a particular run neednever be the software used on the Macintosh"'' used. computers. It should be rioted however that the choice of this mode of operationis Program restrictions on the userare largely the responsibility of theprogram, not the replaced by user restrictionson the pro- machine. gram. The computer is placed in the back- ground, while problems andprogram so- A popular method of implementingthis lutions are brought forward. This lessens style of programming is touse a dendritic computerphobia. program structure. This structure hasa central waiting loop which can detectac- Non-linear program structures, tions by the user, and a collection of sub- menus and more... routines that are selected inresponse to those actions. Additional cross links and In the following examples some Macintosh returning routes between the variousrou- terms are used, for which equivalent items tines and the 'main loop' give the program and methods will be foundon other mod- its final behaviour pattern. Most of the ern micros. running time is spent at the primary node or main event loop waiting for the user to Menus area method of outlining the broad act. choices available to the user atany given 214 Using Computers Intelligently 205 The Structure of Mac Fourier 1----Fourier Trasform Routine Main Event Loop Sub-Sub routines Figure 1: Non linear structure of the program (not all of the returns to the main loop are shown)... time. They are enabled or disabled as re- nain loop. Other choices are offered as quired by the program to eliminate illogi- deemed desirable. cal of invalid options. The rnain event loop can detect a menu choice through monitor- Relative values vs Absolutes. ing a mouse event, and can then branch the program to the appropriate subroutine. When comparing the fields in which Fourier Transforms are used an immediate Further choices can be made with dialog problem quickly arose as to whose conven- boxes and buttons. They can be seen as tions to use. One text outlined no less than buffers. The "CANCEL" button, e.g.,al- seven different conventions in expressing lows the user to abort and return to the the transform and its inverse. The values of 215 206 Using Computers Intelligently ra QuitFunction Mew Modify TransformResolutionDisplay 022 Mudrfy %/Replace £182 Rdd (or mqdify) XP Subtract X- Multiply (AM) X* Divide (use with caution) X/ FM (Sin/Cos only) XF Convolute (slow) Correlate (slow) X® Shift 180° Square wanes: Absolute Value XR Phase (0-360) 0 Square KB Exchange Rea IcolmaginaryNH Amplitude (0-1) Normalise Amplitude MN Frequency I Integrate (sum) XI ( OK )(Cancel) Reverse Fig 2: Menubar and menu, dialog box and buttons. the normalising coefficients was the main Amenu option can be used to scale down area of disagreement. We made the deci- amplitudes for display.Dialog boxes sion to avoid the problem altogether by not make use of 'intuitive' ranges such as using an explicit normalisation. All dis-amplitudes of 0-1, and the centre of dis- played functions and transforms are nor- tributive functions can be placed on a 0-1 malised relative to an appropriate maxi- scale without using explicit channel num- mum and scaled to fit the screen. As abers. This is just one of the steps taken consequence labeled, graduated scales fortowards simplifying operations and giv- the displaysare not used and there areno ing the user more time to think about the restrictions in the inputting of amplitudes. physical issues involved. Reciprocal Areas Lorentzians and Biexponentials --01114616am.---- F(to) Irt ft VT F( ) Fig 3: The Qualitative Uncertainty Principle, note the lack of absolute scales, yet the trend (and so the idea) is immediately obvious. 216 Using Computers Intelligently 207 Levels of volunteered information, construct some arbitary function seen in a towards'qualitative' programs text or otherwise desired requires some creative effort, the construction methods This trend towards a more qualitative are not always obvious. approach became an important aspect of the design philosophy of Mac Fourier. Active exploration of limitations of computer (especially digital) Visual feedback duringprocesses techniques In order to allow the user to see whatwas During development, one of the test func- happening at any given time, the modify tions used to test the algorithmwas a and function operations can be seen evolv- square wave, which should theoretically ing on the screen as they are carried out. give a purely imaginary transform.It Unfortunately the structure of the trans- doesn't! The 'error' can be tracked down to form algorithm itself does not allow aa fundamental problem of using a finite similar approach tc be used. This leaves representation of the function and trans- the transform operation as somewhat of a form, i.e. the cliffonce between a Fourier black box'. However the whole point of series and a Fourier integral. the program is to look at the properties of this 'box' rather than understand itspar- Rather than trying to reduce the effect by ticular algorithm. increasing channel numbers or introduc- Visual Feedback ing sophisticated interpolation smooth- Multiplication (a squarewavo by an exponential). ing algorithms etc, this 'error' was left in the program operation and emphasised in the teaching process as a fundamental property of any digital transform process. General applicability to many fields, efficient use of programming resources Fourier Transforms were selected as the first of the lateral teaching topics because V* they provide a general purpose tool com- mon to many fields of study and therefore Fig 4: Visual Feedback. Simple operations represented a teaching redundancy. In are demistified by being displayed as they addition the usual mathematical ap- are being executed. proaches often did little for the student's understanding of the physical aspects that Allowance forcreativity were the real learning objective. In addi- tion, all the processes critically related to Choices were made as to what basic func- the Fourier Transform, such as correlation tions should be included for the purpose of and convolution, could easily be linked. A constructing functions and transforms. major obstacle to be overcome was to pro- The final list of functions and operations vide a translation for the local dialects used included the modify menu allowed the to explain basically the same conceptst construction of the simulations and prob- each of the different fields. Thus the spec- lems found in all the reference texts. To troscopists' dispersion/absorbsion curve 217 II- 208 Using Computers Intelligently A Weakness of Digital Techniques. Finite Representationerror. Real Square Wave Transform A.L.L.A.A.A.A.s.a.A. PEAL GS' REAL11 I I C=D IMAG IMATATillti Note Zero Values ECON13 REAL REAL114 I .. tii I 1III IMAG IMAG Fg 5: The finite number of frequencies available to form a transform gives rise to the low fence of real spikes in the transform of a square wave. If the real component is removed by filtering etc, and the resulting spectrum is inverse transformed, a square wave is obtained that has zero 'crossover' values; these are not present in the original. As the contributing sinusoidal waves all have zero values, the zero crossing can only disappear in the limit of an infinte series. is related to the sound-engineers log/log sional transformer of the optical images. frequency response curves; the general slit Bringing together disparate fields under a function to the electronic system transfer common rubric is our central theme. function and so on. The similarities in dif- ferent fields can be found just under the There is nothing new in programming a surface of the different conventions. One computer to perform fourier transforms. can goon. Any spectrum is a fourier trans- The important thing here is a tool that can form. In physiology, the ear is treated a be used to mimic a tremendous variety of real-time continuous audio frequency apparently unrelated problems/situ- fourier transformer. A lens is a two-dimen- ations, and then to explore the relation- 218 , Using Computers Intelligently 209 FT-IR Spectroscopy: Real Function Real Transform Inteforostrant IR Sprctrunt: rT 4:Gs Id Moto: The spectrum is ropests4 in the negative troquenciss. Fig 6: Any Spectrum is a Fourier transform. Image Processing ships that exist between them. The user ca n concentrate on the meanig of the transfor- mations to his preJlem and not be con- cerned at this stage with a whole set of notational conventions or spend time understanding the transform algorithm itself. Fig 7: Fourier Transforms used in Image processing Time domain data handling operators in the frequency domain. Fig 8: Many common, and apparently disparate data processing operations are in fact more closely related in No Morin. the Fourier domain than in their original one.Differ- entiation (first order) of a function is equivalent to multiplying its tan iformby the imaginary ramp funct- ion, whereas frequency filtering can be achieved by multiplication by similar real ramps. Filtering can be 4 related to convolution or litrotttrlikt Diansilver firth order Differostlito, deconvolution. 1...... r...... r...... Nolo Oporsior Is Imaginal g 219 210 Using Computers Intelligently Convolution and The Fourier Transform. Fig 9: The nature of convolution can be illuminated by seeingthe operation as the multi- plication in transform space. The lateral design of MacFourier allowsa user to 'enter this cycle at any point. The Phase Problem. Original lattice Beciprocal lattice R Al mammals, Fig 10: The relation- ship between Fourier transforms and the Equally Valid Solutions crytallographic dif- for the lattice, given the Diffraction Pattern. Observable fraction phase prob- Diffroction lem.One car. only Pattern observe the modulus of the complex diffrac- ? b32:11 tion pattern (Fourier Al ernetive lattice Alternative Reciprocal lattioe transform of the lat- tice) and so seveal al- ternative lattices are [1111111j1111111 possible given one dif- REAL fraction pattern. Here only one alternative is shown but others are possible. IMASIINAR, MAMMARY 220 Using Computers Intelligently 211 Acknowledgement The program development was funded under a raculty Research Grant to B.K.S. for the topic 'A Lateral Approach to Terti- ary Science'. 2 2 1 Interactive audio... A challengeor companion to interactive video ? N.A. Shaw Footscray Institute of Technology In tertiary education there have recentlyhave adapted the Morrill Cube (after been many advances, both in terms of hard-Morrill et. al., 1974), to illustrate that any ware and software, in the style and applica-successful intervention in student learning tion of new technology to the teachingand welfare must consider these three process. This paper describes some recent major components: experiences with the use of "interactive audio" techniques, and endeavours toa. the Target of the intervention strategy show why this particular technology has an important niche in the spectrum of alterna-b. the Purpose of the intervention strat- tive teaching methods. egy, and In addition, this paper compares interac-c. the Methodology of the intervention. tive audio with its companion technology, interactive video. Whereas both of theseBy using this model, the technologies of technologies have the potential to enrichinteractive audio and interactive video can the educational process, the special attrib-be located in the spectrum of possible inter- utes of each can only be fully appreciated vention strategies. Furthermore, each of when their appropriate role in the learningthe three dimensions of this model has environment has been carefully deter-several parts, allowing further detail to be mined and these features matched tospe- appreciated during a comparative s'udy. cific teaching tasks. This allows us to clearly illustrate the role and specific advantages of one technique To provide a substantive framework over another in each situation. against which to compare these two inter- active technologies, a context for learning is According to this classification, both inter- set by classifying both methodologies in anactive audio and interactive video are inter- overall scheme of "intervention strategies"ventions which have the following charac- in teaching. teristics: Intervention strategies in teaching a. The target can be individual students or small groups of students. It would sometimes appear that there is a tendency for computer aided learningb. The purpose a can be any one of reme- materials to be created with regard to the dial or crisis intervention, preventative attributes of the technology rather than intervention, or developmental or fur- with reference to how and where the sys- thering intervention. tem fits into the overall educational proc- ess. To assist in developing an overview orc. The method of intervention belongs to framework, Sillitoe and Gorman (1987) the category of educational technology. Using Computers Intelligently 213 By highlighting these separate elements ofThe following brief description._ identify the intervention process, it is possible to the main components in this range of inter- focus upon the special characteristics of the active techniques: intervention strategy which most ade- quately address the situation. For example, 1. Text: The simplest style merely dis- by observing that the technologies are di- plays text on the computer monitor, and rected at individual students or small many of the early CAL lessons operated groups o f students, staff can allow an inves- at this level. The introduction of com- tigation of personal learning styles to influ- menially available packages such as ence the choice of delivery mode. Pro- "Storyboard" have enhanced this style grams which are designed for delivery to a by offering special effects, including a mass audience would not share this charac- variety of intriguing ways of wiping or teristic. dissolving text on the screen and a vari- ety of type fonts. The role of educational technology 2. Text and Graphics:Adding graphics, Within the category of "educational tech- particularly animated graphics, to a nology" there is a broad range of teaching text-based lesson can further improve styles, most of which are machine depend- the presentation of materials as well as entand require little commitment of increase the level of student interest and human resources at the stage of actual les- alertness. Perhaps the majority of CAL son delivery. However, by further subdi- programs in education today are oper- viding educational technology into the ating at this level. categories of interactive and non-interactive, (see Figure 1), we can 3. Text, Graphics and Audio:This is the identify those teaching styles which use domain of interactive audio, where text computers for responding to student input and graphics (or digitized pictures) can and for controlling the delivery of materi- be displayed on the monitor simultane- als, from the others. ously with the sound of voices, music, and other special sound effects. The Figure 1: Two categories of educational added audio dimension offers great technology. scope for teaching those concepts and skills which depend critically upon Noninteractive Interactive sound for full communication. Examples:SlideTape ShowsComputer 4. Text, Graphics, Audio and Video: This Videotape Aided Learning part of the spectrum refers specifically to interactive video, where selected portions of video can be shown to stu- Finally, within that group of educational dents under program control. The abil- tee-,ogies classified above as "interac- ity to superimpose text and graphics as tive", there is a group of techniques which overlays on the video as well as having can be represented as in Figure 2. As can be full sound facilities, offers the CAL appreciated from this diagram, when pre- designer what appears to be the ulti- senting learning material to students using mate set of tools for lesson creation. these techniques each one depends to some extent upon visual stimuli (use of theIt is important to appreciate that in the monitor), while others may also involve theabove discussion, components of the spec- use of audio stimuli and enhanced visualtrum were arranged in an order ofin- presentation such as full colour video. creased degree of sophistication involved 223 214 Using Computers Intelligently Figure 2 VISUAL STIMUU AUDIO STIMULI PROGRESSNE INCREASE IN SOPMTICATION in using the technologies. Withprogres-digital.For example, an analogue tech- sively increased sophistication comesannique could be an audio tape recorder/ increased degree of complexity and higherplayer which stores the data on normal costs in both materials development andaudio cassette, although only a few com- the type of peripherals needed for lesson mercially available players are capable of delivery. Considerations of cost, effective- computer control. In this context, "control" ness, preparation effort, and complexity refers to the ability of the computer to per- are all critical elements which will be used form certain tasks with the tape such as: later to compare interactive audio with interactive video. searching quickly for any desired position on the tape, Having identified the range of CAL tech- playing or recording on the tape niques available, the next step is matching whilst simultaneously monitoring a specific teaching task to the appropriate the position, and choice of educational technology. Know- controlling the volume. ing the special attributes of each technol- ogy is most important when making a suit-The two common digital techniques of able choice.The following section de- storing audio are: scribes interactive audio insome detail, identifying those special features which (i) compact disc, (which is laser tech- justify its place in the spectrum of interven- nology), and tion strategies. (ii) floppy or hard disc (normal mag- netic media). The technology of interactive audio Similarly the sounds stored digitally on a Any device which can record and repro-floppy disc can be retrieved quickly and duce audio information such as voice,reconstituted back into high quality audio. music, or sound effects, and at thesame Experiments have shown that recording 42 time have all its functions controlled by aminutes of high fidelity sound consumes computer, is likely to be clagsified as "inter- about 10 megabytes of memory.If an active audio". A typical program can dis-application could tolerate lower quality play normal text and animated graphics on recording, then up to nine hours of sound the computer monitor, whilst synchro- can be stored in the same 10 megabytes. nized sounds are reproduced as desired. Whatever the audio storage medium there- fore, an interactive audio system must have The method by which the audio informa- the ability to store and quickly retrieve such tion is stored may be either analogue or sounds under computer control. 224 Using Computers Intelligently 215 Special attributes of interactive aspect of the system which helps to set it audio apart from any other learning tech- nique. As indicated earlier, the selection of one style of educational technology over an- c.Visual reinforcement of the audio con- other is determined not only by the nature cepts is achieved by simultaneously of the particular task in hand, but also by displaying text ant_ graphics on the the special attributes which each style can monitor. As the voice explains certain offer. Too often we see one particular tech- aspects of the lesson, key words (or nology being forced to perform a Procrus- whole sentences) can appear on the tean task better suited to another teaching screen, thereby supporting and adding method. constructively to the learning process. So what are the particular properties which Animated graphics can also be used to make interactive audio an attractive, ad- enhance a simulation. For example, a vantageous alternative in certain teaching graphic representing a sphygmoma- applications? Some of the most important nometer can be animated so that mer- aspects are as follows: cury in the column appears to be drop- ping. Whilst the "mercury" is drop- a. The interactive nature of the system is ping, synchronized blood pressure an inherent part of any computer based sounds are reproduced to complete a lesson. Appropriate design of the pro- realistic simulation. gram allows students to control their own rate of progress with the lesson, With the introduction of digital scan- well as their own path through the ning devices, a photograph can be cop- material. This places students clearly at ied onto floppy disc, incorporated into the centre of the learning process, an the lesson, then later displayed on the important attribute which is common to monitor with additional text, graphics, most CAL lessons. and audio enhancements. b. Audio: In cases where special soundsd.Costs: As in any educational technol- are central to the concept or skill which ogy there are production costs involved students are seeking to learn, interac- with the source materials and programs tive audio is particularly useful. As will as well as delivery costs in presenting be discussed later for example, describ- the product to students. Fortunately, ing the relationship between Korotkoff interactive audio is not an expensive sounds and blood pressure readings on medium in which to invest, either for a sphygmomanometer cannot be effec- production or delivery. Whilst interac- tively conveyed without fully repro- tive audio relies upon some device for ducing the genuine sounds. In addition, its source of audio, whether cassette the teaching of nursing students to rec- tape, compact disc or a digital audio ognize foetal heart beats is also greatly interface, the cost of adding an appro- enhanced if these sounds are actually priate interface to transform a normal used as part of the lesson. Even voice personal computer into an interactive communication itself, with the appro- audio system is less than the cost of the priate emphasis and intonation, is rJ- computer itself. tentially more effective than relying on students to read a computer screen full e.Ease of production. Unlike some other of text. It is an emphasis on this audio new technologies such as videodisc, 24 5 216 Using Computers Intelligently interactive audio allows completelec- At least three of these four parts turer control of the product, as well as depend extensively upon audioas requiring only a relatively shortpro- the essential means of conveying duction time.Experience with the information, thereby justifying this medium has shown that feedbackfrom choice of technology over other tech- staff and students whouse the program niques. can be readily incorporated into the lesson. This can be achieved eitherby (ii) "The Foetal Heart" is another inter- modifying the audio tape with simple active audio program with a medical recording equipment,or by creating education theme which depends new digital audio files on floppy disc. extensively on specially recorded sounds. Again the program is aimed After field trials whichmay involve at undergraduate nurses, but it is further development and refinementof also appropriate as an introduction the program, the product can be con- for students of mid-wifelycourses. verted with confidence toa permanent, The program has three main parts: reliable, and fast-access mediumsuch locating the foetus in the as a compact disc. uterus, and identifying the particular "lie", Specific examples of interactive locating the foetal heart, us- audio ing a knowledge of the foetal lie in conjunction witha The following paragraphs brieflydescribe knowledge of the character- the elements of three projects whichmake istic sounds created by the use of this technology. foetal heart, placenta, and other mother-body noises, (i) "Measuring Blood Pressure" This counting the foetal heart program, which has been designed rate, and determining if the for first year studentnurses, has the foetus is healthy. following four main parts: identification of the It will be appreciated that, for ethical Korotkoff sounds (especially reasons, it is not reasonable to allow those associated with the teams of undergraduate student systolic and diastolic blood nurses to examine and prod the pressures). abdomen of selected pregnant reading the sphygmoma- women. However, use of this inter- nometer scale whilst listen- active audio program can simulatea ing to the specific Korotkoff great deal of the clinical situation by sounds, thereby performing using genuine sounds recorded with a blood pressure measure- help from fully trained medical staff. ment. experiencing those mechani- (iii) Spelling:The third example to be cal factors which influence briefly described is an interactive blood pressure readings, audio program designed to help and those tertiary students who have hearingavarietyof significant problems with spelling. Korotkoff sounds from One technique is to read a sentence people ofdifferent age containing the particular word and groups and health status. at the same time display the sentence X26 Using Computers Intelligently 217 on the monitor. In this example, the students on one campus, but hopefully aural, visual and in-context clues are in similar courses offered at other edu- all important parts of the task. One cational institutions. of the greatest problems encoun- tered by students is converting a The Production Team lecturer's comments accurately into the written word. As can be appreci- A successful model for the development of ated, without an audio componentcomputer aided learning materials has to this basic literacy program, the been proposed by Bork (1982, 1984) and strategy would loose much of its adopted here as the production process. In impact and relevance.. essence, the approach encourages the use of a multi-discipline team on everyproject, Criteria for selecting Interactive with the assumption that the integrated audio effect of a group of specialists contributing to the project will result in a better product In common with most computer basedthan the concerted efforts of an individual. learning projects, there is a considerableAuthoring programs, which are designed commitment in terms of time and resourcesto encourage any or all teaching staff to in developing each lesson. Therefore whencreate their own computer aided lessons, academic staff submit a proposal or request tend to discourage this team approach. for the development of a technology based lesson, considerable effort is spent onAlthough only a few interactive audio matching these requirements with an ap- programs have so far been produced at propriate choice of educational technol-Footscray Institute, already a scheme has ogy, ascertaining as clearly as possible theevolved whereby the resources of a variety learning skills involved and detailing theof people are used in the production proc- desired outcomes. ess. Subject expertise is provided by the teacher who is ultimately responsible for Projects selected for development with the the accuracy of the content. Special input interactive audio tool have generally satis-comes from an instructional designer, a fied both the above philosophy and thecomputer programmer for developing following criteria: special effects, a script writer, graphic art- ist, and professionally trained voices re- 1. The subject matter must have a critical corded in broadcast studios. dependence upon aural cues and spe- cial sounds for conveying the desi: ed Production time for a program is typically knowledge or skill, ensuring that only about five weeks. However several proj- those projects considered "appropri- ects are concurrently being prepared and ate" for interactive audio are developed are all at different stages of development. with this tool. Given appropriate conditions, it appears that one new program could emerge from 2. Individual or small group attention is the process about every 10 days. required to cater for a large range of student backgrounds and abilities. Evaluation 3. The project must be of fundamentalProduction of an interactive audio pro- importance to a curriculum offered on gram is only one of the phases involving this campus. Ideally the project will bethis team. Academic staff and the project of value not only a large number ofco-ordinator are also committed to the 227 218 Using Computers Intelligently process of delivering this material to stu-student who stated that she "hatedcom- dents, evaluating the effectiveness oftheputers" and complained that she "felt dis- project, and performingany programadvantaged from her peers because of her maintenance recommended during thelack of experience with computers". evaluation. The interactive audioprogram "Measuring Blood Pressure" hasso farOne class of twenty-two studentswas used been field tested witha significant group ofas a control group in the evaluation. These students (150 first year nurses),and thestudents were asked to attend the skills "Foetal Heart" program isstill underlaboratory in the traditionalway, but after- evaluation. Academic staff and studentswards were given the opportunityto use have reported successfuluse of the technol- the interactive audio program. Comments ogy in the strongest positive terms, anda from the control group covered the follow- few suggested improvements havesinceing points: been incorporated into theprogram. Some wished that they had used the Nine academic staff who participatedin program prior to the skills labora- the evaluation phase reported thefollow- tory. ing observations: Some only understood the correla- (i) The time allocated for teachingas- tion between Korotkoff sounds and sessment of blood pressurewas measurement with the sphygmo- greatly reduced. manometer after using thepro- gram, and (ii) Almost all studentswere able to identify the Korotkoff soundsre- Some heard for the first time the quired for blood pressuremeasure- Korotkoff sounds which they had ment when taking their first blood failed to hear or recognize during pressure reading. Frustration with the laboratory session. learning the concept and acquiring the associated motor skillwas Since using the program with 150 student greatly reduced. nurses, an additional "computer managed learning" component has been added. (iii) Students enjoyed the interactive Each student now logs on witha unique way of learning and the processidentification code, and individual became an interesting,non threat- achievements with the programare re- ening way of learning an important corded. psychomotor skill. At the end of the lesson, studentsare in- (iv)Co-ordinating students so that they vited to record their own comments. Using were able to use the program imme- the the student's identification code, the diately prior to practice in the skillstape recorder seeks a position on the tape laboratory was difficult, due mainly which has been allocated to each student to an insufficient number of interac- and the student is prompted to record tive audio delivery stations. comments for a maximum time of 60 sec- onds. These comments are recorded with- Student reactions to theprogram were alsoout effecting the audio tracks whichcon- solicited, and almost all commentswere tain material used in the lesson. Thismeans extremely positive. The only negativere-of communication with the teacher,to- sponse came from one mature age femalegether with the automatically logged 228 Using Computers Intelligently 219 achievements, now provides a much more greater than a corresponding inter- sophisticated way of analysing student active audio project. reactions and performances. b.Ease of maintenance and develop- Overall, both staff and student responses to ment: Once a videodisc has been the interactive audio program "Measuring pressed, its contents cannot be Blood Pressure" were very encouraging. changed. If application of the disc Based on this first level of experience with in the teaching task reveals some the tool, it can be claimed that interactive design problems, the controlling audio is a potentially useful technology, computer program has to be re- particularly for teaching skills and con- vised to accommodate the error, or cepts which involve some auditory stimu- alternatively, another videodisc lus. must be pressed with the errors corrected. Comparing Interactive audio with Interactive Video: With interactive audio, the project can be fieldtested in the classroom Throughout this paper reference has occa- using either analogue tape or digi- sionally been made to the companion tech- tal (magnetic) format for the source nology, "interactive video". As a versatile, of audio. Improvements and modi- creative, and effective medium, videodisc fications to the content can be made has captured the imagination and enthusi- as a result of the trials, and the asm of people in education, training, indus- medium need only become fixed try and commerce in many countries when the final compact disc is throughout the world. This is an appropri- pressed. ate place to compare the two technologies according to some specific criteria. c.Ease of Production: The level of complexity of a videodisc produc- It should be acknowledged that each of tion, particularly one for educa- these two technologies has similar but dif- tional purposes, can result in exces- ferent attributes, and care should be taken sive lead times. For example the to selectively apply the appropriate tech- videodisc produced in South Aus- nology to the particular teaching task. In tralia for teaching English as a Sec- the discussion below, we have made a ond Language ("The Aussie Barbe- comparison of the two approaches assum- cue") has been more than 12 ing a hypothetical case where both appear months in production. On the other to be equally suited to a given task. hand, interactive audio projects are considerable less complicated, and a. Costs: Interactive video is an ex- small production teams can de- pensive technology, in terms of (i) velop programs within a few the size of a production team, (ii) weeks. facilities and time needed for pro- duction, and the (iii) equipment d. Effectiveness and appeal: If the task needed for deliver), of materials to requires the display of visual im- students. It is estimated here that ages representing the dynamic the production costs of a 12 inch realworld, then, as suggested ear- videodisc project (involving con- lier, video is the natural choice. siderable amounts of original video However, if the task can be ade- quately described with computer shooting) would be about 30 times 229 220 lls;; Computers Intelligently graphics (including still pictures), interactive audio which have been de- then an interactive audio program scribed in this paper. The examples which could be an attractive, effective and have been given from our own program low-cost alternative solution. development area illustrate how matching these features to specific teaching tasks can The learning effectiveness of each result in efficient and effective computer mode of presentation could be a aided learning. suitable topic for future research, especially if such a study incorpo- This paper has also emphasized that sev- rated an analysis of the different eral criteria can and should be used in learning styles of students and the selecting an appropriate educational tech- different modes of assimilation of nology to apply to a specific task. Such information. selection criteria can include the effective- ness of the technology and how it relates to e. Versatile applications: A videodisc the different learning styles of students, as produced for one task can be used well as the flexibility, versatility, simplicity under different program control for and cost of the production processes. new purposes. This versatility of the medium means that greater The work reported here also suggests that value can be extracted from the interactive audio performs well against videodisc than originally intended, most of these criteria, and represents a real although this "re-purposing" usu- alternative to its companion technology, ally makes us of only the visual interactive video, in a number of specific material stored on disc. However, learning situations. Additionally, program with interactive audio programs, designers should note that interactive there is less likelihood that materi- audio and interactive video can comple- als produced for one program will ment one another if a situation could bene- be of any value or relevance to fit from simultaneous application where an another. independent, flexible audio source is used with stored videodisc images. f.Dual Applications: When a normal videodisc player is used to recall The impact and importance of interactive and display a single picture or still audio in computer aided learning, espe- frame, there is usually no associ- cially at tertiary level, is still to be fully ated audio. However, if an interac- recognized. In those instances where the tive audio system is simultaneously specific attributes of interactive audio are under the same program control, matched to both the learning task and to the then it is possible to use natural learning style of students, this technology voice and other sound effects to is strongly recommended to educators in complement the visuals coming the field of computer aided learning. from the videodisc. Acknowledgements Conclusion The author gratefully acknowledges that a Interactive audio is an effective teaching grant for the "Disadvantaged Students process which has an appropriate place in Project" from the Commonwealth Tertiary the spectrum of alternative teaching meth- Education Commission was used to ac- ods, and itis suggested that attentionquire equipment and develop interactive should be given to the special attributes of audio programs.Footscray Institute of 230 Using Computers Intelligently 222 Technology has also contributed to this development by creating a special full time academic positionfordeveloping technology-based educational materials. References Bork A., (1982). University Learning Centres and Computer Based Learning, ICST, 12(2), 81-86. Morrill, W.H., Detting, E.R.; and Hurst, J.C., (1974). Dimensions of Counsellor Functioning. Personnel and Guidance Journal, 52, 354-359. Sillitoe J. and Gorman, J., (1987). The Mor- rill Cube as a Theoretical Underpinning of Participation and Equity Initiatives. Paper presented at the HERDSA Con- ference, September, Perth. 231 Applications of interactive video S. M. Byard and N. A. Shaw Footscray lnstitute of Technology, Language competency is a significant and growing pnblem amongst students at the appropriating the codes and nuances of Footscray Institute campus, and traditional language required of tertiary students. For methods of instruction appear to require sup- example, many of our students have diffi- plementation in order that increasing numbers of students can be afforded the necessary assis- culties with such notions as "tone, rele- tance. This paper discusses one initiative which vance and jargon", the distinction between has been mounted to address this problem us- inductive and deductive argument is rarely ing interactive video instruction. The experi- understood, and the ability to handle ences gained during the introduction and use of this medium on a large scale are presented simple and complex sentence structure, together with some student reactions to the especially for those who rate themselves as trial. weaker than the average student, seems clearly to lie outside their language reper- toire. The Institute has been conscious of There is a high proportion of students atdifficulties encountered by these students, Footscray Institute who are recent immi-and througl.:Institute Language Policy grants. This is, in part, due to the locationCommittee, the Department of Humanities of the Institute in the western region of and the Educational Development Depart- Melbourne which is a traditional settle-ment, has been exploring a range o; strate- ment area, and these people naturally lookgies to address this problem. to FIT as the major tertiary Institution in this region of Melbourne to provide terti-As part of these initiatives, interactive ary education for themselves and theirvideo delivery systems acquired under a children. Many of these students have beengrant from the Commonwealth Tertiary educated to tertiary or senior secondaryEducation Commission for increasing par- level in a non-English language, and have ticipation in tertiary education by disad- achieved a satisfactory grounding in cogni-vantaged students, have been introduced tive skills such as mathematics, engineer-into the regular teaching programme on a ing and science areas.However, thesetrial basis.Featuring the commercially students are "underprepared" for tertiaryavailable programme "Writing For Re- education in Australia because their sults" designed by K. Stuart and D. Limon, mother tongue is not English. In addition,this presentation technique was integrated other students may have had the majoritywith the Written and Oral Communication of their secondary education in Australia,subject taken by students of science and but live with families who have a first lan-engineering. guage other than English. Interactive video systems have been the As a result, there is a considerable fraction subject of previous seminars at this and of our students who have difficulty inother conferences. We present this paper as r.> r) t.; Using Computers Intelligently 223 a report of the trialling of this technique in munications course since the stated objec- our classrooms, and clahr. that ells experi-tives of the HS course focused particularly ence has enabled a better appreciation ofon the need to: this mode of instruction to be gained by members of teaching staff, and has enabled a. understand the basic elements of effec- us to observe the difficulties likely to be tive writing in organizational settings, found by other educational and industrial and organizations seeking to implement such computer-based instructional schemes. b. apply a planned or disciplined ap- proach to writing. The delivery system and software. Clearly, in the light of our earlier comments Hardware involved in presenting the pro-regarding the background of many of the grammes to students consisted of a per-students at FIT, these objectives integrate sonal (compatible) computer with hardwell with the aims of a general literacy disc, a video disc pia} ter, colour monitorcourse at a technological institution. with fast blanking, and the iv4IC-2000 con- trolling interface card. This combinationAlthough a student text had been previ- offered the facility of overlaying text andously set for this subject, our traditional graphics information generated by theapproach to teaching had also emphasised computer with video images coming fromthe use of "in-basket" techniques, real life the video disc. Software consisted of con-exercises and audio-visual material to pro- trolling programmes (together with certainvide authentic contexts for students work- text and graphics overlays) stored oning on language development.Thus floppy or hard disc, and a set of corre-"Writing For Results" could be seen as a sponding videodiscs. complement to eid sting curriculum materi- als. In addition however, interactive video "Writing For Results" was produced by seemed to offer us the possibility of extend- Stuart and Limon for the London baseding these strategies in the letter/memo company of Interactive Information Sys-section of the course by allowing students tems (HS), and marketed in a PAL video to obtain repetitive reinforcement of sound disc format. This training course is mar- principles of effective writing in a collabo- keted as being "...designed to help yourative context which was not unduly write more effective business letters andteacher-intensive. memos".It consists of four video disc modules entitled "The Basics, PlanningConsequently, it was agreed by the rele- Your Writing, Writing with Style, andvant standing Committee of the Humani- When the Going Gets Tough", each struc- ties Department to use 'Writing For Re- tured to take approximately an hour tosults" during the first five weeks of Semes- complete. These provide a comprehensiveter One 1987 when students were to be and sequential course, although eachworking on an outline of communication module is also self-contained. theory and a unit on letters and memos. The educational rationale for this The design of the project approach. We had initially planned that the interac- tive video modules would replace conven- The Humanities Department of the Insti-tional instruction for one hour a week over tute considered Stuart and Limon's mate-a four-week period for a selected cohort of rial suitable for the Written and Oral 0--n-students. This proposal was contingent 233 224 Using Computers Intelligently upon suitable supervised accommodationundergraduates enlisted for the trial were for the equipment and students beingin six groups of full-time day students: five available within the Humanities building,groups from first year Computer Technol- but this proved to be impossible to organiseogy and Digital Electronics and Comput- with relevant teaching space and staffing, and one group of second year students being unprecedentedly loaded during in Building Engineering.One further 1987. group of part -time evening students in the second year of an Occupational Health and Further:mare, because of the heavy time-Safety Associate Diploma withdrew imme- table for students in Science/Engineering itdiately because of timetable constraints. became clear that it would be difficult toHowever, the change in project design obtain sufficient free hours in a four weekdiscussed above caused an added reduc- period for th one hundred (plus) studentstion in the number of groups participating to make full use of the modules outside offrom seven to three, reducing the student timetabled hours. Within the scheduledpopulation from 105 to 45. Further com- class times, even using both delivery sys-ment on this reduction in groups will be tems, it was evident th it a one hour timemade later with the intention of highlight- slot was not enough time for most students ing some pragmatic difficulties regarding to work through a module. Subsequentthe introduction of new technology into experience in use of this equipment in theexisting discipline areas. Educational Development. Department at the Institute now indicates that most ;'tu-Some practical considerations of the dents take six to eight hours to use all fourmethod of delivery. modules, even when working in small groups. As with all valuable equipment, there was With these problems in mind, and becausea necessary requirement for supervised space in which to house the interactive of the added observation that many of the students seemed so keen to experience thevideo system. In addition, we found the material, it was decided that we wouldneed for close supervision of some of the encourage all students to work through thestudents. An attempt was made by staff to modules. We decided that: check with students at the beginning and end of each module and at regular intervals a. all students in designated classes werewithin the hour, which, although adding to the time commitment to the project, gave to be timetabled for the use of the mate- the teacher a flexible opportunity to moni- rial in groups of five while they simulta- tor each student's grasp of the material. It neously followed the prescribed course of instruction, and also had the benefit of checking whether one student was monopolizing the key- board, or observing if students were work- b. all students were to be asked to compare the interactive video-based learninging in languages other than English con- trary to instructions. with the other mode and to complete a questionnaire at the end of the four- The former problem arose since the mod- week sequence of study. ules were designed for individual use, and naturally encouraged single-user re- Student.; involved in the project sponses. Because the majority of the stu- dents were in the first year of their tertiary Initia 'Iv, 105 students were identified ascourse, the expectation that they should participants in this pilot scheme.The adapt to small group work smoothly and I ktWt 234 4tj , Using Computers Intelligently 225 without difficulty is perhaps a little unfair. Student reaction to the use of In addition, the physical requirement for interactive video groups of students to congregate in a con- fined area in front of the monitor led occa- Questionnaires were distributed to all par- sionally to what one student referred to as ticipants in the pilot scheme and complete_ a "tendency toward group larrickinism",by those of the students who were still requiring a certain amount of social control accessible to the project team in April/May by the lecturer. during their next unit of work. The re- sponse rate can be seen from the following Whereas none of these problems are spe- table: cifically due to the equipment or material involved in interactive video, it is neverthe- less an important area of consideration Course of study NumberNumber when planning the introduction of new of students of techniques with large groups of relatively respondents unsophisticated students. Building Engineering 17 11 Computer Technology 28 21 Reactions of staff involved in using Total students still the system. accessible to trial 45 32 The marked reduction in participants in the Table 1: Accessible respondents in trial from seven groups totalling 105 stu- "Writing For Results" survey. dents to three groups totalling 45 students arose partly from the difficulties that stu- dents and staff experienced in organizing A copy of the questionnaire is appended to access to the equipment at times of maxi-this paper. In the questionnaire, we asked mum demand. In addition, part-time staffthe students to estimate how long it took were unable to continue with arrange-them to work through the modules, how ments which took them beyond their con- they had liked working in a small group tractual commitment to the Institute. and whether they had reviewed any of the material since finishing Unit One of the We found that without technical assistance Written and Oral Communication Course being continually available, simple me-(questions 1 to 5). We also sought their chanical difficulties,(like running the(forced) opinion of the most convenient wrong video disc for a chosen unit of thelocation for further use of the equipment programme so that video and text failed to and their perception of the chief ber.efit of synchronize, or failing to notice loose the material relative to other styles of in- leads) were a common source of embar- struction. (questions 6 and 7). rassment for Humanities staff who were not familiar with electronic equipment.Qucstions 8, 9 and 10 determined the This perhaps contributed to the decision of student's previous experience of comput- some staff to withdraw .from the trial.ing, and question 11 asked them to rate While genera'technical support wastheir own performance in English com- available to project, it is dear that staffpared to their peers in the course. Finally, intending to ue such a system and materialquestions 12 and 13 were open questions must become personally comfortable with which allowed the student to comment on the technique by having ample non-threat-what they perceived as the best and worst ening opportunity to practise using thefeatures of the the "Writing For Results" equipment before incorporating it into material. their classes. 2 3 3 226 Using Computers Intelligently The following comments were received asThere was a strong preference among stu- responses to the open questions 12 and 13. dents for future location of the material in These responses are presented with the'the audio-visual section of the library. This reminder that only 45 out of the initial 105probably relates to the limited hours of students were accessible to the project, and opening of the Humanities and Educa- of these, 32 responses were gathered. Also,tional Development Department venues we should comment that the respondentsrelative to difficulties with the students' were the most consistent members of theirtimetable referred to above. For example class in terms of both attendance and sub- during the first semester the EDD was shut mission of written work. at lunch time and after 4 pm, except by special arranger" znt.Despite this more Some of the advantages seen by stu- than 50 students used the interactive video dents were: while it was housed in EDD in 1987. makes the person using the system more Some students and staff noted the relative aware ... he pays more attention to video inflexibility of the IIS programme, espe- because he is interested. cially in the first module which restricted the exit / review possibilities. Other com- Excellent acting and production. Ideas ments were : clearly presented... easy to understand. No scoring incorporated It was a unique and I found effective No questions can be asked. Need a way of learning. general scoring result at :he end of the module. On the whole students responded posi- Doesn't have a printer to print out a tively to the interactive video even though permanent copy of questions and their it took them longer than an hour to work answers. through each module and some had some misgivings about the small-group style ofIt was common for students to have diffi- study: culty catching what they considered to be key details, for example the names of per- Not enough equipment (video, key-sons on the screen, although this might board) so we have to share with manyeasily be addressed by offering users a other students and never touch thescenario introduction rather than the topic keyboard. summary which is currently supplied. Others found the material boring: Preliminary analysis of student pc,-formance. After the first disc the rest became rather repetitive in the film and the questionsIt is the subjective opinion of the project asked. team that as a consequence of using Interac- tive Video, many weaker students made Most rated the material highly. Two thirdssignificant gains in language competence. of the Computer Technology students saidTheir grasp of the context in which formal the material compared most favourablywritten communication takes place and of with the conventional instruction, agreeingconceptual material crucial to consciously that "This material gives a realistic pictureincreased mastery over written language of life in a business firm which I wouldn't appeared to show considerable improve- otherwise have." (Q. 7(a)) ment. .236 Using Computers Intelligently 227 In common with the staff in the Educa- tional Development Department who useConclusion. of this teaching and learning method in parallel Language Development work, weThe results of integrating interactive video suggest that the introduction of interactive material into the "Written and Oral Com- video into language programmes providesmunications" course at the Footscray Insti- an important alternative learning focus for tute of Technology suggests that this the student.This encourages student-method offers a useful complement to centred rather than teacher-centred learn-conventional instruction techniques. Fur- ing, a central component of the culture of thermore, its use in small group settinss in tertiary education. In addition, such mate- the context of classroom learning offers rial appears to consolidate what studentsstudents the chance to make significant have already learned, and when used by a gains in the appreciation of new strategies small group, provide opportunity for use- for formal communication, and in mastery ful conferencing and peer interaction. Asof the principles of formal writing. indicated earlier, we should advised some caution in the use of this technique forTwo key sources of difficulty for Institu- groups of students, particularly at the tions seeking to develop the use of interac- lower level of course offering with rela- tive video technology as a means of instruc- tively unsophisticated students. tion appear to be the provision of suitably supervised access to students, both in Interactive video was specifically designed terms of location and time, and the neces- for individual response, but notwithstand- sity to provide effective preparation for ing the advantages increased utilizationthose who will supervise the learning proc- and the benefits of peer interaction, the ess. forcing of the technique to service several students simultaneously needs increasedBibliography teacher involvement and an added dimen- sion of sensitivity by the teacher to the Milheim, W. D., and Evans, A. D., (1987). dynamics of group interaction. Using Interactive Video for Group Instruction, Educational Technology, The final point to be made in assessing June, 35-37. gains by students in this trial is to note that for many this was an empowering expo- Acknowledgement sure to new technology. A significant sub- group said they I-ad no previous experi- The authors wish to acknowledge the assis- ence in computing (4 of the 21 students intance of Dr. James Sillitoe, Head of the Computer Technology and 4 of the 11 sec-Educational Development Department at ond year Building Engineering students)Footscray Institute of Technology, in the but were nevertheless able to complete apreparation of this paper. significant proportion of the course. As with language skills, we believe that com- puting skills are more adequately learned in a meaningful context, and the long-term benefits of this type of exposure to com- puter technology cannot be overlooked. 228 Using Computers Intelligently Sheila Byard is a lecturer in the Hu- manities Department at Footscray In- stitute of Technology and has been teaching tertiary students and other adult learners for over twenty years. Whilst employed in the Adult Migrant Education Programme she was in- volved in several projects based on self-paced learning principles, and at FIT has been particularly interested in schemes to encourage language devel- opment opportunities for disadvan- taged students. Neil Shaw, formerly a senior lecturer with the Physics Department at Footscray Institute of Technology, is now heading a series of initiatives which involve educational technology for the enhancement of learning by disadvantaged students.Dr. Shaw was the recipient of three significant CfEC Equity grants during 1985-7, and has developed as series of interac- tive video and interactive audio pro- granuws suitable for use in tertiary institutions. 238 Simulation and queuing theory: A combined approach W.W. Read and D.A. Waugh Department of Computer Science James Cook University ,1 proach.In the mathematical approach, CAL applications in the field of mathemat- attention is usually focused on obtaining ics have not been extensive; however, CAL queuing system statistics analytically; techniques have been applied at different levels in applied sciences for some time. simulation may be given some attention, in This paper focuses attention on simulation a much more minor role, than the analytical as an aid to understanding queuing theory. analysis. On the other hand, the systems In courses on queuing theory, the purpose approach usually incorporates simulations of this is two fold - in a purely mathemati- as the main vehicle for obtaining system cal course, tutorial time can be more effi- statistics; theoretical results for some sys- ciently used to increase students under- tems may be quoted with little of the standing of the simulation approach. In a mathematical background given. A stu- systems course more time can be spent on dent enrolling in both types of courses theory, with less lecture time spent on the usually receives a good grounding in most practicalities of simulating queuing sys- aspects of queuing systems. tems. A basic model is presented together with the software design and initial imple- However, a student wishing to obtain some mentation on a Macintosh Plus personal computer. knowledge of queuing systems with a minimal enrolment requirement may miss out on a large proportion of the knowledge Queuing and queuing systems are funda- required for even a basic understanding, if mental to modern civilization, and almost everyone has had contact with a queue at he or she completes a course in only one of the above options. some point in their lives. Who hasn't been caught in a queue at a store or supermarket, It is the intention of this paper to focus and wondered if the service could be ar-attention on the development of a simple ranged more efficiently?Unfortunately,simulation model implemented on a queues appear to be increasing in fre-Macintosh Plus personal computer. The quency (as well as length!) and somemodel is proposed in order to aid both knowledge of systems analysis and queu- approaches, so that more lecture time can ing theory is relevant to an expanding be spent on simulation theory in the mathe- range of degrees and diplomas. This im- matics course, and "pure" queuing theory poses some responsibility on tertiary insti- in the simulation course. In this way, it is tutions to provide pertinent courses on thehoped a student enrolling in a minimum of subject, particularly when enrolled stu-courses will receive as broad a grounding dents are not majoring in mathematics or as possible in queuing systems. systems subjects. A Simple Queuing Model The teaching of queuing theory can be roughly categorized as either a theoretical A large variety of simulation models and mathematics approach or a systems ap-languages (Melamed & Morris,1985; 23 5 230 Using Computers Intelligently Ramakrislulan & Mitra,1982) exist today,typical mistake is using random numbers and it is not the intention of this work to more than once. Another area is in the attempt to develop a completely newsimulation of service time- quite often model or language.Instead, a simplestudents simulate service times when the model will be used to enhance the teachingqueue is empty. The knowledge required of some of the fundamentals of queuingto overcome mistakes such as these is very simulation, rather than obtaining knowl-basic and a minimum of lecture time edge about a queuing system as the ulti-should be (and usually is) spent on it. mate goal.Students completing a fullHowever, a student must learn these course in queuing theory usually becomepoints, and the appropriate place is in a proficient in the use of a model and lan-tutorial. The ease and simplicity of using a guage. However, time must be spent toMacintosh Plus, makes it an obvious ve- gain a reasonable degree of proficiency inhicle to use to teach students about funda- using such a system model, time spentmentals such as these. learning about the model and language, rather than queuing. A student who isThe basic model will be a G/Gim queue enrolling in a minimum of courses could(General inter arrival and service time dis- possibly spend this time more construc-tributions, m servers). Although, for given tively on queuing theory, both pure andinter arrival and service time distributions, simulation. However, the fundamentals ofthis queuing system is easily simulated, simulation must still be learnt. Computer there is no analytic solution (Kleinrock, graphics provide a very efficient teaching1975). The simplicity of simulation will medium (Benzon,1985) and with the profu- allow attention to be focused on the basics sion of such relatively cheap graphics ori-of the simulation approach, rather than entated micro-computers as the Macintoshobtaining queuing statistics from a com- Plus, also a viable one. One method ofplex problem. The features the me del must simulating a queue manually is via a table. incorporate are as follows: The number of people in the queue,current departure times, arrival times etc. are ar-1. Inter Arrival and Service Time Distribu- ranged in columns, and updated as the tions: A selection of analytical distribu- simulation progresses. As noted by previ- tions must be available, such as negative ous researchers (Arganbright, 1985) elec- exponential, poisson, etc. As well, some tronic spreadsheets provide an excellent capacity for user supplied discrete interactive teaching medium for mathe- probability distributions in the form of matical modelling, and a variation of this histograms must be provided. approach will be followed in this work. It has been noted (Sonenberg,1985) that 2. Random Number Generation: Initially, computers are not used very extensively in random numbers are to be selected the teaching of mathematics; this is one individually by the user, with some area in which computers should and can provision for automatic selection later. play a major teaching role. 3. Error Detection and Recovery: A com- It has been the author's experience that prehensive set of common mistakes and students (and consequently tutors and lec- an appropriate response must be ca- turers!) spend a lot of time, usually in a tered for. The response must be avail- great deal of confusion, learning the very able at different user selected levels (e.g. basics of simulation. An example is in the beginner, intermediate, advanced), and use of random numbers to simulate points employ the graphics capacity of the from a given probability distribution- a Macintosh Plus as much as possible. 4 Using Computers Intelligently 231 4. The simulation must be presented visu- ing system can commence. A random ally as well as in tabular form. number may be selected from a table, and used in conjunction with a graphical repre- 5. Design should be efficient and as usersentation of the appropriate cumulative friendly as possible, with a comprehen- distribution function to simulate an inter- sive help capability. arrival/departure time. Alternatively, the inter-arrival/departure can be generated Simulation Environment automatically from the appropriate distri- bution.This is then used to obtain an Initalization - the user is given the choice of arrival/departure t me, and entered into re-activating a previous simulation ses-the worksheet.The worksheet is then sion, or commencing a completely newupdated to the next arrival/departure. session. At any point during execution the session may be saved, under a user defined Error detection and help-upon detection of name, or terminated. In either case a sum-an error, the incorrect entry in the work- mary of the statistics associated with thesheet is highlighted, and the state of the simulation is presented. The level of helpqueue at the time of the error displayed. must be specified as either beginner, inter-For instance, a departure time entered into mediate or advanced. Beginners will bea server's column when no-one is in the modelling a G/G /1 queue, whereas inter-system first causes the incorrect entry to be mediate and advanced users will be model-highlighted.The visual display of the ling a G/G/n queue, and therefore mustqueue then shows a dotted outline of a specify the number of servers in the queue. person being served, indicating a nonexis- The inter-arrival/departure distributionstent service time. A very brief error mes- may be chosen from a range of analytical sage is displayed, with further help an distributions, or a discrete histogram in-option. The three help options range from put. In the latter case, the user is requiredbeginner to advanced, with the higher lev- to supply the appropriate step functions. els progressively briefer versions of the beginner level advice. Options - a graphic display of the queue during simulation is one of the main op- Software Design tions available, with the default being a tabular summary only. Breakpoints can beThe basic system is composed of the fol- set, at which times the simulation islowing meta-modules, (see fig 1), I/O stopped, and system statistics displayed.Controller, Queue Simulator, and the File The simulation can also be "stepped"system. The user interface is designed to through automatically in constant timemake maximum use of the toolbox routines increments, v.ith a statistical summaryprovided by the Macintosh operating sys- displayed at each step. Percentiles for the tem. All input and output is directed various statistics may be set, with the de-through a window of some form, either a fault taken as 5%, giving 95% confidence dynamic display or a dialogue box. A facil- intervals on average statistics. The default ity is provided to save a session in progress service order is first-in-first-out, with other for resumption at a later time. service orders (e.g. last-in-last-out, ran- dom) available.The number of peopleThe Queue Simulator is designed with the actually in the queue initially may also be following sub-modules, (see fig. 2), Com- set, with the default taken as zero. mand Processor, Simulation Controller, Queue Editor, and the Model Generator. Simulation - with all the parameters andThe Command Processor handles the inter- desired options set, simulation of the queu- action with the user through the I/O Con- 4 1 232 Using Computers Intelligently High Level Data Flow Diagram User Saved Session4,10 Figure 1 troller, processing the various commands that a user may select from the various'Controller through the Command Proces- menus. Editing information is passed on to sor. the Editor module while commands that concern the running of a simulation are The machine specific details are restricted passed on to the Simulation Controller. Theto the I/O Controller module so that Queue Editor module controls the user'sporting to another machine will limit the input to the system including the entry ofscope of the modifications to this module. simulation data and completion of the de- tails at each discrete step of the queue being Summary and Conclusions modeled. The Queue Editor gets the re- quired results for a specific distribution byA basic queuing model has been presented, extracting the appropriate formula from a together with the software design and pre- stored library of distribution formulas. Ifliminary implementation on a Macintosh parameters are required by the user, a re-Plus personal computer.The model is quest is made for a valid value for each aimed at assisting the learning of the funda- parameter involved. The Model Generator mentals of simulation techniques, particu- actually controls and models the presentlarly when the student is enrolled in a and previous states of the queue. The de- minimum of queuing courses. In view of tails concerning the queue's status are keptthis, the focus of the design is in the basics in the Queue Status data sink/source. The of complicated systems. The Macintosh Simulation Controller handles the actualPlus has a comprehensive graphics capa- simulation of a queue by the system forbility, and the user friendly features have demonstration purposes. Commands suchbeen exploited to provide an easy-to-use as setting breakpoints and stepping environment for the novice computer user. through the simulation are handled by theThe simulation model is available at a vari- Simulation Controller. The current status ety of levels, so that the more sophisticated of the displays and any necessary modifica- user can avoid some of the details so neces- tions are eventually passed on to the I/Osary to the novice. ,242 Using Computers Intelligently 233 'N Detailed DFD of Queue Simulator Info t Commands 4-QueueData Status Processor Data Display Status Updates Elnfo Parameters Modifications Formulas Data Distributions Initializations i Figure 2 The software is at present in the prelimi-References nary stages of development, and comple- tion and comprehenshe testing with stu- Arganbright, D. (1985). Mathematics and dents is required before any detailed con- MathematicalModelling Using clusions can be drawn. However, the au- Spreadsheets, Proc. Conf. CALITE, thors feel that the model will provide a Melbourne, pp. 141-150. valuable learning tool, particularly when Benzon, B. (1985). The Visual Mind and used in conjunction with a theoretical The Macintosh, Byte, pp.113-130. course in queuing theory. The iruxlel ap-Kleinrock, L. (1975).Queuing Systems pears to be an ideal vehicle for applying Volume 1: Theory, Wiley-Interscience. CAL techniques in the field of mathemat- Me lamed, B. and Morris, R. J. T. (1985) ics, and so may play a non-trivial role in this Visual Simulation: The Performance field. AnalysisWorkstation, IEEE COM- PUTER, August, pp. 87-94. Ramakrishnan, K. G. and Mitre, D. (1982; An Overview of PANACEA: A Soft- ware Package for Analyzing Markovian Queuing Networks, Bell System Technical Journal, Vol. 61, No. 10, pp. 2849-2872. Sonenberg, E. (1985). Possible Impacts of Computers on the Teaching of Mathe- matics, Proc. Coni. CALITE, Mel- bourne, pp. 133-140. 243 Using Computers Intelligently Dr. Read was born in Ayr, Queens- land. He received a Bachelor of Science with First Class Honours in mathe- matics from James Cook University of North Queensland in 1982 and subse- quently completed a PhD at James Cook in 1986. During the course of his doctoral work, he was employed by the Civil and Systems engineering department as a Research Officer. He is currently a lecturer in the depart- ment of Computer Science at James Cook, where he lectures in algorithms and data structures. His research inter- ests include sorting, signal processing, computational mathematics and com- puter aided learning. 244 Instructional systems design: The hidden agenda Rod Sims and Pascal Grant Computer-Based Training Consultants Expert Solutions Australia Often touted as the messiah of cost-effec- ing forms of training, it already is quite tive training, Computer-Based Training successfully in many Australian organiza- (Clip has only recently begun to deliver tions. Rather, the central issue concerns the the goods. Trainers who once believed Quality aspect of CBT courseware. The novice authors using simplistic authoring essential criteria for a successful CBT strat- systems were capable of developing high- egy is not whether one has developed quality courseware, all at minimum cost, hours of CBT slowly replacing traditional now recognize the need for experience and quality. Current emphasis is now placed forms of training and seemingly saving the on the need for instructional design and organization precious dollars. What will experienced authors for effective course- determine the effectiveness of any CBT ware applications. This paper quantifies project is whether the training addresses the real costs of developing aCBTsolution the learning requirements of individuals from the Instructional Systems Design in the field. (ISD) to the actual authoring of course- ware. Specifically, the time and costs in- CBT offers a unique characteristic com- volved with ISD and the variations in pared to other forms of training, it is inter- courseware format are identified as major active. The learner can be tested in a private determinants of the overall costs of CBT environment without intimidation, he/she applications. Through defining a "check- list" for courseware development, the can practice until mastery is achieved and authors provide a guide for estimating the with the use of sophisticated graphics and effort (time, materials and cost) required animations, can be presented with highly for producing computer- delivered train- effective learning strategies. If organiza- ing. From a practical and experiential ba- tions fail to meet the criteria of learning sis, the conclusions of this paper are that effectiveness, given the powerful charac- CBT has and will continue to provide a teristics of sophisticated interactions - CBT significant cost-saving potential for train- can and will be relegated to the bottom of ing. the training cabinet. There is little doubt that microcomputerThe purpose of this paper is to draw atten- based quality Computer-Based Trainingtion to the issue of creating quality course- (CBT) courseware is an extremely power- ware. In order to achieve quality course- ful training medium. This is especially true ware, serious attention needs to be devoted if an organization wishes to train large to the following aspects: numbers of individuals, spread over mul- tiple sites and the subject matter is not likely to be volatile. 1. Instructional Systems Design (ISD) 2. Authoring The issue therefore, is not whether CBT can 3. Project Management survive as an effective complement to exist- 4. Project Personnel 245 236 Using Computers Intelligently Instructional System- Design approach. In undertaking any CBT projectmany or-b) Improving managementof instruc- ganizations have failed to take into &ccount tional design and development by that before any CBT developmentcan take means of the monitoring and control place, an extensive analysis and design functions of the systematic approach. component is essential. Indeed the pre- development stage of creating qualityc) Improving evaluation processes by courseware referred to as ISD can often means of the designated components take more effort and time than the actual and sequence of learning events, in- authoring stage. As an estimate of the time cluding the feedback and revision taken to develop an hour of CBT,course- events, inherent in models of systematic ware development groups consistently instructional design. devote up to 60% of the total project time in the ISD component. d) Testing or building learning or instruc- tional theory by means of theory-based Instructional Systems Development is nota design within a model of systematic new methodology, it grew out of US. De- design. fence training projects in the early 1960's (CYNeil, 1979). Since that time it hasgrownThe purpose of a systematic approach to to become a highly successful and rigorousinstructional design is that "it encourages means by which training needs have beenthe setting of a design objective, and it addressed for a multitude of applications.provides a way to know when that objec- The distinction between design and devel- tive has been met" (Gagne & Briggs, 1974, opment is not discussed in this piper, as thep. 228). authors see the terms assynonymous (Romiszowski, 1986). ISD is a proven and tested method ofensur- ing effective training for any CBT under- ISD has emerged in response to thepress-taking. CBT can only be effective if the ing need for a more effective, efficient anddesign stage has been systematically com- reliable educational system. It ix wolves pleted. ISD offers the opportunity to ensure aspects such as task analysis, behaviouralthat all the various components of the de- objectives, criterion reference testing, indi-sign process have been completed and vidualized instruction, formative evalu-validated in t.:der to maximize le inning ation, and media selection. While ISDwaseffectiveness. ISD will provide a detailed originally developed in the context of mili- list of learning poL,ts, suggested learning tary training, it's use has spread to allpathways and required interactivity. The domains of instruction and is probably the subject matter and training objectives are dominant instructional methodology ex- reassessed and the lesson is storyboarded isting today. in full before any authoring can commence. ISD in the CBT development cycle is syn- Turning tothe costs consideration of onymous to the systems analysis phase ofComputer-Based Training development, any D.P. undertaking. The methodologythe instructional design aspect becomes serves four basic purposes: not only an essential ele client for learning effectiveness, but will afro ultimately de- a) Improving learning and instruction bytermine the resource costs of the project in means of the problem solving and feed- terms of time taken. In costing CBT proj- back characteristic of the systematicects, there are certain variables ensuing 246 Using Computers Intelligently 237 from the ISD stage that will ultimately af- Constraints on development times im- fect costs and resources. These include: posed by commercial realities may indeed limit the design process considerably. This the amount and complexity of the sub-is exemplified b'le low level of business ject matter activity with Int,Aligent Tutoring Systems, entry level of student's perhaps the most advanced form of CBT. availability of subject matter expertiseOne of the basic impediments is that one and documentation hour of intelligent instruction delivered by "system" accessibility for training de- computer may take up to one year to de- velopment velop, as compared to the 150-200 hours for the productivity of the authoring sys- the development of one hour of "tradi- tem tional" CBT using pre-determined paths the instructional strategies to be em-rnd an expository learning methodology. ployed, i.e. simulation, tutorial, game or The bask steps of ISD are shown in Appen- drill & practice dix 1. the complexity of the Computer Man- aged Instruction, whether basic studentTo conclude this section on ISD, any or- management or full database integra-ganization contemplating a CBT project tion. should ensure that the required ISD be- comes a major component of the project. It is not the intention of this paper to cl - Perhaps more importantly, the designer pand further on these cost variables. How- must fully understand the cost implica- ever, it should be noted that the designtions of determining the detailed project considerations need to incorporate a de- specifications. tailed specification of the training require- ments. Authoring A word of caution should be noted con- In the field of Computer-Based Training cerning costs of duality courseware devel- (CBT), authoring is frequently given top- opment. If a designer wishes to employbilling whenever courseware and authors complex instructional strategies with mul- are discussed. More often than not how- tiple branching and implicit student mod-ever, the notions of CBT authoring are elling as the determinant of progress based on misinformed or simply incorrect through the lesson, development limes will statements describing the facilities of a ultimately increase in proportion to theparticular authoring system or language. level of complexity. This section provides a realistic definition of authoring by exploding certain nyths This raises an important question concern- surrounding authoring and detailing the ing the differing levels of quality course- relevant issues for commercial develop- ware currently being used in industry. If ment of CBT. business, government and industrial groups recognize the costs of training as the Authoring is the process of creating com- most important CBT selection criteria, it puter-based displays which are presented will come as no surprise that CBT course- in an instructional sequence according to ware developers need to design and de- specified criteria. velop courses cost effectively. Conse- quently, a compromise must be made be- To be able to author, a number of prerequi- tween instructional effectiveness and CBT site steps must have been completed. First, excellence (Hitchcock, 1987). an instructional event (curriculum, course 247 Using Computers Intelligently or lesson) must be identified, with the as-most commonly used authoring languages sumption that CBT has been validated asare Control Data's TUTOR, PILOT and the most appropriate delivery media. BASIC. For authoring systems, the choice often seems wider than the prolific spread- Second, this event will have been subjectedsheet and wordprocessing packages! to Instructional Systems Development processes to provide a set of sequencedRepresentative of current authoringsys- instructional task-c.. objective(s) for eachtems are WICArs WISE, Control Data's task, (optional) assessment protocol forPCD3, MPT's PC Class, Compaq's Top each objective, evaluation procedures and Class, Microcraft's AUTHOR, Progeni's specification of macro and micro instruc-FORGE, Mcdonnell Douglas's A1S-II, tional strategies for the courseware itself. AIP's COURSEMASTER, MARCONI's MANDARIN and Goal Systems' PHOE- Third, a subject matter expert in conjunc- NIX. tion with an experienced author 'roduce "storyboards" which specify the contentAlmost without exception, these products material and interactions, linked to objec-are marketed as being "easy to use" and tives, which are to be displayed to the stu- requiring no "programming" skills. dent. The storyboard also defines specifica- tions for branching and review. However, as more and more of these sys- tems have been made available to the writ- Having completed these steps, authoringers for evaluation, it is apparent that there can begin, using whatever authoring toolare two realities. the organization has installed. With the wide range of authoring tools available, theThe first reality is that all the systems allow type of courseware developed is highlyauthors to create graphics and text using dependent on the facilities and optionssimilar WHAT YOU SEE IS WHAT YOU provided by the authoring software, andGET (WYSIWYG) editing procedures, and the capability of the author to stretch thethat the resulting displays can be linked system beyond its documented capabili-into a simple sequence or branching struc- ties. ture. For example, the WICAT "WISE" author-The second reality is that to achieve interac- ing system allows the author to generatea tive learning and full use of the media, the complement mode whereby graphics mayauthor will have to resort to programming be overlaid and the colours of the twoconcepts, if not actual programming! graphics mixed to create a third colour. However, experienced authors have dis-For example, one of the authoring systems covered that this mode, when combinedevaluated by the writers allows the author with another mode, can allow the overlay- to create a sequence of graphics using ing graphic to be erased from the screenWYSIWYG editing, but then requires the without affecting the original graphic. Us-author to text-edit an authoring language if ing this technique can result in exceptionalany changes to the original display are student displays. Such use of this techniquerequired. In addition, any controls re- only comes after many months of experi-quired within the delivery system must be ence with the authoring system. programmed in the authoring language. In direct contrast with authoring systems Authoring tools come in two types: author-which allow the author to manipulate the ing languages and authoring systems. Thedisplay objects on screen, this second real- Using Computers Intelligently 239 ity highlights the the misinformationWith respect to the second myth, claims which is published concerning "ease of have been made which suggest that novice use" and no requirement fro programmingauthors have created high-quality course- skills. ware in a matter of hours (Sims, 1986). These claims are simply false, as authoring The most appropriate method for differen- systems take time to master and high-qual- tiating authoring systems is to examine the ity courseware implies interaction. Both design philosophy behind the system - andfactors combined mean that effort is re- of course its year of release. In some in- quired to produce such courseware. For stances, authoring systems have been de- example, a substantial courseware module signed to take minimal use of a microcom-will contain at least one hour of instruc- puters options, with template screen dis- tional material. To create one hour of plays and limited graphic options. Con-courseware takes from 175 to 200 hours. versely, other systems have adopted anThis translates to some three hours per approach towards expert systems, with theminute. facility to automate some of the instruc- tional design logic. Progeni's FORGE andWhat then can novice authors create in a Control Data's PCD3 are typical of this matter of hours? Some interesting displays genre. perhaps, but certainly not high-quality courseware. Exploding the Myths The preceding comments have shown au- There are two authoring myths: thoring to be a skilled task and a significant component of the ISD process. The follow- a.it is easy ing discussion analyses authoring in terms b.novice authors can create high-quality of the commercial development of compre- c. urseware hensive, high-quality courseware pack- ages. Authoring is not easy. Certainly it is far removed from the complexities and struc- After the ISD and storyboarding, the au- tures of programming languages, but it isthor must decide on the various course- not easy. A typical authoring system will ware attributes. For example, courseware take 4-6 months to masZer. Mastery implies standards are essential for the consistency the author's ability to contain the full op-and continuity of lessons. Currently, these tions of the authoring system with little jf standards include a screen border with any use of manuals or help facilities. Expe- corporate logo and module identification rienced authors can however adapt to other on the top line and student control prompts authoring systems more rapidly. The writ- on the bottom line. Similarly, standards ers experience has been that a "good" au-must be defined for text colour, highlights thor can convert a storyboard into comput- etc. erized instructional displays at the rate of approximately five per hour, which ThisOne of the most important standards to varies according to the type of lesson and define is that for interaction, which is the the authoring system used. essential element of successful high- qual- ity courseware. Questions relating to the It is true that the commands used to create number of attempts, automatic exiting and display elements or objects are easy to feedback must be resolved. For example, learn, but the authoring of instructionaldoes the student continue to make a re- material is a demanding task. sponse to a question until correct or will the 249 240 Using Computers Intet..jently judging process terminate after a specifiedtems which provide fewer options. As number of attempts or condition has been such, they can promote the quick develop- reached. ment of courseware, but at a sacrifice for full use of the display screen. However, it is The responsibility for interaction lies in thealso important to keep in mind that over storyboarding activity- it should not beindulgence in courseware development based on a set number of text screens fol- should be minimized (Hitchcock, 1987). lowed by a question, but rather on asking the learner to interact by using knowledgeAs each option is added, the development already gained. Such an approach also sig- effort (ISD and storyboarding) and author- nifies a move away from iestrictive frame- ing effort require more time and therefore, based systems to more substantial, flexiblein a commercial environment, incur more and effective systems. costs. The authoring options below are placed sequentially in terms of ease of When authors are familiar with the fulltransfer from storyboard to display: capabilities of an authoring system, the interaction of the courseware will becomea. Text: Normal size the major component of the instruction,b. Text: Varied size thereby increasing the dynamics of thec. Text: Alternate character set learning experience. c. Text: Coloured d. Graphics: Circles, boxes, lines With storyboard in hand, the author cang. Graphics: Coloured commence translating static displays to the f.Graphics: Rotated, sized dynamic display screen. One of the majorg. Response Judging: Multiple Choice, flaws with many courseware packages has True/False been the failure to use the graphical capa-h. Response Judging: Short Answer (to bilities of the available technology. This 256 characters) suggests that the ISD team were not con- i.Response Judging: Student Interface versant with the real capabilities of CBT. It Devices has always been bewildering why screens j.Feedback: Text capable of highly interactive and dynamick. Feedback: Text, graphics learning events have been relegated to elec- 1 Feedback: Additional response judg- tronic page-turning. ing in. Branching: Linear The success of CBT depends on the effp,.7- n. Branching: Student Controlled tiveness of the courseware. Courseware is a o. Branching: Author Controlled learning product and needs to be interac-p. Animation tive; it is not an information source which,q. System Variables to control display/ if necessary, can be page-turning. °ranching logic r. Data Collection for student records/ Authoring Costs management s. Student Interface Devices: Touch, light The cost of authoring depends on what the pen, mouse courseware entails; as each additional fea- t.Audio and Video ture is added the development time and costs increase. These features are oftenAuthoring: The future dependent on the facilities provided by the authoring system, and as mentioned ear- With the continual development of author- lier, it is cheaper .I more simplistic sys- ing systems and the increase in microcom- `250 Using Computers Intelligently 241 puter capabilities, the future of authoring iscialists and individuals that have had expe- bright. What remains is for the developersrience in several authoring systems. They to incorporate all the useful technologiesare proving difficult to find. such as CAD/CAM, A.I. and Expert Sys- tems with current authoring systems toNotwithstanding the present dearth of provide trainers with software that willexperienced people, the writers believe assist in more efficient generation of effec- there are several essential ingredients to tive training. successful courseware developers. These include: computer awareness, training In the long run however, authoring sys-sensitivity, logical and conceptual abilities, tems by themselves are non- productive.customer relations, diligence and unlim- The important issue is to find the authorsited energies in order to make deadlines. who can use them efficiently and produce training materials which are effective. With Conclusions CBT courseware on hand the trainer has a significant training option. Instructional Systems Development and a mastery of the authoring system are essen- Project Management tial for high-quality courseware. Such courseware is not just a product of a fully- The role of project management is a crucial optioned authoring system, but the result aspect, especially when the number ofof careful planning in the analysis, design, hou..$ to develop one hour of interactivedevelopment, implementation and evalu- CBT is in the order of 170-200 hours. For the ation of the training material. same reasons a D.P. project would be managed from start to finish, CBT shouldWith this approach, effective and relevant be as carefully monibred. courseware is guaranteed. The writers have developed a checklist ofReferences CBT project management guidelines in order to facilitate the completion of course- Gagne, R.M. & Briggs, L.J. (1979).Principals ware within specified guidelines. The ex- of Instructional Design.New York: Holt, perience has been that one hour of CBT Rinehart & Winston. from initial analysis to the final validation Hitchcock, B. (1987). The CBT Trap.Data and field testing of the course can be Training, 4(5),11. achieved within 25-27 days. During the Kearsley, G. (1986).Authoring: A Guide to project a systematic validation is under- the Design ofInstructional Software. taken with the client to ensure that subject Reading, Mass.: Addison-Wesley. matter is correct and instructional strate-O'Neill, H.F. (Ed.)(1979). Issues in Instruc- gies suit the intended audience. tional Systems Development.New York: Academic Press. Project personnel Romiszowski, A.B. (1986).Developing Auto- Instructional Systems.New York: Kogan Project personnel for CuT development is Page. perhaps one of the most difficult aspects to Sims, R. (1986). On the development of deal with. Unfortunately there is a desper- high-quality courseware, in G. Bishop ate shortage of skilled CBT personnel in & W. vanLint (Eds),Proceedings of the Australia. The proof of this is that course- Fourth Annual CALITE Conference. ware development organizations are con- Adelaide, December. stantly looking for experienced ISD spe- 251 242 Using Computers Intelligently Appendix 1: Stages of Rod Sims is an Associate Director with instructional systems design Expert Solutions Australia, specializ- ing in Instructional Systems Design Stage 1: Analysis and Courseware Development. Rod Task analysis and selection has been working in the Computer- Objectives analysis Based Training field since 1979, and Target Population analysis has worked as a lecturer and CBT Training and Resource analysis Consultant in government, academic Work Model analysis and business sectors. With a Masters degree specializing in Computer- Stage 2: Design Based Training, Rod's current research Media selection and syllabus content interests concern the use of authoring Course design systems and the quality of courseware. Prototype and production costing He is currently managing the author- Development planning ing of courseware for government Management planning departments and manufacturing or- Evaluation planning ganizations. Implementation planning Pascal Grant is an Associate Director Stage 3: Development with Expert Solutions Australia, spe- Lesson scripting cializing in Instructional Systems De- Lesson entry and supplemental media sign and Courseware Development. integration Pascal began his career in teaching, Management system production moving to Computer Education and Product integration and testing Computer-Based Training in 1981. Supplemental media scripting Pascal has a Masters degree in Educa- Supplemental media production tion, specializing in Instructional De- sign. His major research interests con- Stage 4: Implementation cern the implications of Intelligent Site preparation and system personnel Instructional Design Systems for the training development of training materials. He Initial data collection and testing is currently managing the design and Revision development of CBT im; lementations Turnover and acceptance within several government and pri- vate companies. Stage 5: Evaluation Formative (pilot study) Summative (after 4-6 months) 252 L.- The use of CAL in remedial mathematics for e: economics students Anne Arnold Department of Economics University of Adelaide For several yearsthe Department of Econontics teaching programs and exercise sets. Un- has been a-...ductfrtg a mtedial math.--matics course for first year students with little mathe- fortunately, little has been done since then, matical badcgrowid. Recently,CAL programs although we still use her programs. were intrcOnced into this course. This paper desaibes how this was done and considers the One of the subjects which seems well suited success of the projectin particular, the reactions for a high CAL content is a first year course of the students, the benefits to the teaching staff and the many spillover effects on other students in remedial mathematics for economics in the Department are explored. The perform- students.Mathematics is becoming more ance of the program will be judged by boat a and more important in the study and teach- survey of students as well as consideration of their performance. ing of economics. Students with a reasona- bly good high school mathematics back- ground are reasonably well catered for Although CAL seems to be gaining morewith the option of doing mathematics in the and more in acceptance, it seems to me that Mathematics Department or in the Eco- there area few areas which may not have nomics Department; students with an aver- received sufficient attention or thought,age high school mathematics background especially when preparing a course. Thesegenerally have sufficient knowledge to include, firstly, assessment of the benefits cope with the mathematics involved in the and costs of such a learning program and core Economics subjects and no inclination secondly,the importance of targeting the to do any more,anyway; but students with programs to the particular group of stu-very little high school mathematics were in dents concernedI was also concerned toa quandary.Whiie they often recognized make sure the students had sufficient their lack of skills and realized how much it backup and personal contact with a lec-hampered their study of economics, there turer/tutor so that they didn't fall behindwas little they could do. This first year or have unsolved problems, without losing remedial course fills that gap. It is only the advantages of CAL by requiring too open to students with very weak mathe- much input from staff. This paper is,then,matics backgrounds and who are concur- concerned not so much with "how to"rently studying or have already passed the conduct a lesson/course or CAL program first year core economics course. These are but, rather, how to assess the program, andusually good students who are highly a discussion of factors to consider. motivated and will work hard and learn quickly.(The less motivated student will CAL has been adopted only to a limited often not take up this option) extent in the Department of Economics at the University of Adelaide.Glenys BishopI have dwelled on the nature of this subject first introduced us to CAL and while shear some length so that I can highlight the was in the Department she set up severalparticular problems in teaching this course 25 3 -4, 244 Using Computers Intelligently and the reasons why CAL is potentially so They are able to go back and re-use the successful here. First, the students do havelessons as often as they like, and not worry very variable backgrounds - while theyabout being anuisancetostaff have had no formal mathematics education members,nor appearing silly to their peers in the latter part of high school, their con. or their tutor by always asking question, petence ranges from none at all to somegetting wrong answers and so on. In addi- familiarity with numbers and graphs etc. tion, the lessons are available to everyone Most of them share a deep seated fear ofin the Faculty - including staff members mathematics and high anxiety over thewishing to give their school age children subject. Second,they learn at vastly differ- some extra help!! My aim, in the long run, is ent speeds. Some seem to pick up the con-to set up all of the mathematics for econo- cepts very quickly whilst others need muchmists course as CAL modules although for more reassurance, practice and time. Third,reasons given later, I don't envisage that they all require their confidence to be builtthe whole course would be taught solely by up. This often necessitates giving them a lotCAL. Progress towards this aim is slow... of practice problems that are marked byespecially as we are about to alter our their tutors or, less preferably, by provid-computer resources away from a VAX ing thorn with solutions. Fourth, they arerunning VMS to PCs. This will necessitate highly motivated (as mentioned above)rewriting the existing lessons and for this and are prepared to work hard at the sub-reason we have not moved quickly on ject and often undertake extra work in theirwriting new lessons. However we have own time. Fifth,since not all mathemati-Iwo new suites about which I wish to talk. cally illiterate students in the facultyThe first suite is a set of lessons on MATRI- choose to do this courses it is useful to beCES; the second a set of lessons on LIMITS. able to open up the course for any studentsThe programming for both of these was to sit in on, even if they don't formallydone by Mr.Tony Kiek, a programmer in enroll. Also of relevance is the fact that wethe University Computing Centre with a seem unable to find a text book whichspecial interest in training and education. covers the course in sufficient depth orWe worked closely together - I provided a detail and with sufficient examples andcourse outline and detailed course notes, exercises. In addition, there is typically Tony did the programming and any subse- only a small number of students doing this quent changes and additions. Each pack- course and it is a labour intensive under-age was setup as a series of about five or six taking. individual lessons, each about an hour long and containing explanations, examples It i., clear that CAL is ideal to meet theseand exercises.I do not intend to give a de- requirements - it allows the flexibility fortailed account of the content or mechanics students to go at their own pace, progress-of the lessons, but will point out that they ing rapidly over the sections they find easycontain provisions for going back to previ- or familiar, going slowly and/or repeating oussections,"fast forwarding" for stu- the sections they find difficult. Many exer-dents repeating the lesson and a very clever cises can be incorporated into the lessons, record keeping system. The records kept which the computer instantly marks andare the date and time the lesson was tells them if they were right or wrong, started,the user's name, which lesson they giving hints as to the answer if required.undertook, whether they finished and how This builds ip their confidence and alsomany errors they made. The only reason for alerts them as soon as they don't under-and use of this record keeping is in assess- stand a concept. 254 Using Computers Intelligently 245 ing the packages - using this informationcomputer before. One or two were scared we can determine whether any of the les-that they would not be able to cope, but the sons are too long, too short, too difficult, ormajority were excited by the challenge of have any particular problems.' envisagedoing some thing different. This of course that we will not need to continue this record put them on the right track for positive keeping once we are entirely happy withlearning. After the CAL lessons, all were the programs. In addition, we could moni-quite happy with the idea of learning from tor students' progress but this may not bethe computer although some indicated that entirely fair as the students seem to preferthey would not like to do all their work this the anonymity and privacy of the computer way. There were a few particular problems lessons, so this is not done systematically. I with these lessons that caused some dissat- will be happy to provide more details ofisfaction - namely that two of the lessons these two packages to anyone interested. were too long. This is very important - although we had test run the exercises Of interest now is the appraisal of the sys- many times, we did not allow for the tem. The amount of time spent writing amount of time the students spent copying these two packages was very considerable,from the screen. This can be rectified by although the second was much easier and altering the lessons, giving the students a quicker once we had the experience of themore accurate impression of how long each first to draw on. For a small class such as lesson takes and warning them to be selec- this the costs of involved in tying up the tive in what they write down. Many com- terminals etc is negligible - at the times ofmented that the ability to take very good the year when the students use these two notes was an advantage of CAL lessons but packages, computer resources are reasona- it also took a lot of time. Specific questions bly slack and there is no problem in thewere asked about the advantages and dis- students finding free terminals. The limits advantages of CAL, as well as these par- package requires a graphics terminal ticular lessons.It seems students see CAL as which is a constraint as many of our termi- a useful adjunct to traditional teaching but nals are not graphics terminals ,but this did do not want to lose all contact with lectur- not present a major problem. These are theers and tutors. Five (out of fifteen) thought costs. What of the benefits? Benefits are of that CAL was too impersonal whereas the three sorts - to the students of this particu-others said it was not. However all but one lar subject, to the staff and the wider spill-thought that the ability to work at their own over effects. pace was a very real advantage.In fact, this was probably the most important advan- Benefits to the student tage to the students - they could work at there own speed, pause when they wanted A survey was conducted of the students to to, go back over some of the sections and find out how they felt about CAL lessons inrepeat whole lessons at a later date. Several general and the LIMITS course in particu-students also liked the active nature of this lar. Unfortunately, because this is only alearning -being forced to think about prob- small class, the number of respondents was lems on the spot, not being afraid to get the low (15) even though the response rate was wrong answer,not tending to "drift off as high (over 75%), and so meaningful statis-you do in a lecture" and even appreciating tical analysis is impossible. One of the re- the challenge of having a readily identifi- sults that startled me was that before they able task. In other words, they had some- started on the course, very few of the stu- thing they would start, do and finish, and dents did not like the idea of learning from as with an assignment,they derived satis- a computer, even though few had used a faction of completing such a defined task. 255 246 Using Computers Intelligently The disadvantages that the students citedlecturers realize considerable time savings were on two levels - firstly aimed at theas we did not need to be there when the particular package and its length and sec-students were doing the work. As these ondly aimed at CAL. These perceived dis- lessons replaced all lectures and most of the advantages are important and must betutorials in this topic, we were saved these considered carefully to ensure studentscontact hours. There was obviously less continue to be motivated by and learn frompreparation and marking time,too.The CAL lessons. The major disadvantagesstudents wanted some personal contact students noted were; they missed the ver-with their tutors so we had one tutorial on bal explanation of, and interaction with a the topic, but I suspect that this was not tutor/ lecturer,they found they lackedreally necessary but more due to habit. keyboard skills and/ or knowledge ofAgainst this time saving I did have several computing so they often wasted time. theystudents call in to see me and seek help - had to get used to reading, writing, typingcertainly more than I would during a lec- and thinking at the same time. they wereture course but, they were told that this was aware that it is easy to put off doing thea new suite of lessons and that their com- session and then to find yourself behind,ments were welcome, and many of their some said they often wanted to ask thequeries and problems caused us to alter computer questions such as "why'? orand improve the wording of the lessons so "what if"? that these problems should not arise in future years. On the basis of 20 students Most of the disadvantages that the students.and five lessons, I estimate the academic mentioned can be easily solved by provid-staff teaching the subject saved about 7 ing them with more help and instructionshours of contact time and 14 hours of before they start and ensuring that they are preparation and marking. able to ask questions of the tutors and lec- turers. None of the students perceived theAs well as the direct time saving,there were problem of trying to write down too much other benefits to the staff - enjoyment, chal- detail from the screens but my feeling islenge, learning new skills and so on.These that they did this - given the opportunity,are largely non-measurable but not insig- they tried to rewrite the entire lesson. Other nificant. problems that I felt arose were that they need to be very honest with themselves and Spillover benefits be preparedtousethelessons intelligently.We had one student whoThere are many spillover benefit, although raced through the lesson in about ten min-many of these may not be apparent utes but made 117 errors!!! initially.Certainly, the availability of these lessons to an audience wider than just the Benefits to the staff students they were written for is an impor- tant factor. I ran a remedial mathematics In their survey responses the students indi-course for Honours Economics at the start cated that they would not have liked a tutorof this academic year and was able to point in the terminal room with them while theypotential Honours students in the direction did their lessons - only 4 of the 15 wouldof these packages. The advantage to them have liked this. They did however like to in particular is privacy and flexibility in the know that help was nearby - the supervisor timing of these CAL lessons.In addition of the economics computing rooms is closejust the meagre computing knowledge that and my office is opposite the room theythe students pick up from this exercise ( and mainly used. This means that tutors andit it is no more than being able to log on, log 2 5 Using Computers Intelligently 247 off, and type a few characters) makes them dents stated advantages of CAL evidence feel that computers are not all that difficultthis. Given that most of thu disadvantages to cope with, and probably helps them cited were either of a technical nature (the when they meet computing in other sub-layout of the computer rooms, length of Jed& lesson and so on) and can easily be fixed,it seems that we should try and incorporate The effects on student learning more and more CAL into this subject.There are three limiting factors, however.First, Apart from whether students actually en-the the students do not want to be left joy CAL or not and whether it eases theEntirely without personal contact and work load of the teaching staff, there is thesome, although considerably reduced, question of whether it improves students teaching commitment is still needed. Sec- learning and understanding of the ondly, the lessons must be technically well topic s.With this particular course, there are presented - what seems to be to the author two aspects to the question;first whether of the lesson to be a triviality can become the particular subject helps with their stud-very important to the students (for ex- ies in economics in general and second;ample, whether upper or lower case is whether CAL is better than traditionalrequired). Thirdly, the time required to set techniques in teaching them the required up a well developed, well targeted suite of mathematics. The first question is not asprograms that the students will find relevant here as the second.With such a interesting,challenging and useful without small number of students and no way ofbeing too long or too difficult provides a controlling between years, it is a difficultvery real constraint. question to answer.However, my impres- sion is that students who had done the CAL lessons understood the concepts better Anne Arnold is currently Senior Tutor than students in previous years.Questions in Quantitative subjects in the Depart- in the exam were purposely kept to a stan- me-t of Economics, University of dard very much the same as in previous Adelaide where she teaches mathe- years and the marks were higher in the matics and statistics, which is where CAL groupthaninprevious her interest in CAL lies.Her research years.However no firm conclusions can be interests are in many areas of applied drawn from this as too many other factors economics. varied.Itis at least encouraging!In addition,if student results are at all corre- lated with their enthusiasm and motiva- tion, then CAL should be effective. Summary The results of the student survey seem to indicate that students enjoy CAL and will learn from it. In fact only one student was not looking forward to the next suite of lessons and most were more favourable to the idea after having completed one pack- age than before.The particular features of CAL make it especially suitable fora course such as remedial mathematics and the stu- 257 The subject matter of CAL technology: Towards a development methodology John Barrett Macquarie University ..IMMMIMIIIM.11511111, This paper suggests that: (1) Implementation functions of tertiary education, the follow- and acceptance of CAL should be within a dear ing statements will probably not be accept- understanding of the nature and purpose of learning in tertiary educational settings; (2) able to all readers but will serve as a starting CAL has a unique and significant potential point for discussion about the goals of any which has been constrained through a lack of tertiary instructional system. conceptualisation and adherence to inadequate theory; (3) Substantial advances can be gained from adopting a "common code" for represent- Some of the teaching aims of post-school ing design, implementation and learning di- education could be thought of as providing mensions; (4) Such a coding schema could be a means for students: developed from cumulative knowledge in the area of instructional design, from CAL research and development knowledge over the past To gain a meaningful understanding of twenty years and from recent developments of the structure of the discipline(s) being the integration of artificial intelligence and studied (after Ausubel, Bruner et al.) CAL; and (5) Theoretical, conceptual and prac- tical considerations for CAL should be founded on the integrative research and debate in the To develop a knowledge base of fac1s, evolving interdisdplinary field of Cognitive procedures, systems, techniques re- Science. lated to the subject area This paper presents some ideas and =der- lying principles which impinge upon To form an appreciation of the history of Computer Assisted Learning (CAL) human effort and individual thinking and achievements which contribute to courseware production and its successful current ideas, research findings, beliefs implementation in higher levels of learning and values systems in a discipline. This in formal education and business/industry training. includes intellectual processes, tools, methodologies and outcomes of en- But, before proceeding to a more detailed quiry and debate which highlight discussion of CAL development problems human creativity, frustration, failure, it is necessary to outline some aspects of the chance interventions, brilliance of rea- role of teaching and purposes of learning in soning etc. tertiary education into which a CAL system could fit. To understand the meta-methodologies of the arts, sciences, engineering, the Articulating Learning Goals in tertiary law etc. education To become motivated or inspired by creative problem solving processes in- Given that there is (and seemingly always cluding acquiring a value set for the has been) substantial debate about the 6 arch for truth, beauty etc Using Computers Intelligently 249 To develop general skills of criticism, has to serve the above purposes in the debate, reasoned argument and skills in first instance. That is, for example, in an oral and written communication. educational setting a delivery mecha- nism with high entertainment value but To develop a repertoire of knowledge/ low instructional purpose has little skill competencies - with a number of place; particular high level performance skills in specialised areas. any delivery mechanism must cater for specific needs of instruction in provid- To fulfil intellectual, aesthetic and so- ing a suitable learning environment that cial/ moral needs and desires (for each can only be truly evaluated in student individual) in a socially responsible achievement terms; way and become aware of social and cultural traditions and contexts. the mechanism for providing instruc- tion must be efficient in terms of the Develop techniques of inquiry and time required for the development problem solving ("research" and evalu- process and the effort required for ation skills) and to undertake, manage implementation and supervision. and complete tasks and assignments by controlling, allocating and applyingWe therefore need a paradigm for instruc- personal resources and the resources oftional research and development which the local environment. covers educational theory for design and provides a suitable learning/ instruction To contribute to the development and computational model. learning of social interaction skills of one's self and others by interactingWe my conclude that our task is to select a through modes of interpersonal and framework for the formulation of a devel- media communication. opment environment which includes: This 'goal list' is by no means complete but how to capture and represent content merely suggests general areas of under- standing that learners should gain from a how to store and deliver content tertiary education program. The list serves to illustrate the points that: how to generate interaction with con- tent and ideas knowledge frameworks of concepts that are in-filled with facts and proce- how to assess the effectiveness and of dures and are central to the learning the learning process and the efficacy process; and efficiency of the instructional deliv- ery system we are analysing human learning and most (all?) goal statements have an af- The computer as an instructional deliv- fective component; ery device emphasis is on cognitive processes Here I wish to emphasise the view that "the rather than behavioural outcomes; computer" is not just another piece of edu- cational technology of the same class as an an instructional delivery mechanismoverhead projector. Not only do we have (human/human or human /machine)new toolsin this symbol manipulating 2 5 9 250 Using Computers Intelligently machine, which is an attribute unique to 2. Problem Solving and Modelling Tools. Go computers, but other aspects such as the a step further than (1) above and de- speed of retrieval, access to vast amounts of velop models of the problem solving stored information, an ability to bridge approaches employing these new cog- geographical and conceptually wide gaps nitive modelling tools. of communication, and the ability to con- trol other devices singling it out as one of3. Packaged CAL.Use someone else's manldnds greatest inventions. In addition, courseware, commercial or otherwise the very nature of the computer as a tool for and readjust the curriculum around intellectual modelling raises it from a whatever is available. number manipulator to a "cognitive part- ner" in research, teaching and dissemina-4. Local CAL development "The Cottage tion in tertiary education. Industry Model".Employ whatever skills are available from those (or that As various tools and processes evolve the person) in the area(s) of instructional creative powers of teachers in post-school tecnniques, curriculum representation institutions and the army of 'human re- (subject area content), graphics design source developers' and trainers will invari- and coding on computer (program- ably demand new software tools for creat- ming/ authoring) ing educationally more complete (e-nd computationally more demanding) learn-5. CAL "R & D Centre Model". Full de- ing environments. There is now appearing velopment tools and resources in a a number of high quality authoring tools managed organisation. This category is and languages, including Course of Action, very rare in tertiary education in Aus- Course Builder and Guide and the Apple tralia but a niche of rapid growth in the Hypercard. But there remains a number of commercial world. serious deficiencies which must be ad-Deficiencies in any of the above develop- dressed if we are to head towards the full ment environments include: capabilities that computers potentially of- fer. appropriateness of the specific course- ware Deficiencies in Current Approaches. application developers knowledge (or lack of it) In previous papers, I referred to the process development tools capabilities demands placed upon the developers of project management skills and support computer-based instruction (Barrett, 1983a resources '83b, '83c, '86a, '8ob). These papers analyse financial support, encouragement, rec- design and implementation variables and ognition for outcomes. constraints in courseware authoring. Ap- proaches to CAL implementation gener-Again I contend (see Barrrett 1983c and ally fall into one, or a mixture, of the follow- 1986b) that significant advances in the ing categories and scenarios of usage in widespread use of CAL in Tertiary Educa- tertiary settings: tion will come about through the develop- ment of more appropriate development 1 Application Tools. Take an applicationtool.., the building of a cumulative library package such as a word proce;Ang orof an accessible resource base and con- statistical package and incorporate it incerted effort in an efficient CAL production the classroom with as little adjustmentcentre and conducted on modern business as possible to the curricula. grounds. 6 Using Computers Intelligently 251 The underlying problem of CAL tional p.ocess and evolves a ma- development chine environment for the delivery of designed instruction or provides The underlying difficulties facing .2.AL tools for learner controlled explora- development and its infiltration into the tions educational system (beyond financial con- siderations) is a lack of agreement on the 3. media dimensions from influ- "superstructure" or subject matter of the ences of graphics design, visual CAL development enterprise. By this I media specialists or communication mean there is no universal agreement of the experts. nature of the instructional enterprise or itsFor too long, many CAL developments particular underlying coding mechanism.have been dominated by the author's our This position has arisen mainly because: own area of interest or expertise to the learning research was forced into a exclusion of one or more of the other ele- backwater on the tide of behaviourism. ments vital to a success courseware pro- The whole approach has been shown to duction. It is extremely rare that a develop- be devoid of any substantial contribu- ment environment contains a person or tion and should be relegated to itsteam having the right mix of educational, proper place with the learning" ofcomputing and media /communications lower order "reflexive" skills. expertise. the principles and research findingsMore importantly, however, development relating to Adult Learning have, in tools have provided little positiveguidance almost every case, been ignored for the integration of the distinctive phases of the interpretation of initial ideas through delivery mechanisms have been mode- design, layout, coding, testing and evalu- led on the mechanical teaching ma- ation. Most instructional developers, and chines and similar media. As previ- probably most teachers remain alienated ously stated this has restricted us to a from CAL because of the lack of easy access limited metaphor blinding us to what is to either development environments (and possible with an entirely different per-tools) or an inability to make changes to spective courseware that is only "nearly" suitable for their particular classroom application. there are at least three diverse forces which influence CAL development and In a similar vain, a distinction remains be- the techniques, tools and philosophies tween tools available to designers and tools front each has marked influence on the available to learners in a run time version of finished product - the courseware. Learners often have a much more restricted environment during 1. educational/ learning dimensiona CAL lesson. Often, developers have which operationally encompassescommented that they "really got to know the design of instruction. A vast in- the subject area after wrestling with struc- formation base has evolved fromture and content or 'programming dimen- research on learning and instructionsions" but then they only pass on to the which should be the basis of anylearner a constrained path with virtually no courseware development "freedom to learn"! 2. the computational dimension whichThrough computer accessibility and their operationally defines the instruc-pervasiveness in all walks of life it is not 6i.(1: .4; 0Ij. 252 Using Computers Intelligently surprising that a whole host of forces havefacilities. The metaphor of "desk-top pub- motivated a diversity of people to developlishing" was realised and the idea material- CAL courseware. Some goals have beenised as PageMaker from the now highly more commercial than educational, somesuccessful Aldus Corporation (Talevich, processes more computing concerns than 1987). instructional goals, and many outcomes amateur rather than professional. They areThe point here is that we in tertiary educa- more of ten conducted as 'back-yard" (readtion need to be very clear of our goals and "isolated campus") activities rather thanpurposes to come up with a suitably vi- processes of modern production centres. sioned tool that will help us achieve our goals. There is an even greater problem than lack of purpose and lack of co-ordinatedre-I believe that current CAL is still based in search and development effort and this the book metaphor, with a foundation in concerns a lack of conceptualisation whichverbal material and text. Developers and is taken up in the section that follows. learners need more of a visual base and a suitable metaphor. One suggestion could Devising a CAL metaphor be a 'space exploration metaphor'. This al- ludes to a 'universe of knowledge' which is Macey olman'F -reatest achievements haveto be explored and mapped. It includes the lfeen built on a dream- a vision. The use ofidea of various systems (disciplines) and a suitable metaphor provides not only the.sub-systems (fields) which require explo- visual imagery and emotional stimulus butration and 'documentation' (understand- also provides a concrete frameworkoning). In this metPrher the CAL software which to hang our ideas. Many peopleacts as an electronic guidance system for know of the chalkboard matrix that gavethe journey of exploring new knowledge. rise to the Spreedsheet metaphor and the production of Visicalc. Similarly, the "deskHowever, there is a far greater and more top" metaphor ha:: been carried over from fundamental deficiency. What we need is a Smalltalk to the Apr le Macintosh, Com- universal code of symbolic representations modore Amiga and Atari machines. which will assist us in recording the map of a knowledge universe. And we need this More recently a team of "stood down"NOT to alter the knowledge structure of the workers from a Seattle newspaper set out discipline into some phony or forced cod- on a market research tour of Washingtoning systen but to provide the most efficient arI Oregon with the genera. itotion of sell- means of synthesising materials into a for- ing the idea of producing advertising ..opymulation for natural learning experiences on microcomputer at newspaper offices. and its producing the documentation. This But, in a moment of discussion the fourcan only be achieved through an agreed particiants had concurrent insight into theupon analysis and classification system underlying problem they faced. It was nothaving its unique nomenclature and in- newspapers that had the "need" for collat-structional/ learning coding system. Our ing advertising materials- they alreadygoal then should be to develop a common had the process, however backward- butcode for learning environments, with spe- what if individuals had access to suchcial attention to computer-based learning. tools? Mr Paul Brainerd, a journalist with aThe representative code should condense long standing interest in computers, real-or provide a "shorthand" representation of ised the goal was to provide the generalthe know Age, procedures, skillsetc. populace with full professional publishingmaking up a discipline or sub-disciplinary 262 Using Computers Intelligently 253 area. The code should provide a means of practical attributes of natural ease of use cuing the various instructional delivery and management. The common code rep- modes, media and interaction processes. resenting instructional content and proc- esses essentially represents the layers of - In search of the Common Code the learner's cognitive processes One could hardly imagine developments activities and processes in the learning in, for example, the arts, engineering or environment science without the existence if an underly- the intentions and operational instan- ing code - possibly mathematics! Similarly, tiations of the author/ instructor agreed standard codes for manufacturing design and production are essential. Even Building a code the computer industry with numerous "standards" has many common underly- The type of coding system suggested in this ing coding schemas. But, something as paper should be built upon three main important as education - specifically in- foundation areas. structional design and learning products" is conspicuously lacking conceptual agree- First is the accumulated knowledge of in- ment or functional commonalty in respect structional design systems technology of representing the teaching/ learningepitomised by the work of Romizowski process. We are in a situation which would (1984,1985). Although, as mentioned ear- be as crazy a.; each draftsperson producing lier, much of this work is based on behav- building designs with no agreed uponioural psychology there is a wealth of syn- symbols or means of expressing specifica-thesised material from systems research tions. This is even more disastrous whenwhich can be balanced by content and we examine current practice in CAL pro-procedures from the two areas that follow. duction. Second, we should take into account the The overall problem of CAL development vast amount of informationbuilt up from at becomes clear if we briefly examine the least twenty years of research and develop- current process of authoring CAL materi-ment of computer-based learning systems. als. This could be labeled bottom-up, ato-These areas of expertise need careful evalu- mistic approach and well entrenched in ation in order to identify what really works. behavioural methodologies. The 'mechan- (See for example the work of Baker, 1978; ics' of structuring objectives, formulating McCann .1981; Rushby,1981,0'Shea,1983; modules and sub-sections and the verySteinberg, 1984; Kearsley, 1987; Merrill, development process from the parts-to- 1985 and 1987; Deer, 1987). the-whole encourages a 'dis-integrated' product. More importantly, the "produc- Third, the CAL coding taxonomy Should tion system" is devoid of any universalityby based on a firm conceptual, theoretical in its coding process or its documentation. and computational footing derived from But it becomes impossible to compare thethe evolving discipline of Cognitive Sci- design, learning processes and possibleence. This new interdisciplinary field cognitive outcomes unless there existsbrings together debate, research develop- some ty,)e of common coding system. ment from Cognitive Psychology, Linguis- tics, Computer Science (Artificial Intelli- It shot..1... be possible create a system whichgence), Neuroscience and Philosophy. is theoretical!) .,und, flab universal agree- Many of the theoretical and practical prob- ment at the coding level and which has lems faced in these disciplines are directly 263 254 Using Computers Intelligently related to such areas as knowledge repre-to include these dimensions in modern sentation, problem solving and machineIntelligent CAL systems but without the configuration and especially human learn-benefits a common code would offer. ing that they form the very conceptual rock on which a new science of instruction canThe ideas expressed here may be consid- begin to built. (See. for example, the writ- ered in terms of a music analogy; the musi- ings of Anderson, 1980; Gardner, 1986;cal notation represents, at various times, Levine. and Rheingold, 1987; .Minsky,the ideas of the composer, sequential and 1986). propositionalinstructionstothe musician(s), 'control structures' to conduc- But what central areas, or design content,tors, psycho-motor commands to players, should the proposed code encompass? Themusic to the audience and sound/ visual/ cognitive elements which must be covered aesthetic images and experiences to the in detail are (at least): individual listener -all by means of a work- able coding process for recording, storage, language-in general and the jargon of aretrieval and presentation and reception. content field or discipline perception, attention and visual im-Like music, our instructional ideas and agery content needs to be represented not only numerical and computational elements for the idea they 'contain' but how they are knowledge structure(s), representationto be delivered, ie. on what "instrument" or and memory .audio/ visual mode or media. Similarly, cognitive strategies in learning the instructional "conceit" must be "or- reasoning and problem solving strate-chestrated" and delivered in such a way gies and development of expertise that makes the whole experience worth- cognitive development requirements while in light of the criteria of the com- psychology of motor performance (asposer/ author, conductor/ teacher and related to competency). listener/ learner. It is readily observed that the above factors are highly weighted in the cognitive do-It should be noted that, even if there are main ant': suggest a coding mechanismsome "intermediate" processes the coding which represents aspects of cognition mechanism either holds true to the original rather than such 'traditional' dimensionsauthors intentions or, at least, allows a range of variations and interpretations by as behavioural objectives, frame sequenc- later conductors and musicians or record- ing or reinforcement. However, while not excluding affective or cultural influences ing engineers. That is, for example, record- ing sessions by different cultural groups or interventions in cognitive performance, these are not the primary focus of this under different conditions allow local vari- ations while still preserving much of the approach. The coding schema should first original ideas. For instruction, provided and foremost reflect the cognitive proc- the original ideas and content are 'classical' esses taking place in the mind of the learnerthey can be preserved and delivered in a s he or she interacts with content mediated more modern modes or styles but still by the delivery system. In operation the convey the original content. suggested representative code has to serve three masters - representing the content of Development processes and the delivery systetn the discipline, becoming the stored and transmitted code of the CAL delivery sys-The suggested common code approach tem and being a representation of the stu-provides a number of practical advantages dents learning. Attempts have been made to instruction: 264 Using Computers Intelligently 255 ALcumulated resources and delivery con- tive. The author would begin by writ- trol or access. The "common code" ap- ing, or capturing the essence of the proach not only provides a common content as a natural language narrative. mechanism of understanding the struc- The next step is to bring this into instruc- ture and deliver specifications of con- tional code by use of a "Tool for Instruc- tent over disciplines and across time/ tional Editing" (TIE). The author would space barriers, it encourages the pool- go through each paragraph by use of the ing of resources on a national and inter- instructional editor marking out defer- national basis. It provides an efficient entially all the "instructional units" and means of authors identifying or analys- control structures which would be ing their own, or others, instructional translated into coding symbols like materials. More importantly, cllssifica- musical notation. It is the wholistic tion of vast amounts of material such as script which holds the key to continuity that appearing on CD ROM (for ex- and integration. If the script is altered, ample the Microsofts "Bookshelf" refer- similar to an ideas outliner (such as ence works - there are now more than Think Tank), the text will change place 150 CD ROM applications) will prove but the underlying structure and repre- invaluable to instructional designers sentation will remain unaltered but and instructors. transposed. Templates. Similarly, templates of accu-The TIE allows for the analysis of content in mulated resources on an international terms of structure of the content with re- basis will advance the implementation gard to facts, concepts, principles, rules and quality of instructional/ learningand propositional organisation, proce- resources. With the appropriate imple- dural knowledge etc; of instructional vari- mentation tools (application 'front ables such as number and type of examples, ends', CAL shells, designers toolkits, pacing, questions/ answers, etc; of learner instructional "sub-assemblies" etc) lec- cognitive variables such as memory encod- turers and instructors should be able toing prompts, consolidation processing ac- concentrate on the quality of subject- tivity, reasoning processing etc. The TIE matter content ald cognitive processes will also be capable of orchestrating deliv- of the learnerwhich should be nowery modes. This is an "authoring front- feasible for any "student model" will be end" which assists in the Instructional constructed in the same underlyingDesign process and, more importantly, code. "generates" the underlying code in 'cogni- tive units'. A New Authoring Process. But, given the intricacies of the task and the seeming Overall, the common code approach will complexity of any coding schema, how encourage the integration of learning can we possibly hope to author even a through coding which relates to "natural ten minute sequence of cognitive learn- language". Thus, this form of authoring ing? The same could be said if we werewill pave the way, or integrate with, devel- about to embark on a new thing called opments in in.elligent CAL programs. Its music! The answer is virtually oppositemain strength c ther than practical consid- to most CAL authoring. The process erations, is that it is founded on cognition suggested here involves working from (thought) rather than beha vioural manifes- the "big picture", top-down perspec- tations. 265 41111111MMEMIIII 256 Using Computers Intelligently References McCann, P.H. (1981). Learning strategies and computer based instruction. Com- Anderson, J.R. (1980). Cognitive Psychology puters and Education, 5, 131-140. and its implications. New York: W.H. O'Shea, T. and Self, J. (1983). Learning and I.reeman & Co. Teaching with Computers. Englewood Baker, F.B. (1978). Computer Managed In- Cliffs, NJ.: Prentice-Hall. struction:Theory and Practice.Engle- Rushby, N.J. (Ed) (1981). Selected Readings wood Cliffs, N.J.: Educational Technol- in Computer-Based Learning.London: ogy Publications. Kogan Page. Barrett, I.B. (1983a). Instructional Demands Steinberg, E.R. (1984). Teaching Computers for Computer Based Learning. 2nd Aus- to Teach. London: Lawrence Erlbaum tralian Computer Education Confer- Associates. ence, Melbourne. Talevich, T. (1987) Paul Brainer '70: The Barrett, I.F. (1983b). Computer-Based Learn- Wizard of Aldus. Old Oregon, University ing: Constraints on Development. Annual of Oregon, 41. Conference of NSW Computer Educa- tion Group. Software Discussed Barrett, J.F. (1983c). Designing Computer Based Learning in Tertiary Education- Bookshelf Microsoft, Australia. Requirements and Realities. ComputerCourse Builder. TeleRobotics International Assisted Learning in Tertiary Educa- Inc., 8410 Oak Ridge Highway, Kn- tion, Brisbane. . oxville, TN 37931. Barrett, J.F. (1986a). Student Model in Intelli-Course of Action. Authorware Inc., 8621 gent Computer Assisted Learning. Pro- Pine Hill Road, Bloomington, Minne- ceedings of the 17th Annual Australian sota 55438. Colleges of Advanced Education Com- Guide. OWL International Inc., 14218 puting Conference (p.102). Northeast 21st Street, Bellevue, WA Barrett, J.F. (1986b). Towards Specifications 98007. for Intelligent CAL. 1st Australian Artifi-Hypercard. Apple Computer Australia. cial Intelligence Congress, Melbourne. Deer, B.L. (1987). AI and The Authoring Proc- ess. IEEE Expert, Summer. John Barrett is a Lecturer in Educa- Gardner, H. (1985). The Mind's New Science. tion at Macquarie University. His New York: Basic Books. research and development interests Kearsley, G.P. (Ed) (1987). Artificial Intelli- are in the psychological aspects the gence and InstructionApplications and use of computers for learning in Methods. Boston: Addison-Wesley education and training.He first Publishing Company, Inc. became involved in computers in Levine, H. and Rheingold, H. (1987). The psychology and education at the Cognitive Connection. New York: Pren- University of Oregon in the early tice-Hall. 1970's, graduating with a PhD He is Merrill, M.D. (1985). Where Is The Author- currently chairman of the Apple ing In Authoring Systems? I ournal of Education Foundation and a mem- ComputerBased Instruction,12(4), 90-96. ber of the initial executive commit- Merrill, M.D. (1987). An Expert System for tee of ASCILITE. Instructional Design. IEEE Expert, Sum- mer. Minsky, M. (1986). The Society of The Mind. New York: Sim.)n and Schuster. 6 Computer Assisted Learning and the future of education Brian W Carss University of Queensland. ilmimwai: Th.! changing nature of educational technology is towards a digital electronic environment, tional aditinuum, and none of the pro- away from the present multiplicity of hardware. posed futures were seen as likely to evolve This change is expected to be completed within in a pure form. The most likely future will the next decade. The challenge that faces educa- tional institutions during this period of change, almost certainly be a weighted combina- is how to exploit the advantages that the new tion of the five scenarios. media offer and at the same time, enable aca- demic staff to feel sufficiently arnfortable with During phase 4 of the project, two panels of it, to want to use it. A "most likely" future is senior educational professionals and prac- desaibed and the implications of this to educa- tion are discussed. titioners engaged in a series of priority - setting sessions in an attempt to develop a set of weightings of influences on pre-terti- ary and tertiary education in Australia for This paper describes a research project each of the scenarios. which was carried out to determine what developments were likely in informationThe 5th and final phase was directed at technology and how would these develop- developing a most likely composite sce- ments effect education at the tertiary level. nario for education 2000 A.D. Details of A Delphi survey was carried out with most procedures followed in the Delphi study, of the major electronic equipment manu- priority-setting sessions and the derivation facturers being the principal respondents.of the composite scenario are described in The most striking finding of this survey Morris (1985). The results of the final phase was to confirm the rapid move away fromonly are presented. film and analogue magnetic recording media to all electronic digital equipment.Composite Scenario The research project was carried out in a number of specific phases. Phases 1 and 2 Academic staff will still operate in 2000 were devoted to a world-based Delphi A.D. in much the same way that they do study (Linstone, Turoff 1975) of media and now, but in a somewhat different style from educational technologists, designers and that of the 1980's.Education will have publishers with the specific aim of deter- become appreciably more sensitive to tech- mining the state of educational technology nological change by that time. The most by the year 2000 A.D. significant change will be on the quality of education and the techniques used, Phase 3 was used to develop possible alter-brought about by the ad ...flees in technol- native future scenarios which incorporatedogy. the technological possibilities isolated during the first two phases. Some of these By the year 2000 A.D., Society will be both scenarivz-, represent extremes in educa- dependent upon and dominated by, Infor- 267 258 Using Computers Intelligently mation Technology (Goldswortk,,..1980).software. Such companies will exert pres- Education being totally committed to thesure both overtly, through their products new technologies and their availability,having been given an "international fla- will be equally totally dependent on andvour", and covertly, through pressures susceptible to the changes in technologyexerted by the way of the "fashion" phe- itself. nomenon operating through the cultural, entertainment and recreational needs of There will be a strong growth in the accep-people in the workforce. tance of education as being a life long activ- ity, with the consequent life-long demandStudents for facilities, resources and opportunities. Because of the huge costs involved in theStudents are not seen as having a great deal constant turnover of the latest technologi-of influence on educational decision-mak- cal advances, the institution will have toing. They will have a greater choice in the accept more responsibility for the provi- range of course offerings a% ailable to them sion of centrally pr 'duces programmesbut this will be matched by the increase in and resources of hardware and software. what is regarded as "core" material, dic- There will be a corresponding decrease intated by a central authority intent on deter- the individualistic approach to teaching. mining minimum standards of perform- ance. Institutions lecturers Institutions will have changed in that it will no longer be necessary for students to be onThe changes for lecturers will be greater campus and attend formal lectures. Stu-than those experienced by students. Most dents may be required from time to time to importantly, there will be an overall loss of attend specific lectures or participate in prestige. Although the freedom to choose small group seminars but for the most part,teaching strategies will be greater because students will rely on gaining access to the of the greater range of techniques available, information resources they need through atheir influence on what is taught will be highly sophisticated communications net- reduced. In other words, they will have less work which will enable them to work from say in what is taught but more latitude on their homes. how to teach it. There will be increased governmentalA noticeable change that will have intervention into education in an attempt emerged will be the reduction in job secu- to link closely the products of the formal rity of academic staff. Increasing financial education system to societal demands foi constraints will cause the Government as increased sophistication in its workforce. the funding agency, and the University as So too, will there be a need to increase the the employing agency to challenge the quality of instruction. This will lead to the concept of tenure. employment of teams of specialists in cur- riculum design and development, commu- Instructional Designers and Other nication and evaluation. Specialists. There will be an identifiable increase in the The increase in specialist influence on influence on the curriculum of multina-learning and teaching techniques will be tional and international corporations in- profound. The control and dissemination volved in the production of hardware andof information will have become a highly 2G Using Computers Intelligently 259 specialized activity. Education will have Implications of this Scenario for adopted the "medical" model where each Teaching new development results in the emergence or retraining of specialists in the innova- To reiterate the implications of these tech- tion. Educational specialization will range nological developments, the complexity from Communications and information and sophistication of the new technology technology through curriculum design,will require that it be handled by special- monitoring and remedial specialities to ists. Specialists not only in the use of the those of the diagnostic and clinical educa- hardware, but also specialists in the educa- tional physician. tional exploitation of the capabilities of the hardware. There will be an increase in the production of one-way distributed media. It should beFew tertiary institutions in Australia ap- remembered that books, journals and all pear to have recognized the need to pro- hardcopy materials fall into this category, duce specialists in instructional design as as do films. As these new materials will bebeing important for their inevitable fu- made available through the multipurpose tures. Concerfor the development of media communication terminal, there isCAL in a few institutions is in response to the opportunity for central control (censor-economic pressures which are present. It is ship) to be applied. hoped that by adopting CAL as a teaching medium, the institution can avoid increas- A surprising growth area will be in the area ing its staff to cope with the increasing of lecture room media. Despite a concern numbers of students seeking tertiary edu- with the individual, large groups operating cation. within lecture theatres will still be seen as appropriate for some kinds of learning.Several universities are experiencing a The cost of optical film-based materials will "blow out" in student enrolments in the have increased to such an extent that digital Social Sciences particularly in Commerce electronic solutions will be sought usingand Economics.At the University of tape or disc.Long-life laser and digitalQueensland in the Faculty of Commerce technology, together with optical fibreand Economics the use of computer man- transmission will see high quality audioaged learning techniques (CML) is a large visual distribution for display on largeenrolment (700) introductory financial ac- screens in lecture theatres. There will be an counting subject has made it possible to added dimension of interactivity broughtmore than halve the number of tutorial to theatre presentations thereby upgrading assistants needed as well as reduce the the quality of instruction. number of academic staff from three to one. This has been achieved without any reduc- It is unlikely that the optical video disc will tion in the quality of instruction.It has have a great impact on education because however, placed very high demands on the the cost of production and the specialized academic staff in having to master the tech- equipment needed for the manufacture of nology as well as be concerned about the the discs. Thus education generally, willrestructuring of the course content. The use become increasingly a system which willof CML demanded a restructuring of the use the latest information technologiessubject as well as the creation of a sizeable supported by highly trained specialists indatabase of test items. The items in the their application. database had to be carefully linked to the ,e.,(16 9 260 Using Computers Intelligently specific objectives of the course.TheseIn the commercial world, in contrast to the were tasks that took many hours of aca-academic world, computer based training demic staff time to complete. :t. ill) is seen very much as a team effort. A team in which the content specialist may be The r tajor impact of the use of CAL in thisa different person from the author. Such a way has drawn attention to the need forteam may consist of an author, a program- staff development. Staff development inmer, a content specialist, a graphics design the use of CAL cannot be confined to learn- artist and a psychologist.In business, ing of an authoring language or authoringwhere costs must be contained, a team system, but must also include developing aapproach to the development of computer facility in instructional design.It is notbased training materials is taken as a must. unusual in the beginning for new CAL authors to believe that writing instruc-If universities are going to take a lead from tional material amounts to little more thanbusiness, there is an obvious need to estab- the transfer of their lecturing notes onto thelish teams skilled in the use of CAL. A CAL computer system. Many would be authorsUnit will have a more precisely defined set believe also, that it is sufficient to giveof responsibilities than the existing Higher students new information as written textEducation or Tertiary Institutes.It will on a screen followed by a question as themore often than not co-exist with a Tertiary only teaching strategy. If the students getInstitute and make calls on its staff for their the correct answer, the lecturer assumesspecialized knowledge in staff develop- that the students know and understand thement and student assessment. content.Such naivete has dogged the widespread adoption of CAL since theWith the futures as described earlier, it is mid-1960's. important that a highly skilled and knowl- edgeable team be assembled to assist uni- Clearly, would be authors need to beversity staff increase their effectiveness as knowledgeable of teaching strategies suchteachers.For them, there is the aimed as Drill, Test, Inquiry, Simulation, Tutorialreward in having used CAL techniques in and Serendipity. These strategies can onlytheir teaching, of having more discretion- be applied in art appropriate mix when theary time to devote to individual students appropriateness of using CAL has alreadyand their problems, as well as to research been determined. Unfortunately, the work. acquisition of teaching strategies is not a normal part of an academic staff members References training, so unless academic staff make a deliberate effort to seek assistance, theirBell D., (Ed) (1968) Towards the Year 2000: teaching programs will be distinctly ama- Work in Progress, Boston: Beacon Press, teurish. pp 320. Goldsworthy A. W., (1980) Technological It would seem reasonable then for an insti- Change The Impact of Information tution which is committed to fostering Technology.Canberra: AGPS, pp 138. better teaching, by whatever medium, Linstone H.A., Turoff M., (Eds) (1975) The should be concerned that its staff be prop- Delphi Method, Techniques and Applica- erly trained in these techniques. Many tions, Addison-Wesley, pp 620. universities in Australia already have or- Morris M., .(1985) The Future of Pre-Tertiary ganizational structures in place to carry out Education in Australia.Unpublished this kind of training. However, the effort Ph.D. thesis, University of Queensland, will still be that of an individual lecturer. pp 461. Using Computers Intelligently 261 Dr Brian Carss is a Reader in Education at the University of Queensland, where he is Dean of the Faculty as well as being the Coordinator of the CAL Unit. He has a particular interest in the application of computers to teaching. He M..; had a long standing association with CAL being one of the small group of staff at the University of Illinois who wrote instructional material for PLATO 1 during the early 1960's. Demystifying tertiary education Neil J. Clark Faculty of Business Studies Bendigo College of Advanced Education Recent work in instructional psychology has ing and grammar are applied. Work loads strongly influenced the development of a new study strategies diagnostic and teaching pro- exceed those previously experienced. Re- gram for tertiary students. The program will sponsibility for learning, previously borne cover some hundreds of strategies, largely con- by teachers, is now thrust abruptly onto ceived of as academic tricks of the trade, and students. As students struggle to cope, associated attitudes. In using the program stu- dents are asked if they use each strategy, or hold often with little guidance, they fall behind particular attitudes, via multiple choice ques- with work deadlines, with an associated tions. After responding they are provided with decline in work quality. Failures and drop- comments justifying preferred answers.Pre- outs ensue. It is a picture familiar to any liminary trials with both computer and booklet presented material have shown excellent accep- tertiary teacher as well as to student coun- tance by 112 C.A.E. students. sellors and, ultimately, to many students. Secondary education is frequently blamed for not "preparing" students for tertiary Tertiary education is a mystery to manystudies, by the public as well as by academ- students in the sense that they cannot cope ics. Even my corner deli. proprietor com- and don't know why. For these studentsmented, "The schools don't learn them the following scenario is typical. They ar-nothing nowadays". The students them- rive at college or university after havingselves are often accused of laziness while negotiated twelve or more years of educa-others blame poor standards of tertiary tional hurdles. Their expectations are fre- teaching. quently buoyant because they have dem- onstrated their competence, over and over, There are clearly many places in the system and there seems little reason to believe that where attempts at remediation might be success will not continue or that there is anyundertaken. One of these areas, currently real need to change their methods. Statisti- attracting interest, particularly in the cal evidence of high drop -out and failure U.S.A., (Weinstein & Underwood, 1985; rates (usually regarded as being some- Dansereau, 1985; Weinstein & Mayer, 1986, where between 40 and 60 percent), ifParis & Wixson, 1983; Brown, Campione known to students, is disregarded, the "itDay, 1981; McKeachie, Pintrich & Yi- won't happen to me" belief prevailing. ButGuang Lin, 1985), is that of teaching stu- then things start to go wrong. Readingdents reading and study strategies. This becomes difficult, both in the amount re-paper gives some preliminary e.c tails of quired and in comprehending what is writ-one such attempt in an Austral' n College ten. Lectures and tutorials are unfamiliarof Advanced Education.T1.effort has means of instruction. Written work whichattracted strong student support and staff previously satisfied teach( ts is now criti- interest but systematic val:dation studies cised or rejected. Higher standa rds of spell-have yet to be undertaken. The aim of the Ai Using Computers Intelligently 263 work has been to teach students strategies Students are also viewed as having a capa- they presently do not use and to identifybility to consider various options for organ- and at least partially remediate many of theising selected input. Such metacognitive defective methods and attitudes which strategies (Brown, 1985; Brown, Armbrus- prevent students from performing ade- ter & Baker, 1986; Lawson, 1984) have been quately. This is attempted via a question-shown to play a vital role in reading and ing of students about their use of a largestudying. Study strategies are thought of number of specific strategies, and of a lesser as ways of undertaking specific tasks and number of attitudes, and then, by com-very large numbers of such strategies, menting on their responses, to indicate the probably thousands of them, are thought to reasons why particular answers are pre-be involved in skilled performance as a ferred. student. A computer presentation of the questions Skilled students, however, are not likely to and comments has been favoured princi-be conscious of their use of well practised pally because of the control thus providedstrategies since consciousness of skills re- in evaluating the adequacy of the studycedes as they develop (Fitts & Posner, strategies program and in making needed1967). It is thus extremely difficult, in the modifications. case of intellectual skills, to find out how skilled performers actually operate. Fur- Conceptual Framework ther, even when a strategy used by a skilled person can be identified, it may not be This work has been strongly influenced bypossible to teach that skill directly to an the cognitive approach to learning whereunskilled person, because the latter may students are seen, not as passive recipientsnot possess necessary entering behaviours. of information provided, but as active agents using previously acquired knowl-Skilled performance, at any given time, is edge and skills to select and process parts ofnot seen, however, as a simple function of that information in line with their existingindividual learning (Vygotsky, 1978; Day, schemata (Anderson, 1985; Wilson & An-1983). Performance takes place in a social derson, 1986), environment from which varying amounts of assistance - from teachers, tutors, other In the activities of selecting and processingstudents, counsellors, libraries and other students must use their working raemories resources - may be obtained. Reliance on which have processing and storage func-these external resources for skill develop- tions competing for use of a limited capac-ment diminishes as students internalise ity (Baddeley & Hitch, 1974; Case, 1978;new skills ere is thus a need to ensure Daneman, 1984). If capacity is taken up, forthat external supports are appropriately example, in attempting to decipher 1,fa- provided and appropriately used - not too miliar words in reading, there will be amuch, not too little - in the development diminished capability to keep track of the and internalisation of skills. argument, to anticipate what comes next, or to ask oneself questions etc. But more than intellective strategies are involved in the success of students. There Students are also seen as having a capacityis also a need for hard work, for effective to reflect on their own processes, to betime management, and for appropriate aware of the existence of various strategiesattitudes, even attitudes about the nature of and to monitor their success with strate-learning itself. There is also a need for an gies, changing them where necessary.accurate grasp of the expectations of aca- 273 264 Using Computers Intelligently demic staff, for social skills in dealing bothbeing asked if they used each strategy or, in with fellow students and with academicthe case of attitudes, held each attitude. staff, and for an understanding of the or-Multiple choice options were provided for ganisational setting within which students responses using a four category format. A have to operate. Further, there is a need for questioning mode of presentation of mate- students to learn strategies for self control rial was chosen because of the known bene- and for handling stress and emotions. ficial effects of questions in aiding reading comprehension, (Duffy, &miller Or Mason, No doubt many students succeed while1984; Andre & Anderson, 1978) and be- exhibiting defects in some of these areas,cause early trials using questions without perhaps by compensating with strengths in comments had been well accepted by stu- others. An example is the student who has dents. poor abilities in many arras but who works very hard. It was thought that students using the program would identify some strategies Students who have been involved with thiswhich they were already using and gain program use the concept of "getting added assurance ar .d confidence thereby. enough things right", and however impre- Second, they would identify some strate- cise, and ungrammatical, that expressiongies which would require little effort to may be, it probably captures the majoradopt since most of the required entering requirement for success. behaviours were present. A third group of American study strategy programs, (Wein-strategies encountered would require stein & Underwood, 1985;Weinstein & varying degrees of effort by students to Mayer, 1986; Dansereau, 1985) have tended learn and would be unlikely to be adopted to concentrate on intellective type strate-without that effort. None-the-less, aware- gies such as reading comprehension andness by students that such latter strategies memory skills, whereas in this program aexisted was thought likely to provide a much broader view has been taken withseries of quite ..pecific goals for strategy cognitive, affective and social skills beinglearning by students. Quite apart from stressed in addition to cognitive ones. specific strategies, it was anticipated that working through some hundreds of strate- Methodologically the program is seen as a gies in a series of sessions would be likely to multi-faceted, compound treatment, veryhave a large and durable effect in that stu- much in the "shotgun" tradition.Such dents would be far more likely to think treatments are favoured by experiencedabout, question and monitor their methods teachers but largely ignored by psychologi-than in the past. The need for students to cal researchers because of the difficulty ofthink about their methods and about what sorting out the respective contributions ofthey are trying to learn has been stressed by various elements in the treatment.See many writers, (Tobias, 1982; Day, 1980; however, (Brown, Palincsar & Armbruster, Biggs & Rihn, 1984). 1984) who argue for a research strategy of first securing educationally relevant andSelection of items for the program transferable treatment effects by the use of multiple treatments and then, at a laterItems were constructed using information stage, of attempting to isolate individual gathered from the following sources:- tr atment effects. a. Strategies which have been used by The items chosen for the program were researchers in leading and studying. expressed in question form, with studentsb. Strategies identified in previous study ;74 Using Computers Irtelligently 265 strategies inventories (Brown & a.Learning new vocabulary - 60 items. Holtzman, 1965; Dansereau et ai, 1975). b.Preparing essays and assignments - 60 cT"nical "How to Study" handbooks items. (Pe...y, 1983; N.Z. Council for Educa- c.Preparing for tests and exams - 60 items. tional Research, Study Habits Evalu- d.Taking examinations - 20 items. ation and Instruction Kit, 1979; e.Reading productively - 140 items. McEvedy & Jordan, 1986). d. The literature on teaching reading Other categories planned include:- strategies e.g. Herber 1978. f.Using tutorials to learn. e. Discussions with and observations ofg. Learning from lectures. students. h. Time management. f.Discussions with a ...ademic colleagues i.Improving motivation. about their strategies. g. Managing stress and anxiety. g.Introspection of my own strategies in reading and studying. Preparation of comments The search for strategies and attitudes was Comments were prepared for each ques- not restricted to efficient ones but also in- tion in the program expla.ning the reasons cluded several which have been identified for preferred answers. The comments were as inefficient for students :r appropriate written in a conversetional manner and for only for an earlier stage of development. the most prst drew on empirical studies or The "telling all" strategy for answeringpsychological theory for their content. essay and examination questions, in which Where that was not possible comments students simply write down all they know were based on teaching experience.In about topics, and the "copy delete" strat- some comments small amounts of informa- egy for summarising texts, in which the text tion were included about empirical studies is simply copied less unimportant details,or basic psychological theory supporting are exam .'es. the program. Important principles were Additionally, attention was given to strate- repeated in comments on a number of re- gies which have been studied with both lated strategies. primary and secondary students but which were thought likely still to be used by some Examples of questions and comments used tertiary students. The "telling all" andin the program are shown in Appendix 1. "copy delete" strategies are examples. Presentation format Finally, some strategies were included which were thought ly to be used by all ,ee alternative formats have been students, to ensure t. at at least some of the planned for computer presentation to en- methods about which they were ques-hance the utility of the program both as a tion :d would be seen as appropriate. research and diagnostic instrument and as a teaching or tutorial device. The classification of strategies a.Questions only - for research purposes. The strategies and attitudes were classified b.Questions only in blocks of 20, followed into categories which were, for the most by a repetition of the questions, this time part, seen by students as major problem with each question being accompanied areas. Tne categories presently completed by a comment and thy. preferred an- are:- swer. This format combines both re- 275 266 Using Computers Intelligently search and tutorial functions. had been learned from the whole exercise c.Questions, each with associatcl. com- was completed anonymously by these stu- ments.This format would be useddents. mainly for tutorial purposes. All but two students regarded the exercise Twenty question blocks, rather than someas a worthwhile one aad indicated they other number, were chosen to balance thewanted the work continued and extended. need to provide reasonably short workSpecifically they wanted a one semester periods against the need to obtain reliablecourse in study methods introduced in the data for analytical purposes and for reports first year of their course. All of the students to students. commented about their disappointment that most of the strategies covered in the In an early compute presentation of theprogram had not been taught in secondary program format (a) was used, while in later schools and most claimed to have given hard copy trials format (b) was used. Inlittle thought to their own methods hith- future computer presentations using for-erto. mats (b) and (c), reports will be given to students after each block of 20 items. TheAlthough a reduction from 50% (n=83) to reports will rela te a student's score to estab-37% (n=88) occurred in the combined drop lished norms and recommend remedialout and failure rate for this unit when actions. These might include reading fur- compared with the previous year, results in ther instructional material.(Clark, 1986a, other units suggested a better group of 1986b) consulting a student cotinsellor orstudents was involved in the present year. working through practice examples of newAdditionally other changes in the teaching strategies etc. a: xamil -Ing of the unit were introduced wlrt may also have had effects.It was Preliminary trials with the program thus .lot possible to assert, with any confi- dence, that the Improvement cited above Almost all of the effort which has been put resulted from the use of the program. into this work so far has been directed at completing segments of the program andThe computer presentation used for this making these available to students many of work was flawed by programming bugs, whom are struggling to cope with theirslow screen presentation times due to work.Initially a 417 item questionnairecomputer overload, poor segmentation of was offered, both on computer terminalsthe program and,in a few cases, poor and on hard copy, to 80 second year Busi-wording of questions and of multiple ness Studi,.-s Degree students.Studentschoice alternatives. None-the-less, despite completed the questionnaire outside ofall problems, the work was strongly sup- class time in 4-6 sessions as part o: theirported by students who also provided course work in a one semester unit, Admin-encouragement for the development of the istrative Studies 1.Subsequently theserevised, categorised version of the pro- students were required to write an essaygram, including comments. saying what they saw as their major prob- lems as students and what they proposedA t the time of writing, July, 1987, 32 stu- to do about those problems. Three hourdents have completed, on a voluntary ba- long lectures and three hour long tutorials sis, some or all of the first four categoric, of were used to introduce some of the psycho-questions described earlier and other sf.i- logical principles involved. Finally, a shortdents are daily seeking access to the pro- questionnaire se Acing opinions about' :atgram. No proper analysis of the perform- 276 Using Computers Intelligently 267 ance of these students has been made be-References yona that of checking that better students are performing better on the program. Anderson, R.C. (1985) Role of the reader's The rr.,ponse of students to this second schema in comprehension, learning and memory. In H. Singer & R.B. Ruddell verFon of the program has been very en- thusiastic, many or them claiming to have (Eds.), Theoretical models and processes of reading, third edition, Newark, Dela- altered their study methods as a result of ware: International Reading Associa- completing sections of the program. tion. Andre, M. & Anderson, T.H., (1978) The Discussions with other staff members have development and evaluation of a self- suggested that there may have been sDme transfer of skills to other areas. questioning study technique. Reading Research Quarterly, 4, 605-623. Baddeley, A.D. & Hitch, G., (1974) Working Future work memory. In G.H. Bower (Ed.), The psy- chology of learning and motivation. Besides working to complete additional (Vol.8), New York: Academic Press. segments of the program, work is now Biggs, J.B. & Film, B.A., (1984) The effects of underway to set up the presently com- intervention on deep and surface ap- pleted parts of the program on 'ill I.B.M. proaches to learning. In J.R. Kirby (Ed.), System 2, Model 50 computer with Model Cognitive strategies and educational per- 8514 high resolution colour screen, using formance. Orlando: Academic Press. the FORGE computer based learning soft- Brown, A.L., Armbruster, B.B. & Baker, L., ware package. The latter provides for the (1986)The role of metacognition in design, layout, graphics, authoring, colla- reading and studying. In J. Orasanu tion and management of an instructional (Ed.), Reading Comprehenskin: From re- program. search to practice. Hillsdale, N.J.,: Law- rence Erlbaum Associates. Once this work is completed, systematic Brown, A.L., PaIincsar, A.S. & Armbruster, trials will be undertaken to assess the worth B.B., (1984). Instructing comprehension of the program, to establish norms, to de- fostering activities in interactive learn- velop reports for students and to modify ing situations. In H. Mandl, N.L. Stein & theprogram whererequirecl. Longitudinal T. Trabasso. (Eds.) Learning and compre- studies to ascertain if study skills improve hension of text. Hillsdale, N.J.: Lawrence during the period of a degree course are Erlbaum Associates. also planned. Brown, A.L. Campione, J.C. & Day, J.D., (1981) Learning to Learn: On training While it would be quite wrong for anyone students to learn from texts. Educational to believe that a relatively brief exposure to Researcher 10,12, 14-21. this program would "solve" the manyBrown, W.F. & Holtzman, W.H., (1965) problems faced by students referred to Survey of study habits awl attitudes, (Form earlier, it does seem likely that, with appro- C). New York: The Psychological Col.- priate course integration, the program poratic n. could find a place in, or as an adjunct to,Case, R., (1978) Intellectual development study strateg'es teaching programs and from birth to adulthood:A neo- student counselling it many tertiary and Fiagetian interpretation. In R. Siegler some secondary institutions. (Ed.), children's thinking: What develops? r'? 268 Using Computers Intelligently Hillsdale, N.J.: Lawrence Erlbaum As- Cognitive strategies and educational per- sociates. formance. Orlando: Academic Press Inc. Clark, N.J. (1986a) Questionnaire on strate- McEvedy, M.R. & Jordan, M., (1986) Suc- gies for writing summaries. Unpub- ceeding at university and college.Mel- lished paper, Bendigo College of Ad- bourne, Thomas Nelson. vanced Education. McKeachie, W.J., Pintrich, P.R. & Lin., Clark, N.J. (1986b) Strategies for locating (1985) Learning tolearn. In G. and/or inventing topic sentences. d'Ydewalle. (ed.) Cognition, information Unpublished paper, Bendigo College of processing and motivation. North Hol- Advanced Education. land: Elsevier Science Publishers. Daneman, M. (1984) Why some people areParis, S.G., Lipson, M.Y. & Wixson, K.K., better readers than others: A process (1983)Becoming a strategic reader. and storage account. In R.J. Sternberg Contemporary educational psychology, 8, (Ed.), Advances in the psychology of human 293-316. intelligence (V01.2). Hillsdale, N.J.: Percy, D., (1983) Study tactics. Melbourne, Lawrence Erlbaum Associates. MacMillan. Dansereau, D.F., (1985) Learning strategy Raphael, T.E. & Gavelek, J.R., (1984) Ques- research. In J.W. Segal, S.F. Chipman & tion-related activities and their relation- R. Glaser (Eds.) Hillsdale, N.J., Law- ship to reading comprehension: some rence Erlbaum Associates. instructional implications. In G.G. Dansereau, D.F., Long, G.L., McDonald, Duffy, L.R. Rochler & J. Mason (Eds.) B.A. & Atkinson, T.R., (1975) Learning Comprehension instruction. New York, strategy inventory development and assess- Longman. ment. (AFHRL-TR-75-40, Contract Study habits evaluation and instruction kit. F41609-74-C-0013). Brooks AFB, Texas, (1979) Wellington, N.Z.: New Zealand Air Force Systems Command. Ccuncil for Educational Research. D: y, J.D., (1980)Training summarisation Tobias, S., (1982) When do instructional skills: A Imparison of teaching methods. methods make a difference? Educational Unpul .shed doctoral dissertation, Researcher, 11, 4, 4-9. University of Illinois. Vygotsky, L.S., (1978) (Mind in society: The Day, J.D., (1983) The zone of proximal development of higher psychological development.In M. Pressley & J.R. processes) M. Cole, V.J. Scribner & E. Levin (Eds.) Cognitive strategy research. Srluberman, (Eds. and trans.) Cam- New York: Springer-Verlag. bridge, Mass: Harvard University Duffy, G.G., Roehler, L.R. & Mason, J., Press. (1984) (Eds.). Comprehension instruction. Weinstein, C.E. & Mayer, R.F., (1986) The New York, Longman. teaching of learning strategies. In M.C. Fitts, P.M. & Posner, M.I., (1967) Human Wittrock (Ed.) Handbook of research on performance. Belmont, Calif.: Brooks teaching, third edition.New York: Cole. MacMillan. Herber, H.L., (1978)Teaching reading in Weinstein, C.E. & Underwood, V.L., (1985) content arms. Second edition, Engle- Learning strategies: In J.W. Segal, J.W., wood Cliffs, N.J.: Prentice Hall. S.F. Chipman,. & R. Glaser,(Eds), Herber, HI., (1985) Developing thinking (Vol.1.). Hillsdale, N.J.: Lawrence and reading skills in content areas. In Erlbaum Associates. J.W. Segal, S.F. Chipman & R. Glaser Wilson, P.T. & Anderson, R.C., (1986) What (Eds.), (N'o1.1.).Hillsdale, N.J.; Law- they don't know will hurt them: the role of rence Erlbaum Associates. prior knowledge in comprehension: From Lawson, M.J., (1984) Being executive about research to practice. Hillsdale. N.J.: Law- metacognition.In J.R. Kirby, (Ed.). rence Erlbaum Associates. `27o Using Computers Intelligently 269 Appendix 1: Examples of ques- student a lot easier and to improve your tions and comments used in the performance. study strategies program The preferred sequence of answers is there- fore, a,b,c,d. Q.1 I believe my teachers and lecturers, rather than me, are responsible for teaching me any new words I need Q.3When I am reading for an essay, to know: assigrArn.mt or report I ask myself frequently, (at the end of para- (a) Yes, I strongly believe that. graphs, sections, chapters etc.) (b) Yes. I believe that. "What is the author's argument in a (c) No, I don't believe that. nutshell in this part?" I do this: (d) No, I strongly oppose that. (a) Almost always. COMMENT: Most lecturers and teachers (b) Most of the time. would be likely to explain important new (c) Sometimes. technical concepts. But they are unlikely to(d) Almost never. have enough time to deal with every new technical term. Additionally, they are un-COMMENT: This is a very useful summa- likely to spend much time in teaching yourisation strategy and a good way of fitting many non technical words. So, for the most in new information with your existing part, learning new words is up to you. knowledge. The action of summarising, in your own words, which is integral to this The preferred sequence of answers is there- strategy, helps tie in the new with the old. fore d,c,b,a. The strategy is also a precursor to the strat- egy of keeping a running check in your mind of a developing argument, where, Q.2 I set myself a target of learning awith each new paragraph, you run over in certain number of words every day: your mind the sequence of arg 'ment up to that point. (a) Yes, that is true for me. (b) Yes, I do that every so often. The preferred sequence of answers is there- (c) No, I rarely do that. fore a,b,c,d. (d) No, I have never done that. COMMENT: The importance of a large Q.4 If I do poorly in an essay, assignment vocabulary and its connection with your or report I consult my lecturer about ability to read and write wall, has been which students did well and ask him stressed in several other comments. So a to arrange for me to read two or three definite plan to increase your vocabulary is of the top efforts so that I can note an excellent strategy, provided that correct differences in standards and in ap- strategies are used (see other strategies for proach to aid me in future efforts. I hints) to ensure that such learning is effec- use this strategy: tive. Think of this: 10 new words a day would mean 3650 new words a year, and (a) Very frequently. over 10,000 new wards over the three years (b) Frequently. of a degree. Such an increase in vocabulary (c) Sometimes. could be expected to your work as a (d) Almost never. 279 270 Using Computers Intelligently COMMENT: Students often learn things (a) Very frequently. better from each other than from their lec- (b) Frequently. turers. Perhaps it is because there is a more (c) Sometimes. relaxed relationship between students. (d) Almost never. Whatever the reason it makes sense to try to get a line on differences in standards so thatCOMMENT: This strategy is one of the you know just how much, and in whatmost useful of all for tertiary level students directions, you need to improve. to learn. The strategy allows good summa- risation, comparisons and helps memory. The preferred sequence of answers is there-An extension of the strategy allows sub- fore a,b,c,d. categories in one or both sides of the table. The preferred sequence of answers is there- Q.5I think the sole purpose of essays, fore a,b,c,d. assignments and reports is to test the knowledge of students of course content in particular subjects: Neil Clark is a psychologist and senior lecturer in organisational behaviour (a) Yes, I am quite sure that is true. and administration at the Bendigo (b) Yes, I am fairly sure that is true. C.A.E. He believes that C.A.L. will be (c) No, I don't think that is true. particularly useful in the develop- (d) No, I am quite sure that is not true. ment and testing of his tertiary :evel study strategies program and of a COMMENT: It certainly is tnie that essays similar program for use with secon- etc. test your knowledge of different sub- dary students. jects. But they often have another objective which is sometimes not fully grasped by students. That objective is to see if students can handle the analysis and discussion of complex verbal issues with the required skill and whether they demonstrate the required depth of thought seen as appro- priate at a tertiary level. Now, if you don't know of the existence of this second goal there is a fair chance you won't give it the required attention - with dire conse- quences. The preferred sequence of answers is there- fore d,c,b,a. Q.6 When I have to remember different theories or approaches to a topic for a test or examination, I draw up tables listing important criteria or characteristics so that they can be readily summarised and compared. I do this: 4 (. Computer enhancing using a function graph plotter Debbie Clayton, Department of Mathematics and Computing Phillip Farrands, Mathematics Learning Cenire and Matthew Kennedy, Department of External and Continuing Education Capricomia Institute Computers are a valued resource in the instruc- pieces of software are used as tools in a tional process, however they need to be used intelligently. One way of doing this is to use the laboratory setting to assist the student to computer as an experimental tool, to centre the understand mathematics. Worksheets or learning activity on the student and provide a experiment sheets are used in the labora- means to aid students' understanding of con- tory to guide the student interaction with cepts.This paper details how a team from Capricornia Institute developed Cap-Graph, a the software. sophisticated graph plotting package. This package b an experimental tool which, with This paper describes the development and appropriate worksheets, can be used by a wide use of such a software tool. Cap-Graph is a range of students to develop and deepen con- ceptual understanding in their mathematics function plotting package, which, wall courses. Cap-Graph has been trialed in a labo- appropriate worksheets has been trialed at ratory situation with an internal mathematics Capricornia Institute to assist students subject for biology students. The results of this studying an introductory mathematics trial and the implications for incorporation of Cap-Graph experiments in other mathematics subject for Biologists. The potential for the cou7ses is discussed. use of a piece of software such as this in other mathematics subjects is also ad- dressed. At last year's CALITE conference, one of the authors, Debbie Clayton, spoke about Rationale CAL trends it tertiary mathematics that she had obser,. ,...d in the UK and USA. One Why develop a graph plotting package? An trend in undergraduate mathematics waseasy to use function plotting program, can away from the traditional dialogue/tuto-be employed by students studying a wide rial mode of the 1960's and 1970's to Com- variety of subject areas. For example, a puter Enhanced Learning (CEL). This termCalculus student can use it to plot a func- was coined by Professor Avi Bajpai of the tion with a discontinuity when studying University of Technology in Loughbor- limits. A Statistics student can plot proba- ough (Bajpai, /985). bility density functions with different para- meters on each graph. This wide range of In.. CEL application, the computer is put to usage options, as well as the plotter's capa- "best" use by being either: bility to graph quite complex mathematical functions makes it an "intelligent" choice an electronic blackboard, of software tool. a numerical utility, or an investigational/experimental tool. The provision of such a tool to students was seen as one approach to assisting them in When the machine is used for investiga-more easily attaining a holistic understand- tior /experimentation, sopi.:aticated ing of the graphical behaviour of functions. 28.1 272 Using Computers Intelligent Cap-Graph aims to assist students to learn CGA (Colour Graphics Adapter) their mathematics by presenting visual EGA (Enhanced Graphics Adapter) representations of the functions that they Hercules (Hercules Graphics Board) are studying. User Int -rface The Software It was necessary to ensure that the software Design Considerations was simple to use, to try to minimise stu- dent time required to become familiar with The Cap-Graph software was developed the program. The target audience for the utilising the authors different areas of program may not have touched a computer expertise. Defining the goal of the project before, and students may be using the and general system specifications was program away from any immediate assis- undertaken by the academic members of tance. the team, whilst the programmer wrote the software. Close consultation was To achieve the aim of creating a software maintained between all three members of package which was easy to use and hence the team. aid students learning; screen layout, user control, feedback and on-line help were all The program was written in Turbo considered an essential part of the user Pascal, and extensive use was made of interface. Possible outcomes of poorly de- the Turbo Graphic:: Toolkit. Tuibo rascal signed or iftappiopliate inierfctcx,s aie was chosen to allow easy portability to a stated by Price (1985) as : variety of machine configurations. The use of the Graphics Toolkit reduced the user refuses to make use of the development time in creating the user system after initial testing; interface. the user learns a few commands One of the aims of the system design was 'parrot fashion' and relies on us- to produce a program which could, or be ing this, hence he/she cannot easily converted to, run on both NEC III make use of all the system's facili- and IBM compatible computers. The ties; or Capricornia Institute has a laboratory of an intermediary is used who then 20 NEC APC III computers. In addition, becomes the local system expert. similiar machines are installed in the Capricornia Institute's External Study Screen Design Centres (13) located throughout Queens- land. Many students own or have access Research into colour screens indicates to machines with IBM compatible graph- there should be a strong contrast between ics. Although NEC APC III computers arefo.slround and background colour inter; MS-DOS machines, the graphics are sities (Issacs, 1986). Research has also been different to the de facto IBM standard reported on colour combinations which graphics. Versions of Cap-Graph have meet a minimum criteria for legibility been developed to run on machines (Oliver, Sullivan & Barr, 1982). The colour which are compatible with the following: combinations used in Cap-Graph comply with these findings. NEC APC III IBM PC, X1', AT with one of the The screen has two main functional areas, following graphics cards which are used consistently throughout the 282 Using Computers Intelligently 273 program (Godfrey & Sterling, 1982). Thescreen. Explanatory feedback is given top line of the screen is displayed as a bluewhen invalid syntax in mathematical ex- foreground on a yellow background. All pressions is used. command key options and error messages are shown on this line. The remainder of the Help screen has a yellow foreground on a blue background. These colours also provide A detailed help facility has been designed good contrasts on monochrome displays. as an integral part of Cap-Graph. Help Main menu options are shown on the sec-provided depends on the users location in ond line. the program. Context sensitive help, avail- able at all times, is considered essential in The screen design is basically one of a split CAL programs (Hartley, 1986). Approxi- screen with pop-up windows. Commands mately 60 different help screens have been are listed at the top of the screen, and the 'Included in the program. graph displayed on the remainder of the screen. Pop-up windows are used to dis-Program Features play sub-menus, help screens, user inputs and feedback on invalid syntax of user entered functions. Sub-menus, displayed Equations are entered in functional nota- in pop-up windows as required, occur attion form. The scaling of the graphs is de- the top of the screen. Help is availablefined in terms of domain and range, and throughout the program and is displayedscientific notation is used, by the program, in a pop-up window at the bottom of the for large numbers. screen. Current research tends to favour the use of pop-up windows to keep theFunctions are entered by initially defining screen uncluttered ffssacs, 1986). the dependent and independent variables. These are restricted to the English alphabet. In addition, the irrational number e is de- User Control fined as a c,nstant, so use of this letter is not encouraged. The program can store 26 dif- Control of the program is maintained by ferent functions. The screen, however, is the user with the cursor keys, the Returnnever automatically cleared so an infinite and Escape keys, and the Help or Fl key. number of equations can be displayed at Escape is always used to return control toone time by editing then drawing the func- the user by aborting the current process tions. Any of the previously defined func- and returning to the main menu. The Re- tions may be edited by attempting to enter turn key is always used to accept input ora function with the same dependent vari- select the highlighted menu option. Theable. A facility also exists to charge the position of th se keys (and different return most recently entered, or edited firm,. ion. key symbols) on the keyboard are ex- plained at the beginning of the program. A wide variety of mathematical functions Menu options can be selected using either can be drawn using Cap-Graph. Both con- the cursor keys or the unique letter indica- tinuous and discrete functions can be dis- tor for each option. played. Discrete functions are entered us- ing the predefined integer component Feedback function. It is possible to display Binomial, Poisson, Normal and Exponential distribu- Feedback to the user is provided during thetions using this approach. Taylor's Series graph drawing/calculating process by -approximations can be easily shown and small line which moves across the top of the the improvement in the approximation 283 274 Using Computers Intelligently with each successive term deafly demon- Constants strated. The constants e = 2.718281828 Cap-Graph can also draw either the first-or pi = 3.141592654 have been defined. second derivative of the most recently en- tered, or edited function. One of the more powerful features of the program is the ability to define a function LA Defined Operators, Functions and terms of previously defined functions. That Constants is, if the functions f(x) and g(x) have been entered, the the following functions are The normal mathematical operators areexamples of valid functions available, namely: h(t) = f(t) + g(t)m(s) = f(g(s)) + addition p(x) = abs(f(x)+g(x)) subtraction * multiplication (unnecessary if unambi- Scaling guously implied) / division The scale of the graph is user defined as a A exponents (5 cubed is enteredas 5'3) minimum and maximum for the domain ! factorial (valid for whole nun 'Ix rs). and range. In addition, there is an au- torange facility which will calculate a range Functions included are: so that at least 90% of all functions stored in the system will be displayed on the screen. sin(x) trigonc metric sine Scaling is limited to numbers between +/- cos(x) trigonometric cosine 999999999 to +/- 0.000000001. tan(x) trigonometric tangent arcsin(x) inverse trigonometric sine Zoom Facility arccos(x) inverse trigonometric cosine arctan(x) inverse trigonometric A facility has been built into the program to tangent allow close examination of parts of the sinh(x) hyperbolic sine graph. This option allows the user to zoom cosh(x) hyperbolic cosine in on a region of the screen. The magnified tanh(x) hyperbolic tangent section is displayed in a window on the arcsinh(x)inverse hyperbolic sine screen which can be moved so that the arccosh(x)inverse hyperbolic cosine original graph is not obscured. arctanh(x)inverse hyperbolic tangent exp(x) exponential function (e^x is The coordinates of the centre of the wia- also valid) dow are displayed, (to ten figures) at the In(x) natural logarithm top of the screen together with the width of log(x) common logarithm the window, for example +/- 0.33333. Us- sqrt(x) square root ing the coordinates it is possible to find sqr(x) squared function points on a graph to a maximum 10 signifi- rad(x) converts the operand from cant figures. degrees to radians u(x) heaviside function Magnification of an adjacent region can be int(x) integer component easily displayed using the cursor keys. The fract(x) decimal component effect of this facility is similiar to moving a round(x) rounds to the nearest integer magnifying glass over a piece of paper. abs(x) absolute value 28 4 Using Computers Intelligently 275 Magnification of the zoom window can be special skills related to the usage of this reset at any time. The centre of the zoom new teaching technology. window is set by defining the coordinates or by using the cursor keys. The Trial Options The plotter and as.. worksheets have been trialed by an internal first ; ear Hard-copy printouts of the screen can bebiology degree mathematics class. During obtained from Cap-Graph. Six differentthe semester these students worked Epson printer modes are supported whichthrough four experiment sheets in a labora will print the screen in different sizes and tory setting. Instructors (the authors) were height-width proportions. present to assist with both mathematical questions and questions about the soft- Computer Enhancement ware. The four worksheets for the experi- ments related to the topics: The development of the software is per- haps the easiest part of computer enhance- Linear and Quadratic Functions Limits and ment. The enhancement of a subject (onceContinuity Exponential and Logarithmic the software tool exists) proceeds in theFunctions Derivatives and Curve Sketch- following manner: ing. As an example of the type of activities in The syllabus is reviewed. Areas where these interactions, consider the Quadratic students experience difficulty with Function section of the first worksheet. The their mathematics and couldbe assisted a'..m of this experiment was to develop the by a visual function representation are student's appreciation of the effect that identified. each parameter in a quadratic function has on the shape of its graph. Here students In these areas, student centred interac- were first asked to plot a parent graph of x tions with the software are constructedsquared. Then, in a systematic way, the by writing guided experiment / work-parameters were changed and the student sheets that can be used in a laboratorywas required to write down statements on setting. their worksheets as to the effect that each parameter had on the shape of the graph. A possible worksheet structure is: At the end of the semester, the computer the aim/s of the experiment enhanc-ment using Cap-Graph and the the necessary mathematical background the stu- guic ed experiment sheets was evaluated dent needs to proceed with the experiment by questionndire. Eleven of the class of nineteen students responded. reference /s to appropriate texts It was found that, in general, instructions /elated to what activities the student should go through with the software the students list e.d going to the micro- student questions related to those activities. computer laboratory and working through the experiments with Cap- Designing these experiment/wor! sheets Graph. They found the program easy to so that the desired learning outcomes, esult use, and felt positive about obtaining a for this guided interaction is rather diffi- graphical representation of the func- cult. For this, educators need to develop tions they were studying. 285 276 Using Computers Intelligently those students who had not used a different colours for different graphs. computer prior to the interactions, now This will not be possible on machines felt more confident about using the with CGA or Hercules graphics. computers, and a system of being able to match the the students considered it did help them graph to the function after the function to learn their mathematics. has been drawn, The students also made suggestions as to how the user interface of Cap-Graph, and an option to allow the definition of constants, how the format/content of the worksheets could be improved. These suggestions were incorporated where feasible. displaying the equation in the normal mathematical form after it has been A Wider Audience entered, exponent above the base, numerator over the denominator. Aftzx the initial development and program testing with an internal class, Cap-Graph a choice of scale to allow log-log and was made available to a wider audience. At semi-log graphs to be drawn, the beginning of the second semester, 1987, copies of the software were located in exter- incorporation of a numerical integra- nal study centres and in the Library at the tion routine. Capricornia Institute. A number of staff access the program using personal com-Feedback from academics is disciplines puters on their desks. Whilst direct incor-other than mathematics indicates that the poration of Cap-Graph as an electronic scope for use of a function plotter is wide- blackboard (in a lecture environment) has,spread. Cap-Graph is capable of being used as yet, not occurred, overhead transparen-as a tool in diverse areas such as engineer- cies of printouts from the program have ing, financial mathematics, economics, and been used. the behavioural sciences. The authors are currently writing a series of Summary experiment sheets to be combined into a booklet form. This material will be avail-This paper has outlined the development able to students as resource material with and trialed use of a software tool that can the software in the Library and in externalenhance student learning of mathematics. study centres. Lecturers, for a variety of subjects, will also be able to direct studentsSoftware tools such as Cap-Graph do have to relevant worksheets as either an adjunctconsiderable potential in not only mathe- or as a formal component of the subject. matics teaching, but also in several other disciplines. The scope of its use is wide, Portability of the software has meant thatlimited only by the imagination and ability compromises needed to be made in theof the instructor to construct experiments/ program design and features included. The worksheets to guide student interactions main constraint is restricting the memory to achieve desired learning outcomes. usage to 256kb. Since many users will. have access to compatible computers withk:...tThe authors hold that constructing soft- this restriction, a more sophisticated ver- ware tools which can be used by a variety of sion is currently being planned. Featuresstudents and in a variety of subject areas is being considered for this package are: an intelligent use of CAL in the 80's. '2, 8 6 Using Computers Intelligently 277 References Bajpai,A, et al (1985), Mathematics and the Micro. Some hints for software develop- ment. International Journal of Mathemat- ics Education in Science and Technol- ogy,16(3)A07-415 Godfrey,D. & Sterling,S. (1982). The Ele- ments of CAL. The How-to Book on Computer Aided Learning. Reston Va, Reston Pub. Co. Hartley,J.R. Placing Expertise in Computer Assisted Instruction. Fourth Calite Con- ference Proceedings, Adelaide 1986,1-10 Issacs,G. Screen design for Computer As- sisted Learning. Fourth Calite Conference Proceedings, Adelaide 1936,147-157 Oliver,D.,Sullivan,C.,& Barr,J. (1982). So you want to be a Prestel Information Pro- vider? A manual of practical advice for providing local community informaqon on Prestel. Aslib London, British Library Research & Development Report No. 5655 Price,J.A. Software Engineering : A Guide for CAI. Third Calite Conference Proceed- ings, Melbourne 1985,44-60. Debbie Clayton, Lecturer, Computer Enhancement of Tertiary Level Mathematics, Remedial Mathematics, CAL User Interface Development. Phillip Farrands, Lecturer, Screen Design, Secondary/Tertiary interface mathematics, effective use of comput- ers in education. Matthew Kennedy, Programmer, Cre- ating effective software tools for stu- dent use. 281 Updating Mathematical skills:An evaluation of a computer based mathematics course Michael J. Crock, Division of External and Continuing Education Darling Downs Institute of Advanced Education a: 111,1 U:k!ating Mathematical Skills is a continuing based LEARNING activities which pro- education course for students who wish to study at the tertiary level but who do not feel confident vide students with opportunities to test the with their level of mathematical ability. The effectiveness of their learning and obtain study materials consist of a study book, the set grades and feedback. The 'managed' as- text "Self-Paced Introductory Mathemati, pect of the system maintains records of and an introductory audio cassette. All assign- ments for the course are completed on student students' use of CML activities, scores stu- answer sheets and computer marked with full dent learning activities and provides this feedback printed at the time of marking. Results data to lecturers. and feedback are then forwarded to the student. Student progress records for all assignments are computer managed. The course has been fully The breadth of the CML activities provided operational for over 18 months with enrolments in the system range from 'conventional' to date being over 300 students. Initial evalu- CML, with its emphasis on management of ation of the course shows a cost-effective utiliza- student records and scoring, to activities tion of computer managed learning, and a posi- tive response to the course structure and effec- more representative of computer assisted tiveness by students. An interactive version of instruction (CAI), involving extended assignments is currently under trial. feedback, branching programs, algorith- mization, and the linking of microcompu- ters and videodisc systems. In any teaching-learning situation a large proportion of the teacher's time is devoted To effectively utilize this CML system, the to prescribing learning activities, markingunit team needs to analyse and structure assignments, providing associated feed-are subject matter in question into appro- back and keeping records. This feedback priate instructional modules, which re- usually includes advise on learning prob- quire: lems that have been diagnosed in light of student performance. A significant part of 1. the precise sequencing of instructional these activities cast be assigned to the com- objectives, which clearly define ex- puter thus providing the basis for a com- pected student learning outcomes; puter managed learning system. In 1983 a com-euter manageti learning system was2. the careful selection of associated learn- established at the Darling Downs Institute ing activities, which clearly define stu- of Advanced Education and the essence of dent work programs; and this computer managed learning is repre- sented in Figure 1. (IK-rker, White and3. the construction of tests, which aim to Taylor, 1984). measure the extent to which students have achieved each of the stated objec- Incorporated in the CIvIL system is the tives. (Barker, White and Taylor, 1984) development, by unit teams, of computer 4288 Using Computers Intelligently 279 By making this initial commitment to struc- formance. (Barker, White and Taylor, ture the subject matter in this manner, the 1984) unit team can then expect to proviri.e the students with the following potential bene- The Updating Mathematical Skills course fits: which is to be evaluated in this paper was developed using the Darling Downs Insti- 1. early identification and resolution oftute of Advanced Education's CML system problems the pinpointing of studentand unit team approach as outlined above. difficulties and the availability- of this information to both students and exam- Updating Mathematical skills iners alike should lead to a rapid resolu- tion of problems; Updating Mathematical Skills is intended for students who wish to study at the terti- 2. a clearly defined pacing mechanism ary level but who do not feel confident with by meeting the deadlines set for the their level of mathematical ability. It can be completion of each test, students can be taken individually as a continuing educa- sure that they are working through the tion course. or it can be taken as a compo- materials at an appropriate pace; nent of the Preparatory Studies Pro- gramme, whicl, is a currently operating 3. immediate individualised feedback external programme developed to provide an assessment of the extent to whichan alternative means of entry to tertiary stated objectives have been achieved;coun-res for mature age students. Using the diagnosis of difficulties and associatedconcepts of self-paced instruction, and provision of a list of learning activitiesmastery learning the course guides stu- which should lead to improved per- dents through a carefully sequenced series COMPUTER MANAGED LEARNING PRESCRIPTION OF .7'LEARNING ACTIVITIEZ, . RECORD DIAGNOSTIC KEEPING/ FEEDBACK TESTING LREPORTING Figure 1: Key elements of Computer Managed LEarning 289 280 Using Computers Intelligently of topics which provides the foundation for dents to a few of the team of course tutors understanding the mathematics that willwho will be monitoring and assisting their be encountered in tertiary studies. progress. The self-paced structure of the course al-The set text (SPIM) contain the required lows students to work through the material content for the course and the study book is at a pace suitable to their needs, permitting provided to guide students through the them to work quickly through familiarmaterial by introducing topics, highlight- materials, as well as allowing them theing problem areas, and providing addi- opportunity to seek additional assistance tional worked examples and practice prob- in areas of uncertainty. The mastery ap- lems. proach ensures that students successfully achieve the objectives of each topic beforeThe structure of the course is based on the progressing to the next topic, which builds structure and sequence provided in the set upon on the earlier material. text SPIM. Originally produced to provide the basis for a mathematics course for envi- The main objectives of this course are: ronmental science students, the style, con- tent and level of the material presented in 1. To increase the student's understand-SPIM has been used to provide a solid ing and confidence in mathematical mathematics base for students enrolled in a skills to a level which will enable a stu-wide variety of courses including science, dent to successfully deal with the mathsbusiness, and engineering tertiary courses. they will encounter in their tertiary studies; and The study book, which has been specifi- cally designed to enhance the material and 2. To develop a positive attitude towardsfacilitate the external teaching of the mathematics which will assist the stu- course, guides students through SPIM's 14 dent in their understanding of basicunits, which have been grouped into six mathematical concepts. modules according to content. With the exception of the single unit Module 1, each The course material includes: module contains a series of two or three related units. The assignments for this 1. An introductory audio cassette; course correspond directly to the modules. 2. The set text:SELF-PACED INTRO- The following figure illustrates the break- DUCTORY MATHEMATICS (SPIM) down of the 14 units into the six modules, ay Dobson and Stokoe; and briefly lists the content of each unit. 3. Updating Mathematical Skills Study MODULE UNITS CONTENT OUTLINE Book; 1 Number systems and elementary arithmetic 4. A set of assignment labels for CML ac- tivities. 2 2 Notation, order and inequalities The introductory tape gives students a 3 Graphical represen tP don and brief introduction to the course as a whole, equations of lines highlighting the important points they should be considering as they work 4 Sets 3 5 Polynominals and use of the through the course, and introduces stu- binomial theorem 290 Using Computers Intelligently 281 assignment for module 2 has three dis- 6 Functions tinct sections, one for each of units 2, 3 7 Inverse f .ztions and and 4 of SPIM.) graphing of functions Appendix A contains an example of a CML 4 8 Exponential and logarithint: functions assignment and the marking sheet the stu- dent would be expected to complete. 9 Trigonometric functions In order to successfully pass this course, 10 Limits of sequences and functions students need to complete all six assign- ments and achieve a pass mark for each 5 11 Introduction to differentia- one. Assignments take the form of com- tion puter-marked tests and are only graded 12 Applications of differentia- pass or incomplete. Student are always tion given feedback on incorrect answers and should they have an assignment marked 6 13 Introduction to integration incomplete, they are required to submit i4 Techniques and applications one of the two remaining assignments for of integration that particular module, thus enabling the student to upgrade their standard of un- Figure 2: Course Structure and Content derstanding to a pass level. Appendix B contains an example of a The sequence of the units has been care-student's marked assignment and feed- fully planned to ensure that students prog-back specific to the responses the student ress through the material in a logical build-has answered. ing block fashion. The difficr",,,, style ar.d length of assign- Assessment ments corresponds closely with the scope and c ntent of questions found in the pre- The assessment in the course is of twotest and sample quizzes in the study book types: and text. 1. Self-Assessment which takes the formThe CML system permits answers to be of pre-tests, exercises, and sample quiz-multiple choice or exact answer, and the zes in the study book and the set textfeedback given can be specific to an exact SPIM. Such self-assessment items inanswer or to a particular numerical range. this course are not required to be sub- mitted and are for the students' benefitIf students receive a pass for a particular to show them how well they are under-module assignment, but for their own per- standing the material. The answers tosonal benefit would like to submit any of all self-assessment items at, given tothe remaining parallel assignments, this students in either the text or the studymay be done and the results do not affect book. the pass mark already achieved. 2. Formal Assessment, which takes theAppendix C contains an example of the form of six CML module assignments. student record report used to monitor stu- Each of these assignments has distinct dent achievement levels and progress sections for each unit of the text. (i.e. thethrough the course. 2 282 Using Computers Intelligently Student enrolments with individuals ranging between three months and 46 years since last studying in Although the teaching year at Darlingthis area. The majority are also interested in Downs is divided into three periods, Au-attempting external studies while main- dents may enroll in Updating Mathemati- taining their current job position. While cal Skills at any time, as either a Continuing most f tudents live in Queensland, there are Education course or as part of the Prepara- also a number of interstate and interna- tory Studies programme, allowing maxi- tional students regist'ered in the pro- mum flexibility. However, the final exami-gramme. (Crock, 1987) nation for the Preparatory Studies students is held only three times per year, in June,To support and run the programme with November and February. Students study at these numbers there are three part time their own set pace, however, a student who tutors in the Updating Mathematical Skills wishes to study at about the same ratecourse. The course also utilizes Continuing which is expected of people enrolled partEducation administrative staff and DDIAE time in accredited courses, such as theadministrative services such as mailing, Bachelor of Business or the Associate Di- student enrolments, student support ploma in Engineering, should complete the centre, and advertising. course in approximately 18 weeks. Methods of evaluation During 1986, 194 students enrolled in the Updating Mathematical Skills course asStudents are given the opportunity to Continuing Education students. Of thiscomplete mitten questionnaires evaluat- number 17 have successfully completeding each component of the Updating the course, while 152 are currently stillMathematical Skills course, and in fact actively enrolled and continuing studies. Incomment on each assignment and its re- this same year, 238 students enrolled in thelated material. Appendix D contains a course as Preparatory Studies students. Ofsample of the evaluation form used to col- these, 20 have successfully completed the lect data. The appropriate course develop- course and gained entry to accredited ment unit team then has the task of recom- courses, and 92 are still actively enrolled. mending programme improvement based upon students perceptions plus the team's So far in 1987, there are 106 students en-evaluation of the students standards of rolled as Continuing Education studentsachievement. In essence, this means the and 207 students taking the mathematicscurrent evaluation of the course with refer- course as part of Preparatory Studies. Outence to meeting the previously stated ob- of this enrolment, 25 have already success-jectives, is based upon student's direct fully completed the course, with 24 of these evaluative responses and student's gaining entry to accredited courses. achievement records. Subsequent im- provements to the material are then incor- The vast majority of both the 1986 and 1987porated into the material where considered students are mature age students who areappropriate by the leader of the course interested in returning to tertiary studiesdevelopment team. This process is carried after a significant number of years since lastout on an annual basis. attempting studies of any description. Il- lustrating this point is an enrolment surveyResults of evaluation which shows that average time since last studying mathe -Mies for the current Dealing first with student evaluation of the group of over 300 students is 14.6 years, course, Table 1, on the following two pages, 6-1r9 Q. Using Computers Intelligently 283 presents the results for the opinionnairespaced instructional package they were for modules 1 to 6. Responses range fromrequired to use, particularly in the first "strongly agree" ( =1) to "strongly dis-three modules evaluated. The results for agree" (-4) and the table shows the mean the last three modules still reflect students' response (M), the sample size (N) and the strong support, yet reflect the increasing standard deviation of responses (SD). The difficulty they must have experienced as final question listed on the table deals withthey progressed through to the more diffi- the average number of hours required to cult concepts. work through a particular module. Questions 4 and 5, dealing with the appro- Table 1: Results for Student Evaluation. priateness of the assessment items and the services provided by the DDIAillustrate Evaluation based on student success ratesa high level of support for the computer is limited at thi, stage as the course is stillbased assignments, and the associated relatively new and there ire still a largemarking procedures including the ex- number of both 1986 and lc 37 students still tended feedback provided. actively completing the course. However if you examine the subsequent success ratesThe final question asked of students related of students who have continued on to ac- to the number of hours spend on a particu- credited courses, having first successfully lar module. If you take into account the fact completed the Updating Mathematicalthat Modules 2, 3 and 4 encompass three Skills course, it appears t1course -.lasunits in the set text, and 5 and 6 two units, provided a solid mathematical back- the average number of hours per unit in the ground, and signalled that students wouldtext range from 6.55 hours for units in be able to cope with tertiary level studies. Module 2 to 14.18 hours for units in Module Of the first 16 accredited course units(6. These are encouraging results as it is many containing mathematical compo-stipulated that the average time required to nents) attempted by success.. al Updatingbe spent per unit should, on average, be in Mathematical Skills students, the results the vicinity of 8-10 hours. obtained were: five As, six Bs, four Cs and only one incomplete. As stated earlier, discussion on the effec- tiveness of the course based on pass rates is Discussion at present limited. While tentative conclu- sions that the course is achieving satisfac- Evaluation forms submitted by studentstory results can be made, statistics need to have been strongly in support of the Updat-be finalized after the initial enrolment ing Mathematical Skills course with respect population of 1986 and 1987 have com- to its format and perceived applicability in pleted their attempts at the course. achieving the set objectives of the course. In particular, while it was expected studentsConclusions would find the course to become slightly more difficult as they progressed through In a course with a current active enrolment to the latter stages, (reflected in responses of over 400, which has hack over 2000 as- to Question 2 of the evaluation) the stu- signments processed during the last 18 dents maintained a high level of interest months, and a course with only three part- throughout the course, shown in the re- time tutors, the cost-effectiveness and time- sponses to Question 1 of the evaluation. efficiency of running the course as a com- Results for all sections of Question 3 illus- puter based education package is self-evi- trate the students' support for the self-dent. Coupled with the positive response 2D 3 284 Using Computers Intelligently from student evaluation, and the emerging development perspective. Australian success patterns of individuals who com- Journal of Adult Education, 25(1), 23-30. plete the course, it would appear the course Crock, M.J. (1987). Preparatory studiesat is successfully meeting its two main objec- the Darling Downs Institute of Higher tives as outlined at the beginning of the Educatio: A case study in alternative paper. entry to tertiary education. Paper pre- sented to ASPESA 8th Biennial Forum, However, there are still important issues in University of New England, July. the evaluation of the course which are cur- rently under examination, .1nd need to 1:1/4. finalized before a complete evaluation of Michael J. Crock is currently em- the course can be tabled. With the introduc- ployed as a LecturerInstn. ctional tion of the interactive micro-computer Designat the Darling Downs Insti- based assessment in 1983, and with the tute of Advanced Education, where completed statistics of the 1986 and 1987 his duties include the design of student populations, it will be possible to print, video, audio and instigate additional evaluation as to the computerbased courseware for effects on students, and the efficiency of external and internal courses, and the various modes of assignment process- he manages the Computer Managed ing. More specifically, it will enable Learning program at DDIAE. Cur- insight into issues related to the utiliza- rent teaching involves remedial and tion of computers in the delivery, proc- tertiary level mathematics. Current essing and monitoring of course material research involves, the effective and on a large scale. efficient utilisation of instructional Bibliography design modelss, and the analysis of Computer Based Education systems Barker, L.J., White, V.J., and Taylor, J.C. with reference to the cost effective- (1984)Computer managed learnint, in ness and the changing educational tertiary education: An organisational requirements for such systems. APPENDIX A ASSIGNMENT 1A: MODULE 1 . SPIM UNIT 1 This assignment is to be a computer marked test containing both exact answer and multiple- choice questions You should place your answers on the answer sheet -mmediately following this assignment. Answer all questions in the format requested, and please note that if an answer is negative you must remember to include a "" before the value. If your answer is positive oo not place a "+" before the value as it is not required. All answers for this assignment should be in integer and/or fraction for.n, and if you should obtain an answer which includes both an Integer and fraction value, it should be placed on the answer sheet in the following form "3 "would be wnt.n P1 -where "&" is tho "and" symbol. If too many boxes tire supplied for a particular answer remember to leave the end boxes blank. Appendix A: (continued on page287) Using Computers Intelligently 285 Question Mod Mod Mod Mod Mod Mod 1 2 3 4 5 6 1.I found the module H 1.741.80 1.81 1.92 1.79 1.83 interesting. N 167 86 43 27 14 12 Si) 0.47 0.40 0.46 0.62 0.58 0.58 2. The module was H 1.782.32 2.322.84 2.71 2.92 easy to follow. N 176 75 47 25 14 12 SD 0.73 0.62 0.63 0.75 0.89 1.00 3A. The presentation 1.78 1.89 1.88 ?.07 2.07 2.17 of the module including: N 160 71 42 27 14 12 - formative SD 0.50 0.43 0.40 0.68 0.83 0.83 assessment items assisted my learning. 3B.The presentation H 1.772.00 2.00 2.33 2.14 2.5 of the module including: N 163 73 42 24 14 12 directions to SD 0.54 0.47 0.49 0.71 0.86 1.00 read assisted my learning. 3C. The presentation 1.68 1.75 1.83 2.15 1.92 2.33 of the module 0 19 including: N 165 73 41 26 1V Study Book SD0.50 0.52 0.44 0.47 1.00 1.07 assisted my learning. 1 3D. The presentation H 1.821.92 1.95 2.23 2.07 2.08 of tha module including: N 154 73 39 26 14 12 - statements of SD 0.47 0.43 0.51 0.65 0.73 0 51 objectives assisted my learning. Table 1: Results for student evaluation 3 286 Using Computers Intelligently Question Mod : Mod Mod Mod Mod Mod 1 2 3 4 5 6 3E. The presentation 2.11 2.00 1.95 1.88 2.12 2.00 of the module including: N 72 27 19 13 d 5 - telephone SD 0.74 0.48 0.62 0.49 0.64 1.00 contact assisted my learning. 3F. The presentation 1.67 1.74 1.74 1.89 2.142.08 of the module including: 160 72 39 27 14 12 pretest SD 0.57 0.500.55 0.50 0.86 0.79 assisted my learning. 3G.The presentation 1.68 1.70 1.74 2.25 2.21 2.00 of the module including: 161 73 39 27 14 12 set text SD0.54 0.54 0.55 0.81 0.80 0.74 assisted my learning. 4. The assessment itemM 1.77 1.78 1.67 1.92 1.78 1.83 is a fair test of my attainment of N 164 77 42 27 14 12 objectives in this SD.0.52 0.4.5 0.47 0.62 0.42 0.39 module. 5. External studies M 1.81 1.94 1.94 2.05 1.92'1.89 facilities provided by DDIAE N 155 68 34 21 13 12 were adequate for SD0.52 0.52 0.42 0.66 0.490.78 studying this module. 6. I spent approx. 8.0319.6527.4737.4828.3328.36 hours on this module. N 162 68 38 25 12 9 SD 5.9610.8616.7327.9723.7624.54 Table 1: (continued) Using Computers Intelligently 287 QUESTION 1 Which of the following are the prime factors of 180? (Select the correct option.) 1. 2X2X3X3x 5 2.4 x 5 x 9 3.1 X 180 4. 3X3Xe :6 QUESTION 2 Evaluate 3 + 311.2(3° X 33Iy + 32 QUESTION 3 alb -lc Which ofthe following is the correct simplification of - ? a b2c3 ( Select the correct option.) ab3 1. c2 a 2. b3ch ab-1 3. C alb 4. c3 QUESTION 4 Solve the equation 4t -10 = 2t - 9 2 9 288 Using Computers Intelligently QUESTION 5 1 4 1 Evaluate(+ 2= 6+)X 2( 3 6 3 QUESTION 6 Simplify 27-' QUESTION 7 1 2 3 3ab If a = b = and c = find what equals. 3 5 5 2bc QUESTION 8 Solve for x: 3x + 7 = x + 11 QUESTION 9 1 3 1 Solve fort: + + 4 = 7 t 2t 2t QUESTION 10 Which of the following is the correct simplification of x`y(x4z + xy3)? (Select the correct option ) 1.x7y4z 2.x-2yz + xy2 3.x' + yz + x3 + y4 4.x6yz + x3y4 2 9 S Using Computers Intelligently 289 COMPUTER MARKED TEST r ANSWER SHEET Place your assignment sticker here FOR OFFICE USE ONLY J TEST 11032- Assignment lA DUE DATE Week 2 1 Please remember that you must include a "--" if your answer is negnive but do not include a " + " if it is positive. If too many boxes are supplied for a particular answer leave the end boxes blank. 1. (1:21) 2. DODEICI (1:22-1:26) 3. (1:27) 4' DOD OD (1 :28- 1:32) 5. CI CI 0 El CI (1:33-1:37) 6. D (1:38- 1:42) (1:43 - 1:47) 7. DODEIEl 8. DOEIDE1 (1:48- 1:52) (1:53 - 1:57) 9' 10. (1.58) ....s59 290 Using Computers Intelligently UNIT : 11032 01 X : UPDATING MATHEMATICAL SKILLS ASSESSMENT : 21;83 : DIAGNOSTIC QUIZ STUDF"T : D87111111 : MR PETER 1111111 MR PETER IIMEM P 0 BOXMOM ROMA QLD 4455 STUDY BOOK: Page 55 - Equation 7 should read a=10 Page 49 - Answer to Question 1 DOES NOT include "-1, -2, -3, etc. and zero" as natural numbers only contain the counting numbers: 1, 2,3, etc. Page 97 - aeference page in text for 'Factorial n' is page 93 of TEXT BOOK: Page 32 - Exercise 2.4, No. 1A: Interval should read: 1-3,5) No. 1B: interval should read: (-1,11 Page 39 - Point Q co-ordinates should read (-1,1/2) Page 46 - Formula for "d" following figure 3.13 should finish with "(y subscript 2 minus y subscript 1)squared" Page 49 - Exercise 3.6, No. 3: should read: y=-0.15x Page 67 - 2nd example at top of page should read: E=12,4,6...1 Page 121- Top of page should read "Exercises 6.6" Page 148- Figure 7.21 is incorrectly labelled. Function should read "inverse function = 1 /x," not "x+2" Page 171- In Figure 8.6, the functions all cross the y-axis at y=1, the graph has the position of 1 on the y-axis marked incorrectly. Page 184- Figure 9.6. Hypotenuse should be "square root of 2". Page 186- Figure 9.11. Point on x-axis should be labelled " -1 ". Page 326- lb Answer should be preraced with "1/3". "END Using Computers Intelligently 291 - a statement as to correctness/validity of the response - feedback is provided where appropriate QUESTION RESPONSE RESULT 1 +12 CORRECT 2 -7 CORRECT 3 -1 CORRECT 4 +4 CORRECT 5 NOT ATTEMPTED 6 +125 CORRECT 7 NOT ATTEMPTED 8 NOT ATTEMPTED 9 NOT ATTEMPTED 10 NOT ATTEMPTED 11 +1/2 CORRECT 12 +29 INCORRECT 13 +7 CORRECT 14 410 CORRECT 15 NOT ATTEMPTED 16 NOT ATTEMPTED 17 +1 CORRECT Question 5 The correct answer is -1/15 because (-1/5) x (-4/3) - 1/3 = (-1/5 x -4/3) - 1/3 = 4/15 - 1/3 . 4/15 - 5/15 = -1/15 Question 7 ., a-1/a-2 is ermal to a2/a1 = vaa/a = a2 Therefore t e final power of "a" = +2. Question 8 a2b-2/a-5bc = a2.as/b.b2c - aa/bic = a5b-3c-1 Therefore the final power of "a" = +5. Question 9 a2b-2,'a-abc = a2.aa/b.b2c = a°/b2c = aab-3c-1 Therefore the final power of "b" = -3. Question 10 a2h-2/a-2bc = a2.a2/b.b2c = a /b2c = a6b-3c-1 Therefore the final power of "c" - -1. Question 12 Remembering the correct order of operations, the solution is (16 + 22 + 32 x 2) +2+ 4' = 301 292 Using Computers Intelligently (4 + 18) + 2 + 11 = 22 .1 2 16 = 12+ 16 = +27 Question 15 The correct solution is: yz/2x = (4/32/5)/(21/3) = (8/15)1(2/3) 8/15 :1/2= 24/30 = +4/5 Question 16 The correct solution is: 4x +7 = 5x +4 4x - 5x = 4 - 7 -x = -3 x = +3 Number of Tiestions : 17 Number answered correctly : 9 Number answered incorrectly : 1 Number with answer missing : 7 Number with invalid answers : 0 Score ( % ) : 52.9 You have scored between 50% and 70% on the diagnostic quiz which suggests that while a little rusty, you have a reasonable understanding of the fundamentals of mathematics. Please feel free to contact D.E.C.E. should you have any outstanding queries on the diagnostic quiz. If you have no qu.!.ries you should now begin studying Module 1. 302 Using Computers Intelligently 293 APPENDIX ASSIGNO6.PUB.SR 04/AUG/87 19:02 PAGE 49 DARLING DOWNS INSTITUTE OF ADVANCED EDUCATION UNIT ASSESSMENT REPORT 86 11032UPDATING MATHEMATICAL SKILLS 01 X RECD RETD RESULT D86410XXX MR MAURICE X (CONED: 09MAY88) AS 01 ASSIGNMENT 1A S 1AJUL86 23JUL86 50/ AS 02 ASSIGNMENT 2A 13SEP86 17SEP86 79/ AS 03 ASSIGNMENT 3A S 29SEP86 03OCT86 70/ AS 04 ASSIGNMENT 4A 160CT86 22OCT86 66/ AS 05 ASSIGNMENT 5A S 27NOV86 05DEC86 30/ AS 08 ASSIGNMENT dA S 11DEC86 15DEC86 30/ AS 07 DIAGNOSTIC QUIZ S 09MAY86 19MAY86 58/ AS 08 ASSIGNMENT IB S *O'DUE* AS 09 ASSIGNMENT 1C *O'DUE* AS 10 ASSIGNMENT 2B *O'DUE* AS 11 ASSIGNMENT 2C *O'DUE* AS 12 ASSIGNMENT 3B *O'DUE* AS 13 ASSIGNMENT 3C *O'DUE* AS 14 ASSIGNMENT 4B *O'DUE* AS 15 ASSIGNMENT 4C *O'DUE* AS 18 ASSIGNMENT 5B S 21JAN87 23JAN87 100/ AS 17 ASSIGNMENT 5C *O'DUE* AS 18 ASSIGNMENT 6B S 21JAN87 23JAN87 90/ AS 19 ASSIGNMENT 6C *O'DUE* D80601XXX MR ALLAN J Y (CONED: 15APR86) AS 01 ASSIGNMENT IA S 29APR86 14MAY86 90/ AS 02ASSIGNMENT 2A S 23MAY86 28MAY86 93/ AS 03 ASSIGNMENT 3A S *O'DUE* AS 04 ASSIGNMENT 4A *O'DUE* AS 05 ASSIGNMENT 5A *O'DUE* AS 06 ASSIGNMENT 6A *O'DUE* AS 07 DIAGNOSTIC QUIZ S 22APR86 02MAY86 52/ AS 08 ASSIGNMENT IB *O'DUE* AS 09ASSIGNMENT 1C *O'DUE* AS 10 ASSIGNMENT 2B *O'DUE* AS 11 ASSIGNMENT 2C *O'DUE* AS 12 ASSIGNMENT 3B *O'DUE* AS 13 ASSIGNMENT 3C *O'DUE* AS 14 ASSIGNMENT 4B *O'DUE* AS 15 ASSIGNMENT 4C *O'DUE* AS 16 ASSIGNMENT 5B *O'DUE* AS 17 ASSIGNMENT 5C *O'DUE* AS 18 ASSIGNMENT 6B *O'DUE* AS 19 ASSIGNMENT 6C *O'DUE* ASSIGNMENT ENRL OUTS SUB RESUMXT %RESP RETD %RET ASSIGNMENT 1A 152 77 75 49.34 75 100 ASSIGNMENT 2A 152 102 50 32.89 50 100 ASSIGNMENT 9A 152 116 36 23.68 96 100 ASSIGNMENT 4A 152 134 18 11.84 18 100 ASSIGNMENT 5A 152 135 17 11.18 17 100 ASSIGNMENT 6A 162 135 17 11.18 17 100 DIAGNOSTIC QUIZ 15: 52 100 65.78 102 100 ASSIGNMENT IB 162 141 11 7.23 11 I:: ASSIGNMENT 1C 152 142 10 6.57 10 100 ASSIGNMENT 2B 152 147 6 3.28 5 100 ASSIGNMENT 2C 152 147 5 3.28 5 100 ASSIGNMENT 3B 152 149 3 1.97 3 100 ASSIGNMENT 3C 152 150 2 1.31 2 100 ASSIGNMENT 4B 152 150 2 1 31 2 100 ASSIGNMENT 4C 152 151 1 0.85 1 100 ASSIGNMENT 5B 152 150 2 1.31 2 100 ASSIGNMENT 5C 152 151 1 0.65 1 100 ASSIGNMENT 6B ;52 149 3 1.97 3 100 ASSIGNMENT 6C 152 151 1 0 0.65 1 100 MO. *DROPPED: 42 303 294 Using Computers Intelligently APPENDIX D ATTENTION Michael Crock, D.E.C.E. OPINIONNAIRE Course 11032 Module 1 Please return this to the Administrator, D.E.C.E., with Assignment 1. Tick the box which hest describes your reponse to the statements made and add any comments you wish to make. Further specific comments on how the module/course could be improved are welcomed and should be added overleaf or on attached sheets. .rongly agree agree disagree strongly disagree 1. I found the module interesting. Comments: 2 3 4 11.10) 2. The module was easy to follow. Ei 0 Comments: 2 3 4 11 111 3.The presentation olthe modulo including; - formative assessment items \ 1112) - directions to read 1.13) - Study Book assisted 1114) - statements of objectives ray understrnding 0 13' - telephone contact IIle, - pretest 41 17) - set text / 5 3 11.12) Commente: Z. The assessment item is a fair test of my attainment of objectives in this Module. 2 3 il.19; Comments: Please continue over 3C4 Using Computers Intelligently 295 trongly agree agree disagree strongly disagree 6. External studies facilities provided by DDIAE were El E C adequate for studying this module. 2 3 4 (1:20) Comments: IL hour on this module. 1 spent approximately r 1 Comments: 7. I would improve the module by Thank you Michael Crock 305 ..111111r. Are computers the write stuff?Computers and writing instruction Marsha Durham, School of Humanities and AppliedSocial Sciences, Nepean College of Advanced Education A variety of computerized writing aids exist, eg for isolated phenomenon for an English topic exploration, spelling and grammar checking, style analysis area word processing. Educators may teacher. It is however, a commonplace consider the aids unproblematic and even desirable experience for a computerteacher! in composition instruction, espedally as students' (Leonardi and McDonald,1987, p. 45) enthusiasm for them seems to ensure their success. The effect of these 'writer's helpers' on students' The computer's abilities for writing assis- writing has not been thoroughly studied. Prelimi- tance are appealing, as it can provide lay- nary studies suggest that the computer may be of outs, previews, spelling checks, style limited value. Writing is both individual and recur- sive, and programs based on a superficial or pre- analysis and grammar instruction, as well scriptive view of writing cannot duplicate the com- as perfect presentation. Yet, as with any plex choices that writing involves. technology introduced for educational purposes, computerized aids for writing To effectively use computers in writing instruction, educators must be prepared to act as mediators be- should be evaluated on their ability to tween students and technology, by (1) helping stu- support pedagogical objectives. Many dents move from the limitations of these programs to problems surround .he technology, with expand their own choices in writing, and (2) working some programs presenting inappropriate with technical experts to create programs that incor- porate sound educational objectives. procedures and criteria as part of their writing instruction. In this paper, I will focus on the impact that writing programs Two major problems have weakened muchhave on student writers, and comment on of the research about the effect of comput-their effect in changing students' writing ers on students' writing: conclusions basedprocesses and their interactions within the cn insufficient or biased data; and reliance writing class. on anecdotal evidence as 'proof. In addi- tion, the overly enthusiastic tone of many Views of the writing process articles about computers and writing pro- motes uncritical acceptance: Although often taught as separate stages, the major cognitive processes in writing You've written several late passes... planning, translating, reviewing and for students too involved in their work monitoring can be implemented at any to leave... (Y)ou lock your door point in one's writing (Sommers, 1985, p. against a roomful of students working 5).Writing is recursive, with writers re- after school. You arrive home only to peating their processes from first idea receive phone calls from students too through to finished product. For example, interested in their assignment to wait although editing is often thought of as a until class... Experiencing such ram- final action, it also happens before and pant student enthusiasm is a rare and during one's writing. 3 6 Using Computers Intelligently 297 Writing is situationally based, dependentabout, what they know about the subject, on such variables as one's interest and and what they must research: knowledge of a topic, perception of audi- ence, past successes in writing, subject The first use of language that a student limitations and time constraints. As well, of composition has to learn...is in the writers have different, individual writing generation of chaos If we don't begin styles.'Mozartians', for example, make there, we falsify the composing proc- extensive outlines, mentally or on paper, ess because composition requires before writing a comprehensive first-final choosing all along the way, and you draft.Another type, the 'Beethovians', can't choose if there are no perceived favor writing a quick, rough draft, then alternatives:chaos is the source of revising extensively, perhaps arriving at a alternatives. If we are unwilling to risk dear pattern only late in the process. chaos, we won't have provided our students with the opportunity to dis- Recent composition research (Berthoff, cover that ambiguities are, as 1.A. 1981; Stock, 1983; Beach & Bridwell, 1984; Richards has said,'the hinges of Flower, 1985; Knoblauch & Brannon, 1984; thought'. (Berthoff,1981: 75) Tate & Corbett, 1981, Hillocks, Jr, 1986) supports the view that writing is more than Too often students curtail thinking too just exhibiting one's learning. It is a way of early, neglecting to generate choices, eg learning, for our meaning evolves as we ideas, questions, inconsistencies, problems make choices in our writing.Unfortu- and qualifications.Researchers believe nately, choices diminish when studentsthat one major difference between good wrIce, as they manufacture 'Engfish' (see and poor writers is their ability to possess Reference Note 1), ie lifeless writing whichand select different heuristics: "Good writ- has lost its ability to 'speak' to readers.ers not only have a large repertory of 'Engfish' results when students are arepowerful strategies, but they have suffi- unable or uninterested in ensuring thatdent self-awareness of their own process to their meaning is reciprocated by the reader. draw on these alternative techniques as they need them" (Flower, 1985, p. 37). Can computers play a constructive role in teaching students to write for meaning?Computers can help in teaching heuristics. Most computer writing programs offerPrograms have been written for the most only limited assistance in teaching thepopular types, eg freewriting, brainstorm- complex choices that good writing in-ing, tagmemics, treeing, Aristotle's topoi, volves. While useful for specific purposes,Burke's dramatistic pentad, and discovery these programs can be damaging if they arequestioning, as well as for specialized writ- introduced as comprehensive tools, withing topics, eg narration, argumentation, students expected to curtail their writing to poetics and fictional characterization. use the program. Computerized heuristics can help students Computerized heuristics who are unsure about their ideas, and who might not speak up in a class dicussion Academia's emphasis on evaluation, style,about topics. The dialogue format of some grammar and punctuation may cause stu-heuristic programs encourages students to dents to 'freeze' in their thinking, and ex-go beyond their commonsense under- plore only the safest ideas. Instead, stu- standing to consider different aspects of a dents need to be encouraged to experiment, topic, and amass more significant and in order to discover what they want to write thoughtful information. The program can 307 298 Using Computers Intelligently also point out where they lack sufficentmethod of having writers describe their knowledge. Students introduced to a wide composing habits, either as they write, or in repertory of invention strategies may learn retrospect. Studying students' modes of to be more flexible in choosing ways to writing is even more difficult because of the explore ideas according to a particularnegative influence of having topics as- writing task. Brainstorming, for example, signed, and the finished work evaluated. may work fora short crea five writing piece, Despite breakthroughs in research (see while a diffemnt heuristic may be moreReference Note 2), not enough is known suitable for a lengthy research report. about writing processes to adequately evaluate the impact of computer use. Using these programs can create problems. If only one heuristic is available, students The hardware and software is so often seen may have the mistaken view that it is rele-as beneficial, that it is difficult to consider vant for every writing task. Students maythat both can have a detrimental effect on become overly dependent on software, students' writing. For example, word proc- and wait passively for cueing. Some pro- essing offers 'clean copy' by immediately grams may be so complicated that students incorporating changes. Seeing their first cannot relate them to their assignment,draft perfectly displayed can give students Others may alienate students by using a a false sense of its importance, and thus lock -'step approach, or an inappropriateundermine the value they perceive in re- tone of response. And, on their own, heur-flecting on their work, and editing it. Not istics programs will not overcome, stu-being able to refer to their original and dents' writing apprehension or their pre-subsequent drafts may rob students of a occupation with 'regurgitating' thesense of achievement. Resurrecting dis- lecturer's ideas. By undermining students' carded information is impossible with interest in human feedback, programs maysome programs, yet this is often necessary isolate students from perceiving their writ- when the writing takes a new direction. ing as a meaningful, communicative act for others. Computer hardware can also inhibit student's writing. A mouse, for example, Educators can thoughtfully integrate a may still be more cumbersome than a pen - heuristics program into the writing class bydl, and therefore editing with it is avoided. pointing out its immediate and long-termThe computer screen, which allows only a advantages and disadvantages, or by com- limited view of one's work, makes it diffi- paring responses of the program with thosecult to assess overall movement. Com- of a real audience. They can ensure thatmands like 'scroll' and 'find', and features students do not rely on the program for all like split screens and clipboards,are help- their writing, but incorporate responses ful, but these still do not allow the simulta- from others, as well as think up their ownneous viewing of major sections that some questions in considering their topic. Thiswriters may need. The screen may prevent critical approach allows students to under-interaction, as students stayed 'glued' to stand the program's limitations, and thetheir work. And, it can inhibit trials and value cf knowing and using a variety ofother types of experimenting, because it methods in assessing a topic. creates a 'public' display of one's work. Even seeing one's work in an electronic Impact of word processing format can be alienating, as one student found: "It's awfully hard for me to look at the screen and want to edit anything. My Researchers interested in the writing proc- paper seems so far away, almost abstract" ess have usually relied on the subjective (Harris, 1985, p. 327). Using Computers Intelligently 299 Absence of definitive knowledge about Once students view the computer as a good computers' influence on writing has not'trialler', that can easily clean up false prevented some educators from assumingstarts, weak writing, and undeveloped that students who use word processing notsentences, they may move naturally to sig- only write more, but edit more signifi-nificant revision. cantly.Preliminary research, however, suggest that students do not fully utilize Text graders the editing functions in word processing. A recent study indicated that although ob-A full-time composition instructor in an served students believed that they revisedAmerican college writing program may more on the computer, they did not. In mark over 350 assignments in a 10 -week addition, changes made were mainly mi-term. Consequently, interest has been keen nor, eg "adding a few details, rearrangingin computerized text grading, which al- words within a sentence, substituting onelows educators to flag common mistakes word for another of essentially the sameand leave the program to provide detailed meaning, deleting a word or phrase, eh. explanations to students. (Harris, 1985, p. 328). What is the pedagogical worth of this soft- In surface revision, the computer is only a ware?Educators may deride that the 'glorified typewriter', used to transcribecomputerized grader provickes extensive handwritten work, correct surface errors,information while freeing them to concen- and make attractive pages. Significant revi- trate on global comments. An instructor sion, on the other hand, includes making can mark 'CS' on an assignment, for ex- "semantic and rhetorical changes that af- ample, and leave the program to define fect the content and organization of a piece'comma splice' to the student, and suggest of discourse" (Harris, 1985, p.323). Writers exercises related to this error. revise significantly when they concern themselves with making their proseThe drawback to the program's ability to reader-based :they actively test theirpinpoint errors is that it may encourage meanings and gauge their readers' com- lecturers to flag every mistake, in content, pr'hension.Students who are writer -style, grammar, spelling and punctuation. base ' believe that it is sufficient if theirNot only do they then spend more time writing is clear to them, unaware thatmarking, but such intensive correction meanings are not so easy to establish: may be counter-productive. Students may see the lecturer's comments as little more The making of meaning is a dialectical than "mutilations of their work and intru- process determined by perspective sions into their ideas", and believe that the and context. Meanings change as we comments show that "they haven't ex- think about them; statement and pressed the approved opinions" ( Sudol, events, significance and interpreta- 1985: 333). tions can mean different things to dif- ferent people at different times. Mean- In addition, students often have little ings are not prebaked or set for all understanding that the lecturer's com- time; they are created, found, formed, ments are to aid them in their future work. and reformed. (Berthoff, 1981, p. 71) Consequently, composition researchers are becoming interested in the value of In creating meaning, the computer can be checking work in progress, rather than useful, as it allows experimentation: ideasfinished pieces.Instead of reading for can be developed, erased, or put on 'hold'. mistakes, the lecturer's interest must nec- 300 Using Computers Intelligently essarily shift to searching for intendedpriate, because they rely on simplistic, meaning, attempting "...the dialogue thatprescriptive statements or rules.In one has been formed in the writer's mind,program for younger children, for ex- transformed in his dialogue on paper"ample, the check consists of asking the (Berthoff, 1981:39). Surface errors are ig-writer the same general question for each nored until the student learns how hisorsentence:"Are you using punctuation cor- her intended meaning has been under-rectly?", "Is this a good stood. sentence?"(Smithson, 1987: 88). Students needing the check program probably One objection to this idea is the amount ofwould not be able to answer these ques- time taken to read several drafts. Propo-tions. nents suggest, however, that intervening earlier in the writing process saves oneStyle checkers pit one's written work from reading lifeless, inaccurate essays.against a prescribed readability formula. Students can be taught how to give valu-Researchers have already questioned the able comments on others' work, and theirworth of using formulae,originally created analysis can be followed by computerized to check the reading level of textbooks, as a checks for spelling and grammar. The finalstandard for evaluating other types of writ- product, accurate and edited, allows the ing.Readability formulae suggest that lecturer to read it without having to com-word length, sentence length, abstractions, bine the role of critic with that of proof-and other stylistic features constitute the reader and editor. work's level of comprehension, and there- fore its worth. These features are inherent Checkers features of writing, but it does not follow that obeying the program's suggestions Software exists that can check text for spell-will automatically produce good writing. ing, grammar, punctuation, word usage, and style. The expectation is that the infor-While readability formulae may be useful mation produced will help writers under-to professional writers, students may find stand their weak points in a work, andthat thi analysis of their writing is pre- revise. In advanced or specialized writingsented in an almost indigestible format. classes, eg technical and professional writ-They must be taught how to translate the ing, this information can be valuable, estab- evaluation into specific strategies for im- lishing standards of 'correct' writing. Forproving their writing, if the program is to general writers, it may be detrimentalbe useful. unless students understand the program's limitations. Future development of writing programs For example, spelling programs, by freeing poor spellers from having to concentrate on Students need direction in understanding mistakes, create new interest in writing. what may be wrong with their writing. But Most spelling checkers, however, havethe choices that they make are within the vocabulary lists too limited for tertiary context of each specific rhetorical situation level writing, and these would need to be for which they write. The program that can augmented for students' use. take account of all the variables in those situations has yet to be created. As grammar and punctuation is complex and inextricably related to context, check- In the future, valuable programs for writ- ers for these elements are usually inappro-ing, would be sophisticated packages in- 310 Using Computers Intelligently corporating all 'stages' of the writing proc- use out-moded prescriptive models, are ess. For example, students could concen- overly difficult to use, encourage a 'piece- trate o- shifting from a writer-based to ameal' approach to writing, or focus too reader-based view as they revise if they stringently on errors. could use a simple heuristic/word process- ing program that asked them to answer Before any computerized writing program questions about their topic, enter their first is introduced, the technology should be draft, then answer further questions about evaluated to see what benefits it offers to how they had developed their ideas. Such students' writing and to ensure that inap- comprehensive writing programs havepropriate learning experiences are not already being developed, but are still lim-being designed to fit the software. Instead, ited for full classroom use. software that teaches students to expand their writing choices, and to understand l'Engfish" may stem from students opting and explore their methods, will allow stu- for the first, easy response to a topic.If dents to see technology as a help, rather students could select from a variety ofthan another hurdle in what may too often heuristics on one disk, they might be en-be seen as the 'mystery' of good writing. couraged to explore not only more possi- bilities within the topic, but also learn what Educators can also assess their ability to heuristic style most suits them. Anothermediate between the program's operating helpful program would be one that assistsmethods and students' educational needs. students in selecting their own topic, basedIntegrating programs with the course's on analyzing their knowledge, interestseducational objectives, includes consider- and abilities. ing how the technology will affect educa- tors' teaching methods, classroom interac- Graders and checkers are reasonable inclu- tions, and other commitments. More writ- sions in writing instruction, if they include ing educators are writing their own com- further information and exercises for stu-puter programs, basing them on the grow- dents who wish to work on specific writing ing body of knowledge and research about problems.If these 'help' items incorpo- the writing process. As computers become rated longer passages of text, not just iso-more sophisticated, perhaps writing spe- lated sentences, students would learn howcialists and technical experts will collabo- to spot problems in 'real-life' prose thatrate to develop advanced, comprehensive they read.Drill-practice programs inprograms that can accept numerous op- grammar, punctuation, and spelling havetions and choices, and thus allow students lost their popularity, but they may offer a to be truly supported by computers as they non-threatening way of providing encour-experiment in creating reciprocated mean- agement and revision of basic writing skills ings in their writing. to particular types of students, eg mature- age students enrolling after a gap in their Appendix A: Computer formal education. Programs for Writing Conclusion Prewriting Educators interested in computers for writ- 1. QUILL Planner ing should be aware of more than the tech-Apple nical aspects. The cognitive and psycho- Prompts by keywords, title, author, entry logical aspects of writing lead educators to numbers, topic question the worth of 'writers' helpers' that 302 Using Computers Intelligently 2. Narcissus & Echo authoring system Reads list of questions into an invention Revising and Text analysis program 1. ALEXIS 3. SEEN (Dutch program) Apple error description and correction at 5 Prompts for fictional character analysis levels 4. TOPOI 2. Catch DEC, Apple, IBM PCPrompts for pur- Apple poses regarding subject topic Grammar and style r.hecking, plus prompts for revising 5. Prewrite 3. CRITIQUE (formerly E2ISTLE ) Apple IBM 370 Prompts for fictional/personal prose, or Detects 14 classes of syntactic errors in expository prose, then suggests sentences, eg 'who' for 'whom', subject- freewriting verb disagreement, improper verb form. Directed to office correspondence 6. Interaction Bulletin board programs for posing 4. Grammatik questions to class or creating database of CP/M, IBM PC, TRS-80 questions. Prompts for errors in capitalization and punctuation, identifies sexist language, Comprehensive; Specific and trite or wordy phrases, and gives sentence and word length. 1. Writing Workshop Includes visual aids /spreadsheets/ Apple printouts possibly used by technical Grades 3-10 writing students for reports. includes word processor, Prewriting Has audible on-going spelling check. (brainstorming, branching, nutshelling). Postwriting (checks mechanics, spelling, 5. Homer and proofreads) Apple Isolates prepositions, 'to be' verbs and 2. Poetics imprecise words, and gives sentence IBM lengths and word types. Tutorials, exercises and authoring system for teaching prosody at tertiary level 6. Punctuation & Style CP/M, MS-DOS 3. The Poetry Processor: Orpheus A-B-C Checks punctuation errors, unpaired (not available as of Feb 1986) format commands, double words, passive voice, misused/overworked phrases 4. WARP Includes drafter, editor, outliner, printer, 7. Style utilities Includes Bias, which calculates ratio of female to male pronouns, plus flags (being developed) sexists words and phrases. 5. Writer's Helper 8. Unnamed programs including creation , analysis Prof. B Broughton & W Horn, Clarkson (readability), revision University 312 Using Computers Intelligently 303 program that flags libelous and slander- 2. Researchers are using computers to ous statements. record every keystroke that students make as they write. The resulting infor- 9. Styled, Styllist mation is then linked with a new type of IBM retrospective protocol analysis, in Checks long words, punctuation, 'struc- which students describe their writing ture' words, forms of 'to be', norminaliza- choices as they watch their changing tions. drafts on video. Details of this research is found in Hawisher, 1986. 10. Writer's Workbench UNDO Bibliography programs: Proofreading. stylistic check, checks for split infinitives, spelling and Beach, R. & Bridwell, L. (Eds.) (1984) New punctuation errors, overly long sentences, directions in composition research. New passives, word phrases; readability. York: The Guilford Press. Berthoff, A. (1981). The making of meaning: Readability Metaphors, models and maxims for writing teachers. Upper M ,clair, NJ: 1. Readability Analysis Boynton/Cook. Apple Bridwell, L.S., et al. (1984). The writing Runs readability tests: Spache, Dale- process and the writing machine: Cur- Chall, Fry, Raygor, Flesch, Gunning-Fog rent research on word processor rele- vant to the teaching of composition. In 2. Readability R. Beach & L. S. Bridwell (Eds.), New Apple directions in composition researchNew Spache, Dale-Chall, Fry Raygor, Flesch, York, Guilford Press, pp 381-398. Gunning-Fog, Smog Flower, L. (1985). Problem-solving strategies Can make on-screen changes through for writing. San Diego, CA: Harcourt, editor function Brace, Jovanovich. Harris, J. (1985). Student writers and word 3. CRES (Computer Readability Editing processing: A preliminary evaluation. System) College Composition and Communication, Developed by US Navy to a xess techni- 36(3), 323-330. cal and instructional material. Flags Hawisher, G.E. (1986)Studies in word uncommon words, misspellings, ii,ng processing. Computers & Composition, sentences, passive sentences, 'to be' 4(1), 6-31. constructions and gives readability score. Knoblauch, C.H. & Brannon, L. (1984) Specialized word lists. Rhetorical traditions and the teaching of writing. Upper Montclair, NJ: Drill and practice Boynton/Cook. Leonardi, E.B., & McDonald,J.L. (February eg Text Critique, RSVP, Camelot, Writer/ 1987). The micro in the English class- Grader/Reader; Report room: Shifting the emphasis from proc- essing words to processing ideas. Edu- Reference Notes cational Technology, 45-47. Smithson, I. (1987). The writing workshop. 1. The term was coined by the well-known Computers and Composition, 4(1), 78-94. American writing educator, Ken Mac-Sommers, E. (1985). Integrating composing rorie. and computing. In J.L. Collins & E.A. 3 1 3 304 Using Computers Intelligently Sommers (Eds.), Writing on-line: Using computers in the teaching of writing. Upper Montclair, NJ: Boynton/Cook. Stock, P., (Ed.) (1983)Fform:Essays on theory and practice in the teaching of writ- ing. Upper Montclair, NJ: Boynton/ Cook. Sudol, R. (1985). Applied word processing: Notes on authority, responsibility and revision in a workshop model. College Composition and Communication, 36(3), 331-335. Tate, G. & Corbett, E. (Eds) (1981)The writing teacher's sourcebook. New York: Oxford University Press. Marsha Durham lectures in COIPTI1U- nication, specializing in written com- munication and social interaction (in- terpersonal, group, organizational). She is interested in different computer uses relevant within the BA (Applied Communication Studies) course at Nepean CAE: CAI for written argu- mentation and revision strategies in writing, desktop publishing, com- puter conferencing and word process- ing. Presently, She is helping to de- velop a program to teach students how to locate and evaluate fallacies of rea- soning. 314 Assessment of student computer programming assignments Helen Hasan School of industrial and Administrative. Studies University of Wollongong 1=1=130111=1. problems as the teaching staff should be Considering the large numbers of university able to design and write an assessment computing students and the scarcity of teaching package themselves. Another bonus for staff, too many tutor-hours are tied up manually this type of system is that the students' marking student programming assignments. The possibility exists for much of the assessment work is already on the system in a com- to be automated since the student programs are puter file. already in a computer readable form. Many departments teaching computing do have some Indeed many departments teaching com- on -line systems whereby student programs are collected and some checking done. However in puter programming do have some on-line most cases the tutor is at least required to check system 'Thereby student programs are col- the source code for such aspects as structure, lected and some checking done, for ex- readability and adherence to specifications. This paper discusses the requirements for an ample Bailes 1983, Whale 1983, Hext & automated system that carries out much of the Winings 1969. However this is usually tutor's task. It describes the performance of such limited to a run of the program against test. a system used fen a course which teaches FOR- data. In most cases the tutor is still required TRAN to second year Diploma students at this unive..ity. to assess the source code for such aspects as structure, readability and adherence to specifications. The teaching of computer programmming at tertiary level continues to involve large The tutor's assessment of a computer pro- numbers of staff and students in a variety of gram is in some respects akin to the assess- departments. The assessment of student ment of an English essay with corre. ond- programs is freque-ttly carried out manu- ing complexities. In some aspects it would ally by tutorial staff. This process tends to be almost impossible to replace human be unreliable due to its subjectivity andjudgement with a computerized process, well as occupying tutor time that could for example in assessing the relevance of better be spent elsewhere. This paper dis- comments Although this is beyond the cusses a system which has automated this capabilities of the type of system used here, process for a course which teaches FOR-in general the prospect of objectivity and TRAN to second year Diploma students at consistency, together with the saving of this university. valuable tutor time should tip the balance in favour - ` .n automated system. With computer assessment of non-com- puting subjects there is often a problem ofThe System Design. combining the expertise in the subject matter with the ability to write the req, tired The first task in designing this system was computer package. With a computer pro-to decide what criteria should be used in gramming subject this should present noevaluating the students' work.It was 313 1 306 Using Computers Intelligently necessary to look at the way tutors wereFORTRAN statements and some features currently operating as well as how the as-within them. Given the student source sessment would reflect the overall objec-code, first the modular structure and then tives of the course.It should be kept inthe code structure wit! n each module is mind that the assignments were given toidentified. Several check.., are then made. the students for both teaching and assess-Indentation and the use of comments are ment purposes. used as a measure of readability. All vari- ables must be declared, suitably assigned For the FORTRAN course involved the and used appropriately as arguments to following categorieb cnterged as the main subroutines, in common statements or as areas for assessment: arrays. Counts are made of all types of FORTRAN statements use, each type of a. Adherence to standard language syntaxdata structure, module etc. b. Use of appropriate data structures c. Adherence to accepted code structureFor each assignment the person supervis- (sequence, selection and iteration) ing the marking must decide on numerical d. Appropriate modular programming constraints for the parameters to be used in e. Readability and documentation the assessment: for example the number of f.Adherence to the assignment specifica-functions and subroutines, the number of 1, tions 2 etc dimensional arrays, the number of g. Validation of the program with testeach type of statement and so on. The data. marker enters these numbers as allowable ranges which are used in the final phase of The next step was to identify what empiri-the system when the mark is generated. cal measurements could be made in theThese ranges should be decided from the categories listed above.Comp.ling theassignment specifications and a look at a program would help test category (a) while few of the students solutions. For example running the program with test data couldan assignment, where a student is asked to be used for category (g). It was not difficultread names into an array from a file using a to implement procedures to first of all col- dummy end of file marker, should have at lect the students' source code, and then test least one while loop, a 1 dimensional array compile, link and run them. and so on.At helped if the assignment specifications were designed with the as- Categories (b) to (f) would on the othersessment method in mind. hand require some analysis of the the text of the source code.For this a substantialThe complete system. program was written to examine the stu- dent source code, producing both a reportThe complete system consists of 5 stages for the student and a mark which, com-eachof which is processed in batch fashion. bined with a mark for the compilation and run, gave an overall grade for the assign-The first stage is the collection process. As ment. An overriew of this program now the students work on stand alone PC's, the follows. tutor sets up a floppy disk running a pro- gram which copies the students' source Overview of source code assessment code onto the tutor's disk. program. The second and third stages involve batch compilation and run processes with rec- Without making a full semantic analysis it ords kept for each student on the success or was necessary for the program to recognise otherwise at each stage. 316 Using Computers Intelligently 307 The fourth stage is a run of the source code 3. On the preferred method of assessment assessment program discussed above. 50% said they preferred the computer generated assessment, 25% said they In the final stage, the output files from the preferred a manual assessment by the last three stages are merged to produce a tutor and 25% had no preference either report and mark for each student. The way. students are issued with this computer generated report which includes a sum- b. Assignment 5 was marked both manu- mary of their modular and code structure, ally and by the computerized oystem. their use of variables, any marks lost for The resultant marks out of 10 are listed bad structure, bad indentation, lack of in Appendix B. comments, undeclared variables and so on. An example of such a report is shown inA comparison of the two sets of marks Appendix A. show the following: Evaluation of the System Performance. Average manual mark = 7.75 Average computer mark = 7.95 During its development in 1986, the systemAverage difference (between marks for was used to assess 5 programming assign- each student) = 0.5 (6.5% of marks) ments given to 30 to 40 students through-75% of each student's marks were within out the course. The system was evaluated 0.5 in the following three areas: 90% were within 1.0 a. The students involved were surveyed The o ly significant discrepancy in the two for their opinions of the system. marks for any given student was for the b. A comparison was made on one assign- very weak students and to a lesser extent ment of manual marking versus the for the best 2 or 3 students. For the weaker computer generated mark. students the computer generated mark was c. A comparison was made of one featureconsistently more generous whereas for of the students' code (indentafion) asthe +.,T. students it was slightly lower. For the system was used. the main bulk of students, the agreement between the two marks was within the 1 The results of these evaluations are as fol- mark resolution of the marking scheme. It lows: may be significant that the automatic mark- ing scheme tends to be somewhat more a. The students were asked 3 questions atconservative than the human marker. the end of the session in which the sub- ject was taught. c.A comparison was made on the occur- rence of indentation errors in some of 1. On the preferred method of submitting the latter assignments compared to the programming assignments 70% of stu- first, before which the students had had dents said they preferred the new sys- no feedback from the system. In the first tem of collecting source code to the old assignment the students' programs had method of getting printout of source 4.7% of total lines of code with bad code and sample runs. indentation.This dropped down to 1.5% for assignment 3, the largest of 2. On the helpfulness of the computer their assignments, and remained under generated report 10% said it was very 2.0% for the remaining assignments. helpful, 90% fairly helpful and 0% said While there is no comparison with it was not very helpful. 3 308 Using Computers Intelligently comparable improvement that may Performance in an Innovative Medical have occurred had students' work been Curriculum Indicators of Performance. marked manually, it seems likely that SRHE Proceedings. students tend to take more notice of Hasan, H.M. (1986) Automatic Assessment computer generated diagnostics than of Student Programming Assignments they do of the tutor's red pen. for the Course AICA 201 User and Sys- tem Documentation. Wollongong Uni- Conclusion. versity. Hall, H.M. and Kidman, B.P. (1975). Em- On the whole the experience with the sys- pirical Analysis of BASIC and FOR- tem in its first year of operation appeared to TRAN Programs. In 0. Lecarme and R. be beneficial both for the tutor and the Lewis (Eds).Computers in Education.. students and it was successfully used forHext, J.B. and Winings, J.W. (1969). An the same course again this year (1987). Automatic Grading Scheme for Simple Programming Exercises, CACM, 12(5). The system was developed spozifically forHille, R.F. and Higginbottom, T.F., (1983). a course of second year students studying A System for Visible Execution of Pas- FORTRAN. The number of students was cal Programs", Australian Computer therefore not large and the students were Journal ,15(2). reasonably competent. The advantages ofMicrosoft(1984).MS-FORTRAN User the system in saving of tutors' time and Guide and Reference Manual. uniformity of assessment would be mostSperry (1984). Guide to DOS 2.11. valuable in courses involving large num- Su, Y.W. and Emam, A.E. (1975). Teaching bers of students and many tutors. A system Software Systems on a Microcomputer is at present being develot-,?d for a first year A CAI Approach. In O. Lecarme and R. COBOL course involving 350 :students and Lewis (Eds).Computers in Education.. 12 tutors. Whale, G. (1983). The Give Report, Kensing- ton: School of Electrical Engineering Both the FORTAN and COBOL courses are and Computer Science, UNSW. undertaken mainly by commerce students Whitlock, L.R. (1976). Interactive Test Con- and so the design of the courses and hence struction and Administration in the Gen- the assessment criteria are in accord with erative Exam System. Doctoral Thesis this.It is doubtful therefore whether the University of Illinois. systems would translate directly into courses in other departments, say for ex- ample Computing Science. However theAppendix A: Sample Report to general principle and design could readilyStudents. be applied to a wide variety of courses and from the experienced already gained I AICA 201 PROGSAleaNG EXERCISE 4 would suggest that it is an option well Student number 11 GILLON worth considering. Compilation in 57.01 secs with 0 erroris). Bibliography Run in 30.48 sec. >> module ASS4 Bailes, C.S. (1983) A Tutorial Introduction to ++ module header ++ 3D variable used. the Submit System, Wollongong: De- structure: partment of Computing Science. Wol- READ -CALL SUB -CALL SUB longong University. variables used: Clarke, R.M. et al (1980). Undergraduate 318 Using Computers Intelligently 309 RESUITS (4, 4, 2) : type: integer, IX: type: integer, NAME Manual Computer Differ - >>> module GAME ence structure: Mark /10 Mark /10 DO -DO -IF -DO - ENDDO JARMAN 3.0 5.2 +2.2 ENDDO - ENDDO JEBARA 5.0 62 +1.8 variables used: ALVAREZ 6.0 6.5 +0.5 IX: type: integer, argument. BOWEN 6.0 6.8 +0.8 RESULTS (4, 4, 2) : type: integer, argu- HEILER 6.0 6.0 0.0 ment. HADIPRODJO 6.5 8.2 +1.7 I.type: integer, PANIZZUTTI 6.5 7.5 +1.0 J: type: integer, GURIEFF 0 7.2 +0.2 CABIN 7.0 6.5 -0.5 K: type: integer, LEE 7.0 8.0 +1.0 >>> module RAND HAVAN 7.5 7.5 0.0 - value not assigned to function. MARIC 8.0 7.7 -03 structure: ZIPPARO 8.0 8.2 +0.2 variables used: WACHNIK 8.0 8.5 + as IX: type: integer, argument. VILLARROEL 8.0 8.0 0.0 BISIUTWIS 8.0 7.7 >>> module TALLY -03 PAINTER 8.0 8.2 +0.2 structure: JASIC 8.5 8.5 0.0 DO - ENDDO - DO -DO -IF - IF - CHHOR 13.5 8.0 -0.5 ENDDO -ENDDO -DO -ENDDO - DO - AYDOGAN 8.5 8.2 -03 IF ENDDO - MURRAY 8.5 8.5 0.0 variables used: CHAND 9.0 8.3 -0.7 RESULTS (4, 4, 2) : type: integer, argu- HIDAYAT 9.5 9.7 +0.2 SJAMSUDIN 9.5 9.0 -0.5 ment. TRINH 9.5 9.5 0.0 TEAM (4) : type: integer, IVERS 9.5 93 -0.2 A(4): type: integer, value assigned, DARMOKO 10.0 9.7 - 03 X: type: integer, DONELLY 10.0 9.5 -0.5 Z: type: integer, I: type: integer, Readability and documentation J: type: integer, K: type: integer, TOP: type: integer, value assigned, My current position is as a Lecturer in NUM: type: integer, not declared the Information Systems section of the You have ') indentation error (s) . above school as part of the Faculty of Counts: READ 1, WRITE 8, DO 8, IF 4, WHILE 0OPEN 2 CALL 2, COMON 0 Commerce. My background however Lines of code: 76, comments 32 is in Physics with a BSc and MSc. I Modular structure: recently completed a post-graduate program: ASS4 ->subroutine: diploma in Computing Science. My GAME teaching interests at present are TALLY mainly in computer programming ->real fun,....ion: RAND and my personal research interests *MARKS: evolve around the better use of com- Design: 2.00 Program:4.50Re- puters in assessment of student com- sults: 3.00 TOTAL: 9.5 puter programs. I am also part of a research group concerned with the Appendix B: Comparison of effective use of computer technology Manual and Computer to applied computer systems training Generated Marks for at the tertiary level. Assignment 5 319 Easing the load of laboratory assessment Phi II Higgins Applied Science Gipps land Institute of Advanced Education :=110111" At the Gipps land Institute of Advanced Educa- call the computer into action.. Each student tion (GIAE) the drudgery of repetitive marking gets a lab. I.D. password and finds a termi- of laboratory reports has been eased by the intro- duction of computer assessed laboratory activ- nal to the mainframe or a PC somewhere.. ity. This has been achieved by restructuring the laboratory activity and literature to be compat- After a successful log-on and menu selec- ible with the mainframe computer. The entire tion for the required experiment he/she student activity is monitored, assessed and re- corded without the lecturer/ demonstrator sees confirmation that the session is valid handling one sheet of paper. and then: This is the prelaboratory section of The lectures are fairly straight forward Lab. X. Do some reading, precis a bit here and there, copy a couple of diagrams onto You should already have worked the OHP's, check last years notes, work practice problems on this topic in through the examples to make sure they your laboratory booklet and checked give realistic answers, check how you fin- with your lecturer if you had any dif- ished the last lecture and how and where ficulty. you want to start the next and that is a couple of hours work for a one hour lec- You will now be given three ques- ture.. tions each of which you must work through and input an answer to be- Next, the laboratory work Students al- fore the next is provided. ready have a set of lab. notes in a booklet, so just a final check on the equipment and a They are chosen at random from a few notes on instructions and announce- large list so your friends answers are ments and that's it 15 minutes easy!! almost certain to be useless.. If you fail this exercise you will be Now, mark the labs! That's 120 lab. reports allowed only one more attempt at 5 minutes or so each 600 or more long, but NOT WITHIN 30 minutes of the dreary, mind-sapping minutes! 10 hours of failed one. If you fail the second at- largely unproductive work! Why can't we tempt you will be directed to see your afford a tutor/ demonstrator/ external lecturer. marker? Anything to ease this slogging You will NOT be allowed to proceed part of the job is worth a try!. with laboratory activity until you: (a) pass the prelab exercise OR (b) have Well here's one attempt: Divide the labora- r?mission from the lecturer. tory instructions into three sections (Pre- Press "enter" to display the first laboratory, Activity, Postlaboratory) and question.. 39+0 Using Computers Intelligently 31/ The three (say) questions are each chosen answers are redirected and the re-presenta- to illustrate a different concept involved in tion of the problem is an option. the exercise. For each question there are several alternative problems illustrating Example Question 3. the same concept and it is one of these problems that is selected at random. What did Rumplestiltskin offer the captive Queen as a means )f free- Example Question 1. (not a typical Physics dom? problem) Type your complete answer in less How many dwarves did Snow White than 10 words at the cursor position keep house to? and press "enter" for your next etc. (Enter your answer at the cursor posi- The correct answer "Three guesses at his tion and press 3 2 312 Using Computers Intelligently straight line results. The gradient of this is apply to almost anything and of course tested against the "true" values and unitsthere is no testing or feedback on report are checked. writing skills. However these disadvan- tages are more than offset by the huge At present the program does not do asaving in time that is achieved and the straight line fit to the students input points feedback on report writing is given on the but just checks the gradient and interceptslaboratory experiments which are not he/ she inputs with the acceptable refer-computer graded. ence range. Further calculations which have been Phill Higgins has been involved in made from gradients, intercepts and other Tertiary Education for almost 20 years parameters etc determined by the student and has lectured in most areas of Phys- are checked using the appropriate algo- ics, some of Chemistry and Applied rithm and data input by the student so that Mechanics and a few of Mathematics an initial mistake is not multiply penalized. and Biology. His main teaching and research interests are in may analysis, The post laboratory work is a single at- Optics and Fluid Mechanics. He has tempt system with usually 2 or 3 questions been an active member on the Commit- provided for the student to solve. Again tee of the Gippsland Branch of the these are chosen from a file of similar ques- Australian Institute of Energy for sev- tions designed to test the understanding of eral years and has recently become the work that has been done in the lab. involved in Nurse Education from which his main, incentive for CAL has The final output to the student is his com- developed. He is a lecturer in Applied puter assessment overall. A detailed as- Science at the GIAE sessment is provided for the lecturer show- ing all stages of the process.. Security is provided at a number of levels: a. Student and account ID. This enables students to avoid others "practising" at their expense. b. Code word for viewing of assessment lists - each viewing of the assessment list is also dated. c. Code word for modification to assess- ment list outside a student session. d. CPU connect times are logged against password and ID - this attempts to iden- tify the "hackers" trying to crack the system... The method has various degrees of applica- bility to different laboratory activities but the prelab and lab. sections can be made to 322 Athena, Andrew and Stanford a look at im- plementation and evaluation in three large projects Geoff Isaacs Tertiary Education Institute University of Queensland Massachusetts Institute of Technology and puter assisted learning projects within Carnegie Mellon University have both em- universities and it seemed to me that there barked on enormous projects designed to im- plement workstation environments for use in might be some profit in looking at how academic endeavours. Project Athena at MIT is some of them attack the problems of the a network designed to service teaching and implementation, support and, especially, learning only, while the Andrew network at CMU is to service both teaching and research. the evaluation of CAL developments. I Stanford University's computer-intensive envi- therefore decided to devote part of my ronment for teaching and learning owes no Special Studies Program in the second half allegiance to either the Athena or the Andrew of 1987 to exploring these very issues in the model. Here we summarise the environments at these three universities and place them in USA. At the time of writing that explora- context. tion lies in the future; however I have now accumulated a considerable amount of in- It is intended that this paper be the background formation about the places I will be visiting to a presentation at CALITE 87, which will take place after the author has visited these three and that information will be summarised establishments in late September 1987. The here by way of background to the paper emphasis of the presentation probably will be presenting orally at CALITE 87. on the evaluation and staff development activi- ties which are being carried out in these three The universities on which I will concen- universities in support of their CAL systems. trate are Massachusetts Institute of Tech- nology in Boston, Carnegie Mellon Univer- When one looks at the implementation of sity in Pittsburgh and Stanford University computer assisted learning in Australian near San Francisco. universities - or even, by and large, in any Why these universities? MIT and CMU are one university - one generally finds a both hosts to multi-million dollar, substan- hodgepodge of efforts of varying general- tially externally funded projects each in- ity, sophistication, successfulness and lon- volving the development of a campus- gevity. The methods and tools used are wide network system designed principally often totally idiosyncratic even betweento support teaching and students. MIT is staff members within the same departmenthost to Project Athena and CMU is host to and suppurt from a central authorityProject Andrew. Both projects are heavily within a university is the exception, not thesubsidised by computer manufacturers rule. A similar tale can be told of the evalu- who seem to see the universities as cheap ation of CAL materials and CAL-oriented resew and development labour and as courses. possible test beds for new workstations and network developments. The projects There are overseas - especially in the are, to a limited extent, in competition, and United States of America - numerous large, both are currently making available to well funded, centrally supported com- educational institutions either free or at a 323 314 Using Computers Intelligently nominal charge software developed inthis plan feasible by the end of 1986, but in their networking projects. Stanford, on the the event, the fall has not been as sharp or as other hand, represents the non-aligned, but great as expected. nonetheless computer-intensive universi- ties. It is heavily involved in CAL develop-IBM, in late 1982, got the project off the ments and has a large central support sys-ground by contracting to spend $US 35 tem for CAL developments, but it hasmillion to create CMU's Information Tech- elected not to implement a network such asnology Center (ITC). ITC's purpose is to the Athena network or Andrew (yet?). develop the software needed to make fea- sible CMU's plans for integrating comput- Carnegie Mellon University. ing into its educational activities. Even CMU's own documents speak ofIt is hoped that the large concentration of "Carnegie-Mellon's much ballyhooed ef-computing resources becol ning available fort to computerize itself as no institutionto staff and students at CU J will stimulate has ever done" (Maguire,1986). The pat-the growth of a widely u Rful library of tern at CMU has been to install a campus -educational software. To help staff in writ- wide computer network (called Andrew) ing such software (by funding, by advising linking many advanced computer work- or by actually getting it written) is the role stations and high end personal computersof the university's Center for the Develop- into one coherent whole. The network and mentofEducationalComputing its workstations will distribute the load of(CDEC). Andrew educational computing, it is hoped, thus lessening the load on the university's largeThe name Andrew is sometimes used to time- sharing mainframe computers anddescribe all parts of the educational com- providing a more responsive computingputing developments at CMU, but most environment for use in the education offrequently Andrew refers to the software students and for research and scholarlybase on which the developments at CMU activity. are built; the best technical deception of this system will be found in the paper by The CMU plan was first formulated in 1982Morris and others (Morris et al, 1986). The and involved the use of advanced functionAndrew software environment consists of workstations having very high resolutionthree components: a high speed (4 to 10 graphics screens and the storage capacity Mbits per second) network, a file system to and computing power to enable the graph-support a very large number (thousands) ics facilities (among other things) to be usedof workstations on that network and "a effectively. These workstations, whichfriendly, graphics- oriented operating might be made by any vendor able to meetframework for users of workstations" the necessary specifications, were to be(Sherwood,1986) - that is, a user interface networked by a system having cabling tofor graphics workstations. every room on campus. Eventually there was to be a workstation for every studentThe operating system on the computers (note, however, that contrary to wide-which form the backbone of the network is spread belief, students at CMU generallycalled VICE ("Vast, Integrated Communi- have not been compelled to purchase acations Environment"). VICE emulates a computer), faculty member and member of time - shoring file system; that is, there is one the administration on campus - a total ofsingle file system for all stations on the some 7000 workstations. It was felt that the network. It is integrated to such an extent falling cost of workstations would makethat users may "access files in a uniform 324 Using Computers Intelligently 315 manner regardless of the specific worksta-only about half of the eventual target tions at which they are logged in, or wherenumber of users. the files were created originally" (Morris et al, 1986). Changes to files are recorded Workstations under And,. ew must satisfy a centrally by VICE every few seconds re-few conditions before being usable in the gardless of where those files have beensystem. They must be fairly powerful (at used. Devices like printers, bulletin boards least 1 mips), need a high resolution graph- and electronic mail systems are built on top ics screen (in the region of one million pix- of VICE and, like the file system, are avail-els), need substantial random access mem- able quite transparently to workstationsory (in excess of one megabyte - preferably users who need not know their physical orthree or four), local mass storage (either a machine locations. Files in VICE being ac-substantial fixed disk in the region of 40 to tually used by a user are in almost all cases 100 megabytes or very fast access to a lo- copied to a cache on the workstation atcally networked disk which emulates a which the user is currently working.local one), an Ethernet or a Token Ring Changes to these files are made on the copy interface; and they must be able to run the in the cache and the file is written back to BSD 4.2 version of the Unix operating sys- the common VICE storage area when it is tem (this last need automatically implies all dosed. of the others except the graphics screen and the network interface). Andrew as cur- The VICE file system is considered to berently implemented does not allow the extremely secure because no user pro- current generation of personal computers grams run on the computers which hold (IBM PC and AT, Apple Macintosh, Mac and control the VICE file system and filePlus and Mac SE) to be connected. A new traffic carries only complete files fromversion - Andrew Plus - will also allow workstation to file server. In an ordinary limited use of conventional personal com- network there would be an enormousputers. Currently four types of Unix work- amount of paging of file fragments be- station are supported; they the Sun 2, Sun 3, tween workstations and servers; however IBM RT and Vaxstation II (see Arms, 1987). it was judged that the teaching and aca - Andrew ;Aas been designed so that it is demic environments could generally berelatively easy to add support for other adequately serviced by a filing systemmakes of workstation which satisfy the model rather than the usual networked prerequisites set out above. Support for database model which allowed continuousconventional personal computers, when access to the database for update andcompleted, will allow them to treat the modification by several users. Where such Andrew filing system as if it were a huge databases are needed by Andrew usershard disk attached to the personal com- they are supplied on an ad hoc basis byputer; anintermediate server (a Unix access to the university's pre-existing workstation) will sit between the personal mainframe and mini computers. computer and VICE and will hold the local copy of the VICE file being used by the Arms 0987) reports that in January 1987personal computer. In early 1987 the IBM there were 17 servers on VICE. Two ofPC Server was complete and operational those servers were IBM PC-la's and thewhile the Apple Macintosh Server was remaining fifteen were Sun 2's and Sun 3's. expected to be available before the end of Their total storage capacity was 17 Gigaby- the year (see Arms, 1987). tes. At the same time there were 3400 regis- tered users with about 450 Unix worksta-The user interface environment running tions connected to the VICE system. This isunder 4.2 BSD Unix on the workstations is 3 2 316 Using ,nputersIntelligently called VIRTUE ("Virtue Is ReachedCMU Tutor Through Unix and Emacs"). VIRTUE con- sists of a toolbox ("an object-oriented It is perfectly possible to use the facilities of multi-media environment for applications Unix and Andrew o develop custom made development", to quote Arms, 1987) which computer assisted learning materials. The handles the interface to the VICE file sys-Andrew window manager has functions to tem and has a window manager that allowsallow user programs to handle text and multiple processes to share the high resolu- graphics within windows and thereare tion bit mapped display, and packages forother, higher level programs to make this manipulation of text and graphics. Cur- process simpler (the Grits database facili- rently Andrew uses its own, homegrownties, a graphics layout organizer and the window manager, however it is plannedlike). However many potential clients of shortly to re-implement the AndrewuserAndrew have neither the patience, nor the interface as layers on top of the more skills, nor the time to acquire them in order widely supported X Windows systemto use the C language and these tools to developed at MIT. This is the first clearprogram their CAL applications under indication of what seems to be aconver-Andrew/Unix. CMU Tutor has been de- gence among some of the large universityveloped by Bruce Sherwood, Associate CAL projects, although communication Director of CDEC and Judith Sherwood of between them has always been consider- the Mellon College of Science, together able. with other staff of CDEC and ITC in order to address these problems (Sherwood, Computer assisted learning at CMU 1985; Sherwood and Sherwood, 1985; Sher- wood and Sherwood, 1986). Naturally computer assisted learning long predates Andrew at CMU. The CMU cata-CMU Tutor is described by its implemen- logue of educational computing projectstors as "an integrated programming envi- (Seid en, 1986) lists in excess of one hundred ronment for advanced workstations", but projects, some dating back to the 1970's, but it has now also been implemented on most from 1984 onward. The target systemsApple Macintosh and IBM Personal Com- for these projects are as follows: IBM Per- puters, although not all programs created sonal Computer (some for PC, some for AT, on advanced workstations can be moved to various display adapters)- 39; Andrew -33; these personal computers (the reverse does Apple Macintosh (various models)- 4;seem to be the case, however). It is an other (mainly HP9836's running Unix in adaptation of the PLATO project's Micro- engineering departments) - 34; total- 110. Tutor language (Sherwood and Sherwood, Where a project was originally developed1985) with extensions to supportwin- on another system, but has now beendows, mice, menus and other Andrew/ ported to Andrew it has been counted onlyworkstation features. It is an incrementally under Andrew. These projects range from compiled language (so lessons can be small BASIC language programs plotting tested very soon after they have been writ- graphs on a personal computer to largeten or revised) which supports all of the scale CAL authoring systems using all the special text forms used by the Andrew facilities of Andrew, such as CMU Tutor.editor in What-you-see-is-what-you-get The bulk of the projects are in engineering, format (bold, italics, font changes, center- mathematics and science areas, but there is ing etc) and includes extensive graphics a significant number in the social sciences capabilities including a graphics editor. and humanities. 326 Using Computers Intelligently 317 Massachusetts Institute of Technol- About half of MIT's $US 20 million contri- ogy bution to Athena is being spent on the sponsorship of curriculum development MIT's big thrust into campus-wide com-projects. A project may be supported by puter assisted learning started almost con-direct funding, the supply of computa- temporaneously with Carnegie Mellon's.tional resources, the supply of program- The MIT plan, announced in May 1983 wasming assistance, the paid release of faculty a five year plan "to explore new, innovative from teaching or other duties. One uses of computing in the MIT curriculum" hundred and eleven projects had been (Lerman, 1987). MIT's plan was realized byfunded as at autumn 1986, almost exactly the creation of Project Athena, a $US 70the same number as at CMU in a similar million project financed by IBM and Digitalperiod. Funded projects need not be on a Equipment Corporation to the extent oflarge scale - even small projects from indi- $US 50 million and MIT (by various fund vidual staff members can be accepted. raising initiatives) to the extent of $US 20 Note that there is a major difference in million. The support from DEC and IBM isphilosophy between Athena and Andrew. support in kind (equipment, software, The Andrew project is aimed at producing maintenance and staff support etc) not inan efficient networked workstation envi- cash. ronment which will facilitate the use of computers in education. It was felt that the Like the developers of CMU's Andrew,provision of such an environment would Athena's proponents would like softwareprovide the critical mass of support and developed in the projects to be runnable onresources needed to generate a significant advanced function workstations from anygrowth in such usage. The Athena project is vendor and have opted for Unix as thetaking exactly the opposite approach. They operating system of choice for those work- are funding and encouraging an increase in stations. the amount of educational computing ac- tivity in the hope that, at the end of the Project Athena project's five year life they will be better able to design or select an appropriate What exactly are the goals of Athena? Ler-environment for educational computing. man (1987) writes: Another major difference is that the An- drew project is aimed at the support of all Project Athena's mandate is to explorepersons on campus - students, academics, diverse uses of computing in support ofadministrators - in the course of their work, education and to build the base of knowl- while MIT's Athena is aimed mainly at the edge needed for a longer term strategic teaching of undergraduate students and, to decision about how computers fit into thea much lesser extent, at the teaching of MIT curriculum. We encourage a wide postgraduates; it is absolutely not aimed at range of educational applications to supportingstaff in their research and schol- understand hoW computing helps stu- arly activities. dents learn; we place computers in di- verse settings to explore how differentAthena's catalogue of projects (Stewart, physical arrangements affect the use and 1986) lists 60 in all.Of these projects twenty- value of computing; we experiment with five are described as being implemented on different types of software to learn about Athena Workstations (Vaxstation II or IBM its utility, reliability, the ease with which PC-RT's mainly), eleven are implemented it can be run in a very large computingIBM PC-AT's, mostly in the "Athena con- system and the costs of supporting it. figuration" (which includes an IBM Profes- 3 ,2 7 318 Using Computers Intelligently sional Graphics Adapter), six are on IBMmost recent of these (Cohen, 1986) details PC's, and eighteen on other systems (gen-the results of the 1985 and 1986 surveys. erally, where it is clear what other systemCohen's report shows that a very high -and then it is one of MIT's large time sharingincreasingly so - proportion of the under- systems). Where a project has been imple-graduate student body has used Athena (87 mented on both a PC and an Athena Work-percent), such use amounting to over three station it has been classified only in thehours per student per week on average. latter category. Interestingly, she found that usage by male and female students was equal in all classes This split of system types reflects accu-and years at MIT, "...in contrast to male- rately the development of Athena and itsdominated and senior dominated uses of own philosophy. Since the aim was to other computers at M.I.T.". At present stu- develop a critical mass of computer as-dents' usage of Athena is much more likely sisted learning activity on campus,to be personal (word processing and the Athena's directors did not wait for the ideallike) than required for courses. Needless to workstation environment before sponsor-say, most of the personal usage is course- ing projects. Rather, they elected whererelated, however. Athena has entrenched possible, to move with an intermediateitself in students' consciousness to such an system - a heavily optioned IBM PC-AT -extent that, according to Cohen, they "re- and, where even that was not possible,gard Athena as a service rather than asan opted to use whatever usable system wasexperiment". This means, of course, that available. Thus some of the projects in the they criticise it as a service rather than asan 1986 catalogue are conversion projects,experimental setup and this seems to be moving projects from Athena AT's or largereflected in their complaints about time sharing machines onto Athena work-crowded or inaccessible facilities, slow stations. This generally seems to involvesystems and system unreliability as well as mainly conversion of the windowing andlow quality printers. It would seem that at graphics facilities to Athena's standard. this stage Athena's massive injection of funds and equipment has not solved the Evaluation and Athena usual central computer service blues en- countered by so many tertiary education There seems to be more evaluative activityinstitutions. To be fair, the Athena to which of the educatiertal aspects of Project Athenastudents are reacting at this stage is largely that there is at Andrew. This is, in part,a one of terminals to a time sharing system; further consequence of the differing devel- it does not yet include the sophisticated opment plans for the two projects.workstations which presumably will Athena's emphasis on CAL software ratherlessen at least some of the problems re- than network software means that there isported above. likely to be more to lx. evaluated. Yet the difference is greater than that. Athena hasIn addition to the large scale evaluations actually created an Project Athena Study mentioned above there is also a significant Group which looks at the effects of Athenaamount of localised evaluation of materials on MIT. To date, however, the Group hasand courses within individual projects published no papers or reports available to sponsored by Athena. Some of this activity outside scholars. The Project also sponsorsis mentioned in Stewart's (1986) compen- directly a longitudinal survey of under-dium of projects, however few results of graduates since 1985 in order to look atthese evaluations have yet reached the patterns of student use of computers. The open literature. -', 6 Using Computers Intelligently 319 Stanford University IRIS supports CAL developments in sev- eral tangible ways. Among these is the Stanford is not at the centre of a large exter- Faculty Author Development program, nally funded project like those at CMU andunder which IRIS has helped faculty to MIT, but it is, nonetheless, a highly com-complete 36 software development proj- puter- intensive educational environment.ects ranging widely over the university. However, possibly because it has no "big Such help may come in the form of varying name" project, it is less well covered in thecomtinations of equipment, design or literature. All of the infonaation presented programming support. Beyond such direct here comes from an article in the Februaryto aspiring CAL teachers, Carter's unit is 1987 issue of Byte magazine (Osgood, trying to develop a range of self h 1p tools 1987). to decrease dependence on IRIS for the preparation of CAL materials per se by The centre of activity at Stanford seems to making it easier for staff to write their own be IRIS -Instruction and Research Informa-programs. The CAT (Courseware Author- tion Systems - and its attitude,reporteding Tools) project, as this latter develop- by its director, Michael Carter, seems more ment is called, is being implemented on akin to the Athena philosophy than theworkstations similar to those being tar- Andrew philosophy: geted by Athena and Andrew. Indeed, it is not the aim of IRIS necessarily to develop What we are trying to do is enhance aca- the required tools in house; they are quite demic achievement by applying computerprepared to borrow software from other technology. Our best bet is to try to focus itprojects, naturally including Andrew and a little here, nudge it a little there, lead aAthena. little bit over here. With so many really smart faculty members out there, I want to Conclusion get them enough devices so that they know exactly what they want to do, and then The real conclusions to this paper do not yet follow them, rather than to control the way exist; they will be given in the oral presen- they use computers. The trick really is to tion to which this written work is the remove the obstacles so that those peoplebackground. It is hoped that the presenta- caz lead the way. tion will concern staff development and evaluation at the three sites named above, Stanford has several "public" clusters of however, on the basis of our preliminary computers for student use (with studentsreading, we are not optimistic that they who are required to use them in theirwill be a great deal to report either from courses getting priority in access. Such Stanford or from CMU. However appear- dusters provide either personal computers ances from the other side of the world can only or a mix of personal computers and be very deceptive, so my forthcoming visit terminals connected to Stanford's time may produce some surprises in these areas. sharing system. There is also a number of classrooms at Stanford with PC's built in, All the above having been said, we can some including large screens so that areach one significant conclusion on the whole class can view the same screen im-basis of the earlier sections of this paper, age. One gets the impression that at Stan- and that is that there is a clear convergence ford Apple Macintosh computers are pre-among al: three of these sites in the soft- ferred over IBM personal computers orware and the hardware they are using. In compatibles. hardware all three seem to be moving away fr,,m conventional personal computers o 2 c, 320 Using Computers Intelligently and towards advanced networked work- one senses that this is proving a greater stations with very high resolution graphicsproblem than expected in the USA also. and network connections. This reflects in part the increasing complexity of the tasksReferences such workstations are being asked to carry out - in many cases supporting a multi- Arms, W.Y. (1987) Andrew-Plus. Carnegie- purpose working and learning environ- Mellon's Integrated Personal Computing ment, not just running a few computer Environment, Carnegie Mellon Univer- assisted drill and practice sessions- and in sity Academic Services, January. part the increasingly detailed graphics Cohen, K.C. (1986) Project Athena Student being required for tasks from those previ- Survey Findings 1985 - 1986, Project ously mentioned to supporting a Athena Impact Study Report No. 5, "friendly" operating environment for the Massachusetts Institute of Technology. computer illit_ .ate. The great unknown in Lerman, S.R. (1987) Questions and Answers this choice of equipment is Apple's Macin- About Project Athena, Project Athena, tosh II, which seems to be the only personal MIT, February. computer available in the marketplace atMaguire, A. (1986) Andrew is Alive and present (and it is only just available) ca- Well and Living in the Computer pableof challenging the workstations on all Network, Carnegie-Mellon Magazine, counts; its future role at these sites may be Spring, 15-19. dearer by the time I visit these sites inMorris,J.H., M. Satyanarayanan, September 1987. M.H.Conner, J.H. Howard, D.S.H. Rosenthal and F.D. Smith (1986) An- As far as software is concerned, the conver- drew: a Distributed Personal Comput- gence seems partly the inevitable conse- ing Environment, Communications of the quence of several sites independently pur- ACM, 29(3), 184-201. suing the same ultimate aims with substan-Osgood, D.(1987)The Difference in tially the same set of hardware. B' at it is also Higher Education, Byte, (2), 165-178. the result of a freer sharing of expertise andSeiden, P. (1986) Carnegie Mellon Univer- of the actual software itself than might have sity Educational Computing Projects, 267 been expected from the earliest publicity pages, Educational Software Library, about Andrew and Athena. This in turn Carnegie Mellon University, Pitts- results from all of the projects being strictly burgh, USA. non-proprietary in their target machinesSherwood, B.A. (1985) An Integrated Au- Andrew would not be looking to use thoring Environment, Proceedings of the MIT's X Windows environment if it were IBM Academic Information Systems Uni- targeted at one machine or owned by one versity AEP Conference, Alexandria, Vir- manufacturer; it seems to have been ginia, 29-35 (June). adopted by Andrew not only because it isSherwood, B.A. (1986)Workstations at superior to Andrew's own home grown Carnegie Mellon. In Proceedings of the product, but also because it is increasingly Fall Joint Computer Conference, Dallas, being supported by equipment manufac- November 2-6, pp 15-17. turers. As an Australian I can only hopeSherwood, B.A. and J.N Sherwood (1985) that th:se universities will be as generous The Microtutor Language. Stipes Pub- with their CAL lesson software as they are lishing Company, Illinois. at present with their systems software. Sherwood, B.A. and J.N. Sherwood (1986) Then my only problem will be to work out CMU Tutor: An Integrated Program- how tr'und the purchase of a couple of ming Environment for Advanced- hundn d advanced function workstations - Function Workstations, Proceedings of 330 i.';',4-.t 1 Using Computers Intelligently 321 the IBM Academic Information Systems University AEP Conference, IBM Aca- demic Information Systems, San Diego, April. Stewart, J.A. (1986) Project Athena: Faculty! Student Projects, Project Athena, Massa- chusetts Institute of Technology, De- cember 1986. 3I The computer based education system: A process of integration? Michael McCrae Department of Commerce, ustralian National University Current computer technology is fast becoming an important and widely used instructional tool. products have appeared, and only limited But managing the education process is much use made of the potential. Why is this? more than just instruction or assessment. This paper looks at the concept of an integrated This paper examines the potential of CBES computer based teaching system. At least three component areas of computer use may need to as an example of an integrated approach to be integrated to produce comprehensive. using computers in tertiary courses. Com- student-centred learning systems - compute. puters cart also be used in the educational based instruction, computer managed learning and computer assisted course authoring. This environment for various administration paper offers an analysis of the advantages of applications, but this paper concentrates integrating these areas to form a computer based only on educational applications. education "system", rather than leaving them as isolated, unconnected 'application' areas. The limitations and requirements of such an integra- The paper is organised as follows. Section tion process are examined and the 111 !y feasibil- two briefly examines the philosophy be- ity of the concept opened for discussion. hind a 'systems' approach to computer based education. The section reviews some Computer aided techniques now impact on of the major functions and components a whole variety of tasks in the educationaltypically attributed to these systems to il- process. The penetration is becoming in-lustrate the concepts of integration and creasingly pervasive and sophisticated, extension (benerality of application) which both in design philosophy and task com- these systems offer. Section three exam- plexity. But the potential for the integrationines the advantages and benefits claimed of computer assistance into a 'system' offor an integrated approach to the educa- integrated software products covering thetional process. Section four outlines limit- total educational process under a computer ing requirements and suggests possible based technological umbrella remains shortcomings. Section five advances tenta- largely unrealised. On the surface the ap- tive conclusions on the future of the CBTS proach could combine integration with philosophy. The aim is to generate discus- extension generating and managing sion of these issues based on participants courses based on sound educational prin-individual experiences within tertiary ciples in diverse areas. Certainly the techni- education situations, rather than provide cal expertise and the necessa.y umbrella ofanswers. technology is there. But attempts to pro- duce complete, computer based educationThe CBES Philosophy systems appear to have lagged behind technological capability. The proponentsCurrent CBES systems grew out of ex- of integrated computer based educationtended computer based systems that are systems (CBES) argue their benefitsdesigned to manage assessment admini- strongly. But to date few comprehensivestration and recording keeping within the 332 Using Computers Intelligently 323 the learning process (CML) (Rushby, 1985). The design concept of CBES is based on the Traditionally, most CML systems are basedprinciple that the process of course design, on a model of student study behaviour, andimplementation and management is com- concentrate on managing reporting and posed of a series of sequential tasks that can assessment procedures in an attempt tobe defined and modeled as an educational allow the teacher 1.1) concentrate on teach-process. The nature of these tasks and the ing. They typically provide support in four use of computer and communications tech- areas: nology to assist their accomplishment is construct, mark and analyse tests for well documented (see, for instance, CBTS diagnostic/assessment purposes; 1983; Denholm, 1984). Since common tasks keep student progress/performance are involved in all course design and imple- records; mentation,computerbased systems design provide individual guidance throughcan be used to manage the learning envi- structured course material; ronment, and to provide the integrated feedback reports on students perform-environment for developing CAI pro- ance and progress to teacher. grams where these are evaluated as being desirable and necessary. The integration of Using extended CIAL's as an interface, the CAI, CAS and CML functions under one CBES concept attempts to integrate course authoring, computer aided instruction,technological umbrella has been desig- nated CBTS or CBES. The emphasis is now learning assessment and course manage- ment into an integrated suite of softwareon integrating separate tasks as part of a total system. packages. Typically three features charac- terise these systems: Historical Perspective 1. The ba sic design ap proac h views course design and implementation as an edu-The historical impact of computer technol- cational process which consists of se- ogy on the tertiary education process is one quential tasks that are purposive i.e.of ever wider and more sophisticated ap- directed towards particular educa- plications. The progressive steps in the tional objectives. This contrasts withprogression include': previous, segemented approaches di- rected at isolated tasks in the education use of computers as an instructional aid process. as an outgrowth of programmed learn- ing (Rushby, 1985); 2. The design approach integrates the growth of interactive instructional dia- separate computer aided tasks into a logue; system design philosophy. The empha- using the computer to aid or control sis is not just task compatibility but inte- design of the instructional process, in- gration into a total system design. cluding course authoring; 3. The previous concept of computer using computers to manage the learn- managed learning is expanded to in- ing process; dude the whole range of tasks which the development of integrated CBES. make up the education process in the design, construction and delivery ofThese steps reflect the development of courses. Attention focuses on using thecomputer aided instruction (CAI), lesson computer to actively manage the wholeand course authoring systems (CAS), com- educational process rather than justputer managedcourse implementation managing CAI and assessment proce- (CMI) and then using computers to man- dures. age the whole learning process (CML). Of 333 324 Using Computers Intelligently course the definition of the tasks performed similar software suites in different insti- by these categories will vary between place tutions, and even within institutions; and author; especially for instance in the Specificity and lack of transferability of scope of computer assisted instruction applications software. Applications ot- (CAI) [See, for instance, Robbins (1983)]. ten limited to specific courses or com- puter systems. McDevitt and Price (1984)provide an alter- native view of this progression. They sug-The developing sophistication of many of gest that the development path taken bythe computer based course authoring aids computer based teaching systems reflectsalso focused attention on the need for the general evolutionary pattern of infor- computer involvement in the 'active' areas mation processing technology. This pro-of course design and management fol- gresses from the Transaction Processinglowed by management of the learning Phase, through the Information Processingprocess rather than just 'passive' computer Phase, to the Data Management Systemsuse in already designed lessons or courses. phases. Many of the essential ideas of integration, extension, system and educational process This proliferation of computer aided taskare present in these developments (Harris, performance has several features relevant1983). For instance Sutantn(1984) gets near to the present discussion: to the concept in his ComputerBased Teaching System in Power System Analy- Evolutionof software for computersis. based course authoring and managing has been superimposed on the moreThe motivations for movement from a traditional'passive' role of computer'passive' role of computer usage for com- aided instruction; puter aided instruction towards a more A progressive process of bringing to-'dynamic' involvement in managing the gether technological advances in sev-teaching process may be complex. But eco- eral areas including information tech-nomics is never far away. Tertiary educa- nology, communications, and network-tion is notoriously short on funding. The ing; expensive nature of CAI soon emphasised Improvements in understanding of the the need for strategies, since CAI can be a analytica Is and design of course presen-costly tool to devise. The following critical tation, design, learning and the educa-questions began to be asked: tional environment; Increasing analysis of educational ob- what should be CAI'd jectives and their effective attainment which students should get CAI through computer assistance; reducing cost of locally developed CAI Analysis of the education process into structuring CAI lessons to maximise separable tasks, and breakdown of the local applications. components of those tasks; The development of increasingly com-All of this implies the need for develop- plex sets or suites of software that werement of CAI strategies aimed at cost effec- compatible with one another; tive results and maximisation of educa- Often built by those in charge oftional impact. An "engine" is required that courses, rather than those skilled in the will initiate and manage the C.process theory of education and learning; for maximum effectiveness within budget Much reinventing of the wheel due toconstraints. Denholm (1984, p. 38) points separate developments of functionallyout that "tt implement an individualised 334 Using Computers Intelligently 325 learning environment with present dayapproach to incorporating computer aids budg Ming constraints or when just effec-within the educational process which is tiveness of effort is sought, CAI is probably important 2. best applied via a Computer Managed Learning Program." The approach integrates the tasks of: designing courses and authoring course But even prior to the development of an content 'integrated' approach to computer aided Instructing the learner (using computer education, Earle (1984) notes that in the aided teaching) early '80's Leibum (1981, p. 277) summa- managing the learning process. rised the general benefits of Computer Based Teaching as :"individualisation, Looked at in this way, the CML interface selfpace learning, immediate responsecan help give the system a generality in and knowledge of results, automatic collec- application to diverse courses, since there tion of performance records/evaluativeare common tasks and task sequencing data, patient, tireless tutor or drill master,within this subsystem that may apply to all active concentrated learning, motivatesapplications. An extended Computer and reinforces learning, cuts learning time,Managed Learning System (CML) be- gives more time to instructor for providingcomes an integrating interface for other individual assistance, relieves instructor ofsubsystems in the total process. It repre- rote/repetitive tasks." Given that thesesentsan applications software product benefits were well recognised, what doesdesigned to manage learning activities in the philosophy or 'system' design conceptsan educational environment. It provides an that underlie CBTE offer that is new orintegrated orchestration of the learning innovative? What differentiates it fromenvironment. The CML in a functional previous developments in computer aided sense becomes the environment in which education? learning activities are managed, pointed towards objectives, and in which CAI takes 'Integrated' Functions performance place. Tilt. functions it supports include: The essential purpose of CBES is that com- course authoring puter performance of teaching and learn- course organisation ing tasks should be integrated into a com- student testing puter based system of education. In practi- instructor management facilities cal terms this means that separate pieces of evaluation of program effectiveness software written to help perform various management of student records tasks in the educational environment could integration with CAI be integrated together into a software 'sys- communication with microcomputers tem'. The design of each task would then development of courseware data base take place within an overall 'system' design for a computer learning environment. TheAn integrated system can have many con- system can then provide a set of tools that tributing applications. It is the possibility of articulateto provide a system with theof joining components together which widest applicability to diverse course andmakes them dynamic and fully of potential. educational processes in tertiary environ-Try (1984) includes the following applica- ments. The system models the education tion areas: process as a schematic whole, rather than concentrating on computerising separate Curriculum planning tasks. It is the system philosophy as an Curriculum management 335 326 Using Computers Intelligently Diagnostic/Prescriptive teaching func- a one-to-one interaction of student tions and Assessment and reporting with computer Catering for individual differences students can proceed at own pace Using the computer as a learning re- presents problems in logical order source training can be individualised Administration These advantages support the learner by: CBES can provide computer support to providing for individualised learning educators and trainers in the development programs and routine maintenance of directing and tracking learner progress multi-resource, multi-learning environ- generation and marking of tests ments. To achieve this system design in- providing feedback of results and reme- cludes the inherent characteristics of: dial teaching program independence content flexibility Benefits for the teacher include: value independence maintaining records terminal efficiency providing course status information real world learning providing a framework for course de- courseware data base independence velopment and organisation multi-level security providing a communication link with learners Importantly, CBES provides the frame- managing progress evaluation and re- work within which instructional deVelop- porting. ment, delivery and learning evaluation course material can be updated conven- processes occur using computer based iently methods. enhances rather than substitute for lec- tures Advantages and benefits records student performance for assess- ment. The benefits of integrated computer based provides practice and remedial help education systems are well documented in until the desired level of mastery is the literature(See for instance Rushby reached. 1985; Chesmond 1984; Digital 1984; Try student's record and criticism can pro- 1984; Sutanto 1984). These benefits may be vide information for improving future classified in terms of: changes in course material the learner the teacher These advantages combine to: the organisation time saved through shifting of clerical and routine tasks to the computer. Learner benefits include: e.g. report preparation, record keeping, curriculum targeted to his needs statistical analysis and marking. learning programs independent of an integrated system means the teacher other learners can focus on using the computer appli- accurate and timely control of the learn- cations and not be concerned about ing experience technical aspects of equipment or soft- progressive insurance of mastery over ware management. steps Overall, more time can be spent plan- efficient learning resources ning, developing and teaching and less technology training on administrative and technical tasks. 3 3 6 Using Computers Intelligently 327 The benefits to the tertiary organisation aimed at effective learning. A process include: where each task or function in the edu- finely tuned curriculum or course plan- cational process meshes to produce an ning educational process greater than the flexible and adaptable course planning sum of its parts. facilities measurement of attainment of course outcomes 2.It emphasises the need to inject learning powerful individualised learning sys- and education theory into course design tems and implementation and evaluation. Many tertiary educators, while experts The desired objective arising from these in their field are not educational experts advantages of using CBES is increased well versed in the principles of effective educational products and better achieve- education. This can be a perennial prob- ment of educational objectives per educa- lem in universities, especially in Austa- tional dollar spent. lia, where techniques of formal evalu- ation of effective teaching are not in Educational Advantages. vogue, as is the case in some other coun- tries such as USA. Thus, tertiary educa- This list of benefits is a summary of those tors expert in computer based technol- found in the literature. They are well ogy, or well versed in its application to known and add nothing new. But several of course situations may use computer the potential educational benefits have not facilities to aid the educational process been explored in detail, and since they do in various areas without attention to arise from the uniqueness of the CBES principles of sound education. The approach are worth discussing and evalu- danger is in perpetuating ineffective ating. Two educational advantages illus- educational processes. trate this point. This problem of coincidence of educational 1. Course design and implementation istheory with computer technology exper- seen as a complete learning processtise is a continuing one where computer composed of educational objectives and aided learning processes are devised by particular elements that should com-individual tertiary educators in their par- bine into an effective learning environ- ticular fields Use of computer power in ment that is purposive; i.e. whose mainparticular courses may give the appear- objective is to achieve effective and effi- ance of more learning for each education cient learning by the student within the dollar spent, but this may be at the expense cost and resource constraints operatingof the continuation of ineffective teaching in that particular environment. There is practices. I am not arguing that this is al- of course, nothing inherently ineffective ways the case. On balarre the injection of in mobilising computer based tech-computer based learning or management niques into specific areas or tasks within techniques into course structures and im- this overall process. It happens all the plementation might be expected to consis- time in computer based course author- tently lead to improved effectiveness and ing and course management te-:h- efficiency in terms of educational objec- liques. But this 'component' approach tives satisfaction if for no other reason than may not encourage a 'system' view of that the change involves an in depth look at the educational process. A view that what is being taught, how it is being taught sees course delivery as a rational, pur- and the learning evaluation processes. The posive integrated multitask process point is that this review and improvement 3 3 7 328 Using Computers Intelligently process maybe optimised where a 'system' Resource Costs view is taken of the educational process which then attempts to base the total sys- Resource requirements may be substan- tem design on generally accepted sound tial, even prohibitive to Departments educational principles. with little computer system capacity; Programming requirements are oner- Effective CBES systems design would seem ous, especially graphics component; to necessarily require the incorporation of Frequent use by substantial student sound educational principles. So that, to numbers is necessary for adequate re- the extent educators use such systems in turn on investment; course preparation and management, they Expense expensive to lease 'PLATO' follow paradigms that take account of ac- terminal; cepted educational principles for improv- Extensive capital investment to obtain ing or optimising the learning process, whole PLATO system. even though they themselves may not possess such skills or be experts in educa-Problems of standardisation tional processes. If the system is based on Curriculum differences inspecific such principles then the degree to which courses between universities require the course convener and presenter is courseware changes trained in theory of educational process Only runs on a Cyber computer system design becomes less critical to the design Software source code not available. effectiveness of his courses-. A CBES. may help to separate or disengage educationalSutanto's conclusion highlights the com- design skills from skills in the particularpromise between resource limitations and discipline area being taught. The former iseffective education. CBES can dramatically embedded in the CBES, the latter remainsimprove teaching however "one cannot the province of the course convener. justify the extensive cost in obtaining PLATO system purely on a pedagogical This advantage of course, is dependentground especially with the shrinking upon the extent to which particular CBESbudget of most Australian universities. incorporate such principles and points to the importance of the input of experts inConclusions education and learning theory into the CBTS design process. If CBTS is designed In the enthusiasm for computer aided tools, solely by information system and softwareit is easy to forget that the computer is only technocrats who merely mechanise exist-one of many tools in the education process. ing course and educational structures, then The learning environment and experience to this extent, these benefits may not becould be deprived of much of its richness if realised, and the system may merely per- the diversity of teaching/learning instru- petuate ineffective teaching, learning andments and experience were replaced solely evaluation procedures. by computer based education. Such criti- cisms of computer dominated educational Limitations processes are valid. However, they are essentially based on a 'conflict' perspective The typical limitations of CBES are echoed of computers in education. The idea that in Sutanto's (1984) reference to the limita-computers inevitably replace or supersede tions and disadvantages of the PLATO,other instruments; even including the and in an earlier but similar similar evalu-teacher. The shadows of 1988 are still not ation by Montanelli (1979). They stress: far away. 338 Using Computers Intelligently 329 Taking an opposite view, it is possible tobudgets this may encourage wider devel- argue that used with sensitivity and appro- opment or participation. I suspect that the priateness CBES may not 'conflict' with, reason most CAS, CAI and CML develop- but enhance the use of alternate teachingments are by interested individual course techniques and instruments. This suppor- conveners and Faculty members is the lack tive role can be particularly argued in rela- of funds to finance such initiatives out of tion to the use of computers as an integrat- Departmental teaching resources. ing tool within course design, construction and implementation. For example, it offers Perhaps a partial solution to this common possibilities for dilemma may be the introduction of less expensivecut-down versions of CBES introducing sound learning and educa-systems for micro computers. Such ver- tional principles into basic design andsions, while cheaper, do not necessarily approach to tertiary courses by emoed- have to sacrifice the power of the system in ding them within course CBES design; terms of tasks performed. The limitations are more likely to come in limitations on Relieving the course convener andmemory capacities, so they do not suit teacher of routine, recording and ad-courses which would have extensive ministration tasks; memory requirements. The key here may be in devising high productivity systems improving the efficiency in course de- for microcomputer applications. Systems sign by giving the convener and teacherthat are inexpensive and compatible witl a basic paradigm to refer to in the de- the type of memory capacity, data base sign process; capability and processing power now be- coming more widespread among Depart- An underlying planning system for thements as they take advantage of microcom- teacher which integrates all thr tasks in puter systems. course construction and presentation, while leaving the teacher free to use Footnotes those educational instruments for pres- entation, learning and review which he 1. Of course these developments may and sees as appropriate. have continued in parallel as well as sequentially. For instance, in the area of Unfortunately, such benefits are often diffi- computer aided instruction, develop- cult to quantify in money terms; and even ments have occurred in at least four more difficult in terms of money benefits CAL paradigms the instructional per period. Cost is superficially more easily paradigm, the revelatory paradigm, the expressed in dollar terms. Consequent, conjectural paradigm and the emanci- especially in the present climate of de- patory paradigm. For details see pleted funding for the tertiary sector, it is Rushby (1984). the absolute cost of educational enhance- ments r,hich determine priority and com- 2. This approach moves one step closer to mitment. This applies to CBES. Unfortu- a concept of a generalised expert system nately, present systems involve substantial of the educational process for tertiary initial outlay either to buy or to participate courses. in a bureau type st.rvice through remote access. If CBES purchase or access could be References made inexpensive relative to the levels of present day Faculty or Departmental CBTS (Australia) Ltd (1983) A Comprehen- 3 3 9 330 Using Computers Intelligently sive Computer Based Education System, ceedings, Second Australian Computer Melbourne: CBTS, Documentation. Education Conference,Sydney, Septem- Chesmond, C. (1984) A Role for the Educa- ber, pp. 214-5. tion Sector in Developing Computer-Based Training Packages for Industry,Proceedings, Ca lite,Bris- bane, pp. 8-20. Digital,(1984) Introductionto Computer-Based Education,Digital Equipment Corporation. Denbo 1m, R. (1984) Applied Economics to Make CAI More Viable,Proceedings, Ca lite,Brisbane, September. Earle, T (1984), Development and Imple- mentation of CML in a Computing Course,Proceedings, CALITE,Brisbane. Harris, M. (1983) What an Authoring System SHOULD be Like,Proceedings, Ca lite,Brisbane, pp. 432-445. Jones, A. and O'Shea, T. (1982) Barriers to the Use of Computer Assisted Learning,British Journal of Educational Technology, 13(3), 207-217. Leibum, M, (1981) Factors Sometimes Overlooked and Underestimated in the Selection and Success of CAL as an Instructional Medium. In R. Lewis and Tagg (eds)Computers in Education, North Holland, p. 277-283. McDevitt, A. and Price, J. (1984) dBase II: A Tool for Computer Aided Learning, Proceedings,Calite,Brisbane, pp. 167-187. Montanelli, R (1979) Evaluating PLATO in the Teaching of Computer Science, Journal of Computer-Based Instruction, 5(3), 72-76. Robbins, G. (1983) Tomorrow to Fresh Woods and Padtures New,Proceedings, CaliteConference, Brisbane, p. 327-9. Rushby, N. (1985) An Introductionto Educa- tional Computing,London:Croom Helm. Sutanto, D. (1983) Assessment of the Use of 'PLATO' in Electric Power Engineering Education,Proceedings, Calite,Brisbane, pp. 371-5. Try, K. (1984) Computers in Education: An Tntegrated Approach,Published Pro- 340 Education resource planning G. C. O'Brien School of Economics La Trobe University In recent years there has been an increased need if current enrolments are not sufficient to to match occupational training at all levels with meet replacement requirements. the perceived or forecast labour requirements in the associated occupations. This has been especially so in the TAFE area. This paper ex- Economists are concerned with matching plores the theoretical aspects of matching occu- supply and demand, but there have been pational tzquIrements to enrolments in the as- few attempts to do this in educational plan- sociated educational programs. It then de- scribes an implementation of this theory in an ning because of the difficulty of matching interactive computer package which matches specific occupations to specific educational occupational requirements in Victoria with programs. TAFE courses which lead to occupational quali- fications. The package has been developed for the Victorian TAFE Board and forecasts the This paper describes such a match and increase or decrease in enrolments required to discusses both the theoretical considera- meet labour requirements in each occupation at tions and the practical difficulties associ- the end of the century under various economic ated with such an exercise. These ideas are scenarios. These ideas can be implemented at all levels and would be a useful tool in University then incorporated into a practical micro- and CAE Careers Units as well as for Education computer implementation which is de- planners. This paper emphasizes the design signed to work on an IBM style personal philosophy and other aspects of particular computer and could be easily implemented interest to Asdlite members on any other machine. The program pro- vides a host of facilities which would be In recent years has been an increased useful not only for planning purposes but need in all sectors of education to matchalso as a resource for career counselling enrolments in specific educational pro- units. grams with the needs of our society. In particular there has been an attempt toProgram Planning Requirements in match the resources put into education to Technical and Vocational Planning the requirements for specialists in particu- lar areas of the labour market some yearsThe system described in this paper, while later when the increases or decreases in generalizable, relates specifically to the enrolments affect the size of work force. It is provision of technical and further educa- often difficult for planners to understand tion in Victoria and to the associated occu- the implications of lags in a dynamic sys- pational groupings in Victoria. tem. Indeed it may prove necessary to in- crease enrolments in a particular educa-The Victorian TAFE system, in common tional program to meet a target level ofwith the rest of Australia, offers a wide employment in some associated occupa- range of courses through schools, colleges tion at some future time even if that future and adult education authorities. These target is lower than the current employ-courses may be classified into six streams: ment level. This could happen for example professional, para-professional,trades, Using Computers Intelligently other skilled, preparatory and adult orlabour market data in planning new pro- further education. In 1985 there weregrams or in expanding or contracting exist- 330,000 persons enrolled in these six ing programs. The trades group of courses streams, the largest stream being adult to a large extent require specialist facilities education with 131,000 enrolees, while and staff who can not easily be reassigned 31,200 were enrolled in the basic tradeto other programs. Hence significant stream. The trade stream is exclusively changes in the mix of occupational require- apprenticeship in nature and accounts forments in the community will have signifi- nearly one-quarter of total contact hours.cant implications for these'facilities This stream leads to trade qualificationsbound' training programs. and registration and is the subject of this study. It was realised that projections beyond the year 2000 would not prove very useful in The basic trade stream comprises some 117 practice, hence the design philosophy was separate courses which lead to 52 pro- to develop a system which examined a 15 claimed trades. Individual course enrol-year 'planning period' beyond the base ments can vary from a handful of studentsyear. On the understanding that there studying at a particular location to courseswould be continued revisions of future such as motor mechanics which can haveoccupational requirements projections for thousands of students enrolled at dozens of trade and technical groups a base set of diffeterit locations. projections were required. These were found in a study of industry and occupa- The size and complexity of the TAFE sys- tion work force forecasts for Victoria under tem in Victoria - an annual recurrentvarious economic scenarios for the 1990's. budget of over $200 million, 17,500 staff (6,300 full time), 32 colleges and 600 sm..:ler In this analysis four different macro-eco- providers - creates significant co- ordina-nomic growth scenario are available. These tion, management and planning problems.are Apart from the size and complexity of the 1.parity with OECD growth system, the high degree of autonomy of the 2. parity with world growth rolleges, the main providers in the trade3. parity with high world growth st.,_ obviously exacerbates these prob-4. parity with low world growth. lems. These different growth scenario together Since 1981 when considerable autonomyencorvass the rangt. pessimistic t was granted to the separate colleges, TAFE optimistic estimates and hence in the final has had to revise its planning processes.model provide a sensitivity analysis for Part of this process is the development of a various kinds of growth. management information system to pro- vide the necessary data for its internal It must be stressed estinu tes of occu- needs. The system discussed here is a key pational requirements are not measures of element in this system. occupational demand as they do not in- clude any replacement demand. These Educational and Occupational Trends occupational requirements indicate only what work force skill composition will be As about 70% of TAFE's recurrent budget is needed if the growth and output levels are devoted to its vocational activities, it isto be attained. To estimate occupational obvious that a more effective and efficientdemand the age composition of the work system will necessarily take into accountforce was used to generate a Markov re- 342 Using Computers Intelligently 333 placement process, and details of this are follows: given below. Type I one program and one related Units of Analysis occupation The fundamental difficulty with linkingType II one program and several related occupational requirements with trainingoccupations programs which lead to these occupations is that there is not a one-to-one relationship Type III several programs and one re- between these two sets. Under any of the lated occupation classifications of occupations there are some which have no associated instruc-Type IV several programs and several tional program in the TAFE system, some occupations. have on associated instructional przlsrair. and others have more than one instruc-A complete supply- demand interface will tional program. On the other hand there are clearly contain, in general, all of the above some TAFE programs which lead to moretypes of units of analysis. Previous at- than one occupation. The concept of a 'unittempts at similar analysis have been re- of analysis' has been developed to allow astricted either to a single education- train- rational linking CT 'crossly alking' of c :u- ing program based supply-demand inter- p-at:on.: groups and instructional pro- faci- (Types I and II only) or to or a single grams. occupation based supply-demand inter- face (Types I and III only). Both of these A unit of analysis is a mapping which asso- formats force the user to impose breaks by ciates groupings of educational/trainingexcluding certain occupations fror-t par- programs, 'saving common curricula and ticular programs. Often the excluded occu- objectives, with the related occupations for pation-program groups are those streams which they prepare program participants. with the largest enrolments. Our approach The development of this mapping or set of makes no exclusions on the grounds of units of analysis is central to any attempt toanalytical expediency, although we have link occupational supply and demandmade some minor exclusions where no data. The absence of such an interface data or surrogate data is available. means that the separate elements of data collected on the supply and demand sidesCareer Progression and Replacement of the labour market are of limited use for program planning. Indeed, such a link is a Unlike some other countries there is a necessary p:e-requisite for any considera- dearth of data in Australia on student pro- tion of st,pply-demand balances for re- gression rates, post-program labour force gions and occupations. Of course there is asentry experience, occupational mobility much art as science in determining thisand labour force withdrawal for trades mapping and it issometimes necessary, onbased training programs and occupations. the grounds of either utility or empiricalSome studies have attempted to describe validity to discard some occupations, and, tradesperson's career paths but unfortu- more rarely, instructional programs fromnately aggregate over trade groups and either the range or the domain of the map-provide significantly different progression ping. These issues are pursued in the report rate estimates. There is a significant gap in to the TAFE Board. our knowledge of the immediate post- trade training entry experience of indi- Units of analysis are formally classified as viduals. A further major gap lies in our 334 Using Computers Intelligently knowledge of progression i at,;:-: by year or mationnot available at the level of indi- level through the apprenticeship programvidual trades but a recent ABS survey pro- vides some base line data which enables us to make some preliminary estimates. In this study the estimates of year-to-ye Ir progression rates are derived from enrol-This survey by the Australian Bureau of ments over the last few years. TAFE has Statistics is an Australia wide multi-stage only collected comprehnsive enrolmentarea sample of private and non-private statistics on a regular basis since 1981 anddwellings. The sample size and the nature this period has also seen a rapid growth inof the survey limit its usefulness at the the variety and size of trade related pro- individual trade level, but it does provide a grams. At this stage it is not possible tovaluable framework. Of the more than one track students over time so our year-to- million trades qualified persons in Austra- year or level-to-level survivorship rates bylia only 56 percent were employed or work- trade stream category are necessarily tf ing in their trade or were using some trade five. TA. is currently supporting pilot skills in their present occupation. Other key longitudinal tracking studies, but the par-findings, from the point of view of our tial information which does exist indicatesanalysis, we P. that : not only considerable variation between courses but also that these progression this figure of 56 percent varied mark- Tatesare sensitive to external labo, ir market edly from trade to trade (from 8 pr.fr- conditions and vary considerably, from cent in footwear, clothing and textiles year to year. Our system enables the user to to 71percent in the electrical trades), 'noddy these progression rates for each trade course, and of course this allows an an estimated 5 percent had never experimental determination of the sensi- worked in their trade, tivity of the predicted required enrolments to variations in these progression rates. the proportion still practising their trade declined sharply with age, The entry rates to the trades following training are even less well understood. It is by age 25 approximately one quarter clear from Sgro's study that demand is have left the trade. sensitive to wagelevels but our estimates of this entry rate, while plausible, have littleOther studies cast little light on these statis- empirical evidence to support them. Once tics, some roviding similar estimates and again these rates are able to be modified byothers contradicting this study. All of these the user and experimentation with differ-other studies were smaller in scope and size ent rates is possible. and provide only partial estimates. Once again the facility has been provided in the There is also a limited amount of informa-computer program to undertake experi- tion available on the career paths of trade- ments by varying the consequent cohort- spersons. It is Irery difficult to obtain any to-cohort progression rates. estimates of sun key variables as: the pro- portion of trades qualified persons who In the ABS career path survey age specific have never practiced their trade, the age retention rates are given for age cohorts, specific proportions of practicing trade-most of which span 10 years. These pub- spersons who stop using their trades skillslished rates for each trade have been sub- and the age specific rates of permanentjected to a sixth order filter with the appro- work-force separation. Certainly this infor-priate boundary conditions in our study. 3 4 4 Using Computers Intelligently 335 This process provided estimates of reten-to the first cohort, j = 6 to the second etc.,) tion rates in the first few years in the trade and k denotes time. In the present formula- and subsequent analysis matched the re-tion k =1 corresponds to 1987, but this is sults obtained using ten year cohorts. Thiseasily changed. approach has the disadvantage of being relatively arbitrary but appears to giveThe dynamics of the process are deter- robust results in conformity with the pub- mined for each industry by the matrix of lished data. There is, however no way of transition probabilitieswhose jl -th entry extracting from this data transition proba-is the survival probability from state j to bilities from the training program to the state 1 at time k, and the inputs to the system trade. The question of later entry to the by the vector whose entries correspond trade is, of course, no problem since we areto new entrants to the system into the levels only concerned with modelling effective of the TAFE course or into age the cohorts retention rates. Sensitivity analysis, eitherof the trade occupation. experimental or theoretical, provides some measure of how critical these potential The process is then completely specified by inaccuracies are to the final estimates. A the equation which enables us to predict series of cross-sectional analyses at differ- the number of tradespersons or appren- ent times could also provide some esti-tices at any given level or in any age cohort, mates but to our knowledge no such studyemployed in industry is at anytime from has been done. the present: Modelling Throughput Making the calculations determined by the preceding equation is considerably simpli- The methodology behind our analysis isfied because of the special nature of the general systems theory, a branch of mathe-Markov transition matrices and, as matical analysis which includes the theorymentioned above, we initially make the of Markov processes. Initially we assumeadditional simplifying assumption that the that the parameters which describe thetransition matrices are invariant over time. system are invariant over time, but this isHowever the advantage of the above gen- relaxed in the most general implementa-eral formulation is that entrants to the in- tion of the analysis. In practice it is a rela-dustry other than first year TAFE trainees tively straightforward matter ti, determinecan be incorporated. There are some fur- whether this restriction to an autonomousther technical difficulties associated with system is critical. matching single year cohorts in the TAFE program with a mixture of five and ten year Systems theory relates the output of anycohorts in the trade but those details will system, in this case the number of qualified not be discussed here. tradespersons in any future year, to the inputs to the system, in this case the num- From this general formulation it is not dif- ber of entrants to the appropriate TAFEficult to determine retention rates and age prograrit or programs. The state of the sys-specific profiles for each of the trades and tem is the number of tradespersons ormore importantly, in tem.., of later eco- apprentices of each age in each trade group,nomic analysis, the average working life in which is denoted by where the subscript i the industry for each of the trades in ques- denotes the industry, the subscript j de- tion. These calculations also provide a veri- notes the level in the instructional program fication of the estimated transition proba- for j = 1,...,4 and the number of years in the bilities in each trade by the following con- industry for j = 5,...,15 ( j = 5 correspondingsideration. The system can alternatively be 336 Using Computers Intelligently modeled on the basis of retention rates and commencement of the pi ogram, and enrol- the transition probabilities in equilibriumments in the last four years do not affect determined as the Frobenius-Perron maxi-target occupational levels. It seems sensible mum eigenvalues of the dynamic systemtherefore to reduce enrolments in the last determined by the retention matrix. Thisfour years to replacement levels. However alternative formulation provides an heu- in some cases this would produce drastic ristic verification of both the analysis andenrolment changes which could not be the data. accommodated by the system. This analysis is then used to provide esti-Structure of the Program mates of required enrolments in the TAFE instructional programs associated withThe implementation of the above theory each unit of analysis to achieve specifiedand associated practical considerations has target numbers of tradespersons in theresulted in a Turbo Pascal program which associated occ.-pational group at someserves an eightfold purpose. The program specified future time. There are obviouslyhas been designed to be used interactively many different adjustment processes butand provides a fast, user-friendly opera- only two are implemented in our formula-tion. It is completely menu driven and tion of the model. These are requires no user manual, at least for the target audience of TAFE officers. a. a constant annual increase or de- crease, as appropriate, in first year enrol- ments, that is, where E is the constant an-The first objective was to provide an easy nual increase in first year enrolments, andintroduction to the concept of units of analysis and to allow these to be accessed b. a constant annual rate of increase orfirstly from one of the associated occupa- decrease, as appropriate, in first year tions, either from the older Au stralian Clas- enrolments, that is, where r is the con- sification of Occupations (NCO) or from stant annual rate of growth of first yearthe new skill- based Australian Standard enrolments. Classification of Occupations (ASCO), sec- ondly from any of the associated TAFE Both of these growth scenario are useful to actional programs, and thirdly di- the planner. The first requires a constantrectly from the unit of analysis itself. The annual increase in accommodation, equip- ogram displays the complete identifica- ment, staff and resources while the secondtion associated with the unit of analysis, requires a constant annual percentage in-detailing the official codes and titles for crease in accommodation, equipment, staffeach element of the map. Help screens are and resources (when growth rather thanbuilt 4-to the system to provide sufficient decline is predicated). Both approachesinformation for the first-time user. have their adherents in planning. However in the case of a substantial adjustment to the The second objective was to provide a enrolment numbers being necessary it is complete easily accessible catalogue of all important to understand that the second of the ASCO and NCO trade classifications scenario will, in general, create the largest and all of the trade related TAFE training overshoot. This is particularly the case for programs. This catalogue allows the user to those units of analysis with a four yearbrowse through these lists for whatever apprenticeship. In this case, increased en-purpose or to select an associated mapping rolments do not begin to affect the stock ofor congruence from any entry in the cata- tradespersons intil tour years after the logue. 346 Using Computers Intelligently 337 tion of additional units of analysis, if it The third objective was to provide on-line proved necessary. access to a statistical data base relating to the total number of tradespersons in anyConclusion trade group, the age distribution of these tradespersons, the career transition proba- The program as demonstrated at Calite-87 bilities and survival rates across age co- fulfils the first seven of the objectives speci- horts, the enrolments at each level of each fied above although the author has deemed TAFE trade course in the base year, the it expedient not to provide access to the transition probabilities from level to levelparameter modification module at this and from the final year to the trade. stage. The nature of non-autonomous dynamics is not easy to understand and The fourth objective was to forecast target some slells are required to interpret data occupation levels for each occupationfrom the simulation of such systems. The group in 1995 and 2000 on the basis of four program has been designed to allow the different economic scenarios for Australia,easy incorporation of additional occupa- and to relate these targets to the units oftions, instructional programs and further analysis.. data but some modification to the source code will be necessary. The fifth objective was to analyze the given data to de' --mine the required enrolment Acknowledgement levels over some planning period to meet a selected target number of tradespersonsThe author would like to thank David employed in the selected trade, either in a Wilkinson from La Trobe University for scenario of constant growth or of a constant testing the program and identifying sev- growth rate. eral bugs. The sixth objective was to determine theBibliography replacement requirements for a particular target occupation under a chosen growth ABS Report, (1984). Career Paths of Persons scenario, the average age and the age distri- with Trade Qualifications Australia, Sep- bution of tradespersons at the terminal tember to November, 1982, Canberra: time. It is important to emphasize that since Australian Bureau of Statistics. this analysis is to a large extent concerned Frugoli, P.A., (1982). Guide to Forming Units with disequilibrium adjustment processes of Analysis, Washington: National Oc- then replacement requirements, the aver- cupational Information Coordinating age age and the age distribution of trade- Committee. spersons will vary from year to year. Langley, P.C. (1985). Employment and Occu- pational Trends in Victoria, 1985- 2000, The seventh objective was to allow all of the Adelaide: TAFE National Centre for parameters in the model to be modified by Research and Development. the user either for the purpose of conduct- Langley, P.C. and O'Brien, C.C., (1986). ing experiment s or sensitivity :alysis or of Modelling Trainee Throughputs, West- modifying the parameters as calculated ern EconomicAssociation Annual from other studies or available data. Meeting, San Francisco, July. Sgro, P.M. (1985).Apprentice Demand The eighth objective was to enable the and Workers Compensation Rates, addition of further occupations and train- Melbourne, Victoria:Department of ing programs and the consequent defini- Labour. 338 Using Computers Intelligently Dr O'Brien is currently the Dean of the School of Economics at La Trobe University. He commenced work- ing with computers at the Univer- sity of Queensland in 1963 when he was a graduate student in mathe- matics. He then worked at the Uni- versity of New England and the Australian National University, before joining the Economics De- partment at La Trobe University in 1972. He has worked at New York University and Warwick Univer- sity. Current research interests are in economic dynamics and model- ling and he has underway a large project modelling the socio-economic aspects of soil ero- sion in the Philippines. Teaching interests are wide with a special emphasis on Operations Research and Mathematical Economics. The research described here arose from his recent consulting work. 348 If it is useful then use it: Some experiencesin the use of computers in the teaching of Chemistry. Michael Page School of Chemical Technology South Australian institute of Technology. Twelve years ago I purchased some BASIC list- cros, and a multitude of methods of author- ings of drill and practice computer programs. This year I am involved in the production of an ing. I have been encouraged, discoura interactive videodisc.This paper describes helped and obstructed. I have, as Marvin some of the experiences and knowledge gained Minsky indicated to us two years ago, suf- in the intervening years. It describes computer based education packages which have worked fered the relearning process. At the end of and some which have not, when CB.E. has been it I realize we have come but a short way in useful and when it has been a disaster, which the use of computers in the education authoring languages have been found to be scene. The information revolution is only helpful and which have been a waste of time. Examples will be given of drill and practice, just beginning and CD-ROMS, Interactive utilities, tutorials and simulations. And finally- Videodiscs and satellite communications has it all been worthwhile, and where do we go are opening up a whole new horizon of from here? possibilities in education. Some of these possibilities we are aware of, but others we have no more knowledge of than the office At a symposium reported in the Journal ofclerk of the nineteenth century had knowl- Chemical Education entitled "Will Com-edge of the word processor, database and puters Replace Teacher Aids?, Professors?, spreadsheet. Just as they are tools of busi- Labs?, Should They?", participants wereness so are similar packages tools for the leaders in the field of computers in educa- teacher, and just as business and commerce tion. They alluded to the computer as a tool proceeded with the abacus and steel knib- and as a new medium of instruction. Therebed pen so the teaching process should was general agreement that the computerproceed with the tools that are available provides new ways of approaching prob-now. We cannot afford the luxury of wait- lems, but the consensus was that the com- ing. Waiting for the ultimate teaching tool. puter is a tool and that good teachers can- not be replaced by a computer. So this paper is not full of the latest in teaching technology which will turn the I first became interested in using comput-educational establishment on its ear. In any ers as an aid to teaching about twelve yearscase education is too conservative for that. ago after a visit to Australia by Prof. BitzerWhat it intends to do is to help take a little of PLATO fame.This, not surprisingly,of the suffering and some of the drudgery coincided with the purchase of a CDCout of the teaching process, and in so doing Cyber computer by the S.A. Institute oftake some of the drudgery out of the learn- Technology making the use of a computering process. It might even provide ideas to possible in the general teaching sphere ofmake students think (a revolution in it- the Institute. Since then I have followed a self!?) and provide time for the teacher to tortuous path through macros, minis, mi-deal with individual student problems. 349 340 Using Computers Intelligently Philosophy cal work. In our large first year practical classes ene of the terrible time consumers My philosophy as to whether to use com- for the instructor is "marking practicals". puter software materials is based upon theStt.dent calculations are notoriously unre- following proposition. liable. So if one comes across a "wrong" result at the end of a practical report the Use it if it: questions arise; is it bad laboratory tech- nique, use of the wrong "formula", or enriches the course and if time is avail- straight out arithmetic (calculator) incom- able petence? Since in my view the instructor is saves the teacher time and doesn't dis-there mainly to teach correct laboratory advantage the student technique time should not be wasted on the motivates the student to do well other two. So a suite of programs has been saves the student time by speeding up written to check the students' calculations. tasks or supplying information. Such a calculation check is required before a report is submitted for assessment. The Purposely left out is the one of cost effec-student is prompted to input his/her ex- tiveness. This is always a contentious issue. perimental values, and then the final value Are libraries, laboratories, photocopiers which the student has calculated. etc. cost effective aS teaching tools or just traditionally desirable? I would question The program then tells the student whether the purchase of a large computer system the calculation is correct to within the limits just for computer aided learning tutorials set by the instructor, is close to that value or and drill and practice, but there are sois very wrong. It does not say whether the many more things that can now be done final value is correct in a practical sense. with computers to make life easier for theWhen I first introduced this over 50% of the student and educator. One must consider students were making calculation errors at all these things if one is to consider cost the first attempt. Of course they all blamed effectiveness. Frankly I consider such anthe computer until they recalculated and exercise as useless as trying to evaluate the got it "right". By the end of the first term cost effectiveness of a specialist book col- few errors were being made. lection in a major reference library. Not having to search for calculation, trans- One last point in the philosophy of using position or formula errors can reduce computers in education is the fun of it. Godmarking time by up to 50%. knows one needs a sense of humour to be in the education profession at all and if usingOne of the spin-offs of this was the discov- computers can contribute towards the out-ery that students have a bad habit of round- let of that sense of humour so much theing off intermediate values during a calcu- better. Learning should be fun it is the lation. This can cause large errors in final unlearning that causes the suffering andvalues even when using calculators. So that comes later. now advice is given early in the course on the inadvisability of rounding and the The Programs importance of the order in which computa- tions are carried out. Those who do not a. Time savers listen suffer the consequences. 1. Calculation checks It is also noticed that students usually check These have been used to date in two main their calculations in small groups, and er- areas. The first is to check students practi-rors are often picked up by their peers and 3 5 0 Using Computers Intelligently 341 corrected quickly, and at a time when thethe testing process until mastery is ob- corrections are relevant to the task in handtained.Students soon realize that time and not say a week later when reports are spent using the programs saves much time marked or handed back. later. The second area in which this techniqueThis emphasizes an important principle of has been used is for assignment checking. the use of computers in education. The use This has allowed more assignments to beof the computer must be seen to serve a set and marked resulti ng in better detection useful purpose and not just "nice to do". In of specific weaknesses in first year stu-this case passing the test first time after dents. proper preparation is a tangible goal with concrete benefits. 2. Individualized Assignments. Again not a new idea, but a great time saverThe tests themselves are not given at the and student motivator. Sets of individual-computer terminal. The reason for this is ized numerical problems are producedthat the taking of tests and the immediate using a random number generator. Thesemarking of them gives an important con- are printed complete with an answer sheettact between instructor and student, and for each student. A set of answers for eachallows subtle misconceptions about the set of problems is printed for the instructor.subject matter to be dealt with at the time. It takes little time to mark the answer sheets4. Utility Packages for each student and blind copying is elimi-Commercial programs such as word proc- nated. Peer help is encouraged since, firstlyessors for reports, statistical packages for it saves the instructor time, and secondly, processing laboratory results etc. are being experience has shown that lazy" studentsused more and more by students as they do not get the problems done for them.become convinced of their time saving at- Also answers cannot be passed down from tributes. One longs for the day when all one year to the next, a practice commonreports, essays and assignments will be with students for generations! submitted perfectly typed with no spelling errors. Such techniques as the above have been used for some years at the University of a.Simulations Texas at Austin to "cope" with 800 first year chemistry students. These are used to illustrate theoretical concepts when they are first encountered 3. Drill and Practice. so freeing laboratory work from being Now once again finding favour with edu-timetabled to fit in with lecture material. cationalists, drill and practice is used exten-These simulations are not intended to be sively as pretest sessions for a self pacedsubstitutes for work in the laboratory. first year Chemistry course. Each of the 20 units of the course requires, apart from the A simulation of enzyme kinetics is used as completion of an *assignment, a test to bea follow up to a laboratory experiment passed based on the contents of a chapter ofwhere time constraints do not allow all the the recommended text. The pass mark isexpen mental variables to be investigated. about 90%. The computer sessions include all the questions that a student is likely to beA simple nuclear chemistry simulation is asked in a test. Incidentally no two tests areused as a substitute for a laboratory experi- exactly the same and a student can repeatment. In this case it is desirable that the 3 5 342 Using Computers Intelligently students appreciate the unique laboratorylook at duplicated notes if they think an principles, but facilities to mount the ex-exam question is hiding in there some- periment are not available for reasons of where. cost. We must not think that the use of comput- Apart from the value of allowing studentsers in education is going to have a dramatic to apply theoretical knowledge (or lack ofeffect on the behaviour and attitude of stu- it) these simulations have highlighted the dents towards their own education. Of need for students to gain experience incourse a well produced piece of course- planning their own experimental work. In ware presented at the right time in the right most scientific laboratory sessions all the context will stimulate and motivate stu- planning has been carried out by the in- dents, but no more than the same material structor in order that the experiments well presented at the right time and in the should "work". In other words the studentright context by a teacher. The computer is is left with a recipe to be followed slavishly still a tool and a tool which is rapidly losing without real understanding of the con-its awe to the student population. In fact cepts. This delights the student and makescomputers are getting a "bad press" be- practical courses run smoothly, but leaves acause popularly they are being blamed gap, The computer simulation is an ideal more and more for the errors of their users. vehicle for training in organization and planning as a wrong plan does not wasteSo in general I only use computer gener- valuable materials and equipment and, ated tutorials for specific cases of difficulty, since simulations produce results rapidly, and even then find some resistance to their time is also saved. use. Simulations are also used as animatedAuthoring Systems and Languages examples in lecture and tutorial classes allowing a freedom to "see what happensAt present I use PCPILOT on an IBMXT if", and to encourage student involvement.compatible microcomputer with a hard disc, and colour of course.I have had There have been simulation packagesexperience with several other authoring which have not been successful. One rea- type systems namely STAFF and DAL on son for this has been that the simulationmainframes, and PILOT, ENBASIC, Q, itself has been too close to reality and hasQCAL, PAL and MICROTEXT on micro- wasted time in repeated animation when computers. quick results was the objective. Again we come back to the timesaving aspect ofThe reason for choosing to use PCPILOT is computer based education. partly subjective as is the choice of author- ing systems by most people. In my case I c.Others needed a system for the IBM microcompu- ters which are available for student use at No mention has been made of computer the I also needed a system with generated tutorials. My experience so far isgood graphics, a choice of colours and that these are resisted by the average stu- fonts, good string handling routines and dent as a waste of time. They regrettably good mathematical capabilities. I was not can see no point in doing a program be- so concerned with computer managed cause it "it is good for you". Of course thosecourses although PC-PILOT has this capa- students do not read their text books andbility at an elementary level. references for the same reasons, and only 352 Using Computers Intelligently 343 I was not interested in an automatic pro- want to do and upon your personal prefer- gramming system as my previous experi-ence for the type of system you wish to use ence with such facilities had shown them towhich will be coloured by your previous be restricting in the long term and lacking experience. in flexibility. Such "filling in the blanks" systems look very attractive to a newcomerIt is important however that, when evaluat- to authoring, but unless one can break outing a system; one makes sure that it is the of that stage easily one soon becomes frus-latest version as software development can trated with the restrictions imposed. Actu-be rapid, and a system which in its first ally PC-PILOT does have such a facilityform may be primitive can, in version two, with the advantage that it produces the be just what one is looking for. Evaluations source code which can be manipulated of authoring systems could be one area for later as one's experience grows. After some consideration by ASCILITE with perhaps two years working with PC-PILOT I haveevaluation copies of the software and now developed a collection of screens andmanuals available for short term loan by routines which I use extensively giving memembers along with a list of user contacts. considerable flexibility. Such a service would be invaluable to the new author and the more experienced user PC-PILOT is a system which uses a simplelooking for alternatives and updates. text file created with a very good screen text editor. This contrasts with other authoringAnd in the future... systems which are menu driven. Which is the best system I think is a matter for theMy personal future will be to write more user to decide. I prefer the text file probablyand obtain more of the types of courseware because I am used to it and can easily visu-which has worked in the past. New things alize th1/4.! lesson screen from the text file to write not mentioned above are pre and screen. I have rftif,0 used a ram-disc and now post laboratory tutorials and tests, past a hard disc to switch quickly from the textexam paper help and tips,and a suite of file editor to the PILOT driver and so see thesupport programs for disadvantaged stu- lesson as it develops. I can also pull in from dents, and I have to program and test that my library collection, screens and routinesvideodisc we have just produced. which develop the lesson rapidly and with few errors. These routines include suchI shall also continue to buy, borrow (but things as matching tests, multiple choice never steal) as much CBE material as pos- question screens, true/false routines etc.sible because this is infinitely faster than and a lesson shell which sets up special writing it. And if the material is not exactly characters for chemical symbols, windows,as I would like it I am quite prepared to timing and marking macros. So I have slightly change what I do to fit the material, probably reached the point where to and if the material is good it will not do any change to a new and possibly more efficientharm. system would be .resisted because of the mass of material already developed andThe future of computers in education is the chore of learning a new routine. assured, but the future of computer based education or computer aided learning or I have tried to concentrate on the positivecomputer assisted education or whatever it aspects of the choice of an authoring sys-is called will continue to be a rocky one. My tem, and do not wish to dwell on the nega- experience over the years has seen com- five aspects of other systems, since as I have puter based education greeted with enthu- said, the choice depends upon what yousiasm by a few, treated with apathy by 344 Using Computers Intelligently many and met with opposition both open and subtle by a surprising number of edu- cators. It is a pity that not enough teachers see computers as the tool for better educa- tion, and a tool to be mastered not ignored or thrown aside. The rapid technological advances in the fields of information exchange and storage are about to have a dramatic effect on the way we live. A flood of videodiscs and compact disc storage devices are about to appear and educators should be ready to take advantage of the wealth of informa- tion soon to be at our fingertips. This tech- nology will never replace the teacher, but if used with intelligence, will give the teacher the time and freedom to give students the quality of education too often missing in our Schools, Colleges and Universities. The immemorial problems of teaching endure but at least we have been vouchsafed a marvellous new tool with which to chip away the darkness. Derek Davenport. References Will Computers Replace TA's? Professors? Labs? Should They? (1984).Journal of.Chemical.Education,. 61,(1), 26-35. Castlebury, S.J., Culp, G.H., Lagowski, J.J (1973). Journal of.Chemical.Education,. 50(7),469-472. M Page is Senior Lecturer in the School of Chemical Technology and in charge of Associate Diploma Course in Ap- plied Chemistry. Teaches General and Analytical Chemistry and his Research interest is mainly computers in educa- tion Community interests Theatre (performing) and portrait photogra- phy. C`r, 1 , , t ,121ti el Computer managed learning in electronics D Parsons, School of Engineering R Hunter, Division of External and Continuing Education Darling Downs Institute of Advanced Education The availability of computer managed learning data base contains unique identification of material has been extended to students enrolled in Associate Diploma in Engineering course all current assignments, weighted and non - and, in particular to students enrolled in elec- weighted, together with descriptors, dates tronics. The CML component comprises test and assessment weightings. All CML test instruments, facility to report home-based prac- suites form part of this data base. The tical exerdses, and ciradt construction simula- tions. Two modes of operation are available - student records interface allows updating interactive in study centres and home, and mail- of academic progress. The system itself in mode. Some evaluation of student reaction to stores responses to each test item, giving the CML component has been made. the opportunity for item analysis. Produc- tion output is in print and diskette formats. Acceptance of distance education modelsTest suites are designed to incorporate has added impetus for new approaches toitems of multiple choice, exact answer, teaching.This impetus, together withanswer within u range, answer within a expanding availability of new technology, sequence, one of a number of answers, has generated new bases for the support of some of a number of answers formats or learning. The Darling Downs Institute ofany combination of formats. A feature is Advanced Education (DDIAE) commis- th1.2 guidance of students depending upon sioned its computer managed learningtheir performance and the supply of feed- (CML) system in 1984. The system added aback specific to student's response. The powerful component to the range of media figure below details the basic structure of available for the production of learning each test suite. packages. Students elect to use the system interac- Cornputa lvlanaged Learning. tively using microcomputers and diskettes or via a mail-in mode which is processed on The term computer managed learning iscampus. Mail-in operation provides the applied to embrace the facets of testing and same feedback for tests and experiment instruction. The application has allowed reporting as the interactive mode. The the system to support not only assessmentcomputer assisted instruction (CAI) pack- and instruction but to manage reports of ages are available for interactive use only. laboratory exercises. Adoption of a nar- Student access to the interactive mode is rower definition would have limited thelocated on-campus, at study centres and, to application of the medium. a limited degree, at home. Selection of the mode of operation is made at the beginning The system interfaces with three data bases of each semester. - student records, assignments and CML content. The latter allows on-line or batch Operation at study centres and on-campus input of module content. The assignment utilises three diskettes. A student diskette 355 346 Using Computers Intelligently Figure 1: Key Elements of Computer Managed Learning COMPUTER MANAGED LEARNING PRESCRIPTION OF LEARNING ACTIVITIES / RECORD DIAGNOSTIC KEEPING/ FEEDBACK REPORTING TESTING Figure 2 : CML Interface with Data Bases which contains enrolment data, security, agement tool. The areas of instruction and system access records and test results alsotesting are clearly identified. Progress is contains the operating system. A contentrepc:ted instantly; feedback is specific; diskette holds test suites, experiment re-remedial guidance is immediate. Within porting formats and CAI material.A the constraints of assignment management "mail" diskette has access and result data self-pacing is possible. The remainder of transferred to it and is used to update filesthis paper addresses the application of the on the student record base. principles to Electronics, in particular to the areas of testing, reporting and instruc- The CML system embodies fundamental tion through simulation. principles. The medium is used as a man- 3 '-i 6 Using Computers Intelligently 347 UNDER LAKE PRESCRIBED LEARNING ACTIVITIES TAKE CML TEST JUDGE ANSWER TO ..-....--..---...... ----..-.--.-----,....--.-Ipk. EACH QUESTION NO DETERMINE * CORRECT? ERROR TYPE 47 YES PROVIDE PROVIDE FEEDBACK FOR FEEDBACK INCORRECT FOR CORRECT RESPONSES RESPONSES 4, 4, NO PRESCRIBE '-----.'------YES FURTHER PROGRESS LEARNING TO NEXT ACTIVITIES MODULE-----...,- TAKE PARALLEL CML TEST ------, CONSULT LECTURER ....-----"..._ Figure 3 : CML System - Student Activity Interface 351 348 Using Computers Intelligently Figure 4 : Location of Study Centres optional questions as preparation for ex- aminations. It provided them with a means of self-assessment, and, in the case of the LOCATION OF STUDY cviw rb OtADOOLAND summative tests, forced them to work tLor- oughly in searching for answers. In spite of the appreciation expressed by the students, they felt that the amount of feedback provided on incorrect answers was not as great as it could have been. Some security against students using re- sults from other students in previous years is given by providing several tests, and changing the one specified as summative from year-to-year. Experiment Reporting. While many students are employed in situ- ations in which equipment is available to Testing them, there are some who are very limited by lack of such equipment. The electronics Testing is done by providing banks ofexperiments have therefore br.or_ designed questions, most of which are multiple-to be done on the kitchen table, with very choice. About two-thirds of the questionssimple and inexpensive equipment, typi- are numerical, or calculation bases, the restcally just a power supply and simple multi- being of a more descriptive nature.Inmeter. This system works very well and electronics, a lot of reference must be made few students have difficulties, even though to graphics such as circa., diagrams andmost are working alone. timing diagrams, and also to detailed data sheets such as those for integrated circuits. The experiments are usually the assembly It is also important that students see these of a circuit and a series of measurements on, in their printed form, since this is the nor- or changes to it. Very careful instructions mal way th.;:y are available.So, evenare given to ensure that the circuits do thoughts it is possible to provide .,ese on work, and in ninety percent of cases, stu- scree a, students are referred to printeddents have no trouble achieving basic cir- material. (This is also necessary for stu-cuit operation. The importance of doing dents without access to a computer). something practical like this is in seeing it work and determining why, rather than Students require one to two hours of effort just getting it to work. It is the understand- to do one of the twenty two tests provided. ing of the circuit and the principles behind Three of these are nominated to be summa- it which are important, rather than the tive, with a total of twenty percent (A the practical skills. marks for the unit allocated to them. The remainder are optional. This approach was Eleven of these home experiments" have endorsed overwhelmingly by studentsbeen provided in one unit, and sixteen in who were questioned. They appreciatedanother. They /pically take one or two the fact that test questions covered every hours to do and in total are allocated up to part of the work, and they did use thethirty percent of the marks for the unit. 358 Using Computers Intelligently 349 Figure 5 : A Design Problem 3 Design of true R-S Flip-Hop Draw here in the neatest possible way, aoiding crossoversas much as possible R 11 S 15 13 10 17 0 12 9 Q 4 is 0 16 6 5 2.Flip-Clop design Node number I 2 3 4 5 6 7 Connected to LICIOCIDOODE El 42 20.231) Node number 9 10 11 12 13 Connected to OD DO DO OD DE OD 1214.2 451 About half the students feel that they took 1. The performance of electronic devices too much time to do, but still wanted them varies greatly from one unit to the next, to be a compulsory part of ths! work. Most (and from one student to the next due to students expressed some satisfaction with differences in power supply etc). Cor- the practical familiarity gained with elec- rect operation of the circuit can be deter- tronic components, and with the sheer mined by observing correct rends in experience of working with circuits. results rather than absolutely correct numbers. Given the philosophy that doing success- fully is the major objective, the marking 2.It is useful sometimes to have a "design procedure becomes (90% of the time) con- your own" approach. The results can firming a lot of correct numerical results. vary widely and correct design can be The use of a computer to mark in thiscase determined by analysis of reported is essential to allow the experiments to be as design values. Figure 5 shows an ex- extensive as possible.It also allows of ample of this. course detailed feedback comments to be made, based for example on specificre- sults. Because understanding is the goal, ques- tions are attached to each experiment There are at lea' wo aspects of electronics which call for analysis of the work just done where the al-ty of the computer to doat a practical level, and interpretation of calculations on students particu- results. Appendix 5 is an example of such larly helpful. a question. 35? 350 Using Computers Intelligently Student reaction was sought to the experi-Figure 6 : Attitudes to Experiments ments to determine attitudes about the extent to which doing the experiments: 100 Helped helped understand the material in the Understanding unit lot increased their interest in the subject were necessary for a good understand- 100 ing increased increased confidence in handling elec- 20 Interest tronics. A lot Not ...oh Figure 6 shows a distribution of the opin- 100 ions offered in each case. Ne e nary for 20 Unde-stending Simulations. A lot ot much These are an attempt to overcome some of 'Co the limitations of the equipment shortage at home. They also encourage a "try it and increased Practical see what happens" approach to the practi- Confidence cal experimentation of the kitchen table. If lot Not 'Nth these practical experiments do not work, no-one can tell the student why. With the simulations, the computer indicates what from a study centre, and most used the is wrong. system several times. They felt that the exercises did help them to understand the For the range of circuit types (usingdiodes, circ-its, and to become familiar with stan- transistors and operational amplifiers) the dard values of components. The writers student can do the following on the screenpersonal impressions are that there are of a study centre computer. some people who respond to this kind of activity, and some who do not. We will be 1. Analyse a selected circuit with valuesin a better position to assess the value of provided which do produce a workingsimulations when more students have eas- circuit. ier access to a computer. 2. Design a circuit by selecting a particularImpressions arrangement, and the values of compo- Presented are some personal observations. nents. Totally incorrect values are pro- hibited by the system. A badly-de-The effort by staff into these computer- signed circuit which still works is al-based facilities is huge. People involved were lecturer, instructional designer, ad- lowed E.o that the analysis will reveal its limitations. ministrator, programmer and production staff. Of these, the lecturer is the only one who hopes to receive a direct benefit in 3. Any of the above circuits may be altered and again analysed. terms of reduced working load. Even so, with the typical numbers of students con-con- More than half the students trying thiscernedrn (70-90) the lecturer would feel the need to run the material largely unaltered purely optional exercise lived over 10 km for several years. 3 6 0 Using Computers Intelligently 351 The major benefit is to the student who has many more opportunities for formative exercises, receives work back more quickly and receives much more detailed com- ments on his work. There is some anecdotal evidence that more students are spending time ringing lecturers directly with specific questions about work given. References. Parsons, D.J. (1983). Practical Work in the External Training of Electrical Engi- neering Associates in Queensland, Australia. Int. I .Elect. Eng Educ., 20, 101- 110. Parsons, D.J. (1983). The Value of Practical Work in Teaching Electronics to Exter- nal Students.Australian and South Pacific External Studies Association, 6th Biennial Forum. David Parsons is a lecturer in electrical engineering at the Darling Downs Institute of Advanced Education in Toowoomba, Queensland, where he teaches electronics. A major part of his teaching is done externally, and has involved developing home-based teaching materials and strategies. Bob Hunter is Associate Head (Re- sources) of the Division of External and Continuing Education at the Dar- ling Downs Institute of Advanced Education. He is an instructional de- signer working on the development of computer managed learning. 3 61 ,ralmEm...=1 Formative evaluation systems for CAI John A. Price, Accounting and Business Law, University of Melbourne, A major problem with CAI systems is the qual- for the most part the teacher-made ity of their instructional materials. CAI is an electronic form of teaching and the normal tests are summarized to show the score opportunities for teacher observation and feed- or marks of individual students. Only back are not available. A formative evaluation rarely does the teacher use them as the module can be included in a CAI system to help basis for modifying instruction.(p.135) to overcome this. It requires little in the way of additional programming work to incorporate a formative evaluation module into a CAI sys- Student marks are translated into grades tem Yet CAI authors have been slow to recog- and used to separate students into the more nize this potential. The result has been the use of and less able, so that the more able may multiple-choice questions in CAI which are not adequately tested. The author describes a CAI progress to the next stage or to a higher system that he has developed which includes a stream. However data on student achieve- formative evaluation module. The system runs ment has another valuable educational on an IBM -PG application.It can provide the basis for improving instruction. Bloom et al. (1971) urge educators to use the student-tests as Two phases in the evaluation of educa- the basis of "quality control" of instruction tional material and processes are com- and not: "as it is most frequently used in the monly recognized. These are: formative existing educational systems... the grading evaluation (i.e. while a CAI system is being and classification of students" (Bloom et al. developed) and; summative evaluations1971, p.7). This view of evaluation which (i.e. when the CAI systems is completed).considers student grading leads educators Formative evaluation rather than summa-to make decisions about students rather five evaluation is used in the development than about teaching material and methods. and updating of instructional material. Decision making in the educational process The author designed and implemented a must do more than grade students. This CAI system which incorporates a forma-view is supported by Kandaswamy (1980) tive evaluation module. In this paper thewho contends that it must involve: principles of formative evaluation and its application to CAI are discussed. Forma- making decisions about which compo- tive evaluation is discussed in the first sec- nents of the training material are to be tion of this paper. modified in order to make the materi- als instructionally and motivationally Formative evaluation stronger. (p.19) Bloom, Hastings & Madaus (1971), theProgramme developers determine the ef- authors of a major work on the evaluationfectiveness of their material at various of learning and teaching processes, notestages in its creation. During programme that: 362 Using Computers Intelligently 353 development the CAI system is used and development of a curriculum to im- the responses to the multiple-choice ques- prove it. It serves the needs of develop- tions are recorded. The information could ers. (p.17) be used in a step-wise development of the programme. In step one it could be used to The information indicates to teachers and determine the validity of the test questionsinstructional designers the need to im- and in step two, to assess the strengths and prove the teaching material or process. weaknesses in the preceding instruction. Areas of weakness, such as lack of under- standing or practical ability, are identified. The means of providing information to A teacher modifies his/her approach while students, teachers and the instructionala lesson is taking place based on student designers so as to improve the teaching/feedback. One of the limitations with an learning pi ,cess is called 'formative evalu- electronic teacher such as CAI is that ation'. This term was first used by Scriventeacher observaigm is limited. However (1967) in a paper on the methodology ofthis can be overcome by incorporating a evaluation. Bloom et al. (1971) note that: formative evaluation module into a CAI system. A CAI formative evaluation mod- it is (Scriven's) view that formative ule can: evaluation involves the collection of appropriate evidence 'during' the con- assist the initial design and creation struction and trying out of new curricu- stages; lum in such a way that revisions to the improve the instructional content; curriculum can be based on this evi- check the logical sequence; dence. (p.117) help to select questions for CAI drill and; In formative evaluation data on student improve the communication aspects. progress and learning is used to determine the efficacy of the instructional materials The improvement of the drill component is and teaching processes. Formative e%alu- the first stage in the application of forma- ation is an important tool used in the trial tive evaluation to CAI.This aspect of implementation of a new programme atimprovement of CAI is the subject of this each stage in its development. Bloom et al. paper. In the discussion to follow, the CAI (1971) note that in the application of forma- system and the functions of the recording tive evaluation: and formative evaluation modules are described. This corresponds with the top it may be determined at each step in down design methodology used in the (this) process whether the process is construction of the CAI system for this effective or not, and if not, what project (Price, 1985; Price, 1987). changes must be made to ensure its effectiveness. (p.8) The CAI system Measuring student progress, such as test CAI is a method of teaching which uses a results, provides information that is used to computer to provide instructional material improve the instructional material or proc-and interacts by asking the student ques- esses. It involves tests which are as much a tions. CAI tutorials are characterized by test of the programme as the:re of the the use of multiple-choice questions. A student. As Weiss (1977) notes: CAI tutorial instructs the student and asks formative evaluation produces infor- a question on the material jwt presented. mation that is fed back during the For this reason they are frequently de- 3 6 3 354 Using Computers Intelligently scribed as 'instruct and drill' systcrns The recording module When students respond they are provided with immediate feedback as to whether Gerhold (1980) found only a small amount their answer is right or wrong. If they are of interest shown by authors in keeping wrong, they may be provided hints as user statistics in more than six years of CAI to the correct answer, encouragement and use at the University of Western Washing- so forth. The communications between the ton. If records are kept then they should be student and the CAI system are controlled useful, and not just record that a particular by the system interface. This interface alsostudent has finished a lesson or topic. The allows communication with other users, simplest form of recording system is when such as program authors, teachers, men- the number of correct responses is recorded tors and administrators. All these may at against the students name. An example of some time need to access the CAI system to a CAI system which includes a simple re- obtain student information or update pro- cording module is illustrated in Figure 2. grammes. In the CAI system to be de- scribed in this paper ther? are two types of users. These are teachers/ administrators and students. The system menu provides a route fore tch user to access .he CAI system and is illustratea in Figure 1.The CAI system controls access to .iata in the CAI system by a hierarchy of privilege. This is an important consideration because of the need to protect the integrity of student records and teaching material. Figure 2. CAI design with a simple record- Create View Message Run View Message ing system (based on the "progress" data- lessons progress system and run progress system base of McDevitt Er Price 1984). Spread. Word sheets Calculator Databases processor To implement a formative evaluation module the recording module must store Graphics Summary Help Spelling Statistics more than just the on-line information information checker provided to students, such as the number of correct responses. It is necessary to store Figure 1. CAI system menu. the student ID and that student's answer to each question on the first and second at- Typically the teachers' menu allows thetempt. The author developed a database creation and editing of tutorials and view- that held this additional information. It has ing student progress, while students can more attributes (fields) than the earlier only run tutonals and view their own prog-database recording system developed by ress. The ability to provide progress re- McDevi tt & Price (1984). CAI systems have ports to teachers and students depends ona variety of question types such as mul- the CAI systems recording module. This is tiple-choice, true/ false or fill-the-gap. Thu discussed in the next section. database is sufficiently flexible to cater for u 4 Using Computers Intelligently 355 most types of question. The author has a.its too (easy) obviousand a high propor- named this new database the "extended" tion are correct; database. It stores the student's responses b.its too (hard) difficult and a high pro- to each question on the first and second portion are incorrect; attempt. The teacher/ administrator has c.both good and poor respondents score access to the data and can process it on a equally indicating uncertainty and weekly/ monthly basis with a formative probably ambiguity or no exact correct evaluation module. The formative evalu- response; ation module is discussed in the para- d.negative correlation, suggesting an- graphs to follow. swer supplied by the item's author may actually be wrong and; Formative evaluation module e.if one or more of the distractors attracts very few correctresponses indicating The writers of multiple-choice questions that it is not working and is probably should consult a text such as Thyne (1964) implausible. for general advice on test construction. However Allen & Yen (1979) contend that: An item analysis print-out is illustrated in Figure 3 and shows thr-items from a anyone who develops and uses tests recent batch of CAI drill.There are two should (also) be familiar with the pro- important statistics shown in the item cedures for test development and analysis print-out. These are the item diffi- evaluation. (p.142) culty and the item biserial correlation. An item difficulty close to 1.00 indicates an This does not mean that teachers need to easy question. Conversely an item diffi- understand mathematical statistics. It culty close to zero indicates a very difficult means that they can use statistical meas-question, which few students 'get right'. ures in the selection of multiple-choice The item biserial correlation gives the cor- questions or "items" (as they are called in relation between performance on an item the measurement literature). The forma-(question) and on all other items (ques- tive evaluation module of the CAI system tions) in the drill. developed for this project is an item-analy- sis program. Below the author demonstrates the appli- cation of item-analysis statistics for the Item-analysis techniques and computerthree items. Teachers and test-developers programs (e.g. CSHE, 1979) have been can successfully use item-analysis without developed to test multiple-choice ques- needing a full understanding of statistics. tions used in examinations and tests where students are being assessed (Nedelski, Item 1 is too easy. Nearly all the stu- 1965). In CAI the quality of the drill compo- dents were correct on this item. How- nent is still important even though stu- ever there is a good correlation between dents are not being assessed. The tech- performance on this item and the rest of nique of item-analysis provides informa- the drill. tion that is used in the formative evaluation of the drill component of CAI. An item- Item 6 is a poor question and should be analysis program identifies questions in a modified or dropped. The .48 item diffi- CAI system that are ineffective. culty is quite acceptable. However the biserial correlation of .187 indicated that A multiple choice question is ineffective for there is a low correlation between per- any of the following reasons: formance on this question and perform- 3 6 356 Using Computers Intelligently ance in the remainder of the drill. Concluding remarks Item 22 is a good question. Both of the Micn_.. -nputers have increased the oppor- level of difficulty and the biserialcorre- lation statistical measures tunities for the creation of CAI (e.g. Price, in are satisfac-press) and improved authoring systems tory. This, question should be retained. have made program creation and modifi- cation easier.However as CAI authors There is a need for at leastsome easy ques- tions in a CAI drill. This is become more adventurous they need tools near the begin-which will improve the content of their ning of a lesson where theyserve to give material. Formative evaluation isone such students confidence. Item l is clearly such tool. a question. Figure 3: Item analysis of three multiple-choicequestion. + + ITEM 1 difficulty .920 1 2 3 4 5 numberof responses 88 1 2 4 81 0 top score UPPER 1 0 0 0 17 0 UPPER 2 0 0 0 18 0 UPPER 3 0 0 0 18 0 UPPER 4 1 0 0 16 0 low score UPPER 5 0 2 4 12 0 TEST SCORE MEANS 23.0 15.5 16.3 30.4 0 CONV SCORE MEANS 9.9 6.5 6.5 12.8 0 Biserial correlation .88 ITEM 6 difficulty .489 1 2 3 4 5 number of responses 88 6 15 24 43 0 top score UPPER 1 1 0 4 12 0 UPPER 2 1 3 6 8 0 UPPER 3 0 4 6 8 0 UPPER 4 1 4 4 8 0 low score UPPER 5 3 4 4 7 0 TEST SCORE MEANS 26.2 26.6 29 8 30.4 0 CONV SCORE MEANS 10.9 11.6 12.8 12.8 0 Biserial correlation .187 ITEM 22 difficulty .682 1 2 3 4 5 number of responses 88 6 60 5 13 3 cop score UPPER 1 1 15 0 1 0 UPPER 2 0 17 0 1 0 UPPER 3 0 15 1 2 0 UPPER 4 3 9 0 3 2 low score UPPER 5 2 4 4 6 1 TEST SCORE MEANS 24.0 32.0 20 6 24.5 23.7 CONV SCORE MEANS 10.2 13.9 9 2 10.4 10 2 Biserial Correlation 690 + + 3 G6 Using Computers Intelligently 357 References ness. Englewood Cliffs, N.J.: Prentice- Hall. Allen,M.J. & Yen,W.M. (1979). Introduction to measurement theory. Monterey,Calfornia: Brooks/Cole Pub- John A. Price known as Tony is on the lishing. staff of the Accounting and Business CSHE (1979). Testan users guide: Departmen- Law Department of the University of tal handout. Tne Centre for the Study of Melbourne where he is involved in Higher Education, University of Mel- computing literacy training. He is also bourne. in the process of writing up a thesis for Bloom,B.S., Hastings,J.T. & Madaus,G.F. the M.Ed. degree of the University of (1971). Handbook of formative and summa Melbourne. His studies at the Centre five evaluation of student learning. New for the Study of Higher Education, the York: McGraw-Hill. University of Melbourne relate to the Kandaswamy,S. (1980). Evaluation of in- use of instructional video in comput- structional material: A synthesis of ing literacy training. He has been in- models and methods. Educational tech- volved in computer training and the nology, June, 19-25. production of CAI systems at the Uni- McDevitt,A.A. & Price,J.A. (1984). dBase II: versity of Queensland and for the A tool for Computer Aided Learning. Royal Melbourne Institute of Technol- Proceedings of the Second Conference on ogy. He has publications in computer Computer Aided Learning in Tertiary Edu- education journals such as Collegiate cation, Brisbane, 167-187. Microcomputer and the AEDS Journal Nedelsky,L. (1965). Science teachingand test- and the Journal of the South-Western ing. Chicago: Harcourt, Brace & World, University of Finance and Economics, Inc. Peoples Republic of China. Originally Pricej.A. (1985). Software engineering: A form Wales (U.K.) he has Bachelor of guide for CAI. In J.A.Bowden & Science degree from the University of S.Lichtenstein (eds.), Student Control of Wales and a Masters in Economics Learning Computers in Tertiary Education degree from the University of Queens- (pp.35-60). Melbourne: IBM. land. Price,J.A. (1987). Teacher produced CAI using software engineering. Collegiate Microcomputer, 5(3), 239-243. Price,J.A. (in press). Developing computer aidedinstructionsystemsfor microcomputers. International Associa- tion for Computing in Education Journal (formerly AEDS Journal). Scriven, M. (1967). The methodology of evaluation. In R.Tyler, R.M.Gagne & M.Scriven, Perspectives in curriculum evaluation. AERA Monograph series on curriculum evaluation, No.1. Chicago: Rand McNally. Thyne,J.M. (1964). Principles of examining. London: University of London Press. Weiss,C.H. (1972). Evaluationresearch: Methods for assessing program effective- 367 Automated Storyboards a plausible authoring aid? S. Sampath and A.J. Quaine Department of Computer Science Australian Defence Force Academy The effectiveness of an educational package 3. Prepare the detailed layout of the in- generated by use of any authoring system structional materials on paper accord- depends on the methodical development of the instructional design. There are number of ing to the sequence in which the subject stages involved in a complete instructional matter is to be presented design. One is the development of the de- tailed instructional material, as a series of in- 4. Prepare the material for presentation structional events to be presented as a se- quence of frames. The most common ap- using an authoring system. (Rorniszow- proach for a systematic development of this ski, 1986). material is to prepare the detailed layout of each frame on paper. These layouts are tra- There are a number of steps involved in the ditionally referred as storyboards. The data entered via electronically prepared story- development cycle (Romiszowski, 1986; boards has been used [81, to author lessons Bork, 1984) for each of these stages, and it is using the WISE authoring system. Extending important that each step be reviewed for this idea, this paper surveys its application to other authoring systems and suggests that 'the effective production of the educational storyboards can be generalised as a distinct package. data type to be processed electronically on a wider environment, and so provide a general It is possible with a number of authoring tool for effective instructional design. The development of such tools offers promise of systems, for the instructional designer to efficiency in the authoring cycle, and further document the objectives and the topics of author support through electronically acces- the subject matter electronically. Systems sible high-level documentation. such as PCD3 (Plato's courseware design system for PC'S) and Course of Action Systematic development of instructional(courseware design system for the design is the most important part of pro- Macintoshpersonal computer) allow the ducing Computer Based Instruction (CBI)author to build a logic map of each section material. Basic development stages of an of the lesson, thus providing a very useful instructional design system for any com- authoring tool. The third stage of the in- puter based courseware can be summa- structional design very much depends on rised as follows: the authoring system being used. Most experienced instructional designers, au- 1.Identify the objectives of the subjectthors, and authoring system developers matter to be taught according to thesuggest the storyboard approach to pr target population with the help of ex-pare the detailed layout on paper for eit perts frame or event according to the sequence in which it is to be presented. The layout of 2. Establish the hierarchy of topics to bethese storyboard forms depends upon the presented according to the set objec-authoring system (reference note 1). Hav- tives ing prepared these storyboard forms, it 368 Using Computers Intelligently 359 would be ideal if the majority of the infor-It is clear from the table that development mation entry could be passed on to a typist time of the CB: courseware is highly vari- for straight forward data entry. Since most able and depends on the complexity of the of the authoring system requires a certain course being developed. The courseware level of expertise in using the software, it isprepared by AIP and Telecom involved not easy for a novice user to enter the data. more complexity than those prepared by Expert Solutions. However, the time We have discussed in a previous paper quoted by Telecom seems high. The follow- (Quaine & Sampath, 1986) how the data ing reasons could be contributing factors to entered via electronically prepared story- the magnitude of the time required. board forms has been used to author les- sons using the WISE authoring system. The 1. Setting up the screens with responses aim of this presentation is to extend this and feedback is quite a difficult task idea to other authoring systems and sug- using WISE. To obtain the required gest a unique storyboard form that can be effect, the free response frames must be adapted to any authoring system. This will combined appropriately with calcula- provide a general authoring tool for effec- tion frames. tive instructional design and offers prom- ise of further efficiency in the authoring 2. Setting up calculation frames in WISE cyee. requires special programming lan- guage skills, similar to those required for programming in Pascal. Authoring efficiency vs Authoring tools 3. The user design did not specify the preparation of detailed layout of the frames on storyboard. The quality of CBI material depends on a variety of factors. However, as mentioned The authoring time in this case could be by Anver et al (1984), the three major factors considerably reduced if the front end soft- contributing to the quality of production ware tools developed for WISE (Quaine & are: Sampath, 1986) had been used. AIP authors quoted that, in the complete development 1. Experience with instructional design cycle of the courseware, starting from 2. Experience with authoring software design and analysis, 20% of the develop- 3. Experience with subject matter. ment time is spent on creating relevant graphic images. The level of complexity of Of these factors they have shown that expe- the courseware pr d by Expert Solu- rience with authoring software providestion using PCD3 and WISE was compa- the most cost effective method of improv- ra'.1,.. However, the authoring time using ing authoring productivity. This finding I is almost 30% less than using WISE. concurs with the survey done by Lommel Probably the availability of additional tools (1986). However, the provision of tools like in the contentand strategy editors of automated storybozrds as an additional PCD3, could be the reason for the reduced authoring aid can reduce nu_ need for ex- authoring time. Macleod (1983) states that pertise in using authoring soft rare. Tablethe total development time varies irom 1 shows the results of our survey of differ- 100-300 hours (60-180 hours for preparing ent authoring systems, the authoring time script and 40-120 hours of programming) required to prepare one hour of course- for test only courseware to simulation type. ware after the design and analysis of the An entire semester's chemistry course was instruction materials. developed at the university of Delware 36 5 360 Using Compute's Intelligently AUTHORING USERS COURSE AUTHORING TOTAL DETAILED SYSTEM NAME TIME DESIGN SCREEN HRS HRS LAYOUT PAL University Simple of linear courses 70 NA None Pro logic Queensland Expert Project PCD3 Solutions Management 80 200 Storyboard Sydney Expert Using WISE Solutions Community 120 260 Storyboard Sydney Services Dept of Electronics PC CLASS 100 240 Storyboard Army Training AIP The 200 500 Course Master Melbourne Electricity Storyboard Project Telecom Form WISE 400 NA None CBT branch Filling Melbourne Table 1. (Garton et al,1984).They quote a total a new breed of authoring systems that use development time of 350 hours (105 hours events as building blocks, and each frame for scripting, 195 hours programming andcan be considered to con,,st of number of 50 hours of testing) for an half hour lesson.events or may be a single event. Merrill These results indicate that any additional(1987) argues that the frame is a passive software used as an authoring aid can pro- concept, and proposes a new terminology duce cost effectiveness. From this surveycalled an instructional transaction, and we do feel that a genera' purpose story- claims that a transaction is an active con- board form can be a distin:t data type, thatcept. However, the concept of the frame can be introduced for pr iparation of CBI was earlier thought to be the correct model materials, and processed electronically in(Merrill, 1985). The concept of the presen- the development of the oN erall pedagogical tation screen is clearly the unit of interest to design of instructional material. the student, but from the author's point of view the display details could be a collec- General purpose stonj&ard forms tion of frames, screens, pages or events. However, the key issue is that the material The storyboard forms used in oor previousneeds to be prepared and documented in paper (Quaine & Sampath, 1986) were the form of a storyboard. Table 2 shows the specifically designed for WISE authoringbasic unit of current authoring systems. system. Those forms cannot be directly used for ot: .fr systems because of dissi mi- All these systems provide the facility to set larities at the user interc:ce. Neither do they up different types of testing using the ques- use the frarnc .lame as the basic building tion styles like single or multiple word, free block-. some do use named screens, orresponse, true/false, match the list, and pages, which are similar to frames. There is multiple choice. The systems Author and 370 Using Computers Intelligently 361 Authoring System Basic unit for presentation Menu, Language or Both Ai5 -1 Frame - template Menu AUTHOR Frame - template Langauge CourseMaster Event Both Course of Action Screen ICON oriented PCD3 Event Menu PC CLASS Page Menu PLATO Screen Langauge Pro logic PAL Page Both SAM Page Menu WISE Frame Menu able 2 CourseMaster allow the author to set up2. True or False response mathematical question types. 3. Multiple choice response 4.Free text response This summary is supported by Data 5. Gaye response, by matching a list. Training's (1986) Annual Survey of CBT authoring systems, which suggests that the Other common leo.tures of these authoring basic building blocks of the instructionalsystems are that they allow the author to material can be classified into three types :give feedback for the student's responses andto nave complete control of the 1.Menu type. Allows the s. ident to select student's progress through the lesson. So it a particular section of the lesson. is not dill icult to extend the id. a of process- 2. Presentation type.One that does not ing the stun/boards electronically, which require any response from the student.we used for WISE authoring sy stem. The 1. Question £-'an. ;weJ type. Ac cording to tile storyboard proLessing of this system con- student's response a particular action is sisted of three processes: taken. 1.Createsbs -- to create storyboards. The question & answer types can be any of 2.Lessonoverview -- to produce data the following types: flow of lesson logic 3.Printsbs prints the detailed layout of 1. Single word response stoi yboards. 371 362 Using Computers Intelligently SBNO :1 LESSON NAME : EXAMPLE FRAME TYPE : PRESENTATION FRAME NAME NEXT FRAME PREVIOUS FRAME rinAE SCREEN SPECIAL OPTIONS Brief Desc. {MESSAGE LINE} SBNO : 2 LESSON NAME : EXAMPLE FRAME TYPE : QUESTION & ANSWER QUESTION TYPE : One Word FRAME NAME : -- NEXT FRAME PREVIOUS FF "E MAX. Attempts ANSWER Time Dest. Fran TIME SCREEN TYPE OF RESPONSE SPELLING TOLERANCE SCORING WEIGHT SPECIAL OPTIONS Brief Desc. {MESSAGE LINE} Figure 1: Storyboard part 1 The only process that has to be changed toevent based. The second level of menu se- make this system adaptable to other sys- lection will be to choose one of three basic tems is the Createsbs, :n order to inuude types we have mentioned earlier i.e menu, the different features offered by the other presentation, question & ans\i er types or systems. Three levels or menu selections anyother type (other type could be special could be allowed in this process. The first options provided by other systems). If the level will ye to esti.l.lish whether the au-user chooses the question & answer typ_ thoring system to be wed is frame based or then it will display the third level of the El 7/ Using Computers Intelligently 363 Name: First Welcome to lesson Fr Name : Second Summary of the lesson I Fr Name : Presl Objectives of the lesson Fr Name : Pres2 Summary of each item in the lesson Fr Name : Menul Student is allowed to select one item out of 4 Fr Name: Pres3 Fr Name : 011/F Fr Name : Pres4 Fr Name : Q2multil Definition of Testing Student's Summary of Testing Student's Quadralaterals Knowledge Geometrical shapes Knowledge on shapes on Quadralaterals 1 I Fr Name: End Menul Menul IPres3 Pres4 Guidance to continue Menu 1 Figure 2: Logic flow of a lesson menu which will allow the user to choose Except in the case of PCD3 authoring sys- the type of question he wants to create.tem, all the other systemsprovide the Figure 1 shows part 1 of the modified story- facility to create a set of graphic libraries board form for a presentation frame and and these can be called in any lesson. Most one word answer and question type ( for of these systems can indicate the exact loca- event type of authoring system the word tion of the cursor and thus any graphic frame will be replaced by event). Similarlyobject from any library can be positioned at other storyboard forms could be designed that coordinate on the display frame. AIP for other frame or event types. employs special graphic designers to create the graphic libraries requiredfor the The field Brief Desc in part 1 of the story- courseware. The CourseMaster authoring board form is to allow the user to enter system provides some of the facilities of the brief description of what is to be di,played storyboard processing we have developed inthe frame.This description will be for WEE. Their system can also produce printed for each frame when the process the hard copy of the flow of lesson details in Lessonoverview is run. the textual format.Thiselectronically accessible documentationis within the The output of Createsbs can be ;aput to authoring system. The method of story- Lesson oven iew to produce the layout of board processing that is being suggested in the lesson as shown in Figure 2. this paper is independent of any uu thoring 3 7 3 364 Using Computers Intelligently SBNO :1 LESSON NAME : IXAMPLE FRAME TYPE : PRESENTATION FRAME NAME : FIRST TEXT TO BE ENTERED POSIT! (X,Y) GRAPHIC OBJECTS LIBRARY USED POSITION (X,Y) TO ENTER MORE TEXT OR GRAPHIC OBJECTS PRESS-10, SBNO 2 LESSON NAME EXAMPLE FRAME TYPE. QUESTION & ANSW FRAME NAME SECOND QUESTION TYPE One Word ENTER TEXT TO BE DISPLAYED: TEXT SIZE GRAPHIC OBS ,rS LIBRARY USED POSITION (KY) FEED BACK' TO ENTER MORE TEXT OR GRAPHIC OBJECTS PRESS --00- Figure 3: Storyboard part 2 This type of approach will provide excel- system and can be used as front end tool for lent documentation in the development any system. The ideal situation in a large cycle and smooth running of the whole production environment would be to pre-project. As well said by Palmondon and pare the necessary graphic images by the De,chenes (19E6), consistent documenta- graphic designers as done by AIP, and tion cannot be kept during the develop- enter the sequence of lesson details via part ment cycle of a courseware without auto- 2 of d^ c.t.zryboard form, shown in Figure 3. mation of the must mundane task of edit- ing. 34 Using Computers Intelligently 365 Lesson Preparation ciency in the authoring cycle, and further author support through electronically ac- In the case of WISE lessons, keystroke files cessible high-level documentation. Such for the lessons could be generated by the documentation can be ported to other sites preprocessors using the data entered in the of development and with slight modifica- storyboard forms. Using the input stream tion to the material it could even be adapted redirection facilities (WICAT, 1983) ofto another system. As suggested by Bork WMCS, these keystroke files are input to(1984) this documentation can be sent to WISE to generate the lessons automati-reviewers and expel is which will enable cally. Similar preprocessors can be written the material to be reviewed before it is for the authoring systems like AUTHOR collated into a courseware using an author- and FSL , an authoring language undering system. With most of the authoring development (Quaine & Johnson, 1986), systems it is tedious to modify the linkage and data entered in the storyboard formsbetween the frames once a lai ge number can provide the variable input to theseframes have been produed. hi the story- authoring languages. It could be investi-board board processing the links set up can gated whether the keystrokes for creatingbe easily modified and reviewed thor- each frame or event type could be captured oughly before it becomes CBI courseware. using the operating system facilities ofThis labour intensive process of instnic- other authoring systems. If it is viable, then tional design and development can be preprocessors can be developed similar to made more cost effective by further devel- the user interface tools for WISE. Whetheropment of Automated Instructional Sys- the preprocessors are essential or not de- tems similar to the Courseware Instruc- pends on the users of a particular authoring tional Toolkit developed by Courseware system. Even if the preprocessors are not Inc Kearsley, 1986.(This is a prototype available it will be easy to transfer instruc- system and the current version runs on tional material from well documented sto-IBM PC/XT.) A number of organisations ryboards are moving towards using CBI for training purposes. These types of automated sup- Discussion port tools can provide a consistent metnod- ology, , re-usable design and cost effective Most of the authoring systems aim at pre- computer based teaching programs in a paring similar type of instructional materi- training environment (Howes Sr Williams, als.Thus providing the automation of 1986). storyboards as an authoring aid to these Acknowledgements systems can standardise the production of CBI material. This need not be considered We acknowledge with gratitude the assis- as a front end tool only for the authoringtance given to S. Sampath by the members systems. This can also be used as a front of the CBT branch of AlP (Melbourne), tool for developing audiovisual instruc-Expert Solutions, Australia (Sydney), Tele- tional materials similar to the IBM's Story- com (Melbourne) and TAFE College board software package (Robbins, 1986). (RMIT). All offered useful infomation and 0.ir survey of different authoring systems providedthe opportunity to investigate does indicate that storyboards can be gen- different authoring systems and sample eralised as a distinct data type to be proc- courseware. essed electronically on c wider environ- ment, and so provide a general tool for Reference Note effective instructional design. The devel- 1. Expert Solutions Australia, Private opment of such tools offers promise of effi- Communication. 3 7 5 366 Using Computers Intelligently References Based Instruction, 12(2), 90-96. Plamondon, R. & Deschenes, J.(1986). Anver, A., Smith, S., & Tenczar, P. (1984). Course design using software engineer- CBI Authoring Tools: Effects on pro- ing methods, Computing Education, ductivity and quality, Journal of Com- 10(4), 417-427. puter-Based Instruction, 11(3), 85-89. Quaine, A.J. & Johnson, C.W. (1986). Frame Bork, A. (1984). Producing Computer sequencing control in PAL, Proceedings Based Learning material at the Educa- of CALITE 86, pp234-244. tion Technology Center, Journal of Com- Quaine, A.J. & Sampath S. (1986). User In- puter-Based Instruction, 11(3), 78-81. terface Tools for WISE, Proceedings of Data Training (1986). Annual survey of CALITE 86, pp245-253. CBT authoring systems, Data Training,Robbins, G (1986). Cal redefined : a win- May, 32-68. dow on available resources, Proceed- Dear,B.L. (1986).Artificial Intelligence ings of CALITE 86, pp13-22. techniques: Applications for course de-Romiszowski, A.J.(1986). Developing velopment, Educational Technology, Auto-instructional Materials. London. 26(7), 7-15. Kogan Page, 54-94. Garton, R., Reed, M.J., Reed. G. & Stevens,WICAT Systems, Inc. (1983). WMCS, Sys- E. (1984). Developing a CBI Course: The tems Programmer Reference Manual, process, Proceedings of 25th Interna- Vol. 3. tional ADCIS conference, pp142-143. Hillelsohn, M.J. (1984). Bench marking authoring systems, Journal of Computer- S. Sampath is a Lecturer in the Austra- Based Instruction, 11(3), 95-97. lian Defence Force Academy with a Hofstetter, F.T. (1985). Prespectives on a research interest in Computer Ah_ed decae,e of Computer-Based Instruction: Instruction and who teaches Informa- 1974 84, Journal of Computer-Based tion Systems and Program ring Lan- instruction, 12(1), 1-7. guages. HowL.s, R.F. & Williams, D.O. (1986). A methodology for developing Com- A.J. Quaine is a Senior Lecturer in the puter-Based teaching programs, Com- Australian Defence Force Academy puting Education, 10(3), 347-352. with a research interest in Computer Kearsley, G. (1986). The Courseware In- Aided Instruction and who teaches structional ToolkitTm: A rrototype Operating Systems and Programming Automated ISD System for Personal Languages. Computers, Proceedings of 266 Interna- tional ADCIS Conference, pp199 -204. Lommel, J. (1986). What matters in an au- thoring system?, Data Training, July, p3?-46. Macl:od, C. (1983). The Design and Devel- opment of Computer-Based Training Programs, Training and development in Australia, 10(2), 11-16. Merrill, M.D. (1987). Prescriptions for an authoring system, Journal of Com- puter-Based Instruction, 14 fl), 1-10. Merrill, M.D. (1985). Where is authoring in authoring systems, Journal of Computer- 6 Computer Managed Learning Its application to increase student achievement using formative self-assessment Jon D. Stanford, University of Queensland Howard P. Cook, Queensland Institute of Technology. Computer basedassessment was introduced format was successful in promoting stu- into an introductory economics subject, EC110, dent learning. The introduction of CML in in the Department of Economics, University of Queensland, during Second Semester 1986. 1986 led to much more widespread adop- Over 600 students used computer generated and tion in the Department of Economics in marked random tests for a number of assign- 1987. ments. A commercial Computer Managed Learning The particular circumstances occurred (CML) software package was used to issue and because one of the authors, Jon Stanford, mark student tests and to keep records, generate had returned to lecturing the first- year reports and analysis of student and question course in 1985 and was interested in im- bank performance. Because tests were answered by students away from the computer, only a proving student performance and student small number of terminals were required The motivation; as well he had a strong back- distinction is made between Computes Man- ground in instructional design, use of aged Learning and Computer Assisted Learning (CAI). computer simulations, and use of various media and course evaluation at the Univer- The effectiveness of the CML system is evaluated sity level. The oth( r author, Howard Cook, by using evidence from the responses to a Stu- on secondment to the Computer Assisted dent Questionnaire and by the end-of-year ex- amination. Learning (CAL) Unit, University of Queensland, to advise university staff on The paper concludes that CML provides a pow- the application of CML had a strong back- erful mechanism for add-^ssing such issues as ground in applications of CML for indus- learning effectiveness, 11nation of resources, student motivation, support to students, learner trial training. Co-incidently the Faculty of control and cost effectiveness. Commerce and Economics, University of Queensland, had just installed a VAX It is recommended that tertiary institutions give computer which was available for use. The serious consideration to the implementation of CML on a wide scale in order to improve student CAL Unit had purchased a commercial performance and staff productivity. This can be CML software package (CBTS) which oper- done at affordable cost. ates on the VAX range of computers. Thus all the ingredients required for CML imple- mentation, hardware, software, systems This paper ..escribes the implementation ofsupport, instructional design and over 600 CML in a large first-year economics course students were available. The implementa- at the University of Queensland in Second tion worked well because the ,iuthors were Semester 1986. Although the introductionable to work co-operatively and to enlist of CML arose from a particular set of cir- support from many sources. cumstances and although CML was ini- tially thought of as a trial run for 1987,CML The problems of orgari;;Ang, teaching and was very effective and even in its first -uc managing a large first-year course are sub- 3 7 368 Using Computers Intelligently stantial. Most students in the course enter learning process. Management of learning university direct from school and manyis the process of course design, specifica- find the transition difficult. Many students tion of learning objectives, issue and mark- find sitting in large lecture theatres alienat- ing of tests, recording results of tests, pro- ing and in such an impersonal environ-vision of feedback and remedial support, ment lose motivation. Academics are gen- analysis of mot ule effectiveness, analysis erally swamped by day-to-day concerns ofof learning performance, preparation of administration of a large course: makingreports to students and lecturers (Miller sure student handouts are ready on time, 1986). dealing with student enquiries, liaising with tutors, setting tests and examinationThere is an important distinction between papers, maintaining records of studentComputer Managed Learning and Com- performance and progress; are unableputer Assisted Instruction. Computer As- to take a detached look at the course andsisted Instruction is taken as the learning what they should be doing, and are unableprocess in which instructional material is to plan rationally for course developmentstored in the computer and released to a and for course innovation. Australian uni-student seated at a computer terminal. versities have been under financial pres-Computer Managed Learning is used to sure for some years and are unable to de-refer to the system where instructional vote increased resources to overcome these material is available only externally to the problems. computer and where the computer plays an important management function. It is The solution to some of these problems lies possible for a CML system to incorporate in the adoption of a new approach whichComputer Assisted Instruction but it is will encourage students to take greatergenerally not necessary or even desirable. responsibility for their own learning, to set and achieve their own learning goals and to The Course acquire good learning skills which will stand them in good stead during theirThe course, EC110 Introductory Econom- professional lives. Consequently the im-ics, is a standard principles course which plementation of a CML system within a accounts for 25 per cent of the workload of large course is more than the use of a com- a full-time student. The course is designed puter in teaching; it necessarily embodiesfor continuing students in economics; the the adoption of a different approach toDepartment offers as well a terminal first learning and teaching in which the role oryear course for students who do not intend the teacher is that of facilitator of learningto proceed to the study of economics at and manager of learning experiences. Assecond-year level. Jon Stanford often remarks "the first thing The course, EC110 Introductory Econom- to learn about computer managed learning ics, is required for all students enrolled in is that it has very little to do with comput-the Facu ty of Commerce and Economics ers." (unlike most ether Australian universities both the B Com and B n degrees are Definition of CML offered by the Unive:sity of Queensland) and is available to students in other Facul- Computer Managed Learning is defined ties, especially Arts, who wish to enrol for here as the application where the computer a non-terminal course in economics. Total does not have an instructional role and nominal enrolment in 1986 was 640. whene the function of the computer is in the control, administration and testing of the The major features of the course enrolment ?.s1 be summarised as follows: Using Computers Intelligently 369 approximately 50 per cent of students Second Semester: Best Five Gass Quiz Marks 25 are enrolled for the B Corn degree; (Part A & Part 13) approximately 25 per cent of students Assignment - Part A 5 are enrolled for the B Econ degree; Assignment - CML Test 5 approximately 25 per cent of students End-of-Year ExaminationI'art A 30 are enrolled for the B A degree or an- End-of-Year Examination - Part B 35* other degree; " This mark was scaled according to the student he entry requirements for the B Com performance in the final two CML tests. degree are significantly more stringent than for the other degrees; The procedure is illustrated in Table 2 on approximately 75 per cent of studentsthe assumption that the maximum marks have entered direct from Queensland on the CML tests had been obtained: Grade 12; a small but significant proportion of Table 2: Illustration of Scaling of End-of-Year Exami- students have experience of successful nation Marks, Part B, EC110, 1986. tertiary study. Unadjusted Mark Scaled Mark CML Scaled Mark The course is taught by large lecture format Olt of out of 25out of S5 to three self-selected groups, two of which 35 i 4arks % Mark are scheduled during the day and the other 35 100 25 10 35 100 scheduled in a block in the evening. The 30 86 21 10 31 89 course is taught throughout the year i.e. 26 75 19 10 29 83 over the two semesters, and is examined by 25 71 18 10 28 80 23 65 16 10 26 74 two formal unseen examinations, each of 20 57 14 10 24 69 two hours duration, in November. 18 50 16 10 23 66 15 43 11 10 21 60 The lectures are given by three academics; 10 29 7 10 17 49 one of whom teaches the First Semester and the other two each teach half of Second Semester. The end-of-year examination isThe first CML test was in lieu of an assign- ment; the other two were voluntary; marks divided into three parts to reflect this divi- were obtained on an "all or nothing" basis. sion of lecturing effort. During 1986 there were no small group tutorials available in the course; students The final grade for the course is determined by a combination of progressive assess-were required to complete a weekly class quiz and the results of this quiz were dis- ment and end-of-year examination with a cussed in a "mass tutorial" following a relatively high weighting towards the end- of-year examination. Although the deter- lecture. mination of the final grade is relatively complex it can be conveniently explained Learning Objectives for the Course in Table 1. Previous experience with students in EC110 (and with later year economics stu- Table 1. Contribution of Each Form of Assessment to dents) has indicated that, in general, stu- the Final Grade in EC110 in 1986. dents could perform at high levels in tests Form of Assessment Number of involving verbal recall but that in tests Marks requiring the display of other skills they First Semester: performed at significantly lower levels. Best Five Class Quiz Marks 25 Two Assignments 10 Areas where students have performed at End-of-Year Examination 65 low levels are: 370 Using Computers Intelligently a. identifying the empirical counterpart ofsecond tier skills should not imolve repeti- concepts understood in verbal terms; tion of tasks used in mastery of first tier b. drawing logical inferences from verbal skills but should be designed more explic- statements or from simple data (either itly to achieve the particular learning objec- hypothetical or real); tives of the second tier. c. drawing strong conclusions from ver- bal statements or from empirical data; This design of the course uses principles of d. performing relatively simple calcula-learning derived from the Keller Plan ap- tions; proach to learning and from the general e. developing detailed critical analysis ofprinciples of a Personalised System of In- theoretical or empirical topics. struction. Testing through the CML system was de- Contribution of the CML System to Achieving signed to improve student performance in Learning Objectives areas a, b, c and d above and it was recog- nised that other responses were required to A CML system is able to make a significant improve student performance in area e; in contribution to assisting students to this area students have to acquire skillsachieve the learning objectives of the which involve analysis, selection, discrimi-course in the following ways, by: nation, synthesis, criticism and evaluation. generating unique tests for each stu- Thin implies the use of a two tier approach dent to learning in which the competence prin- marking the tests immediately ciple was adopted for the first tier where allowing the lecturer to monitor indi- basic level skills are acquired. This ap- vidual student proach was seen as appropriate in the progress through the course course, EC110 Introductory Economics, recording student marks and other where competence in basic level skills data comprises the great majority of student providing various reports effort. The competence approach required providing detailed statistical analysis the division of course material into rela- of student performance tively self-contained modules and the pro- generating detailed statistical analysis viding students withclear cut learning of the answers to items in the test bank objectives and specific instructions as to so that testing procedures can be how to achieve those objectives. An inabil- modified in the light of experience and ity to achieve these objectives at high levels of revealed difficulty. (i.e competence) at the first attempt was not regarded as "failure"; instead learners areA CML system offers significant advan- "recycled" until they attain competence.tages in that computers have a distinct This implies that learners may proceed at advantage over people in doing repetitive their own pace in achieving learning objec-tasks, in recording data and in undertaking tives although institutional and logistical large calculaCons and in collating informa- constraints place limitations on self-pac- tion. ing. The testbank consisted of the following The achievement of learning objectives at style of questions: the first tier provide the ground work for acquisition of skills at the second tier level true/false but the tasks set for learners to acquire multiple choice 380 Using Computers Intelligently 371 matching completion Yen /SA Brazilian Cruzeno/SA calculations. June 1935 310 3560 !tine 1986 270 4920 Some typical questions and their relation- The Japanese yen has dr-,riated ship to the learning objectives are given: against the $A?" Matching Question: This question relates to learning objective b. i.e. "drawing logical inferences from "Consider the following statements: verbal statements or from simple data (ei- 1. Exports of coal from Australia to ther hypothetical or real)." Japan; Calculation Question: 2. Import of an !BM computer to Aus- tralia; "Assume you are a portfolio manager with $A ### to place in the short term money 3. Payment of interest on an overseas market and that the spot Dutch Guilder/ loan owed by the private sector in $A rate is #.## and that money market rates Australia; in Amsterdam and Brisbane are ##% and ##% p.a. respectively for a six month term; 4. Payment of an age pension to anwhat six month forward exchange rate Australian citizen living in Greece; would leave you indifferent between plac- ing funds in Amsterdam or Brisbane? 5. Purchase of an airline ticket by a (Express your answer to two decimal resident of Hong Kong from Qantas places e.g. X.XX)" for travel from Hong Kong to Sin- gapore; [Note: The numbers represented by ### are randomised variables so that each cal- 6. Purchase of oil by a Liberian regis-culation is unique.] tered ship in an Australian port; This question illustrates the proposition which category would these be clas-known in economics as " the covered inter- sified as, namely (G) goods, (S) serv- est parity theory of foreign exchange" and ices, (I) income, (T) transfers. (enter is important for understanding the rela- your answer as either G, 5, I or T.tionship between money markets and tor- Note: Some categories may not ap- eign exchange -iarkets. The calculation is pear, others may appear more than fundamenteto understanding the con- once.)" cepts involved. This question is relate...I to learning objec- Implementation of the Project tive a;i.e."identifyingthe empirical counterpartof concepts understood in The project used a VAX cc "Tutor and the verbal terms." software supplied commercially by CBTS with a testbank specially written by Jon True/False Question: Stanford. The course material was divided into three modules: "Given the following hypothetical data International Trade and the Balance of on exchange rates in respect of the Aus- Payments tralian foreign exchange market: aForeign Exchange 38i . 372 Using Computers Intelligently The Monetary System and Monetary600 students who used the system; there 2olicy. were queues at peak times especially lunch time but these were no longer, and nearly A further subdivision of the modules may always shorter, than those in the student have been desirable but the limitations ofRefectory. The Computer Room was open the time available for the project (the final 8am-10pm, Monday to Friday and 9am- seven weeks of Second Semester) and the 5pm Saturday and Sunday. large number of students made this im- practicable. Students prepared answers for the test away from the computer; some answers Students were required to take three ran- required consultation with the tex t books, domly generated tests; one for each mod- lecture notes, handouts and required calcu- ule. Each test consisted of approximately 15 lations to be performed. When ready to questions being a mixture of true/false;answer the test students logged on to the multiple choice; matching and calculationsystem and gave their answers when . with each type of question carrying a differ-prompted by the computer which pro- ent weighting. The mastery level was set as vided opportunities to review and change 75% for the first test, 80% for the second test answers. When the answering process was and 8c% for the third test. Students werecompleted the computer marked the test, alit wed i. maximum of three attempts atinformed the student of the percentage each test; each test had different randomlyachieved and printed out the correct an- drawn questions. swers to incorrectly answered questions. If 0 students obt?..ined the required percentage Student names and identifiers (the first sixthey were able to draw the next test; if not, digits of their student number) werethey could draw another test on the same loaded onto the VAX from student mastermodule. files held on the University's IBM main- frame avoiding the need to enter such dataThe lecturer was able to monitor student manually. progress from reports supplied on demand by the CML system, to respond to individ- On a terminal students were able to: ual messages from students and to give messages to the entire class. draw a test have test marked Discussion of Student Qu,;:.-t:)nnaire examine the learning objectives set/change a password Details of the Student Questionnaire are send a message to the lecturer available in a detailed report of the project; look at their records and status. some issues raised by students which re- quire further elaboration are discussed If students drew a test the computer gener- here. These are: ated one which was then queued for print- ing at one of the two teletype printers and system problems logged the student off the system. If stu- test design and construction problems dents took one of the other options they cheating remained on the system for longer but for a formative versus summative testing. relatively short period. The second surprise from the Project (the The system had 14 VDU terminals and thisfirst being the extremely high level of stu- proved to be generally satisfactory for the dent support for CML) was the relative 9,4:, t) (..;,.. Using Computers Intelligently 373 small number of system problems. Most It is possible under the testing procedures related to printer and paper supply prob-for a student to cheat in the ordinary sense lems; given the demands on the printersof the term i.e. to obtain answers from (over 3500 tests and 3500 results were another student and use them for their printed in six weeks) and the poor quality own. However, the system has safeguards printers used, the problems were wellagainst this. within acceptable bounds however annoy- ing and frustrating they were to the stu- First, each test is unique (calculation ques- dents concerned. The solution to this prob- tions have been randomised so that even if lem is simple - buy better quality printers! two students recei.:e the same question the (This has been done at low cost in 1987 and numbers are different) so that collabora- this area of difficulty has been significantly tion is made difficult. This is particularly reduced). true in the first year of CML when all the questions are new and there is a high level The level of problems with the test designof uncertainty about the correct answer. and construction was also low; some prob- lems were easily fixed - errors in questions Secondly, while there are advantages to and solutions were eliminated as soon ascooperation between students, it remains they became known and some test items true to say that students are in fact compet- were refined in the light of immediate ing for grades so there is an incentive for experience; however, the temptation tobetter students to withhold information. tinker with the testbank was resisted until the CML testbank analysis had sufficient Thirdly, to cheat in this way requires the data. cheater to be able to identify other students with superior and correct information and Some students commented that questionswho are willing to release this knowledge; were "ambiguous"; this may have been empirically we can say that such abilities to true of some questions but most of the discriminate are rare. "ambiguous" questions showed that some students were unable to discriminate be- Fourthly, and perhaps more importantly, it tween finer shades of meaning, closely re- should be understood that the CML system lated concepts and valid and invalid infer- and associated computer held records pro- ences. vide an easy ,neans for the lecturer to main- thin surveillance of student performance A small number of comments related to theand to identify students whose perform- possibility of "cheating" because t1'._ tests ance on the CML tests appears to be incon- were done away from the terminal without sistent with other test results and student supervision. Since the testing was designed characteristics. as formative rather than as summative, students were encouraged (at times re- Fifthly, assessment of the end-of-year re- quired) to consult textbooks, notes andsults suggest that very few students passed references to answer test questions cor- the course by cheating in the CML tests. rectly. As well, some test items generated This issue is being examining closely in the spirited (at times heated) debate in small study of student performance now being groups of students. We think that suchundertaken. activity is entirely appropriate at a univer- sity and consider that it is a desirable fea- The student questionnaire ture of the CML system if iF encouraged such behaviour. QUESTION 1: Before using the CML sys- tem did you have any previous 383 374 Using Computers Intelligently experience with computers? enrolment and comprised the following by degree enrolment: QUESTION 2: What did you think about computer based (immediately Per Cent available) testing? B Econ (incl. B Econ/LLB) 20 B Corn (incl. B Com/LLB) 53 QUESTION 3: What did you think was theB A and other degrees :.8 greatest advantage of the CML system? This is representative of the total enrol- ment. QUESTION 4: Did you experience any frustrations with the CML sys- In answer to Question 5: "Would you like tem ? to see the CML system in the entire ECHO courser 91 per cent of respondents said QUESTION 5: Would you have liked ther in answer to Question 6: "Would you CML system in the entire EC110 like to see the CML system in other course? courses?" 87 per cent of respondents said yes; while in answer to Question 8: "What QUESTION 6: Would you like to see theis your verdict on the CML system? Should CML system in other courses? the Department proceed with this sys- tem?" 96 per cent of respondents said yes. QUESTION 7: Do you have any other comments aboutyourexperienceFrom this evidence we conclude that the with the CML system? student satisfaction with the CML system was at a high level and that student re- QUESTION 8: What is your verdict on the sponse to the system was positive. CML system? Should the Depart- ment proceed with this system? f:rom the general discussion at the re- sponses to the Student Questionnaire we QUESTION 9: Would you like to have more conclude that generally the CML system control over the CML systemconstituted a positive learning experience (e.g. drawing tests whenever you for most students. liked; proceeding through the course as fast as you liked)? The results of the End-of-Year Examination QUESTION 10: Would you please tell me The unadjusted and unsealed results f something about yourself? the unseen end-of- year examinations c. shown in Table 3 below; the examination QUESTION 11: Any other comments? papers were set by each academic respon- Evaluation of CML sible for lecturing each segment of the course but were not necessarily marked by The evaluation of CML was carried out by that person, in fact most of the papers were assessment of the respJnses to the Studentmarked by other members of the Depart- Questionnaire, the results of the End-of-ment. Year Examination and the cost of CML. Grades in the University of Queensland are The responses to the Student Questiontu.ire awarded on a seven point numerical scale from one through seven with a grade of The response to the Student Questionnaire four being a Pass and grades of five, six and was over 75 per cent of the nominal student 364 Using Comp.ders Intelligently 375 seven being superior passes in ascending 1.the results in Part B are consistent with ordergrade of three (usually awarded to previous experience in Keller Plan and papers with a mark just below 50 per cent) PSI cn.arses; is a non-qualifying pass while grades of two and one are unambiguously failures. 2.the results in Part B discriminate clearly The examining conventions at the Univer- between students with passing and fail- sity of Queensland as expressed in the ing grades which we regard as a desir- guidelines issued by the then Professorial able feature in its own right; Board (now the Academic Board) are that grades should approximate the percent- 3.in 1986 this discrimination meant that ages indicated in Table 3. Final grades for very few students would have im- the course are determined by results of proved their adjusted mark in Part B as progressive assessment as well as end-of- a result of the CML tests as can be seen year examinations. by consulting Table 2; Table 3. Results End-of-Year Examination in EC110 the results of Part B may reflect a Haw- Introductory Economics, 1986. thorne effect due to strong student ac- ceptance of CML as disclosed in the Percentage of Students Percentage First '.,per becend Paper results of the Student Questionnaire so Mark Grade Part A Part IP that the superior results in PaB may be 0-40 1 11.5 47.6 14.8 due in part to inducing students to 41-49 2/3 20.4 10.1 4.2 devote more time and effort to the work 50-64 4 43.3 2L3 193 of Part B. 65-74 5 17.6 10.1 18.5 75-84 6 6.7 8.5 24.4 85+ 7 0.7 2.4 18.8 5. given these qualifications and other matters which cannotbe tested, we Segment of the course to which the CML system would conclude that the results in Part applied. P are consistent with the view that stu- dent performance in Part B of the cour se In order to assess the results in Table 3 it is has been improved through implemen- necessary to consider the 'a priori' expecta- tation of CML. tions one would hold in respect of the three components of the end-of-year examina- The Cost of CML tion. Since the segment of the course to which the CML system applied and to The costs of CML fall into the following which the Second Paper Part B results categories.hardware vests, software apply was the last segment of the course costs of establishing the testbank and the and because some knowledge of the seg- operational procedures and material run- ment covered by Part A is considered nec-ning costs. essary as a pre-requisite for Part B, our 'a priori' expectations are that results in Part B Hardware costs would be no better than in Part A or in the First Paper and may be significantly worse. Under the University of Qu...ensiond's re- source allocation and accounting proce- In the light of this expectation, our interpre- dures, the costs, both initial capital and tation of the results in Part B, which are running, are not allocated to operational statistically significantly different fromunits' budget and the hardware is reat,:d those in Part A end the Fir, t Paper, is as as a common resource so it is not post ibl 2 to follows: know how much of the hardware cost is J 376 Using Computers Intelligently properly allocable to this CML project. For approximately $30. These costs have to be the purposes of making cost comparisons compared with the marginal funding avail- we assume that the rental costs of the VAX able to the Department in respect of each for the period of the project were $25,000. additional student; in 1985 this was $2500 per student or over $600 if allocated by the Software costs straight line method to EC110. The all-up cost of the CML package, sup- That CML is cost- effective is emphasised plied by CBTS in 1985, was $18,000. when it is remembered that the establish- ment costs (first year costs) are high, and Costs of establishing the testbank and the second and subsequent year costs are much operational procedures. less than the initial year costs. In addition, in later years, CML can be routinised so that Major costs of the project were the cost of it can be maintained and expanded with establishing the testbank, which Jon Stan- less skilled and dedicated staff than is re- ford estimates as taking at least three weeks quired to establish it in the first place. In full-time work, and the costs of setting up 1987 CML is being used in three first-year the project which look two months ofcourses and one third-year course; the (1!..ts Howard Cook's time. Ensuring the project of expansion in this way are much less than worked meant that both authors had to the establishment costs. forget any ideas of 9 to 5 days, and during the period CML was used by students, both Other issues worked weekends and most nights of the week. CML enables more students to be taught effectively with given resources; it enables Material running costs a saving of resources used in the repetitive areas of marking simple tests and in record- Costs for consumable supplies to keep the ing results and other data; these resources project running (paper and ribbons) werecan be more effectively used in other areas trivial. Total costs were ILss than $500 and of course management and design. CML amounted to an average cost per test gener- also enable students to gain more control ated of approximately 11 cents. This costover the pace at which they progress also included the processes of marking the through the course; if this feature of CML test, issuing results to the student, keeping were expanded we estimate that a signifi- records and providing data for analysis ofcant proportion of students could complete student and question bank performance. the course requirements in less than a semester. Evaluation. In order for these benefits of CML to be Our evaluation of CML suggests that it is captured completely, the University needs cost-effective; if we allocate all costs, except to change certain features of its organisa- for the time of the authors in developing the tion. One horror story will suffice to illus- project, to the first year of the project. wetrate this point. The complete student rec- estimate the average cost per student isords in EC110 were maintained on com- approximately $70. If the cost of the soft- puter in 1986, the University administra- were is amortised over three years, the tion was offered the lists in any one or all of average cost falls to $50; if the cost of the the following ways: hardware is amortised over the expected in hardcopy printed by computer in any life or the hardware, the average cost falls to format; 386 Using Computers Intelligently 377 on floppy disc; progress through the course more on file on the University mainframe quickly and at a higher level of achieve- transferred via Kermit from our floppy ment; or from the VAX. 9. In order to reap these benefits organisa- We were amazed to find that the only for- tional changes will be necessary. mat they would accept was their forms completed by hand! Acknowledgements Conclusions The authors would like to acknowledge the assistance of Dr B.W. Carss, Dean, Faculty From the experience of this project of using of Education and Co-ordinator, CAL Unit, CML for formative assessment in a large University of Queensland, Mr C.W. first-year economics course in 1986, weFarmer, Manager Employee Relations, conclude that: QEC, and Mr F.J. Miller (then with QEC) in arranging the secondment of Mr H.P. Cook 1.It had improved student performanceto the CAL Unit. The Head of the Depart- by as much as 40 per cent on the final ment of Economics, University of Quec ns- examination in 1986 (we can also note land, Dr A G Kenwood, was very suppor- that it significantly reduced the percent-tive of the project. age of students who obtained less than 50 per cent in the examination as com-The particular assistance of Mr M.J. Wood, pared with the results of in 1985); Faculty Programmer, Faculty of Com- merce and Economics, University of 2. Students had positive responses toQueensland, whose contribution in set- CML; ting-up and maintaining the CML hard- ware and software had significant bearing 3. Students found that CML supportedon the success of project. their learning; however, many re- quested increased learner control from Reference CML; Miller F.J., Cook H.P. andClark C.Q. 4. Student motivation in the coarse in- (1986). The Use of Computers in the Main- creased; tenance of Workforce Competence, Electric Energy Conference 1986, Institution of 5. CML was cost-effective; Engineers Australia, Brisbane, 257-258. 6. The costs of establishing CML are rela- tively high; but CML offers significant Jon D. Stanford, Senior Lecturer, De- cost advantages in later years of use; partment of Economics, University of Queensland. Major Research interests 7. A major cost of establishing CMI is the are Monetary Economics, Economic use of skilled and dedicated staff time; Policy, Public Sector Economics and Economics Education. In the area of 8. CML can improve productivity in Economics Education previous proj- teaching a large first year course in two ects include use of the media in teach- ways: through allowing academic staff ing economics, course and instruc- to manage more students with given tional design, course design and evalu- resources and in enabling students to 387 378 Using Computers Intelligently ation in distance education. Current projects include a study of student performance in first year economics and the application of CML in teaching Monetary Economics at third year level. Previous work in Economics Education has been published in Teaching at a Distance, The British Journal of Educational Technology, ASPESA Newsletter and in 0. Zuber- Skerritt (ed), Video in Higher Educa- tion, London, Kogan Page, 1983. Howard P. Cook, Manager-Projects, Queensland Centre for Computer Managed Learning (Q-CML), Queens- land Institute of Technology. Research interests include the use of CML to manage the learning process and the application of resource based learning packages to accommodate individual learning needs particularly within the industrial/commercial environment. Previous CML experience includes co- researcher of a National Training Council project in CML in a large de- centralised industrial organisation. Current activities include the develop- ment of a commercial CML system operating on PCs. Computers in tertiary education in the northern territory A review Jean Stevens Department of Computer Science Darwin Institute of Technology IMIIMINIIIMIW MelllN AIMMAMIMML71110, There is a general lack of knowledge in the other Territory. The primary teaching facilities of States of Australia about current developments the Institute are provided by a VAX Cluster in the Northern Territory. Those who visit the Northern Territory are surprised to find the and, in addition, many exciting develop- pervasiveness of computing throughout all lev- ments have taken place with a number of els of education. Computing technology has joint educational facilities being set up been embraced by a population characterised by isolation from the rest of Australia. between DIT, government departments and other organisations. These include the This paper reviews the facilities,availability establishment of an Information Technol- and use cf computers in tertiary education both ogy Training Centre, a Computer Educa- on campus at Darwin Institute of Technology, the University College of the Northern Terri- tion Centre and a research and develop- tory, and the Alice Springs College of Technical ment group, the Special Interest Group - and Further Education, and off campus through Computer Oriented Nurses (SIGCON), the Northern Territory Opel College and the specifically concerned with the relation- Northern Territory External Studies Centre. ship between computers and the nursing profession. The Northern Territory has a population of 160 000 people who are isolated from the Information Training Technology rest of Australia. In spite of its small popu- Centre (ITTC) la tion and the huge area it covers, cc raput- ers have become widespread throughout The Information Technology Training all levels of education. The Northern Terri-Centre (11TC) is a joint facility provided by tory has five organisations which provide Darwin Institute of Technology and the Advanced Education, Technical and Fur-Northern Territory Treasury, Division of ther Education and facilities for externalComputing and Information Technology studies in computing.Government de- (NCOM) which is responsible for provid- partments are closely associated with these ing suitable computer training courses for activities, and in spite of the difficultiesall Northern Territory Government em- imposed by isolation,there is a healthyployees. The Centre provides up-tc-date interest in research and post-graduate IBM equipment for use by Institute lectur- studies. and Computing in the Northern ers and students and training in microcom- Territory is very much alive and well. puter and mainframe applications for Public Sector employees. Darwin Institute of Technology (DIT), originaily established as tl,e Darwin Com-The centre is situated in the Faculty of munity College in 1974, provides the cen- Business and consists of a Mainframe Ter- tral focus for a large proportionof theminal Laboratory and a Personal Com- nany computing activities in the Northern puter Laboratory. The Personal Computer 389 380 Using Computers Intelligently Laboratory is equipped with 10 XTs, each brary in conjunction with the Education with single disk drives and a hard disk, and Department's Library Services; the Mainframe Laboratory has 12 telex ter- evaluation of new technology and ma- minals connected to the Northern Territory chines to determine their applicability Government IBM 3090 computer. All of the in schools; courses offered in the Centre concentrate advisory visits to schools and regions on giving "hands on" experience with the to give curriculum and systems advice relevant system or package. and inservice training; coordination and support for regional The courses utilising the 3090 are mainly education centres; for Public Sector employees, who use the information dissemination through Government networking facilities.The publications; courses offered cover a diverse range of liaison with national project teams for materials including EASYTRIEVE+ and development of the uses of computer SPSSX.Courses utilising the Personal technology as an educational resource; Computer Laboratory cover material such resources for award and non-award as SYMPHONY and SMART and also a computing courses offered by D.I.T. SMARTLINK course, which is designed for and mainframe to microcomputer communica- fund-raising courses. tions in the GAS system. These courses are usually one or two days in duration andThe Computer Education Centre operates form a sequence which are taught by eitheras a functional unit of the Division of DIT lecturing staff or by NCOM staff. DIT Computing and Mathematics and is situ- computing students have access to theated in the Faulty of Applied Science. It ITTC laboratories where appropriate. consists of an Apple lie Teaching Labora- tory, an Apple MacIntosh Teaching Labo- The Computer Education Centre ratory and a Student Access Room. In addition, the Centre houses a range of per- The Computer Education Centre is a joint sonal computers, such as the DEC Micro- facility provided by Darwin Institute ofcomputer and the Commodore 64 which Technology and the Northern Territoryare available in a number of Territory Department of Education.The CentreSchools, together with a number of otter, coordinates the use of computers in educa-machines under review. tion in all Northern Territory schools and is responsible for providing the followingThe Centre is currently staffed by two services to the Education Department, Senior Education Officers from the Depact- ment of Education, who are attached to D.I.T and the community: D.I.T. as members of the academic staff, and one lecturer from the Division of inservice training in computer use andComputing and Mathematics. Lecturers curriculum applications; from the Department of Computer Science pre-service training in computer Ise work closely with the Centre and utilise the and curriculum applications; facilities on a regular basis for teaching post-graduate studies in educational purpoE es. computing; training for all members of the school community, including parents, in theSIGCON : The Special Interest Group educational uses of computers; Computer Oriented Nurses. software evaluation; The impetus for the formation of SIGCON maintenance of a software lending li-came from a group of Registered Nurses 390 Using Computers Intelligently 381 who had completed a computing unit as ment. ASCOT also runs seminars and part of a Diploma of Applied Science. The workshops for outside users such as Gov- interest expressed in,and the lack ofernment departments and fee paying knowledge of, the computer as a tool for courses. Modems are provided for access Health Care led to some initial research to other tertiary centres for external study being undertaken in this area which has purposes. now expanded into the areas of knowl- edge-based systems and applications ofThe Northern Territory External Studies intelligent CAL systems. Practising Centre situated at DIT coordinates most of nurses, lecturers in Nurse Education andthe interstate tertiary studies undertaken lecturers in Computer Science comprise theby Northern Territory residents, including r re membership of SIGCON. the activities of a large number of comput- ing students. The Centre provides tutors Health care practitioners in the Northern for many of these students and is currently Territory are faced with unique problemsco-ordinating the activities of a group of that require specialised training, particu- post-graduate computing students. larly in areas such as Aboriginal Health and Community Health Care in remote areas. The newly established Northern Territory Members of the nursing profession regard Open College is designed to provide post- the computer as a tool that could aid suchsecondary courses for isolated residents training through CAL techniques and the and uses the computing facilities of rural use of knowledge-based systems and they schools to provides a number of short feel that participation in the development courses in computer studies out of school of such tools will enhance their p;ofes- hours. The Open College has two centres in sional standing. the Darwin Rural Region and other centres in Katherine, Tennant Creek, Alice Springs The group operates under the auspices ofand Nhulunbuy. DIT and the Northern Territory Branch of the Australian Computer Society.It hasThe University College of The Northern access to the facilities of the Department ofTerritory (UCNT) has only been iecently Nursing and the Computer Educationestablished.It is situated in Darwin and Centre at DIT and is currently negotiatingcurrently has a MicroVAX II and three with Royal Darwin Hospital for research teaching laboratories. At the moment these facilities. In addition to the research activi- are used only for one semester introduc- ties SIGCON publishes a regular newslet- tory courses. ter and has run several workshops and seminars for its members. Although many advances have been mach in computing throughout the Northern Other Teaching Facilities in the Territory, there is still potential for future development.It is anticipated that the Northern Territory School of the Air and The Northern Terri- tory Secondary Correspondence School The Alice Springs College of Technical and will provide facilities for remote users of Further Education (ASCOT) has three computers. A link between Darwin Insti- computer laboratories, one based on the tute of Technology and the University Burroughs B26 minicomputer system andCollege of the Northern Territory may be the other two based on personal comput-established to fully utilise and extend the ers. These are utilised by award coursesfacilities of both institutions and the Com- offered in Business,Secretarial Studies, puter Education Centre intends to extend Management and Hospitality Manage- itsservices tothe regional euucation 3 382 Using Computers Intelligently centres through communication systems such as Viatel. The pervasiveness of com- The author has taught in a variety of puting throughout all levels of education institutions and countries and been will become even more evident in the fu- Involved in computer education for ture. over twenty years. The current posi- tion held by the author is that of Lec- Appendix: Computing facilities turer in Computer Science specialising at Darwin Institute of Technology in Technical Systems Analysis, Project Management and the Applications / On 30th April 1987, Darwin Institute of User Interface. She is the convener and Technology had approximately 1500 stu- technical advisor to the Special Interest dents enrolled in Adv,inced Education Group, Computer Oriented Nurses (degree or diploma level) courses and 3500 (SIGCON) which operates in the students enrolled in Technical and Further Northern Territory. This group is Education (TAFE) courses. In addition to involved in computer education and students enrolled in award courses 3500 research into the use and development students attend fee paying courses at the of CAL systems related to the nursing Institute every year. Darwin has a popula- profession.The author is currently tion of 72000 so that the educational needs developing an expert system which of nearly 12 percent of the population are will enable nurses to gain advice on met by the Institute. areas of nursing specificly relevant to the tropics. This syste:r should be of Some 350 students enrolled in award great assistance to both nursing train- courses have accounts on the main frame ers and nursing staff in isolated areas. system and a similar number utilise the The knowledge required of nurses is various micro-computer facilities. much mere general than that required of doctors, although not as in-depth. A Darwin Institute of Technology has a major component in the development VAX/780, a VAX/750 and a micro-Vax of the expert system is in discovering which are cluster4 together and are util- ways of extracting knowledge from ised by staff and students. people with no technical background in computing and in developing ways to access iarge amounts of general in- formation in a reasonable time. There- fore the research falls firmly in the areas of knowledge engineering, knowledge representation and pattern recognition. 3 ti9 Computer assisted examination for fundamental programming courses Tao Li, Department of Computer Science, The University of Adelaide Luyuan Fang and Larry Christensen, Dept. of Computer Science, Brigham Young University, Chongsan Yu, Dept. of Electrical and Computer Engineering, University of Wollongong In this paper, we present several types of ques- can be error free and the grading can be tions which are suitable for computer assisted completely computerized.Since we are examination of fundamental programming courses. Computer implementation and analy- discussing exams for computer program- sis of the questions are also addressed here. ming courses, there are also certain ques- Human factors in designing the examination tions which are natural to computer exam questions are also considered. The choice of but not paper test. questions and the appropriateness of imple- mentation are discussed. The analysis and argu- ment presented here can justify our design and However, not all examinations can be unplemen ta bon . completely computerized in our current technology.Exam questions involving natural language descriptions are espe- It has always been a hassle for students and cially difficult to automate. Therefore, we teachers at the end of ear c.rm or semester. must choose several types of questions Students are busy at preparing for exami- which cover a large variety and are easily nations and the teachers a rebusy at making computerized. The simplest are the yes/no and grading examinations.Very oftentype of questions and multiple choice students have to come to a specified class- questions.These questions have been suc- room at a scheduled time to take the examcessfully used in TOEFL and GRE. But and the teachers must be present at the these questions are not sufficient for funda- exam. After the exam, teachers and graders mental programming courses. Examinees must spend a large amount of time to de-may get a certain number of points simply sign a grading scheme and to grade theby guessing and making random choices. exam papers. This has been quite a labori-These questions are also tosimplistic to ous and error prone job. Students oftenprevent cheating and plagiarism if the come to complain about their grades afterexam is to be taken by many people at the exam papers were graded. This raised different times. our interests in implementing a computer- assisted examination system to deal withIn this paper, we consider several types of above problems. questions besides the yes/no type and the multiple choice type. We will also analyze The aims of computer-assisted examina- the scoring probability of the questions in tion are to give prompt processing of exam each type. The advantages and disadvan- data for administrations and to reduce thetages of each type will be considered. labour and errors for grading exam papers. Examples of each type will be given. Computer-assisted examinations can pro- Computer implementation of the exam vide flexible time and place for examina- questions will be discussed. tion. In addition, properly designed exams ::i i) ,e5 384 Using Computers Intelligently Types of Examination Questions an understanding of the problem, the ex- aminee has 1/n chance of gettinga correct In this section, we describe several differ-answer. Since the examinees normally scan ent types of exam questions. The scoringfrom the first answer to the last answer, it probability of each type is also analyzed.would be advisable to place the correct An example will be given for each type. Foranswer for each question in the last few completeness, we also include the yes/noanswers. Following is an example. type and the multiple choice type. When we describe the types of exam questions, Question 2. A recursive function isa we also consider some 'human factors', subprogram which which measure the effectiveness of the questions while being used against stu- A. has multiple parameters dents. B. can always be converted into it- erative loops YES/NO Type Questions C. calls itself indefinitely D. can be implemented using stacks A question in this type requires a simple answer, either a YES or a NO. Even without Only one answer is correct in thiscase. an understanding of the problem, an exam-With a random choice a student has 25% inee has a 50% chance of getting a correctchance of getting the light answer. This is answer. Following is an example questionbetter than the yes/no questions.The of this type. analysis is as follows. Question 1.In a digital computer, (Character) Multiple choice questions information is represented by a se- are still simple and easy to grade. They quence of tones with continuous fre- are also useful for testing basic under- quency increments? standing. But this type of questions is trickier than yes/no type questions. Answer: yes( )/ no( ) (Prob and HF) A prol-ability of 1/n is Analysis for yes/no questions is given be- still susceptible to cheating and the low. From here on we will use 'character' answer to a specific question is easily for characteristics and 'prob and HF' for memorised Ind passed onto another probability and human factor. person. (Character) This type of questions isEven though the yes/no type and the simple and easy to grade.It is verymultiple choice type have certain disad- useful to test student understanding ofvantages, they still prevail in exams. This is basic concepts. due to the simplicity of such types of ques- tions. (Prob and HF) The 50% probability makes it susceptible to cheating andMultiple Match Questions pure guessing. This type of questions usually contains a Multiple Choice Questions description and two lists of items, each item in one list must be matched toan item in the An answer to this type question usuallyother list. Sometimes we may arrange to requires the examinee to choose one from ahave more than two lists for matching. set of n answers provided to him. WithoutSuppose that we have two lists, each of Using Computers Intelligently 385 which has n items. If the examinee is totally THE DIRECT OUTPUT TYPE. In this type ignorant about the question, the probabil-of questions, a program segment is given. ity p(k,n) for him/her to get k correct an- The student is supposed to trace the execu- swers from a total of n answers is approxi- tion of the segment and to report the output matelyl/(ek!). This is because sequence. Following is an example. p(k,n) = total # matches of k coned and (n - k) wroj_gl total # of matches Question 3. Trace the execution of the and the number of permutations with ex- procedure f(5).Show the output se- actly k matches (PfeifferSchum, 1973) is quence produced by the write state- n-k f$ ment. ( k)(n-k)! i! which is approximately n! function f (n:integer):integer;, var t: integer; n n-k (4) begin )(n-k)! El.() ( k I! )(n -k)! if n <3.= 1 then begin write (1); f := p(k,n) = 1 end rd else begin t n*f (n-1) ; write (n); f t end; Apparently, the larger k is, the lower end; probability of getting it right. Following is an analysis for multiple match questions. The student is required to list the sequence of numbers output by the function with a (Character) Multiple match questions parameter 5. A sequence of five numbers is are simple to do and easy to grade.produced by f(5) and the student must Partial grades are given according to the trace the program to produce this se- number of correct matches. quence. Analysis is shown below. (Prob and HF) Probability is 1/(ek!) (Character) This type of questions may The probability shows that scoring by range from easy to very hard depend- random guess is very difficult. These ing on what are given in the program questions are nct susceptible to cheat- segments. Partial grades can be given ing and plagiarism because memoris- according to the numbers of correct ing so many matches is quite difficult. answers and wrong answers in the output sequence. Some questions can Although muliiple match questions appear be very challenging. similar to multiple choice questions, they have quite different human factors and (Prob and HF) Probability for this type probabilities. relies on many factors and is not easily calculated. But it is obvious thatgetting Program Trace and Selective Types a correct sequence can be quite difficult. It is also hard to memorize the entire One type of exercises, the trace of a pro- sequence and is thus less susceptible to gram run, is particularly helpful for stu- cheating. dents in fundamental programming courses. The student must trace a program THE SELECTIVE CHOICE TYPE. In th:s segment lineby line to find out what are the type of questions, a program segment is output data. There are three typical types given together with an output sequence. of trace questions, the direct output type, The student is supposed to trace the execu- the selective choice type. and the selective tion of the segment and to select the ele- match type. of these types will be ments in the sequence which are produced examined next. by the segment. Following is an example. 3 9 5 386 Using Computers Intelligently Question 4.Which of the following is very hard. Also this type of questions is numbers cannot be produced by the not susceptible to cheating. write statement in the following pro-THE SELECTIVE MATCH TYPE. In this gram segment. type of questions, several program seg- ments are given together with several out- i : 0;j:. 1; put sequences. The student is supposed to for k :-1 to n do (* where n is greatertrace the execution of the segment and to than 1 *) begin select the elements in the sequence which. are produced by the segment. An example temp : i; i:.. j;j : i + temp; write (i) is given below. end; Question 5. Three programs shown below. Output numbers : A table is also given for filling in the an- 1,2,3,4,5,7,8,11,13,37,21,25,30,34 swers. Each row corresponds to a specific value of i (the index variable) and each The student must identify, from the outputcolumn in a row contains the output for one sequence, the numbers which can be pro-of the programs. Fill in the circles in each duced by the write statement in the pro-row with the corresponding program gram segment. The probability p(k,m,n) ofnames (A, B, or C) to indicate the programs getting k correct answers, from a sequence which produce the outputs. of n numbers among which m are correct answers, is program A; program B; program C; it m n-m var i,x,y,zinteger;var i,x,y,z:integer;vatcy,rinqp; Zi.k(k )( IA) begin begin begin p(k,m,n) = for i:=1 to 5 do for i:=1 to 5 do for i:=1 to 5 do n it begin begin begin x:=round x:=trunc 1i = k () x:=trunc(sgrt(i/ (sgr(i/2.0)); (sgr(i /2.0)); 2.0)); y:=x mod 3; y:=x mod 3-1; y =x mod where k m since there are a total of m 3 + 1; correct answers only. The derivation of the z:=xy; z:=xy; z:=y; probability is neglected to save space. end; end; end; Analysis is as follows: end. end. end. (Character) This type of questions also ranges from easy to very hard depending indexxyz xyz xyz on what are given in the program seg- i=1 1 ments. Partial grades can be given accord- 0-1-1 0 1 000 ing the numbers of correct answers and i=2 1 23 1 1 2 101 wrong answers. Grading is easy. i=3 213 22 4 123 i=4 404 41 5 123 (Prob and HF) i=5 6-15 12 3 606 n m n-m The st,..i..tent must identify, on each row, the II.,k(k)(I-k ) p(k,m,n) = program which produced the output in a column. Each row requires a match and the n n probability for getthig kicorrect answers in Ei = k (i) the i-th row is 1/(eki!). Let k denote the total It is bit difficult to comprehend this for- number of correct answer. Then k .5.n X m mula of probability. But it suffices to say where k = ki+ k2 +...+ kn. Since the matching in each row is independent of the ^ther that getting a completely correct sequence rows, for a question with m programs and 396 Using ComputersIntelligently n rows of answers, the probability p(k,m,n) ging syntax errors are relatively easy and of getting k correct answers, is approxi-modern comp:lers normally report very mately detailed error messages. But it is more 1 n 1 difficult to debug semantic errors. Hence, p(k,m,n) it is essential to include such questions in n +irk Al the examinations. An example of debug- ging question is given below. because the number of combinations for getting k correct answers out of n rows is Question 6. A recursive function FIND- equivalent to distribute k nondistinct ob- MIN is shown below. The function tries to ject into n distinct cells (Liu,1968] and the find the minimum value element in an result is (n+"). array X[1..N]. The value of the minimum k element is returned. There are two (seman- In calculating the above probability, we tic) errors in the following program. Check have assumed that the probability for each the program and mark the erroneous state- combination of kt, where En ki= k,is the ments. Also, give the correct replacements SaMe. t for the two statements. type INARRAY = array[1..N) of integer; Following is an analysis of this tyi e of questions. function FINDMIN(var X:INARRAY; Nan teger):integer; (Character) This type of questions are var min-of-rest integer; begin I usually difficult to do. Partial grades Erroneous can be given according to the numbers Statements of correct answers. Grading is easy. (Prot and HF) p(k,m,n) is approxi- if N=1 then FINDMIN := X[11 [1)1 else [2)1 mately 1 n 1 begin l:311 min-of-rest := FINDMIN(X,N-1);[4)1 +kk -1)i =1 e14 if min -of- rest provide a simple editor for on-line editing. References However, there is no easy and fair way to grading.Either a simple no points/full Liu,C.L.(1968), Introduction to Combinatorial points grading is employed or an off-line Mathematics, New York: McGraw-Hill, grading scheme is employed. We have also Chapter 1. implemented this type of questions in our Pfeiffer, P.E. and Schum, D.A. (1973), Intro- computer-aided examination system. duction to Applied Probability, New York: Acader -4c Press, Chapter 2. Implementation The types of examination questions and Dr. Tao Li is a lecturer of Department of some other types of questions have been Computer Science, University of Ade- implemented on a WICAT graphics work- laide, where his current research in- station. To improve human-computer in- cludes Artificial Intelligence, Expert teraction and help the student during the Systems as well Parallel Processing. exam, we have included many features to simplify the use of the system and to make Mrs. Luyuan Fang is now at Depart- it more user-friendly. ment of Computer Science, Flinders University of South Australia We have taken fully advantage of the graphics facility of the WICAT system. Mr. Chongsan Yu is a visiting associate Along with each exam question, there are professor of Department of Electrical also various instructions for using the facil- and Computer Engineering, Univer- ity displayed. The user can use the system sity of Wollongong. His research inter- without referencing any manual or guide. ests include Expert systems on process This provides a suitable environment for control, Computer Networks and Data examination. A few screen imagines are base management systems. shown in Figures I through 4. Summary A set of questions suitable for computer- aided examination are presented. These questions are also analysed to show their appropriateness for computer implemen- tation. The human factor consideration and the probability for anti-plagiarism are also discussed. These questions have been implemented on a graphics workstation. Such a system has demonstrated many advantages. It is e step toward a full automated system for computer-aided education. We are looking at other types of questions to be incorpo- rated in our existing system. Some intelli- gent tools will considered for future exten- sion in this area. 398 IIMIONYMINNI.111=11111111. Using' Computers Intelligently 389 QUESTION #1 I 0; 1.Which of the following numbers cannot 1; be produced by the "writein" statement in for k 1 tondo this program? begin temp I; Please mark the numbers that cannot be i 1; produced. j I + temp; writeln(1); RETURN end; MOVE MARK BACK 2 3 1-5-1 8 11 I 13 [17 21 25 30 34 coNE1 2. What is the name of the series' of numbers this program generates? Figure 1: Screen image of a multiple selection question QUESTION #1 0; 1.Which of the following numbers cannot 1: -1; be produced by the "writeln" statement in for k 1 tondo this program? begin temp I; Please mark the numbers that cannot be i j; produced. j + temp; writeln(i); RETURN end; MOVE MARK BACK 4 8 Egj 1 7 I 13 17 Mg Pr 130 F341 LDCNE 2. What is the name of the series of numbers this program generates? Figure 2: After some choices are selected. 399 390 Using Computers Intelligently QUESTION #2 For each concept in the left set, find a best match one from the right set, and put the relative number In the proper blank. 1 BINARY VALUED 1. Translate & excute program 1 ISOFTWAREs0 r2. CPU 3. Top down decomposition IINTERPRErER0 14. Sequence of steps which produce a result I I=PM DEVICLID ri7S:';Specify Instructions I 6. Printer 1 7. Translate programI 8. Digits 0-9 1 ALGRITHM0 9. Pascal programI i m p I10. Psuedo-code I 11. 0's & 1's1 SENEONDARYSTOR 0 12. Floppy disks DICIE 113. Carry Instructions L14. Keyboard I Figure 3: Screen image of a multile match question QUESTION #2 For each concept in the left set, find a best match one from the right set, and put the relative number in the proper blank. rBINARY VALUE i. Translate & excute program IsOFTWARF_S I2.CPU] 13. Top down decompositior1-1 Iitfill9PREIEFOO I4. Sequence of steps which produce a result 1 I OUTPUT DEVIL= 5. Specify instructions I 6.Printer I HARDWARES 1 7. Translate program I I8. Digits 0-9 1 LALGRITHM10 19. Pascal programI COMPILER TO. Ps uedo-code j I12. Floppy disksI ISENCONDARY STORAGE0 Cpl 13. Carry Instructions I 14. Keyboard Fgiure 4: After several matches were made 400 1 Computer Assisted Learning at the University of Otago Marian Vlugter, CAL Consultant / Programmer Computing Services Centre, University of Otago NM The University of Otago has made a commit- years. We will then look at the factors that ment to the development and use of CAL tech- influenced this and at the situation at the niques in the University. It is the first university in New Zealand where positions have been present time. Finally, we will look at some created with specific responsibilities for Com- of the directions and plans for the future. puter Aided Learning: one in the educational area attached to the Higher Education Devel- opment Centre and one in the technical area CAL then and now attached to the Computing Services Centre. This paper will give a historical overview of In 1985 six departments had made a firm CAL at Otago, describe the current situation and outline the changes that are taking place. commitment to using CAL with classes (see Table 1). The Commerce Department used a computer laboratory to teach their stu- Over the last two years CAL at Otago Uni-dents about the use of programs in the versity has changed. In fact, some greatbusiness world. Teaching laboratories strides have been made and CAL at our weie being set up for use in Psychology a nd university could be said to be growing up Computer Science. Several departments fast. This may be attributed to three factors: were either evaluating the possibilities of financial support CAL or ready for development work. Six establishment of advisory services departments received a microcomputer for 6growing popularity of the Macintosh teaching development purposes. Four of computer these were not yet involved in using CAL. The Anatomy Department was carrying on We will first look briefly at the growth ofan experiment (started by the Orthopae- interest in CAL over the last two or three dics Department) in the evaluation of inter- active videotape. Table 1.1985 Department Computer Type of use Ariacatlictics Apple n 3imuiaiions Clinical Biochemistry VAX problem solving exercises Economics Apple II simulations Education Apple II teaching about CAL Mathematics Apple II demonstrations and simulations Microbiology Apple II laboratory experiments Pharmacology Apple II laboratory experiments and demonstrations 40j ,., 392 Using Computers Intelligently By 1987 sixteen departments are usingFinancial support some form of CAL in classes (see Table 2). It is interesting to note that of the 9 newBecause it was noticed that there was an users, 6 use the Macintosh and that 2 of theincrease in applications which expressed older established users have switched to oran interest in developing Computer As- have added the Macintosh. The Physiologysisted Learning units and which requested Department has replaced their oscillo-computer equipment for use in teaching, a scopes with a laboratory full of Macin-working party was formed in 1984 in order toshes, using locally developed, very so- to obtain an overview and evaluation of phisticated, software. There are plans tocurrent and possible future activities in extend the use of the Macintoshes fromCAL and to advise on the implications for laboratory tools to incorporate CAL exer-funding policy. cises as well. There are similar plans in the Physics Department. It is expected that byThe working party visited 20 departments this time next year another eight or tenand groups within the University of Otago departments will be using CAL applica-and reported in June 1985 that there were tions and that those who started off care-sufficient interesting developments of a fully and convent'-,naIiy will have ven-practical sort to warrant a modest increase tured out in more diversified use of CAL. in support for CAL in the University. Three The reason for this growth should belevels of support were suggested: mainly attributed to the three factors men- tioned above and we shall now look at each expert advice and information in the of these factors individually. early stages Table 2. 1987 Department Computer Type of use Anaesthetics Apple 11 simulations Anatomy Macintosh drill & simulation Clinical Biochemistry VAX problem solving exercises Dentistry Macintosh cinll Economics Macintosh simulations Education Apple 11 teaching about CAL French VAX drill Geology Macintosh numerical analysis simulations (spreadsheet) Mathematics Apple II & demonstrations and Macintosh simulations Medicine (Wellington) VAX drill Microbiology Apple 11 (more units) laboratory experiments Pharmacology Apple 11 laboratory experiments and demonstrations Physics Macintosh simulations & concept learning Russian Macintosh chill Textiles IBM drill and tutorial Zoology Macintosh drill and simulations 452 . ,-,,,,,,la.,---,...mc.smx Using Computers Intelligently 393 practical support (hardware, program- cal area attached to the Computing Serv- ming assistance) when ready to proceed ices Centre. The major objectives, identi- establishment of CAL units (consisting fied by the CAL sub-committee, were: of 5 -10 machines) when ready for class use fostering of communication between departments with an interest in CAL The wot king party also made the following ensuring the educational soundness of recommendations: the design of CAL applications encouraging the efficiency of software the appointment of an educational design adviser and a computer programmer to give advice and assistance in CAL Fostering of communication provision of some out-of-hours CAL computing facilities for students In the early stages many departments or a one day exposition of CAL to fostersolitary persons in a department had communication among people inter- worked in isolation without being aware of ested in CAL what others were doing. The first CAL exposition in 1985 soon made it clear that in It is remarkable that each one of these sug-some cases efforts had been duplicated, gestions and recommendations was ac-been needlessly laborious or that some cepted. An exposition was organised people had been 'reinventing the wheel'. immediately in the winter of 1985. For many people this was the first occasion to To overcome this problem it was decided to become aware of CAL activities at the uni- form a "Software Society". This group versity and many of those working withmeets several times a year to discuss areas CAL became aware of the others for theof CAL use, design and development. first time. Out-of-hours computing facili- Meetings may consist of demonstrations, ties for CAL were offered by adding spe- presentations, discussion or the exchange cific machines to the microcomputer labo-of technical (programming) information. ratory in the Computing Services Building.Seminars and lectures publicised among The appointments of an educational ad- the wider university community, aimed viser and a programmer were approved. both at regular users of CAL and at those The working party became a permanentpeople who are still in the information sub-committee which now makes recom-gathering stage, are also given several mendations on all applications relating to times a year. computer support for CAL. Because it was considered impossible for the sub-commit- The consultant/programmer attached to tee to address all educational uses of the the Computing Services Centre acts as a computer, it decided at an early stage to central advisory service and clearing house concentrate on one specific application:of CAL related activities and information. Computer Assisted Learning. Anyone seeking information or advice regarding CAL will in the first instance Establishment of Advisory Services contact one of the advisers. Depending on the stage of development, it will be the task In the winter of 1986 the two CAL advisers of either the educational adviser or of the commenced work, one in the educationalprogrammer (and sometimes of both) to area attached to the Higher Educationkeep in touch with and offer assistance to Development Centre and one in the techni- the department in question. 403 394 Using Computers Intelligently Educational soundness of CAL applications itself does not have to be. As a result less software has been developed than had It is primarily the task of the educationalist been hoped, because the time and effort to advise people on the educational impli- required from the staff was not avail- cations of CAL, on the effects it may have able or was not seen to be sufficiently on the curriculum, on the various ways worthwhile. CAL could be approached educationally, on the educational aspects of software one-off applications: are they cost effec- design and on the evaluation of CAL use. tive? People are becoming more aware Where a department is contemplating or of the costs involved in CAL, both in planning the introduction of CAL for the hardware and in person time. Since the first time, the educationalist and the pro- benefits are only rarely or sporadically gramming consultant will work together evaluated, it is hard to justify the invest- with representatives of the department, ment. On a personal level, teaching staff locate and evaluate or design and program can always be assured of the results of the software to be used, offer advice on the time invested in research, not of time financial implications, where feasible set invested in unevaluated attempts to up a pilot study and evaluate the project. improve teaching. More efficient software design The Macintosh: a friendly machine, but difficult to program. Since its appear- The earliest programs used in CAL were ance, the Macintosh has been seen as an often programmed locally by the teaching ideal computer foi CAL and at Otago staff, generally written in BASIC, usually the use of this computer in the CAL area run on the Apple II computer. In most is overtaking the use of others. In the cases, those involved with the program- next section we will look at the reasons ming felt that the task was too time con- for this and the consequences for the suming and too distracting from teaching CAL programmer. and research. It was hoped that the ap- pointment of a professional programmerGrowing popularity of the Macintosh would make it possible to develop morecomputer software more efficiently. The main reason for the great popularity of The results, a year later, may in some waysthe Macintosh lies in the very carefully be considered extremely encouraging, in designed user interface: "The Macintosh is other ways disappointing. In 1987 there are designed to appear to an audience of nine more departments using CAL exer- nonprogrammers, including people who cises. However, seven of these almost ex- have previously feared and distrusted clusively use simple drill and practice exer- computers." (from The Macintosh User cises. There are some very specific reasons Interface Guidelines, Inside Macintosh, for this: page 1-27). In general, developers of Macin- tosh programs follow the guidelines given software design: a time consuming task.byAnnla anti }ha racillt is rnric;ctorus, cn The most vital part of the development that users do not have to learn a completely of a CAL application is the design of thenew interface for each application. The software and the integration of the uselearning time for using the Macintosh is of it in the curriculum. This is a task that very brief, some say non-existent, and this has to be performed by the teaching fs particularly important in CAL were large staff, even though the programmingnumbers of mostly naive computer users 404 Using Computers Intelligently 395 have to become familiar with a programOne of the problems with existing,ady- they may use only two or three times. made programs is that they are seldom just right. The most often heard comment is: However, the Macintosh is considered one "Yes, it is a really nice program, but.... and of the hardest machines to program and if only...". For this reason the most effective even for an experienced person, learningprograms are authoring languages or au- how to program 'the Macintosh way' may thoring tools. Since authoring languages take three months or more, assuming one is are the easiest to use and require the least already fully conversant with program-input from the teaching staff (sometimes ming itself. This fact does not add to the nothing more is done than the typing in of cost effectiveness or ease of development ofold exam questions by a typist!) they tend CAL programs! At present four depart- to be a first choice. ments are developing applications for the Macintosh and Macintosh expertise isHowever, more and more suitable soft- gradually building up. ware tools are coming on the market and potential CAL users are becoming more On the plus side is the fact that in the Unitedaware of the possible uses of tools such as States the Macintosh has been adopted by spreadsheets, data bases, modellers. Com- many universities as the ideal machine forments have been heard from one or two use with students and a steady trickle of departments, which at present mainly use software suitable for CAL purposes hasthe computer for drill exercises, that the resulted from it. At the same time an in-limitations of drill exercises are becoming creasing number of very imaginative pro- irksome. Now being totally familiar with grams is becoming available, which are the computer and some of its potential, suitable for use as authoring or softwarethere is a strong tendency to branch out. tools for CAL units. Three or four departments are using spreadsheets on the Macintosh (Excell, Another approach, and one followed con- Multiplan, Jazz) for CAL exercises and sistently by the Mathematics Department, more departments are at present experi- is to use standard modules and to developmenting. The program STELLA, a model- ?ful modules and development toolsling program for the Macintosh, has been first, as building blocks for later specific slow in taking off, but is generating a grow- developments. A modular programminging interest as a simulation and modelling language such as l'ascal is most suitable totool. Animation and pictorial databases are such an approach and since Pascal is the other directions currently under investiga- 'native language' of the Macintosh, mod-tion. ules are developed in that language. A large range of modules, some most ecifi-There is a policy to buy software suitable cally useful for mathematical manipula- for CAL centrally, so that it can be evalu- tion, has now been developed. ated by the CAL advisers. Generally, new programs are announced in the Comput- The present situation ing Services Centre Newsletter, so that departments may request demonstrations Asa result of thc factors oed above, theor a loan of the program for evaluation CAL programmer/consultant has spent before deciding to buy it. Those known to less time actually writing application pro-be interested in a certain program are per- grams and more time trying to locate andsonally contacted. In general there is an encourage the use of existing CAL pro-atmosphere of cooperation and informa- grams, software development tools and tion sharing and the result is that efforts are more efficient design methods. less likely to be duplicated. -*--711111MINIMINNE111 396 Using Computers Intelligently The educational adviser has lectured andwith small groups at first, often second or given workshops on the educational impli- third year students, and then to gradually cations of CAL. He has been encouraging extend the use of CAL, as it is found effec- the careful planning of projects, the use of tive, to the larger classes. The logistics of pilot studies and evaluation of the exercise,this will have to be carefully thought out: using recognised research methods, thus whether it is feasible to accommodate large making it possible to publish the resultsenough computer laboratories in each and to consider this work part of ongoingdepartment using CAL or whether large research. The underlying objectiveof public access laboratories, possibly net- course is to evaluate and possibly improveworked and requiring management, might the effectiveness of CAL and its overall be needed. This question of access to hard- place in the curriculum. ware effectively forms the ceiling of future growth in CAL. It is becoming harder to define CAL, as people become more familiar with the useSome developments that are expected in of microcomputers as tools in everydaythe near future and some of the investiga- activities. When is the user learning and tions currently going on: when is the user merely addressing the computer as a tool? The CAL sub-commit- It is expected that more departments tee may in future be less able to pin down will turn away from authoring lan- CAL and be compelled to broaden its crite- guages and drill exercises and that the ria. It has been suggested to refrain from use of 'application building programs' ( using the abbreviation CAL, and rather such as spreadsheets, modelling pro- refer to "the use of computers in teaching grams, animation, hypertext, data and learning": UCTL. Quite a mouthful bases) will increase. and it has not as yet replaced the easier term CAL. The University will be investigating the use of Interactive Videodisc in CAL. Some future directions Funds have been set aside for this pur- pose. CAL has sometimes during its history been seen as a time saver, a personnel saver, a An area of research is the use of Soft- panacea, or simply the bandwagon to get ware Engineering techniques for the on as fast as possible. Although occasion- design and develcpment of software. ally such thoughts do seem to be enter- tained by some of those showing an interest Discussions are going on for the estab- in CAL at Otago, the centralized and struc- lishment of a pilot study for the use of a tured approach to CAL usually manages to set of computer writing tools for an identify and to dispel them. The growing English writing skills course interest in CAL is therefore in general ac- companied by a careful consideration of all Because of the benefits seen in a central the factors involved, educational, technical point of information it is cons defied and financial. important to set up a national contact group for tertiary CAL in New Zealand. The last factor especially will be the decid- ing one in whether or not the current trend Summary of a steady growth in the number of depart- ments using CAL will continue. There is a In the early days of CAL at the University of tendency to develop and evaluate CAL Otago departments were working in isola- 406 Using Computers Intelligently 397 tion, often on an ad hoc basis. Most pro- grams were developed locally, written in BASIC for the Apple II. Because of a central- ized approach, starting off with a univer- sity wide investigation and followed by both financial and advisory assistance to departments wishing to use or develop CAL, the interest in CAL has increased steadily and the approach to CAL has matured, become more structured. The three levels of support recommended by the CAL sub-committee have been oper- ating successfully: initial advise and analy- sis of the needs followed by allocation of a development machine together with pro- grarrurung and educational advice and assistance and finally making available a set of microcomputers, allowing the intro- duction of class use. The writing of stand-alone applications has decreased. The use of ready made pro- grams or programming tools has in- creased. The approach to programming has become a modular on: and Pascal is the main language used. The use of Software Engineering methods in program develop- ment is an area of research. Gradually the Macintosh 114 is becoming the preferred CAL machine, because of its 'user friendli- ness'. A department contemplating the use of CAL is carefully accompanied in its inves- tigations with advice. Where possible a pilot study is set up and increasingly the introduction of CAL will be evaluated and the results used to improve the effective- ness and the appropriateness of it in the everall curriculum siruciure. 407 A Computer Based Testing and Evaluation System Bill Chia: Computer Centre and Mon Khamis: School of Education Nepean College of Advanced Education The Design and delivery of many Tertiary could support and enhance the teaching of Education courses are based predominantly on traditional presentation methods using lec- these courses. Computer based system,are tures, seminars tutorials and reading materials. one of the possible range of techniques. However, there are other techniques available to support these courses. Over the last few years, there has been A computer based testing and evaluation sys- considerable discussion and consideration tem has been developed on models which can given to methods of development of incorporate techniques such as diagnostic feed- computer based teaching and learning back, decentralised learning, self testing, moni- materials. Some have argued for retaining toring or progress and mastery :earning. The the flexibility of General Purpose computer system can provide reports to support summa- tive evaluation, test analysis and item analysis. languages such as Pascal, while others Furthermore, the data can be presented in a have pointed to advantages of specialised form that supports ongoing formative evalu- authoring languages. Yet it has also been ation of the course on which tha test is a part. pointed out that even the authoring lan- The approach adopted in developing this com- guages provide too little in the way of puter bases system has been to make it straight- higher level constructs (Merrill 1985). forward and powerful enough to be used by Authoring languages may be valuable to teaching staff who may have Limited experience with computers, yet sufficiently general and authors if they lack programming skills flexible to meet the needs of different subject and have too little time to gain them. Sys- areas or different teacher's goals. Within the tems involving higher level constructs confines of the system, the individual student would be helpful to them and to any who can be given some degree of control. The system can be adjusted to give guidance to the student have limited experience in instructional on the basis of performance.It may readily be design. expanded to allow other facilities to be incorpo- rated. Authoring systems, as distinct from au- The system was developed with the aid of the thoring languages,may provide some PLANIT authoring language and has been suitable models, though they are often only applied to subjects in the Schools of Business, Nursing and Health Studies, and Education. available on specialised hardware. The The paper describes the system and its applica- majority of academics in tertiary education tion to a course in Research Methods in the are likely to find that their needs aren't met c-.11-c! cf .tNerser.A F !mplica- very well by the materials they can readily lions for the design of courses at the Tertiary level are discussed. obtain. Commonly, the materials differ in emphasis from that desired and often re- quire a particular brand of hardware which Many Tertiary Education courses are basedmay not be available and cannot be justified predominantly on traditional presentationon the basis of needs of one part of one methods using lectures, seminars tutorialscourse. Most would be daunted by the and reading materials despite the fact thatprospect of doing their own programming there are other techniques available whichand would not have the time anyway. 4 G 8 Using Computers Intelligently 399 At Nepean C.A.E. a system has been devel-The system has been applied to a range of oped around an existing Authoring Pack-subject areas. It has been used for Behav- age (PLANIT) to provide simple-to-useioural Science, Management Studies and computer based assessment and evalu-Auditing in the School of Business; Meas- ation. The resultant system has a numberurement and Evaluation, Research Meth- of the features that a range of academic staffods, Educational Psychology in the School would find useful. With these features theyof Education; Human Growth & Develop- can get the benefits of several types ofment and Behavioural & Social Science in computer based testing without the need the School of Nursing and Health Studies. to learn a programming language or an A set of tests in biological science are cur- authoring language. rently being developed. The system makes use of options which can be selected by the Academic user and Description of the System. applied to their own course materials. The individual ideas in the system are rela-Test items, in the form of multiple choice tively simple, but they help to bridge thequestions, are gathered and arranged into gap between what a lecturer may wish to groups on the basis of subject matter, diffi- achieve with the aid of computer, and theculty level or some other criteria. Thee level of practical programming knowledgequestions become the basis of a kind of "on- orskills which he or she may currentlyline" test bank. From this batik of items, possess. This system allows easy develop-questions are selected according to a pat- ment of some computer based assessmenttern as decided upon by the academic user. and evaluation procedures for enhancing The items are made available to the student learning. It can be used as part of ongoingvia a centralised computer. The student assessment or for final assessment. can take the test at his or her own time (within a predetermined period, at selected Lack of programming skills may be onetimes as decided by the academic). The reason that academics might avoid com- academic can retain much of the control, or puter based support for their students. make it available to the student as he or she Limited knowledge of instructional strate- thinks fit.Options such as "immediate gies and instructional design may be an- feedback", "progress reports", " printed other. The system we have developed al-report or certificate" can be decided upon lows the combination of various optional by the academic. A most important aspect features to produce a number of coherentassociated with assessment and reporting testing models. The options may be set relates to its use for "measurement of prog- specifically or patterns of options may beress" and "mastery of content". The advan- accepted by default. The user interface is atages which are available to foster and set of menus which feed the informationenhance learning can often be lost or re- through a 'preprocessor' to generateduced because of the difficulties of provid- PLANIT Language -lesson code'. This les-ing the appropriate feedback and guidance bvp Lvde ;5 quite quite easy to understand to the Student. By means of the system and can, if desired, be modified through a reported here, tracking of student re- standard editor or through the PLANITsponses and performance data within the System itself. The lessons, however, usu-various subject matter areas or various ally run satisfactorily without modifica- tion. In this way, the power of 'high levelrequest. Results for an overall test or group decisions' is readily combined with theof tests is gathered automatically and can flexibility of 'low level' detailed adjust-be retained or discarded, as the academic ments. may require. 409 400 Using Computers Intelligently How to use the system Of the above, the data entry is provided by The academic users carry out the following the Computer Centre staff; the rest is pro- steps. vided by the system. 1. Provide the questions, together with anApplication of the system to an Educa- indicator for the accepted answer ortional Research Methods Course answers. 2. Group the questions by subject matterFormative and summative methods have areas, difficulty levels, educational ob-been used to assess teacher education stu- jectives, or other criteria. dents in an educational research methods course. Traditionally, the formative exer- 3. Decide what testing model and optionscises comprised of workshop reports and a are to be utilised, and what access is tomajor assignment. The summative proce- be made available to the students. dure was an end of semester exam. These evaluation procedures did not provide 4. Decide what reporting functions are tostudents with appropriate motivation, and be carried out when and if required. the no,essary feedback or performance to plan ongoing study strategies to master the 5. Trial the resultant test course content. The students have viewed these procedures as major obstacles to be 6. Inform the students of the test's availa-overcome at certain times in the semester bility, any special conditions, and howand did not provide a positive structure for to access it. the average students to assess their level of performance and plan study activities to 7. Obtain reports and use accordingly. compensate for deficient areas of knowl- edge. Services Provided Certei features of the computer base A evaluation system were used to develop a The services provided for the academic are: formative assessment exercise to overco, 'le some of the cnticisms of the tradiF mai Data entry of the 'raw' question items. class tests and to provide students with an opportunity to obtain feedback in a non Conversion to PLANIT Language for- threatening environment. This feedback mats allowed them to identify areas of course content they were familiar with and other Delivery of the test to the student. areas needing further study, emsthcring of report data Structure of the class test Reports to the student: this includesThe educational research methods test progress reports, end-of-test reportscontained a bank of 50 multiple choice and summarised results. items with four possible options for each item. The items were grouped into five Reports to the lecturer: This includesareas. The students completed 25 ques- detailed reports (if required) and re-tions randomly chosen from the 50 item ports summarised by class, by topic orbank as follows: by course. 1 0 , 41 Using Computers Intelligently 4 0 1) nature of educational research student can then be directed to read certain 8 questions - 4 chosen chapters of the text relating to the items the 2) hypothestesting student got wrong. 10 questions - 5 chosen 3) data collecting procedures The level of mastery was not incf irporated 14 questions - 8 chosen into the test. A useful feature would be to 4) scaling procedures set a mastery level of say 75%. If the student 7 questions - 3 chosen did not achieve this then he would be di- 5) interpreting Statistics rected to learn certain areas of content be- 11 questions - 5 chosen fore retaking the test. The correct response for each questionwasThe diagnostic procedures are useful to incorporated into the system in order tostudents and course presenters. If certain provide the student with feedback on testitems were causing difficulty to students performance. The students obtained anthen the concepts associated with these analysis containing the total number ofitems need further reinforcement and clari- questions attempted, the number of cor- fication.Lecturers can ascertain certain rect, and Incorrect answers and their re-areas of knowledge which cause students spective percentages. more difficulty and plan the course to cope with this. The students attempted each question chosen once. They were not permitted to The non threatening environmentcan be a choose another option if their first choice major motivating factor in student learn- was incorrect. However, this would be aing. If students are given the opportunity useful option within the test structure.to undertake the test a number of times Students could have a second attempt, and until they have achieved a certain level of if this was incorrect then the correct optionperformance then they are more likely to would be given. continue to learn because no judgment is made on their performance by the lecturer The time element was not considereda or the... peers. A further modification of the major factor in the development of this mastery process may include a barrier for class test.Students had two minutes toeach section of the test. The studen s must respond to each of the 25 items chosen.achieve a fixed mastery level in eachsec- However, the time element can be useful tion of the test prior to any new items being learning stimulant, and can be adjusted presented. according to the number of attempts made by each student or students can be given an The options can be included in the system opportunity to adjust the response time to so that if certain questions containing basic suit themselves. concepts were correct then extension ques- tions could lie presented. Unless this oc- The item bank of 50 questions wasverycurred then no new questions would be limited.This could be substantially in- pr sented and the student would be in- creased in order to test content areas more structed to study sections of the ttt prior to exhaustively, particularly if students at-retaking the test. tempt the test several times. An available feature which was not incorporated into These options allow the student to practice the test is the analysis of scores for eachexam strategy and identify strengths and section of the test and the provision ofweakness. The student would developan feedback on performance in each area. Theoverall strategy to sit for the final exam. 11 33=IMIMI 402 Using Computers Intelligently The course could be modified to reduce..leoptions may be used for the assessment impact and the threat of summative evalu- system or for test-cum-tutorials wruch ation and cha lge the orientation to infor- provide some extended support material. mative eva1.lion. The options availableSome of these OPTIONS are available i- in Plan should be used to allow theteractively from a menu, others arepro student three attempts within a givenpe- vided by making a few adjustments to the riod of time. The results of the t1rat-lesson file. Options currently available or tempts could be aggregated and convertedproposed, are set out below. into a final score for the summative evalu- ation.The benefits which could accrue OPTION: Feedback or none from this procedure relate to the conver-DESCRIPTION: Right-Wrong feedback, sim of sboi t term memory to long term with or without Answer Correction, memory Stucielts are lore likely to recall with or without Answer EXPLANA- information if they have an opportunity to TION for some or all questions, learn it for mastery and understanding with or without Specific Feedback on n.her tl.an recall. This could be achieved particular answer chosen, by modifying the test structure to change with or without Branching (increases the correct option for each item. This may complexity). prevent the student recording the question and its correct response, and force the stu- OPTION: Immediate Retry dent to develop an understanding of the DESCRIPTION: variations are: concepts associated with each question Not alloT,o4.1,permitted, or forced (1, instead of recalling factual responses. 2, 3 or more times) POSSIBLE USES: Test, Self Tests, Mastery Appendix: Models and Options OPTION: Time Limit (on or none) DESCRIPTION: Maximum time allowed There are a number of models which have per question. May be set by author established their usefulness. These form set by student (if permitted by author) some of the components available for use within a course. set by lesson performance OPTION: Random Order Preentation. MASTERY LEARNING DESCRIPTION: The questions may be FORMATIVE AND SUMMATIVE presented in random order. ASSESSMENT They can also be randomly selected DIAGNOSTIC FEEDBACK from defined subgroups of questions. STUDENT CONTROL The number to be chosen from each SELF TESTING subgroup can be se by the author. MONITORING AND ADVISING POSSIBLE USES: Generate a different test REPORTING ON TI-M BASIS OF for each student. CONTENT, DIFFICULTY, NDIVID-The choice could be set by the student or by UAL, CLASS and COURSE. lesson performance criteria. The facilities which have been made avail-OPTION: Progress Function. able may be thought of as TOOLS for con-DESCRIPTION: Gives a summary of structing models. Thec.:. are available in the student's prrgresE (No. seen, right, form of OPTIONS which allow specific wrong, percentages). setting of various alternatives to suit the During and after test, or at end only teacher or learner's particular goals. The (author's choice). 412 Using Computers Intelligently 403 POSSIBLE USES: Extend to give advice to (sub)section or around a (sub)section. It the student. Ativice may be based on may be placed under author control, sub-categories ofquestions such as student control or student performance content, specific objectives, difficulty control. etc. POSSIBLE USES: Remedial or fast track. SUITABILITY: Tutorials or self test. EXAMPLES, RATIONALE: Guide or moti- COMBINATION OF OPTIONS: Options vator. may be combined to produce particular environments for learning. OPTION: Finish at any point. e.g. GROUPING BY DIFFICULTY, PROC- DESCRIPTION: Available at the option of ESS the author. Can also specify, resume ate.g. GROUPING BY CONTENT, ... RE- the same point or resume at the start, or PORTING. specified points only. POSSIBLE USES:Allow or withold studentReferences control. Depends on material Merrill M. D. (1985). Where is the Author- OPTION: Retake the test ing in Authoring Systems? Journal of DESCRIPTION: Immediately or after a Autumn, given period (unlimited or restricted Computer-Based Instruction, 12(4), 90-96. number of times, or disallowed) Mevarech, Z.R., Time Engagement and POSSIBLE USES: Self Test or mastery test- Achievement in Computer Assisted ing or assessment. Instruction, Educational Technology, July SUITABILITY .EXAMPLES, RATIONALE: Student control, or not? 1986,38-40. OPTION:Printed Certificate or Report. DESCRIPTION: Prints a Certificate or Bill Chia is a principal tutor (comput- Report if preset score or percentage ing) at Nepean College of Advanced Education.His current position is ranges are attained. Academic Computing Advisor and POSSIBLE USES: Extension to give sub- Liaison. He is a member of the organis- group scores. ing committee for the CALITE-87 SUITABILITY, EXAMPLES, RATIONALE: Mastery, self test goal setting and Conference. "reward". Dr. Mon Khamis is a lecturer in the Provide written advice to student. School of Education at Nepean College of Advanced Education. His current OPTION: Branching and Return. research interest is in the area of DESCRIPTION: Branch to sections which "Competencies for the successful can be selected interactively. classroom teacher and the successful POSSIBLE USES: This allows review or school Principal". preview of test material. It can also be used to provide supplementary infor- mation, Hints, Data, Worked examples, context sensitive advice or help. OPTION: Branching (No automatic re- turn). DESCRIPTION: This allows branching to a 413 Profile of a successful CAI facility Identifying useful technology Graham R Pars low, Biochemistry Department T Robert Haynes, Advisory Centre for University Education University of Adelaide .MMNIN II Success in CAI (Computer Assisted Instruc- generation mass storage devices, particu- tion) depends on the use of appropriate tech- larly in regard to video images. nology, the selection of suitable material for presentation and management which allows adequate student access to courseware. The What do we learn from technology ex- Biochemistry Department at Adelaide Univer- sity has been using CAI for a decade with tinction? gratifying results while spanning 3 generations of personal computers. The present Biochemis- try Department facility of ten Olivetti M24 Technology which is fated to survive is personal computers is able to save the course responsive to individual interaction and revision requirements of over 300 students powerful enough to perform truly useful who use the facility voluntarily. The principal tasks. teaching tool is the "Q" Instruction Package servicea by a data base of some WOO questions (the Medical Biochemistry Question Bank). At Novelty is an undoubted component of the pioneering forefront, optimistic predic- short-term success. Sputnik aroused amaz- tions can be made for videodisc applications ing interest, although it was almost func- and the authors will discuss their involvement in the production and application of the first tionless. It is difficult in the early days of a Australian University videodisc. technology to decide whether state of the art technology is mature or simply a por- tent of truly practical machines to come. The reason that CAI deserves serious often- The authors are of the view that contempo- lion is that it is a medium which reaches a rary personal computers and video disc majority of students in a manner that istechnology will continue to contribute to different from other teaching contacts anduseful teaching. technologies. Saving teaching money or liberating staff time are not valid argu-In the early 70s colleagues at Melbourne ments in support of CAI. As with otherUniversity Biochemistry Department reaching approaches, the essential ingredi- documented the successes of their on-line ent for success is a belief that CAI has afacility for undergraduates to use a tele- valuable contribution to make. Education printer terminal to analyses enzyme kinet- specialists look for absolute evidence ofics. In the Adelaide Biochemistry Depart- pedagogic value, but the case is very much men'Sinclair ZX-80 in the late 70s fasci- that you will get out of CAI what younated students who saw it translate genetic believe you will. code sequences into protein structures, all in 4K of RAM. The idea of using this equip- In the oral presentation we will illustratement today is laughable, but it does illus- our views by reference to the hardware andtrate the motivational value of using what software we have been associated with. We is perceived as state of the art hardware. will speculate on the applications for new Novelty in these cases obscured long-term 414 Using Computers Intelligently 405 utility. "tele-conferencing arrangement". This perceived need for some aspect of one-to Those of us who rode in on the pioneeringoneness is epitomized by disastrous at- wave of personal computers in the late 70s tempts to use shared listening stations of and early 80s thought that Apple IIs,six head-phones linked to a single cassette Commodore PETS and BBC microcompu-player in the Biochemistry Department at ters would perform each and every educa- Adelaide. This head-phone share arrange- tional task capable of imagination. It has ment was a solution to the then high cost of not proved to be so, the 8 bit computers ofthe technology, and encouraged by the the early eighties have been relegated toenticing glossy from the manufacturer the status of trans,...onal technology. Theshowing happy students in an at.nosphere pioneer personal computers set new stan-of mutual learning ecstasy. dards for user friendliness however, and this has been their legacy to present genera- The tare-slide teaching facility in the Ade- tion computers. laide Biochemistry Department is a survi- vor of technological approaches to educa- The teaching suite of Dr.Robert Beynon tion. It still fulfils a valuable role in non- (Liverpool Biochemistry) has built an envi- lecture teaching because tapes, with ac- able international reputation for CAI using companying written handouts, can impart Commodore PETs and Apple Ils. Thisa higher density of new information than work has been negated completely bycomputer screens. The cassette players changing to a network of contemporaryused for teaching no longer represent nov- MS-DOS computers - an entirely newelty in technology, but an everyday ma- teaching software base is being created. chine. This story mirrors the demise of Exidy Sorcerer computers in the Biochemistry The elements of cassette tape which com- Department at Adelaide University. mend it include one standard - a cassette tape can be Which aspects of teaching technology shared internationally can survive the hype of immature it is personal - cheap enough for 1 tape expectation? player per person. it is unintimidating in its controls. The large scale funding of unsuitable tech- nology has done long term damage to Computers have not reached this degree of general funding prospects for teaching.establishment and are therefore duly re- garded with scepticism for educational The flirtation with educational television is all but dead. Television monitors droopapplications. The wait and see approach however is self defeating - the time to enter dustilly towards disinterested audiences in the learning curve for computer applica- lecture theatres and teachir_aooratories tions is now when computer technology is around the world (many of them still Black reasonably affordable and adequately and White). Mass oriented educational powerful. television lacks the ability to cater for the pace at which an audience is responding to the message and has suffered accordingly. How powerful do educational micro- An apparent success in thi.. use of television computers need to be? is the real-time link between London Medi- cal Schools for teaching in which multiple Computer power is not a real issue. The screens and camera options are used in a essential question is "will the computer 415 r 406 Using Computers Intelligently allow me to extend my teaching ata level there is the need to either master diffiLalt which is satisfactory for my courses." Tech- airaloring software or master the technical nical capabilities of the machineryare in-resource people who have. creasing logarithmically, but our educa- tional requirements are not. Wenow haveNot powerful enough is disastrous affordable machines that have adequate speed, memory and physical compactness too. to be useful. The UK has retarded its development of Contemporary powerful computers (mainCAI by so whole-heartedly adopting the frames) are not an answer to any real edu-BBC micro. Nigel Gardner of the Comput- cational problems in our experience be- ers in Teaching Initiative (CTI) in the UK cause they are made by computer-inhas succinctly argued this case and it isa people for other computer-in people. Pow-compelling conclusion. The grounds for erful computers are usually expensive and this are technical and human. Technically begrudgingly available from theirmanag- the BBC computer is, in its mostcommon ers. For students big computers mean rigidconfiguration, incapable of more than sim- scheduling and limited time of access. Anplistic modelling and shallow gimicry. The intimidating Big-Brother log-on procedurecommitment of funds has been so extensive then exasperating delays. Ifa studentthat the coffers are bare for re-equipping spends ten minutes pondering the truth ofwith better technology. In Australia at a statement, the expectation is none-the- least, funds have not been dissipated by less that the computer will respond in-any initiative of the Federal Government to stantly when a key is pressed. For teachersintroduce computers to education. PROTEINS ARE TYPICALLYSPHERICAL WITH CHARGED SIDE GROUPS EXPOSED TO THEAOUEOUS MEDIUM. THE POSITIVE CHARGES ARE at. %JarGROUPS xt .. .. OFARGININE 'IA:. ANDLYSINE.kt6' . THE NEGATIVE RGES ARE Ct.14:1%TARBOXYLGROUPS THE CHARGE ON OF GLUTAMATE AND A MOLECULE IS THE SUM ASPARTATE. OF POSITIVES AND NEGATIVES. Ee55 ang '.. _ke.g Figure 1: A screen from a "Q" session computer tutorial to teachaspects of pH. Computer tuition on this subject exploits the ability of a computer to interact with students with unfailing patience and consistency. 416 Using Computers Intelligently 407 How many students should work A successful recipe Adelaide Biochem- simultaneously at a station? istry Assuming voluntary self-paced usage itTen Olivetti M24 computers (IBM-PC may be one, two or four. One user only per compatible) have proved adequate to pro- station is appropriate when the material is vide CAI for 300 undergraduates. Each unit new or difficult, otherwise with multiple has a colour screen and a 20Mb hard disk use the advantage of freedom from peerwith courseware selected from a boot-up judgment is lost. Even the best of friends menu. Questions are available throughout may not work well together at a computer each term relating to lecture coursework. if their talents are unequal. Groups of fourThe motivation to take the sessions derives are quite successful for medical studentsfrom taking revision questions together. Medi- cal students as a group have characteristi- relevance (the questions may be asked cally high social cohesiveness plus aca- in an exam) demic competitiveness and enjoy audi- local authoring (Course lecturers are ences to their performance (at least when responsible for question generation or they are scoring highly). This argues for selection) keeping enough space between stations to objective self-assessment of progress. allow people clusters to form (regrettably few places of learning are built with suit-Students can make their own time to use able space). the CAI materials. This can be at times that are uncommitted or quiet time in a labora- What material is best presented by tory period. computer? The local generation of material is impor- The mindlessly repetitive tasks of teachingtant, in part because it shows students that are best suited. For example a tutorial ontheir teachers have made time to prepare the Biochemical importance of pH at Ade- the material. The exclusive use of third laide begins by asking: party courseware is courting disaster. The "Q" Instruction Package has been the basis When pH is high, is H+ concentration for rapid generation of local material at Adelaide. The "Q" Package makes session 1. High authoring essentially an exercise in text 2. Low ? preparation. This low-brow question is insultingly basicTowards ultimate teaching technology when asked by a human tutor (despite the with videodisc widespreaa ignorance of the answer). This ccmputer tutorial then progresses throughColour enliv ens a computer screen because some elementary mathematics of impor-it can impart additional information such tant chemical principles. When humanas separating factual information from tutors attempt to explain these principles aquestion and answer segments of a text predictable dull glaze clouds most stu-display. Graphics too can contribute to dents and learning ceases. The quality ofenlivening computer tutorials. It is in the effort of a human tutor teaching boringarea of graphics and animation however basics deteriorates with repetition, so thethat the most atro.uous misuses of CAI computer is noticeably superior. authors' time are committed. Figure-1 il- 4 i 7 408 Using Computers Intelligently lustrates a graphic from the pHtutorialAn application for teaching written-Chi- mentioned before which is adequate for thenese at Adelaide required a purpose writ- purpose intended and took relatively littleten program. This application exhibits time to draw with a mouse assisted graph-many of the good points of successful CAI ics package. Graphics are in generala pat-with interactive videodisc and will be dis- ent failure of computer technology to ade-cussed. quately represent real objects staticallyor in animation. This is the role for videodiscs. Conclusion Videodiscs contain a library of 55,000 static A successful CAI facility which is likelyto pictures in high resolution whichcan in- survive pressures of obsolescencecan be clude animated segments (likea videobuilt from current microcomputers. The tape). The feature that sets videodisc apartsoftware must be clearly relevant to the from video tape is that each frame of infor- course of instruction. Simplistic computer mation can be rapidly called under the control of a computer. Voila! graphics are generally unsatisfactory, and you have it interactive videodisc is the solution to in- the ultimate educational technology oftheteractive instruction im olving detailed present day. The interactive intelligence of pictorials. computers linked to high quality pictures stored on a videodisc. At Adelaide University wewere fortunate to have accumulated a substantialresource of slides and film/video whichwe were able to transfer to video disk in 1986.This has been linked to the "Q" Instruction package for teaching in Obstetrics and Gynaecology and in Geology (figure 2). RS232 cable Figure 2: Using the "Q" Instruction Packagewith videodisc is a rapid means of authoring. 418 Intelligent CAL, or intelligent CAL users? Rodney Staples School of Electrotechnology Royal Melbourne Institute of Technology For some time now, the development of CAL has been focussed on hardware, software, and course- The skill to use this new technology flu- ware. This paper shifts the attention momentarily to the users. Regardless of the sophistication of the ently does not come naturally or easily. It technology, the long term success or failure of CAL must be developed, and inspired. Once depends on it being intelligently applied. Indeed acquired in theory it must be practiced. people must choose whether the technology should be applied, and, if so, how and when. Just as the Intelligence" built into many new CAL People are Important systems does not happen by itself, people must be adequately informed and prepared if they are to The importance of people in the develop- exercise their judgment intelligently. ment of CAL is shown clearly in the litera- The paper considers how this human resource ture. A few examples, by no means iso- might be developed, since the human "intelligent lated, may illustrate this. user" is better placed to seek new and innovative applications than is any currently extant computer In an international study of factors influ- program. encing CAL development Hawkins (1977) reported that "In all countries the over- Since the study of computer assisted learn- whelming response was the importance of ing (CAL) began, research effort has the initiative taken by one, or at most two to centred on hardware, software, and three project members.... Most frequently courseware. Much of this research is de-commented on was the importance of an rived from work in programmed learning,energetic, dedicated and knowledgeable and training. More recently, systems forcentral figure who could give leadership learning are derived from the study of arti-over the range of activities, and an enthusi- ficial intelligence (AI). astic, cooperative group prepared to put in a strenuous work day over a long period of So far, computers have not been pro- time." grammed to be particularly creative. They canuot yet make the quantum leap of Sprecher and Chambers (1980) studied 200 understanding which leads to a new use forCAL development sites. They observed: an existing tool. They cannot invent new 'They heavy development group was tools to solve existing pro,...,ms. Humancomposed of faculty who placed a high users can. value on CAI [Computer Aided Instruc- tion] and who perceived their department, Thus more is needed than inanimate, even their campus computer center, and the top intelligent, technology if CAL is to be nur- administration of the campus as favou: :rig tured and grow as a discipline. Intelligentit." users are at least as important as the tech- nology. It is the users who will ultimatelyThe importance of giving users flexibility determine a project's success - or failure. to adapt the system to their needs was part 419 wirr.., 410 Using Compufers Intelligently of the philosophy behind the developmentClearly then it is important that potential of Computer Managed Learning (CML) atusers, not just the committed developers the Southern Alberta Institute of Technol- and enthusiastic leaders, must be given the ogy in the early 1980's. Stephen (Distefanoopportunity and encouragement to gain & Stephen, 1982), commented: "The experience with CAL. strength of the system [is) that it lets users be creative, and new applications andBut is it enough for the users to be trained or methods of use appear as users a iapt theexperienced in the intricacies of the hard- system to their particular needs.'Thisware, software, or courseware? Probably implies that the system, which has been not! successfully marketed throughout the world, was specifically designed to ac-The influence of the teacher is profound. knowledged that people rnust still exerciseClark (1983) suggests that the positive judgement on the use of the system. achievement effects for media more or less disappear when the same instructor pro- Ausburn (1982), reporting on a study tourduces all treatments. Hedberg and Char- of the US in which she investigated thelesworth (1981) suggested the same phe- training approach adopted by the US Airnomenon. They observed "that the success Force, also identified understanding andof a media-based system in the class-room commitment displayed by administrationdepended at least as much on the learning and users as significant. She highlighted activities organised at the receiving end as training as vital in establishing both under- on the content that went into the transmit- standing and commitment. ting end."Clearly the teacher who is managing a learning program, and specify- These examples serve to illustrate thating alternative resources to meet technol- committed and informed leadership isogy in determining the effectiveness of the necessary for CAL to be made to work. media in producing learning achievement. Thus any training or experience must in- They also suggest that developers andclude consideration of the students who users must be informed if they are to bewill use the system too. innovative in their use of CAL. Training of some sort plays an important part in theMy own informal observations of students development of the people who will imple-working with tutorial type CAL material ment CAL. They are all positive examples.suggest that students can quickly become bored or frustrated, frequently not com- But there are also factors at work which pleting the set material, and if given the contribute to resistance to the use of CAL. choice may not return to class. Kearsley Zemke (1984) found that:"... one distin- and Hillelsohn (1982) support these obser- guishing characteristic of the 'computer- vations. This happens when the CAL de- phobe' cannot be explained away by our signer didn't fully appreciate the diverse sample: He is absolutely to be foundneeds of student users. In many applica- among the one in five trainers with nary a tions of CAL this phenomenon is not visible micro-computer at work in his department.because the students are motivated by Not only has he virtually no experience some form of compulsion. Certainly then, with computers, he has never taken a CBT it requires some understanding of the prob- [Computer Based Training' course.In lems of helping people to work with CAL short, the computer reluctants in our midst learning resources if they are to be effective. are most easily distinguishable from the computer advocates by the single charac-Chorover (1984) echos an increasing view teristic of inexperience." that: "It is only in the context of a suppor- ,4.1o, q 4 V Using Computers Intelligently 411 tive educational community - a human tisewill develop and leadership will environment conducive to learning - thatemerge. This is not always true. the hazards of automation [in education] can be avoided." Inappropriate uses, orIn 1980 the British Columbia (Canada) uses which lead to boredom or frustrationgovernment contracted with a private in a learning context will be quickly reme-company to develop CAL materials for died by a teacher with experience using the local school divisions. This contract also medium. The teacher must decide if, or included strategies for developing the skill when, the tool should be used. People must of participating teachers.This project dictate the use of the technology, not the sought to involve teachers by offering five technology drive the people. hundred dollar incentive grants to teachers who participated in courseware develop- Even an expert ("intelligent") CAL, which ment. It hoped that by generating software has been specifically designed to capture in quantity, some quality software would the specialist expertise of both teachers andemerge. This grant approach "produced subject specialists, needs some user skill if no software of usable quality."In this it is to be effectively integrated into the example, the "Mongolian horde" approach learning environment. did not work. The question which is yet to be resolved is A similar scheme was a major focus of the how best to provide this experience. Australian government initiatives in com- puter education in secondary schools. Developing the Experience While this initiative undoubtedly devel- oped the skill of many teachers, it barely There are four ways in which the necessary scratched the surface of the need, and was experience base can be developed in a laydiscontinued last year after only two years population: in place. Other initiatives in New Zealand suffered the same fate: It seems that now Mongolian hordes approach; the same few faces who have always as- Osmosis; sumed a leadership role in secondary Formal courses; computing are continuing to do so. Again Strong leadership. the "Mongolian horde" approach didn't work too well. "Mongolian hordes" approach The most successful implementation of the Solving problems by original development"Mongolian horde" strategyis the con- has been mainly achieved by employing certed effort in England to get two teachers large numbers of people to work together with CAL experience into every secondary at a solution. Coyne suggests the "Mongo- school in the country. These teachers were lian horde" approach, throwing largeto be the nucleus of experience in the numbers of people at a problem, willschools from whom others could learn. probably be paid for ... with the worst skillThis program has taken many years to shortage ever, and this shortage will berealise its goals, and has been expensive, accompanied by escalations of salaries,but has demonstrated that a systematic dissipation of management effort in mit--training program to develop "grass roots" gating the increased staff mobility, and ateacher's skills can work. But even given its debilitating feeling of instability pervading success, the program has shown that "a lot the industry." The expectation is that givenof the educational software [resulting from enough people working in an area exper- the program] is rubbish." 4 2 412 Using Computers -,ent/y The "Mongolian hordes"strategy requirestematic resources for gaining the experi- a great deal of money, and considerableence quickly. In tertiary institutions which commitment from the administrationto require teacher training, experiences with support the development program untilit CAL could be includedas a compul3ory can become self sustaining. The Englishpart of the course. In the more usual situ- experience shows that, because of there-ation where such courses are optional, they sulting increased awareness of CAL,usersstill need to be available whenusers are eventually became more discerning intheirsufficiently motivated. But thecourses in choice of suitable courseware,but it is athemselves are usually not the motivation brute force strategy which doesnot system-for formal trainings. atically strive to develop leadersor breadth of skill among users. Because of the reluctance to makeuse of formal courses, a people development Osmosis strategy based on them will take a long time The English experience has shown to reach most existing teachers. If the staff that if turnover is small, the time required to get there are enough people in thesystem with adequate skills in CAL, others will be the expertise to where it is neededmay be en-unacceptable. couraged to seek the skill too.thus the experience of a few will flow through the system by osmosis. This strategy reliesonStrong Le4dership sufficient skill being available for theexpe- rience to be visible and clearlyacceptableClearly what is needed to excite teachers within the system. Withoutencourage-about the intelligent use of CAL, isstrong, ment from administrators for peopleto visible, and enthusiastic leadership. This undertake specialist courses,or withoutmust be two fold. rewards for gaining the expertisesome other way, there will not be enoughcorn- First those who have made CAL theiraca- rni tted, experienced people to "rxed"thedemic speciality must display theirsuc- osmosis process or to motivatethe com-cesses to others through conferences like puter rellietants. this one, in seminars, or evenover coffee or lunch. These same experts must be ableto Formal Courses discuss the problems likely to beencoun- tered by novices, and the relative merits of Formal courses in CAL methodsare thea range of possible solutions. Using this cornerstone of the successful English staffalready established human resource in the development programs in CAL methods.institution as the seed from which the A major impediment to using formal osmosis process can spread may mean that courses to develop CAL expertise in terti- it is not necessary to presentcourses just to ary institutions is the commitmentre- generate the initial experience.Rather, quired. Most academics are, quiterightly, such visible examples might encourage the committed to developing skills in theirCAL reluctants to sample the formal discipline or in administration. They havecourses and explore other learning re- little time left, and are given little incentive, sources for their own development. to develop specialist CAL skills. Thusaca- demics are reluctant to undertake stillmore Second, administrations should recognise training than they already have, especially that the people who can provide the leader- if it is outside their primary discipline. ship already exist in their organi, ations. Despite this, formal courses still havea This resource must be nurtured. Adminis- place. They can be concentrated andsys- trators need to provide the incentives, both 422 Using Computers Intelligently 413 to keep the experts interested in pursuingThe incentive to develop the necessary their new discipline, and to provide theskills must come from the institutions if incentive for novices to emulate their morethere is to be a significant development of experienced colleagues. For this reason,existing staff to effectively apply CAL. For successful work with CAL should be recog- this skill base to be developed systemati- nised as valid work for promotion andcally, strong leadership is essential. This other rewards. leadership must be accompanied by users of CAL who are encouraged by example to The advantage for administrations in improve their own skills.It is these uscrs adopting a positive attitude to building onwho will eventually make CAL work. the existing commitment is considerable. ItAdministrations cannot expect that adopt- is not then necessary to generate a sea ofing a "Mongolian horde" approach, or novices in the hope that strong hadership,expecting the skills to develop by osmosis, or even adequate experience, win emerge.will be effective without strong, effective The resource is already there. The commit-leadership, backed up by formal training ment is already there. the ability to make where necessary. the quantum leap leading to new and inno- vating uses for the technology is in place.Is it possible that, despite the intelligence All that is needed is the support to maintain built into the technological tools, the real and encourage the work, and to encourageintelligence in CAL needs to be first in the the spreading of the expertise tc rewcom-minds and hearts of the human users? ers. References Administrative Leadership Implications Ausburn, L.J. (1983). Report of Overseas In- vestigation. Victoria, TAFE Board. Cherover, S.L. Cautions on computers in In a climate in which economic and politi- education. Byte, June, 223-226. cal pressure to provide vocational prepara- Clark, R.E. (1983). Reconsidering research tion and broad personal development at an on learning from media. Review of Edu- increased level is being placed on acade- mia, and no less on industry, there is a cational Research, 53, 4, 445-459. Coll, J. (1984). Cited in The Age, 7th August, strong incentive to look to technology as a pp.39, 42 and in a private seminar at savior. There are considerable economic RMIT. advantages for looking to CAL to increase Coyne, L. (1987). How the skill shortages efficiency. But should this cause us to abro- can be solved, in Bennett, N. Computer- gate our responsibility to people and put world, p.16. our faith and trust entirely in technology? I Distefano, J. and Stephen, A. (1982). Com- think not. puter managed learning - An overview. Corn 3, May, 31-33. CEGV. We must put at least as much effort intoHawkins, A.C. (1977). Computer Based developing the skills of people who will Learning..; Survey of the Factors Influ- use the technology intelligently. People encing its Initialization and Development. should be systematically encouraged to The Netherlands: Utrecht State Univer- seek new and innovative uses for CAL. sity, p.55, 61. People are better able to do this than any Hedberg, J.G. i:..nd Charlesworth, S. (1981). extant computer program. People are the The discrepancy between the promise and key to successful development and use of the performance of technology in education. CAL. Inquiry and Action. Papers presented 425 414 Using Computers Intelligently at the Annual Conference of A.A.R.E., 56-&1. Kearsley, G.P. and Hillelsohn, M.J. (1982). Human factors considerations forcom- puter-based training. Journal of Com- puter Based Instruction, 8, 4, 74-84. Murray, S. (1986). Commonwealth initia- tives in computer education and infor- mation technology. In T. Salvas (Ed), Computers in Education: On The Crest Of A Wave. (pp.64-67). Melbourne, CEGV. Prokopanko, C. (1982). Computing in Cana- dian schools. Proceedings of the Fourth Annual Conference. CEGV, La Trobe University, 9-17. Sprecher, J.W. and Chambers, J.A. (1980). Computer assisted instruction: Factors affecting courseware development. Journal of Computer-Based Instruction, 7,2 47-57. Staples, R. (1986). CAL Anew era, or an expensive white elephant. Proceedings of the Fourth Annual Computer Assisted Learning in Tertiary Education Confer- ence. Adelaide, CALITE, pp.272-281. Tough ultimatum to academia. Editorial, The Age, 23 September, 1987, p.13. Zemke, R. (1984). Training and computers: After the courtship. Training, 2, 10, 48- 58. .,724 Computer-basedtraining: Networking of autistic and other retarded students in sheltered training establishmenta R. A. Coldwell, Information Technology Division Royal Melbourne Institute of Technology One of the major problems concerning education Integration into normal schools of retarded people an Australia is often their geographical isolation. Where they live in urban areas and, consequently, facilities can be pro- The "integration policy" - as it is now called vided for them. They can be given spedalised by bon' parents and the newspapers alike - training. Intreasingly, we are finding that both autistic children and adults are having to be pmbably emanates from an administra- moved to these locations and, often, to institu- tive need to economise, on behalf of taxpay- tions away from their parents. The writer argues ers, regarding the cost of special schools to that, given the use of existing technology, chil- dren call now be integrated into the populations the community at large.However, the of their loci schools and their teaching supple- argument for integration rests mainly in an mented by use of computer-based learning via egalitarian myth which suggests that communications networks or even by use of handicapped children should have the modern and telephone lines. This, in turn, en- ables them to become integrated into TAFE and same form of education as other children. other courses and, later, into sheltered work- This is supported by a behavourist argu- shops which are computer-based. ment that handicapped children will, in mixing with other children, become as normal as they are likely to. This argument This is a working paper which reports some is supported by the parents of handicapped of the difficulties with trying to break- children but refuted by those of others wit° down a few of the barriers that we set-upbelieve that, by interacting socially with against using technology to bring educa-handicapped children, their children, too, tir.,n to those who, for some reason, are notwill become handicapped.These argu- in a position to claim it for themselves (seements, given by parents and others on both Coldwel11984). This may seem like a curi-sides, are simplistic. There is, however, ous statement to make in modern-day some behavioural validity to both sides of Australia but, nevertheless, itis -.mild. the argument. Oftea we set-up bathers in our minds - inhibition: which fashion how we classify Integration at the classroom level people and, con equently, how we treat them. This can be a staggeringly rationalWhat this paper addresses is some of the process.One influence on the form ofpossibilities regarding using computer education, ttr t intellectually handicapped aided learning as a mechanism for integra people receiv, the workings of govern- tioa (see Coldwell 1983). Before discussing ment bureaucracy.Government policy this, however, we should look briefly at suggests, for example, that disabled chil- some of the difficulties that occur, at pres- dren should be integrated into ordinaryent, in integrating handicapped children schools. into ordinary classes. 423 Using Computers Intelligently Normally, a teacher in an ordinary schoolA simple instance will demonstrate one can manage, say, thirty children fora while, major difficulty with teaching autistic chil- at least, before, due to tiredness, anarchydren. If we imagine that an autistic child occurs.In a special school, a far lowercould ask a question, the following dia- teacher to child ratioof, maybe, onelogue might occur : teacher to five handicapped students- is considered normal. Integrating one handi- Child: "What's your name?" capped child into a normal class of, say, Me: "Roger!" twenty children can cause pandemonium. Child: "What's your name?" A teacher, if she is able to manage, reaches Me: "It's Roger." exhaustion quicker and, clearly, needs the Child: "What's your na ne?" help of an aide to manage the handicapped Me: "I said 'It's Roger!" children. This refutes the false argument Child: "What's your name?" concerning funding economies. Chaos still Me (now in frustration): "My name is occurs. But the handicapped children are Roger Coldwell." being managed, often, by untrained aides and the concentration of the other childrenIf we examine my responses, we see that I is still disrupted. This would, one assumes,have given four different replies to the affect the progress of all of the children both same question from the child. To an autistic handicapped and unhandicapped. child, this is fickle; hence their confusion with normal social interaction. Some brief thoughts on autism Their commuizication ability seems to be It is necessary to explain what kind of dis-able to cater wit". logic but not with the ruption occurs.To do so, I will try toillogical component of (and certainly not describe one of the worst instances ofthe humour in) our communication. Let us communication disorder that there is. Thisimagine a similar communication with a is called "autism". computer : Autistic children are often mute, with- Child: "What's your name?" drawn, hyperactive, and uncommunica- Computer: "Fred!" tive and, therefore, unable to read or write Child: "What's your name?" because they have difficulty in attributing Computer: "Fred!" meanings to alphabetical, numerical, spo- Child: 'What's your name?" ken, sound and body language symbols Computer: "Fred!" (see Coldwell 1985a). Their response to etc., etc. social interaction often appears to be non- existent. Seemingly through frustration,This constant play on logic seems to be some develop rages. Often their behaviournecessary for an autistic child to see a pat- is interpreted as anti-social whereas it is, tern in the learning process. It is found in we believe, merely asocial.Often theirthe user friendliness of the response of a actions are repetitive and, perhaps, ritual- computer of enquiries to it.Computer ised. They may do the sa ne thing continu-aided learning for autistic people is, there- ally for hours seemingly with no reason.fore, strangely successful although it has The target of their fascination is, some-yet to be demonstrated to the satisfaction of times, a washing machine, a door handle or the psychological world which has, in a similar piece of equipr .-nt (see Montes-general, takenahostile standpoint towards sori teaching theory). its use as a learning mechanism for autistic children. 426 Using Computers Intelligently 417 Computer aided learning for autisic travel to me with a parent.This is an children incredibly time consuming process for both sides and involves expensive travel For the last thirteen years, the writer has for people who already have a large part of been working with autistic, Down's syn-their single-person income eaten away by drome, epileptic and a few other types oftravel costs and other related expenses. intellectually handicapped children (see(Appreciate, here, that the mother's time is Coldwel1198510987b). This has involved often spent totally on supervising the autis- him experimenting originally with highly tic child.) structured computer aided drafting (CAD) systems (e.g. RUCAPS) and, later, simpleNetworking schools educational graphics-based systems (e.g., LOGO) (see Coldwell 1986).The former Our first move to counter the geographic were over-structured and the later are tooproblems involved in communicating with primitive to be of practical use as a com-schools is to use Plato on the Vicnet Com- puter aided learning too".. A reasonablemunications Network to communicate, compromise is the new range of micro-from the Royal Melbourne Institute of computer based CAD systems like "Au-Technology, with the Warrnambool Insti- tocad". The benefit of using this sort oftute of Advanced Education.At the software is itsiogical though flexible struc-Warrnambool Institute, there are Depart- ture (see Coldwell 1987a).(These have ments of Teaching and Computing and an now, of course, been surpassed by total external Studies Unit which is, at present, systems, like Plato, which are available on giving distance courses to paraplegics. our communications network.) (The Department of Teaching already has a course in special education). This is an I normally work with a child with a dual initial phase which will precede exten, ling control so that the child uses a mouse while the network out to a school in the I communicate with the child via a key-Warrnambool area to enable a teaching board through the microcomputer. Theaide to connect-up with the Warrnambool process that I follow involves the child Institute as a Centre for western Victoria. developing a set of their own drawingsThis will enable direct contact between through a stage involving hieroglyphicRMIT and children in classrooms in west- meanings into use of alphabetical symbols. ern Victoria or, alternatively, enable similar Hence, they draw a cat, store it as "c" for contact between them and someone at the "cat", retrieve it and use it to create pictures Warrnambool Institute. Where parents of together with "dog", "house", "mummy" handicapped children have microcompu- etc. Eventually, I save the symbol with theters at home, they, too; may be able to full name of "cat" needing them to extend connect-up to the Vicnet communications their vocabulary to gain its use. Initially, network. retrieving three "cats" will involve them pressing the "c" key three time-. but, even-Centre for social computing tually, they will use the number "3" to do soapplications (CeSCA) etc. The process involves their ascribin, meanings to the symbols. Using a pre-se Computer aided learning focal points are library of symbols would not do this ofneeded in tertiary institutes throughout course. Australia to create centres for social com- puting applications. Networking schools This whole process has severe limitations to integrate handicapped children is but however. Normally I travel to them or they one of their possible functions. Another is 418 Using Computers Intelligently the creation of software that is needed tomany disabled students will be amongst assist teachers to communicate, in dialoguethem. Some part of industry will, we hope, with children in one-to-one interaction, develop sheltered workshops to enableour while much of the repetitive learning isdisabled graduates to work, alongside managed by child-machine communica-their more able colleagues, in industry. tion. References What have we ("one regarding this? Throughout the couple of months, I have advertised in Departments of Com-Coldwell, R.A. (1983), Computerised Learn- puting of the tertiary institutes throughout ing a sociological perspective.Proceed- both Victoria and Tasmania (with astound- ings of a Conference on computer-a .ded ing success) asking people to write soft- learning in tertiary education, August, University of Queensland. ware for intellectually handicapped chil- Coldwell, RA. (1984), Prerequisites for dren.The advertisement has also been com- puter-use reproduced in the "Bulletin" of the Austra- some analogous theoretical lian Computer Society; it has appeared on systems. Proceedings of the 6th Annual noticeboav is of computer rooms of com- Computer Education group of Victoria mercially orientated organisations as far Conference. May, La Trobe University, Melbourne. away as Adelaide and been reproduced h. most of the newsletters of Associations forColdwell, R.A. (1985a), You don't teach people with different types of disability in adults computing in the way thar you teach Victoria. (As the writer was working, a children. Proceedings of an Australian moment ago, the phone rang with a call Corikputer Scdety Conference, Phillip from Brisbane.) I have also appealed to our Nand, and later published in the Aus- own faculties for obsolete microcomputer tralian Computer Society Bulletin, April, pp.15-16. hardware and received a mixture of differ- ent machines that we will be using to createColdwell, R.A. (1985b), Autism, Montessori software for the use of intellectually handi- teaching theory, computer aided learning capped and disabled children who are and autistic communities. Proceedingsc the Calite '85 Conference on Computer being integrated into schools. Aided Learning, University of Mel- bourne. The function of this installation could also Coldwell, RA. (1986), The role of computers be as a computer laboratory for the use of in assisting autistic children: some case disabled students of the Institute ("Dis- studies. Proceedings of the IBM Inaugu- lab") who are disadvantaged, compared ral Conference on Educational Trends, with other students, because of their dis- April, Melbourne. ability. It can also be used to coach mem-Coldwell, RA. (1987a), Computer aided bers of staff who are apprehensive about graphics and retarded children. Proceed- entering their own computer laboratories ings of the Second National Autocad as novices. Conference, March, Melbourne. Coldwell, R.A. (1987b), Computing and dis- Most important of all of our plans is to use ability.Invited to read paper by the communications networks to communi- Australian Computer Society at the Sci- cate with children in schools throughout ences Ciun, melbc.r.1.-.1.e, which was later Australia to familiarise them with the Insti- published in the Australian Computer tute. In communicating with the RMIT, via Society Bulletin, August, pp.4-5. modem, they will be one step closer to tertiary education. We hope, too, that 428 Using Computers Intelligently 419 The author is a sociologist and a tempo- rary lecturer with the Information Technology Division of the Royal Mel- bourne Institute of Technology in Mel- bourne. He is also a Research Associate of the University of Melbourne where his post-doctoral research has been funded by the Department of Commu- nity Services Victoria and the Sidney Myer Fund of the Myer Foundation. Recently he has received funding, too, at the RMIT from the Apple Education Foundation. He plans to liaise with the Lincoln Institute of Health Sciences, which is being merged with LaTrobe University, to develop his work further in this field. 429 Decentralized learning on the tertiary campus: CAI to increase student music literacy skills Vanda Weidenbach Nepean College of Advanced Education Traditional approaches to teaching Music ing skills should be of a higher order. Fur- in both schools and tertiary institutionsthermore, because learning has been have not led to universal Music Literacy. spread over time, these skills are more Graduate teachers on entering primarylikely to be maintained. service, aware of their inadequate music skills, generally avoid teaching the subject. Ma trial for this hypothesis, the writer has As a consequence, the status is perpetu-begun a study with a group of first year ated. To redress this situation, there is anstudents at Nepean C.A.E. for whom a urgent need to institute innovations, at thecontrolled number of trials on a specific tertiary level, which will ensure the acqui- program will be conducted. Student prog- sition of adequate music teaching skills of a ress will be individually monitored and on- level sufficient for the primary classroom,going changes emanating from student especially since the introduction of the 1984 data will be made. A program designed by Music K-5 Curriculum which makes the the writer for primary school children will teaching of the subject mandatory. be used for the initial trial. One accepts that many of our students have limited com- Organizational structure of many teachingputer skills at this level and the program programs does not permit the learning of has been designed for the novice. After this music skills to be evenly spread over timefirst study, it is anticipated that, depending so frequently what is learned in isolated on student response, the program will be blocks is not maintained. In addition themodified for the mature student. The very nature of music practical skills, inwriter is also currently investigating the particular, demands time for consolidation status of C.A.I. in other teacher training and maturation of coordinated move- programs within Australia. ments. Music education in the tertiary sector It is proposed that if students take one hour units spread over their entire program One of the major concerns of music educa- using Computer Managed Instruction to tors in the tertiary sector is to determine the supplement formal block semester lec-most efficient way to make use of the lim- tures, eral benefits may result. It is ac- ited time allocated to the teaching of music cepted generally that students learn at dif-in the preparation of teachers for primary ferent rates. Consequently, when individ- service. The majority of students entering ual control over how often to take instruc- pre-service primary training exhibit lim- tion is made available, and when programsited musical skills in terms of music literacy that only permit progress after masteryand practical musicianship. Since 1984, the levels have been achieved are used, result- release of the N.S.W. Departmer t of Educa- 430 Using Computers Intelligently 421 tion Music K-6 document has made theIt is acknowledged that as adults, we do teaching of music mandatory, so there is anengage in such processes. The recent Re- urgent need to ensure students graduate port on the Music Industry by Dr. Hans with adequate skills to implement the cur-Gulberger (October, 1987) has highlighted riculur.t. In the past, the teaching of musicthe extent to which Australians are in- has proven resistant to traditional ap- volved in music. His data show it to be a 1.6 proaches. billion dollar industry employing more than 60,000 individuals, as well as a grow- The proposition that all individuals haveing industry, with the numbers of music the potential to develop music skills to ateachers and musicians having doubled level which enables them to participate in since 1%1. This increase seems to parallel understanding basic concepts and express- the growth of music through the mass ing musical ideas through creative endeav- media and the exposure society now has to ours is indirectly supported by the Music various forms of music which includes K-6 document. Those of us committed topopular, jazz, classical European and to music education have long held the view some degree classical non-European idi- that innate musical ability is not essential to oms, as well as folk music. The degree to achieving a level of skills commensuratewhich individuals are involved in music with the level demanded in any other sub-varies and the level at which one may be ject area. For example, all children receive involved, diverse, but the size of the indus- education in the language arts which will try suggests that the greater majority of us ensure a facility to communicate through- have some involvement which further re- out their lives and there is no expectationinforces the proposal for universal music that this learning will necessarily result in education. literary skills of an exceptionally high or- der. The same proposition holds true forIn the past, the "musically gifted" have the learning of music skills and therefore been the main recipients of individualized one should not expect that children receiv- music instruction, frequently receiving ing a music education will necessarily be-tuition from the private sector. Less atten- come professional musicians. tion has been paid to the educational main- stream. The main reasons for this have been For those whom psychologists would firstly, the lack of sufficient music special- identify as musically "gifted", a general ists and secondly, the generally limited music education in the fundamentals ofskills of the regular classroom teacher. literacy, creativity, performance and audi- From a financial point of view, the proposi- tory imagination and analysis, will provide tion of specialist music teachers for all a sound foundation. For these individuals,schools is unpalatable to the funding bod- studies in music of a more intensive and ies. The acquisition of high order music rigorous nature may result in the acquisi- skills in the areas of instrumental and vocal tion of high order music skills leading to performance skills, composition, analysis, either committed amateur or professional aural skills and the like, which form the musical status. But our major thrust should basis of a music specialist's expertise, take be to ensure a fundamental grasp of music years to attain. The cost therefore to pro- literacy skills for all children. Therefore, duce sufficient music specialists for the just as language studies are taught, in orderprimary classroom would be considerable. that individuals may communicate in a social environment, so should music skills, The alternative, which has been promoted of a level adequate to enable children to as having a sounder philosophical basis, is engage in a variety of music activities,to train primary teachers with a level of thoughts and discussions, be leamed. 431 422 Using Computers Intelligently music skills which will enable themtoproviding intensive training is available. understand and manipulate musicalcon-Until recently these conditions could not be cepts sufficient to satisfy the demands ofmet. However, if students are givenaccess the curriculum. (As a side issue, it isanto an interactive teaching medium, avail- unfortunate fact that accountability ofable throughout their program, and foras teacher training courses in this regard is often as is required to reach a mastery level, non-existent. I am not aware of any analysiswe see a more promising scenario. which has tried to correlate the skills of primary teachers when the exittheirLooking at the second condition, that of the courses with the demands of the classroom,very diverse skills levels of students enter- taking into account the existing primarying pre-service primary educationpro- curriculum.) Nevertheless, this should begrams, we find that Introductory Studies in the aim of preservice primaryprograms. Music Education is not only boring for students who may have taken music atyear And here we find another difficulty whichtwelve level, but also a waste of student is related to the general structure of teachertime which might well be used in either education programs whose units, forcon-extending their existing skills in musicor venience, are in large blocks. For example, remediating deficits in other areas. At the in my own institution, First year studentsother end of the spectrum, thereare those have a one semester Introductory Course instudents who do not possess the most Music at the beginning of their training, butrudimentary music understanding and for there is a period of two semesters beforewhom additional teaching is essential.If they meet the subject again by which timethese students have access to an interactive the skills achieved in the earlier encounterteaching medium, on an individualized are greatly diminished. Music is a subjectschedule, they will all benefit in terms of area which demands continuity, time totime spent to reach a predetermined crite- consolidate and exten I both the intellec- rion. tual processing of musical concepts and practical motor skills for instrumental So, in summarizing this preamblewe con- development. Two final considerationsclude the following; need to be addressed; rate of of student music plays an important part in adult learning and the entry skills. Taking the society first, educators accept that students do all children have the potential to acquire learn at different rates and we know from music skills the extensive studies undertaken in special all primary teachers need basic class- education that even the severely andpro- room music skills as prescribed in the foundly handicapped will acquire skills, curriculum given appropriate teaching procedures music learning during preservice train- and programs. For these individuals, ing needs to be spread ove. time intensive teaching over time will result in the students learn at different rates so ac- acquisition of skills not so long ago thought cess to learning according to need will of as unattainable in such a population. be more efficient and effective than group instruction If we apply the same reasoning to the both C.A.I. and C.M.I. have the poten- acquisition of music skills at the tertiary tial to provide such access level, it is feasible that trainee teachers will attain a level of music skills which willWe therefore hypothesize that; satisfy the needs of the classroom, if their decentralized learning on the tertiary learning can be spread over time and campus, used in 432 Using Computers Intelligently 423 conjunction with more traditional ap-drill and practice for speed and fluency, to proaches, should lead to increasedprovide simulations, games and problem music literacy for trainee teachers andsolving activities for generalization and ultimately, for all primary school chil-extension, make the microcomputer a ver- dren. satile instructional medium. Its facility to allow interaction between teacher and Computers as music teachers learner, further enhance its qualities as an educational medium. At the very outset we wish to emphasize that Computer Assisted Instruction in Many advantages, well documented for music is not expected to solve all the prob- other subject areas are relevant to the teach- lems of Music Education in Teacher Train-ing of music and, in addition, there are ing. However, dearly it does have the po- special features which are particularly tential to assist tertiary music educators inappropriate to music learning. overcoming some of the existing difficul- ties. No-one would suggest that all music In general terms, the computer is able to; courses can be taught more appropriately, pre-test students so they enter the pro- or even at all, using the new medium any gram at an appropriate level more than one would suggest that any one present music sounds, notation and method of teaching is justifiable. An eclec- written text tic teaching style which ranges from direct analyze student responses, both written instruction procedures to exploratory and musical learning experiences is eminently suitable provide objective, instantaneous feed- for the varying components of music learn- back ing. There are those factual aspects c (music as a consequence, increase potential which remain unchanged such as pitch, learning time intervals, history, analysis and the like give immediate and appropriate rein- while other endeavours such as composi- forcement tion or an understanding of orchestration, keep student records and identify, for are eminently suited to a discovery learn- the lecturer, those students in need of ing approach. (How many compositions or additional help orchestral arrangements anguished over enable students to be actively involved by students during training, never reach in the teaching/learning process their ears through instrumental realiza- provide motivation through moving tion? ) graphics and sound provide further motivation because of Synthesized sound capabilities of the the confidentiality of learning microchip will make the acquisition of spread learning over time these skills in particular more efficient us- give students control over the amount ing C.A.I. as students manipulate their at- of learning required for mastery tempts while simultaneously receiving block students from progressing until aural feedback. mastery is achieved give student access to learning outside The microcomputer is also eclectic in the the lecture room various teaching procedures it can present, move students to new levels or offer again eminently suitable for the learning of remediation, as required. a wide range of music skills. Its ability to provide students with programs which present text and musical notation through will ensure a level of basic skills pre- tutorials for skills acquisition, to provide requisite to their courses, perhaps in the 433 424 Using Computers Intelligently form of a bridging unit. of allowing explorations into the domain of permit lecturers to proceed at a fastercreativity. "The clearest trend in C.A.I., " ( ra te becau se of acquisition o f these skills Upitis, 1983) " is that computers are being thereby leading to further extension ofregarded less as a vehicle for delivering skills programmed instruction and more as a means for allowing children to actively In more specifically music oriented tasks, manipulate variables and solve problems." the unique characteristics of the computer offer; Generally, music educators would agree the provision of moving graphics to il- that the most efficient way for children to lustrate certain concepts of musiclearn music is for them to experience it in which is particularly relevant, given the the most practical ways whether through transient nature of sound movement activities, singing, instrumental the provision of sound itself, now avail- experimenting, or creating their own com- able with most microcomputers positions. Seymour Papert's theories con- the ability to present monophonic and cerning the acquisition of mathematic skills polyphonic sound sirruiarly suggestthat children should the attachment of the MIDI (Musical "do" maths rather than "know" maths and Instrument Digital Interface) throughthrough LOGO this has been made pos- which certain synthesizers, keyboards, sible. The little known LOGO music sys- sequencers and samplers may be con-tem, was developed primarily to teach nected to the microcomputer composition. the ability to reproduce at an ever in- creasingly sophisticatedlevel, the Logo is a computer language written in the sounds of conventional instruments ascomputer language Lisp which has been well as previously unknown timbres much used in programming artificial intel- the printing of musical notation andligence. ( Baugh, 1987). Despite this, its scores principal function is " not so much to allow the ability to analyze information andchildren to engage in computer program- keep data bases" ming but rather to act as an environment in the ability to receive sound stimuli from which they may learn about the particular the student, analyse it and report to thesubject matter in hand and also to learn student about their own thinking." (Stevens, 1987) Thus in terms of problem solving activities, There are also direct benefits for the teacher Logo encourages the child to explore his because the computer allows him to; critical thinking skills. The two primary 1. pre-program systematic instruction types of Loge are Terrapin Logo, devel- 2. ensure relevant sequencing oped at M.I.T. and the other, L.C.S.I. Logo 3. revise and update the program accord-from Logo Computer Systems. The former ing to student performance includes Commodore Logo, P.C. Logo for 4. individualize aspects such as reinforce-the I.B.M. P.C. and Krell Logo and Terrapin ment schedules according to student Logo for the Apple 11 series, while the latter needs is used for Apple Logo, Apple Logo 11 and 5. be relieved of repetitious teaching SpriNa Logo for the Apple 11, Atari Logo, 6, keep records automatically IBM Logo and Logo for the Apple Macin- 7. spend more time on class managementtosh. (Baugh, 1987) Perhaps the main dis- and organization of programs advantage of Logo is that one cannot use traditional notation or note names to define Not only is the computer able to providepitch or duration with numbers or letters direct teaching, it can also provide a means 434 Using Computers Intelligently 425 being used to assign these elements. Forthat music is an art rather than a science. more detailed descriptions of the use ofHowever, the demands of programmers in Logo music, see Stevens, (1987). developing software may lead to improve- ment in this area as specifying precise There is now a number of high level lan-learning outcomes is an essential part of guages which can be used by teachers to instructional programming. create their own programs. Super-PILOT allows one to program sound, graphics andOne further aspect not previously consid- text relatively easily using a series of mne- ered is that of student motivation. Certain monic, alphabetic commands to control thecomputer programs have been found to be operation of programs. Music teachers'highly motivating, an important aspect of entry into this part of the process "mayall learning. Recognizing the powerful result in a fresher approach to some teach-motivation factor in video games, Chaffin ing problems." (Placek and Jones, 1986). Itet al (1982) called for the identification of will certainly encourage music teachers to those variables associated with the motiva- be analytical in lesson planning. Clearly, tional aspects Of the games so they could be music c.a.i. is more than drill and practiceapplied to educational settings. Malone's of theo7y as so many music teachers be- study (1981) resulted in the general classifi- lieve. cation of motivational aspects under three categories- challenge, fantasy and curios- Continuing research on the effects of using ity. Fredericksen et at (1982) described computers to teach music is an importantthree computer games designed to im- issue, not only informative on the specificprove reading skills which incorporated outcomes of studies but also for the general the motivational features of video games. conclusions it produces. Placek's studyThey cited; (1974) with students majoring in elemen- 1. clear cuts goals, tary education presented a C.A.I. lesson on 2.fast pace rhythm, recording data on total time spent, 3. i.runediate feedback amount of program covered, amount of4. variable levels of challenge as impor- time on each main routine, number of at- tant features. tempts and corrects and errors, and other In designing software for music, considera- aspects. The post intervention data withtion should be given to this issue. students found they enjoyed the lezz.,on and learned the basis behaviours taught, and asShould we as music teachers use comput- well, they identified some deficiencies in ers to teach our precious art? Is there justi- the program. However an even more im-fication for bringing technology into the portant outcome related to the develop-music room? Can the computer teach as ment of goals and objectives for musicwell as the teacher? Can it possibly teach education generally. some aspects in a superior way? "Because of the deterministic character One crucial aspect of music learning is that of digital computer-, identifying and of repetition. The skill intensiveness of the defining musical behaviours seems to subject whether it be for developing the be a natural outcome of planning, pro- motor skill of instrumental playing or the gramming, and producing music les-more intellectual processing of recognizing sons for computer aided education sys-and analyzing sound patterns, demands a tems." high level of individualized instruction and repeated practice which in the past has In the past, many music teachers have been been difficult to provide. Without the di- reluctant to state objectives on the grounds 4 3 5 426 Using Computers Intelligently rect intervention of a teacher the microcom- clans and music teachers to more easily puter can: access the computer. During the 1960s and present a stimulus 1970s, music specific attachments were accept a response developed which were expensive and evaluate the response sometimes cumbersome. Since 1983, the give feedback development of MIDI,(Musical Instrument give immediate and appropriate rein- Digital Interface), has revolutionized com- forcement munication between computer and vari- move to the next instructional sequenceous musical instruments. With suitable or branch to further explanations software, a synthesizer connected to the record results computer via MIDI can be used as a real- And repeat the task endlessly. time music input device for composition or performance, for learning keyboard skills According to Hofstetter (1979) "there is no or music theory as well as other music other discipline for which microelectronicsactivities. Through the music keyboard, are better suited than music education.notation can be displayed and edited, then Although still in the infancy stage, micro- played according to the user's directions. A computers are able to produce music nota-hard copy of the resulting music manu- tion, generate aural stimuli, judge re-script can then be printed. In the case of sponses, and diagnose student learning."learning piano keyboard skills, the com- Given appropriate software based onputer is able to direct the student to per- sound instructional design and with ap- form and provide feedback on the accuracy propriate teaching procedures, many as-of the student's performance in much the pects of music knowledge can be learnedsame way a teacher would. with minimal teacher intervention. There are various devices which can be Whether the new technology will be inte- connected similarly. These include synthe- grated in music education is not in ques-sizers, both monophonic and polyphonic, tion. "The challenge to education in theelectronic keyboards, drum machines, 1980s is how best to use computer course-samplers, sequencers, and even special ware to improve instruction." (Wilson and sound effects devices, each adding diver- Fox, 1982). sity to the computer's regular functions and providing f:-.cilities not readily avail- If music education is to survive, it mustable in the regular music room. Some accept this challenge. Not a panacea for thedevices have MIDI inbuilt while those problems of music education, the com-without may be interfaced by an independ- puter, because of its interactive nature, andent MIDI. There also exists a range of syn- the versatility of teaching procedures, is athesizers for non-keyboard musicians medium unlike any other teaching aid, and which are similar to an auto-harp but hav- holds great promise. ing controls similar to those used by wood- wind players or guitarists.Expanders, Peripherals to enhance performance inwhich are sound devices having no direct music learning controls themselves, can be played by another keyboard or synthesizer to extend its tonal range. Mixers, compressors and In addition to the microcomputer unit itselfequalizers may also be used for specific which includes an attached qwuty key-purposes. For further information on these board to access information, there nowperipherals see N.S.W. Department of exists a number of attachments known as Education publication Music and Comput- peripheral devices which enable musi- 436 r Using Computers Intelligently 427 ers (1987), or Stevens ( 1985, 1987) andThe first reported use of computer assisted Wells and Kemp (1985). instruction in schools was in 1959 in New York State so we have had 30 years to Music editors for microcomputers, thoughacquire some understanding of the poten- not originally created for music education, tials and pitfalls of the technology in gen- also have a valuable contribution to make.eral education. Since the late 1960s, there An article by Mercuri( 1981) gave a compre- has been recorded a variety of rn :sic inter- hensive description of nine such programsventions using the computer as the teach- listing some 37 characteristics which pro-ing medium. Much of the earlier instruc- vided useful information. 'The music edi-tion was given at universities in the U.S.A. tor may be used as a motivating tool forand included such areas as aural training, children to pursue additional and moresight-reading, rhythm perception, analysis varied musical activities." (Upitis, 1983). and composition. In the past, particularly at primary level,Allvin's work at Stanford to teach sight composition and creative musical pursuitssinging (1967), Placek's much reported have not been high on the list of prioritiesPLATO system at the University of Illinois although the new K-6 Music Curriculumto teach rhythm tasks, and the work of has identified this as an important area. It Lamb and Bates, University of Canterbury may have been due to the lack of musicalin England, which concentrated on aspects knowledge and experiences of the class-of melody, harmony and rhythm were all room teacher that this occurred. By theforerunners to later studies which have early 1980s it had become apparent thatgrown indiversity. Ned Diehl's work music editors could give children opportu-developing peripherals for the IBM 1500 nities to experiment with musical ideas and computer to teach articulation, phrasing patterns. Having such programs availableand rhythm for wind instruments, pointed in the classroom may not only assist the to new directions in music teac "ing, (Deihl, children but also the teachers to gain expe- 1971). rience and confidence in such musical ac- tivities. By 1974, PLATO had produced courses on the following topics; Tests and Measure- The most recent curriculum do ..:-..:ments on ments in Music Education (Colwell and music learning both within Australia and Weimer); Instrumental Methods Series - overseas, place emphasis on the learning ofwind and percussion (Peters); Percussion music through listening, performing and Terminology (Fairchild); Violin and Viola composing. The vast array of electronicFingering Drill (Lind); Instrument Recog- musical instruments in their various forms nition (Holden); Micro-teaching in Music including those which can be interactiveEuucation (Dvorak); Critical Incidents in with a computer to provide C.A.I., offer to Music Education ( Cooksey); Hand Signals the music teacher and the student ready for Music Teachers (Flohr); Elementary access to a range of instruments and soundsMusic Fundamentals (Placek and Peters); as well as learning experiences not previ- Part Writing (Rickman); and Jazz Chording ously available. It is important therefore (Rucinski.) Programs being developed at that we accept the new technology, capital- that time evidence versatility. izing on its benefits in enabling children to work with another learning tool. One area to receive considerable attention is that of aural training. Such is the intensity Applications of computer technology in required to master the skill, even prior to music education the advent of the computer, non-teacher 437 428 Using Computers Intelligently centred programs were devised using thea 'fried to have students develop more tape-recorder. Kuhn at Stanford (1974) and 'sensitivity and accuracy" in performing. Hofstetter at the University of DelawareMusical notation displayed on the screen (1975) developed computer programs in together with pre-recorded musical ex- this domain. Studies by Herrold (1973) and amples were devised for flute, clarinet, Kilian and Lorton (1974) documented the oboe, saxophone, tn.mpet and baritone. effectiveness of Kuhn's programs. By 1975, Significant gains especially in phrasing with the help of record keeping by thewere evidenced. Peters (1975) developed computer, Killam et al, were able to re-for PLATO a series of programs for flute, search into how students develop aural clarinet, oboe, saxophone, bassoon, trum- skills and how to establish a hierarchy of pet, horn, trombor.e, euphonium, tuba and learning concepts. Hofstetter's work on thepercussion.Using tutorial format, stu- GUIDO (Graded Units for Interactive Dic-dents were automatically moved to more tation Operations ) system led to researchadvanced materials or branched to reme- by showing that, compared with a controldiation according to their needs. group, students using GUIDO for ear- training, performed signifiLantly higher.Programs in fundamentals of music theory, Computerized aural training (Lees, 1980)employing tutorials and drill and practice concluded that "the computer programproliferated at the tertiary level in the was in virtually every aspect vastly supe-U.S.A. Studies in Set Theory (Wittlich, rior" to her own tape-recorded program. 1974); Composition, programs by Nelson And further, the computer's ability to diag-and White (in Hofstetter, 1975), Appleton nose individual stuc.ant's difficulties andand Alonso (1975); Analysis ( Hofstetter, present examples of that type, was a par-1972); Whitney (1975); Information Re- ticularly significant innovation. trieval(John, 1969, 1971); Automated Music Pe nting (Lelena Smith, 1973) further Taylor (1982) using MEDICI, an interactiveillustrate the variety of music c.a.i. in the computer-based dictation system with ter- 1970s. Bamberger (1974, 1979) explored the tiary students, sought to compare student use of LOGO with a group of children who gains using traditional piano dictation andwere able to manipulate sounds of differ- the PLATO System interfaced with theent pitch and duration by using letters, Gooch Box, a ,..,"' music synthesizer.number and words. The resulting music is While all student shu,ved gains in skills,then played by the program. those using MEDICI did not produce sig- nificantly higher results.However, the Through the 1980s there appeared drteth- ^omputer offers advantages of releasingora of courses to teacn music fundamen- the teacher for other functions, reduces tals. Lamb, (1982) developed an interact:. repetitious teaching and allows students tosystem which allows musical concepts to work accordik ig to their own needs andbe modeled graphically. His program timetable constraints. Woods (1986) inmakes less demands than the LOGO sys- descrqsing a rIllation of young children tem while still offering similar games using of whom 30% were unable to sing in tune,traditional notation. At the same time, stated that less than 3 hours of c.a.i. instruc- work on computer managed instruction tion was sufficient to remediate their prob-began to appear. Recognizing that in indi- lem. vidualized performance-based instruc- tion, students do proceed at different rates Diehl's students (1971, 1973) using theand experience different forms of learning I.B.M. 1500 showed significant gains indifficulties, it was acknowledged that, if instrumental performance. His studiesteachers monitored student progress, they 438 Using Computers Intelligently 429 could assist individuals in need. Arenson's of journals can now be found discussing development of the PLATO-based C.M.I.music C.A.I., there still exists a dearth of led to the conclusion that if students are reported research on specific learning out given self-paced C.A.I., teachers have comes. more time for individualized instruction on a needs basis. Currently in Australia, articles are few and research almost non-existent. The technol- Arensort (1982) examined the effect of a ogy is here, the software is becoming more competency- based education program onreadily available and affordable as well as the learning of fundamental music skillsimproving in some aspects, peripheral with a group of tertiary students using thehardware has reached quite sophisticated PLATO system at the University of Dela-levels, justification of its potential value as ware. His results supported the premisea teachin; medium widely discussed, but that drill and practice procedures in this development is not widespread. area when provided by the computer pro- duced significantly superior results inWith the establishment of the PLATO learning than for the control group. (Programmed Logic for Automatic Teach- ing Operations) at the University of West- We conclude that programs to teach a wiz e ern Australia in 1985, using the GUIDO range of skills have been successful. betprogram- there is an expectation that on- how extensive is its us- In a study of 429going re trch into t he effectiveness af the music departments in the U.S.A., Jones programs will ensue. In a recent publica- (1975) concluded the following; tion (MacPherson,1985), some indication Despite evidence of the computer'sof the extent of the programs to which value for instructional uses, the educa-students can be exposed was given. tional community has been reluctant to accept and develop computer assisted As with any subject requiring the acqui- instruction for music. Reflections of this sition of literate or intellectual skills, the reluctance are manifold: computer is of most use when provid- (1) few music educators are involved in ing a graded series of drilled exercises. teaching by computer systems, In traditional aural programmes at the (2) few students are involved in re- tertiary level, and as reflected on search in computer assisted instruc- GUIDO ear training lessons, these drill tion as learners, activities are mainly associated with the (3) few music educators and graduates recognition of intervals, melodies, har- are involved in research in com- monies and rhythms, and by writing puter assisted instruction, them down via dictation type exercises. (4) few quality .:ourse materials are The ear training curriculum of the Micro available, and GUIDO System consists of five main (5) no formal mechanisms exist for shar- lesson types i.e., Intervals, Melodies, ing computer assisted instructional Chord Qualities, Harmonies, and efforts in music. Rhythms. As an outcome of Jones' recommendations, In recognition of the possibility that this a National Consortium of Computer- type of learning might not suit all students, Based Musical Instruction was formed in a two year study has been implemented to the U.S.A. As worthy as the aims of thisevaluate student progress and attitudes. body appear, many of Jones allegations still Mr. MacPherson enunciated the problem hold tnie. While articles in a wider variety faced by most music educators at tertiary 439 430 Using Computers Intelligently level, that is the wide range of individual each student could assist in raising the differences in dictation ability, especiallyatlevels of music literacy for trainee primary first year level. "When using the computer, teachers. If units were provided outside the students receive individualized attention, lecture room for which students could gain with evaluation and feedback of theirre-credit in their own time, such learning sponses being immediate. Each time a stu- could be spread over time thus ensuring dent masters a particular unit in GUIDO maintenance of both practical and theoreti- the computer automatically writesa star cal skills. next to the appropriate unit on the student's floppy disc.By releasing the In 1987, a group of 120 first year students at instructor from personal tuition, it permitsNepean C.A.E. undertaking the Bachelor him to become more of a"diagnosticof Education Program in Primary Teach- consultant" who can pay greater attention ing, were screened for basic music literacy to specific student needs. skills. A sample of the students has been selected to u. ea program designed to teach There is an urgent need in Ath..ralia to take those skills presented in the pre-test. In up the challenge of such research because1988, after a predetermined number of it is only through empirical evidence that trials, the total population will be re-tested rational conclusions, which will ultimately and data of the two groups compared. benefit students, can be made. There existsWiri:e it would be unwise to predict the some evidence of the success of C.A.I. to outcome, there is sufficient evidence from increase students skills, in overseas stud- other research studies to hope that learning ies. For example, the New England Conser-outcomes will be significant. vatory of Music in the U.S.A. (Lafferty, 1986) reported that students taking c.a.i. in Conclusion aural training increased their skills by 50% over students who did not use c.a.i. Such Music educators have long sought means promises are too exciting to ignore. to increase the level of music skills across all sectors of education. If we are to capitalize Stevens and Sillcock (1985) began a piloton the perceived potential of c.a.i., we need study with members of the Defence Forcedata on which to base conclusions, and School of Music Lo assess and remediate thelongitudinal research which can be used level of aural skills of musicians in the for continued development. The new tech- Royal Aust n Army and Navy bands. nology has more to offer than any previous To date the results have not been reported teaching tool but it needs the help of music but that such studies are beginning at local teachers, musicians, curriculum experts institutions augurs well for future endeav-and programmers who are sensitive to the ours in researching c.a.i. outcomes in music importance of instructional design within learning. the framework of the various modes of presentation the computer is able to offer. Nepean study Without sound, validated design, the computer will not live up to our expecta- The foregoing evidence led this writer totions. Instructional design as an heuristic suppose that c.a.i. in music on the tertiary problem-solving process, depends on in- campus could be used to answer the needs telligent decision making on the basis of of primary teachers in training and that, in incomplete data which must be followed addition to its use as a group teaching by rapid evaluation and revision according mode within the lecture room, individual- to the outcomes. ized instruction of the amount required by 4 4 0 Keyboarding for information: Don't letyour thoughts get away from your fingers Uni Carnegie and Michael Gerrard, School of Business South Australian College of Advanced Education The rationale for the development of a corn- procedure for each student, and some were ; uter-based keyboard training program for first-year Bachelor of Business students is de- simply too expensive. Hence the authors scribed. The textual material was developed for chose to develop a new program with the use by young adults, but has been used success- following limited aims: fully with people whose ages range from 5 to 70 years. A sample of the relevant research into the acquisition of psychomotor skills is provided. Use of all fingers at 15 words per min Details are given of the construction of the after 5 hours computer program which drives a learning package which has, in its first year of use, pro- vided the means whereby in excess of 150 stu- Simple to start and use for students new dents have become functionally literate at the to computers computer keyboard. Immediately available at any of 40 mi- cro computers at the College, without Micro computers are now being used ex- the need for a floppy disk tensively by undergraduates in our busi- ness courses, not so much as a topic to be Easily distributed to external students studied bat rather as a versatile tool to be who use a variety of MS-DOS machines. used to support studies in accountancy, finance, administration etc. One of the authors (Carnegie) has been teaching keyboarding for several decades Most new students are not fluent in usingand had developed a successful method any type of keyboard, but they do need to used with typewriters. This method intro- develop fluency quickly. Since it is d ifficult duces a few new keys at each lesson with- to delete any of the more 'academic' topicsout any tedious drill exercises, students from the curriculum, keyboard skills must type complete sentences right from the first be acquired largely outside formal staff- lesson. The new computer program simply student contact time. translates this proven method into a more modern technology. These constraints lead to the desirability of students learning with a tutorial programThe first students to learn by this quick running on the micro computers whichmethod were technical college lecturers they will be using. While many such pro- learning on manual typewriters so that grams exist for a wide variety of comput- they could prepare lesson material and ers, a quick survey of those readily avail-handouts to be run off on a spirit duplicator able for MSDDOS machines revealed un- - such was the technology of 1964! They desirable features in each case. Some were were predominantly men from the Engi- too tedious, some required an installationneering and Building departments. They 441 432 Using Computers Intelligently attended for a total of 5 hours- just one movements week. By that time they had learnt the whole keyboard, including the numbers Finger dexterity. Controlling tiny objects (which are often left to last and then forgot- with the fingers ten) and gained some basic skills in setting out material. Fleishman's analysis of the stages ofprac- tice on a complex co-ordination task such Since then the method has been usedto as keyboarding further emphasises that teach full-time students aiming to become this is not a simple skill thatcan be left to pLlfessional typists, as well as peoplerang-chance. ing in age from 5 to 70. Some of the people were authors whose flow of ideas was (Illustration - Stages of Practice) constrained by their inability to get their thoughts on to paper quickly enough. OneWhen children learn to ridea bicycle or to person came specifically to learn enough to surf, they usually study what theexperts prepare a list of customers and their re- are doing. They watch and imagine them- quirements for Christmas turkeys. Howselves making the same movements. This much easier it would be for that turkey kind of mental rehearsal has been shownto farmer now, storing the detailson diskbe usefulcstablishittg neural pathways from year to year and updating the lists! for the sp ontaneous eYecuCon of precise movements. Some principles of skills acquisition When the learners pluck up thecourage to Keyboarding is a complex psychomotor launch into action, there is alreadya mental skill. It demands the mental co-ordinationimage of what should happen and they of muscles to produce meaningful alpha-experience kinaesthesis- what it actually numeric text. The complexity of theproc-feels like. They soon learn the feel ofsuc- ess is demonstrated by the factors identi-cessful movements and may practise for fied by Fleishman (1966): long periods of time to perfecta particular technique as they gradually refine the skill Control precision. Highly controlled ad-and develop the automaticity which is the justments of large muscles hallmark of a skilled performance. Multi-limb co-ordinating. Co-ordinating The effectiveness of conscious mentalre- hearsal in relation to passive observation in the movements of more thanone limb shown in the following analysis by Cratty Response orientation. Fast visual dis- (1973) of several research studies. crimination with appropriatemove- ment (IllustrationEffective pr- lice) It has been asserted (Austin & Pargman, Reaction time. Speed of response to 1981) that the most fluent performance is stimulus achieved by the person who is able to leave the execution of the skill to the unconscious Arm movement speed. Rapidity ofgrossself. The conscious self attends solely to the arm movement higher order activities associated, inour case, with word processing or computer Rate control. Following a moving target programming. Manual dexterity. Skilled rapidannClearly, if keying is automatic, with stu- dents having no hesitation in locating the 4 4 ') Using Computers Intelligently 433 required keys, they are going to be able to their intellectual content, but the pattern of initiate the actual movement more quickly. the finger reaches required enables stu- This may be of marginal importance in dents to unwittingly establish the correct computer programming, but where people movements. are word processing to create original documents, then fluency in keying be- Meaningful practice is known to facilitate comes a critical factor. learning and spaced practice is generally more beneficial than massed practice, as Denis Glencross, Associate Professor o; interest and enthusiasm are maintained Psychology at Hinders University, hasand fatigue avoided. Nevertheless, it has carried out a considerable amount of re-also been shown that practice sessions search into the acquisition of typing skills.must be both regular and frequent. He has shown that 'hunt and peck' opera- tors show a marked improvement in both If, as teachers, we know there are common speed and accuracy when they retrain aserror patterns, we are bound to try to elimi- 'touch' operators. nate them. Using the program Because it was common for students to confuse E with I and D with K, for example, The computer program contains 24 stan- a conscious effort was made in the program dard lessons, each introducing one or moreto establish one key before introducing the keys, plus three explanatoi, ;essons and key struck with the same finger on the other two optional lessons. The student chooses hand. E is introduced early in the program the required lesson from the menu in figure because it is the most common letter in the 1. English alphabet and is often miskeyed. I is the key in the same position on the other Each standard lesson begins by identifying side of the keyboard. In this program the D- the new keys and the fingers to be used. E finger reach is established before the K-I The students press each new key, and then finger reach. type seven sentences which use these keys. Any errors are politely identified for imme- While the program adopts what the Ameri- diate correction, and encouraging com- cans call the 'skiparound' method of intro- ments are displayed if the typing is accu- ducing the keyboard, it was felt that there rate. was value in establishing the baseline of the guide keys (sometimes referred to as the The top half of the screen displays a dia- 'home keys'). The value of establishing this gram of the keyboard with the new keys for base in the first lesson is seen as more the current lesson highlighted and theimportant than possible problems associ- home-keys underlined.This technique ated with, for example, miskeying D for K. encourages the student's eyes away fromThe construction of the sentences ad- the Keyboard, thus reinforcing the learning dresses this issue. of the key positions without the student getting into the habit of 'hunting and peck- Preliminary results ing'. In 1987 the new program was the first soft- Sentences are used from the very first stan- ware encountered by 270 business students dard lesson, because of the sense ofin the Introduction to Computing unit at achievement that this gives to students. the South Australian College of Advanced The sentences are, of necessity, limited in Education. Their first 75minute work- 4 3 434 Using Computers Intelligently shop showed them how to switch on theConclusion computers, start the program, select les- sons, quit from the program and swi tch offIt is because of the overwhelming weight of the computer. An average student coveredevidence in favour of 'touch keying' that the first 6 or 7 lessons during this work-we require our students to work through shop. the program described in this paper. We do not want to be accused of allowing them to Students were encouraged to complete allbecome cripples - people whose thoughts the lessons within the following two weekscannot be expressed fully or effectively in their own time; no further contact timebecause they are inhibited by their lack of was devoted to keyboard skills. The pro- keyboard fluency. gram was readily accessible for 75 hours each week, stored on the hard disks of 40This computer program appears to meet micro computers. Although it was up toour requirements of enabling all students the individuals to progress through the to acquire modest keyboard skills without program at their own rate, they were ad- infringing significantly on contact time vised to try to have one or two short ses- needed for other topics. sions (15-30 minutes, 2-4 lessons) each day. Griffin and Keogh (1981) show that student A few weeks later 197 of the students re-perception of demands in movement situ- sponded to a brief questionnaire. 33% of ations may well be the key to understand- these had previous keyboarding experi- ing students' movement confidence, since ence and 36% admitted to spending less individual perception will determine how than two hours with the program after thepersonally important and useful participa- initial workshop. The remaining 31% weretion becomes. Movement confidence is considered to have made a significant at- particularly important for handicapped tempt to use the program to gain keyboard-students who attempt to achieve effective ing fluency. movement behaviour in spite of movement inefficiencies related to their handicapping No attempt was made to measure their conditions. typing speed objectively, but all those us- ing the program believed they could type atAre we not culpable, if we allow our stu- least 10 words per minute and almost half dents to become functionally handicapped felt that they could type more quickly thanbecause we do not enable them to perceive they could write legibly. Casual observa- the value of touch-keying? tion later in the semester showed that the majority of these students tended to use all References most) fingers, whereas the majority of the computer programming students in Austin, J.S., & Pargman D. (1981). The Inner previous years used only one or two fin- Game approach to performance and gers. skill acquisition. Motor Skills: Theory Although these results are largely anecdo- into Practice, 5(1), 3-12. tal, they have encouraged the authors to Carnegie, U. (1987). 10-Hour Typing Pro- continue with the program in 1988 and gram. Melbourne: Longman. monitor the students' performance more Carnegie, U. (1985). Key in for Information. objectively. Other institutions or individu- Adelaide: Unisco. als wishing to use the program may obtain Cratty, B.J. (1973). Teaching Motor Skills. it from the authors on payment of a nomi- Englewood Cliffs: Prentice Hall. nal licence charge. Fleishman, E.A. (1966). Individual Differ- 4 4 I Using Computers Intelligently 435 ences. In E.A. Bilodeau (Ed.), Acquisition of Skill (Chapter 6). New York: Aca- Dr Michael Gerrard has worked as a demic Press. computer programmer and consultant Fleishman, E.A. (1975). Toward a Taxon- for ten years since completing his doc- omy of Human Performance. American torate in business management at Psychologist, 30, 1127-1149. London University.Now teaching Glencross, D.J., & Bluhm, N.M. (1986). In- computing to undergraduates in busi- tensive Computer Keyboard Training ness, he stresse:. the need for students Program. Applied Ergonomics,17(3),191- to develop practical skills by using 194. microcomputers throughout their Griffin, N.S., & Keogh, J.F. (1981). Move- course. He is a member o, the Austra- ment Confidence and Effective Move- lian Computer Society and regularly ment Behaviour in Adapted Physical presents short courses for the Austra- Education. Motor Skills: Theory into lian Society of Accountants. Practice, 5(1), 23-35. Uni Carnegie studied motor skills ac- For a detailed psychological investigation quisition at Oxford University and has of the complexities of keyboarding, includ- an MA in Education from Lancaster ing word processing, readers are referred University. She lectures in office to administration and is currently super- vising a number of student research Cooper, W.E. (Ed.) (1983). Cognitive Aspects projects related to office administra- of Skilled Typewriting. New York: Sprin- tion. She is a Fellow of the Faculty of ger-Verlag. Teachers in Commerce (UK) and a member of the Office Automation Association of South Australia, the Business Education Teachers Associa- tion of South Australia and the Austra- lian Association for the Education of the Gifted and Talented. 453/34211111111611ZURENIUMMIN21511111i01311SUNIMIN Esc 1 2 3 4 S 6 7 II 9 0 - <- fa6 ONERIY010P11 CtrIASOFBIIJKL ...... Shit 11ICVONM, . Shft t Alt Space Bar = = Caps [CASK KEYBOARD SKILLS Press Return key to start the highlightedlesson] US116 THIS PROGRAM C key U key KEYBOARD FINGERING P key kty Rote keys ASDFIL 6 611 3 i keys 2key E K .6 Shift keys 11 key I key N key N try 6 6 - keys CHECK 90161 POSTURE key V 6 4 keys ! key R S, keys Y 6 1 keys T i keys 7 key SPEED TESTS 0 I ' keys I try NUMERIC KEYPAD Metter) 6 Mosher/ 1 key WIT FkON PROGRAM Use arrou-keys to highlight your nut lessonfor Fl for Help) Figure 1 Development of CBT using AUTHOR Tony Mc Sherry, Director, Microcraff Pty Ltd The history and evolution of the AUTHOR authoring system, design considerations and by the author(s) of CAI and not require decisions. The creation of simple lessons and computer expertise. Of course, this does tests is described and demonstrated with par- not mean that AUTHOR users are com- ticular emphasis on the system's word proces- puter illiterates, most of us now use word sor, graphic design system and animation package. processors without being able to program and a certain amount of learning is neces- The state of CBT today sary for effective use the system. I should also be careful not to alienate the computer Some educators believe that the field ofexperts who will wish tc, use AUTHOR Computer Assisted Instruction (CAI) is rest assured that AUTHOR supports a well documented, successful and subject to powerful procedural language with a vari- immutable laws. In fact, the field is in itsety of commands to allow programmers to infancy and its short history is strewn withexploit all the power of the system. expensive failures. While some experts try to create a strait-jacket by interweavingThe evolution of AUTHOR Learning Theory and Instructional Design, I believe that there is still a large amount ofAUTHOR sprang from very mundane trail and error experimentation and data roots. The precursor to AUTHOR was a collection to be done before any Grand limited parser that recognized 3 question Design will become apparent. The volatil-types, presented them to students and re- ity of the field is also enhanced by thecorded results. Tests were written using constant introduction of new hardware. the system text editor (remember line edi- tors?) and results could be examined and AUTHOR provides the subject expert withsorted in a few limited ways. Nevertheless, the ability to directly produce CAI and this was a major step up from my previous allows educators to put their theories into involvement with CAI - developing tests practice. using a general programming language, specifically BASIC-PLUS on aPDP-11 AUTHOR minicomputer. Nor were the tests them- selves at the edge of theoretical knowledge AUTHOR is an authoring system. Its task- they were simple tests for motor mechanic is to allow subject experts and educational apprentices. In retrospect, this type of test specialists to develop, present and manage site was a blessing. We had no Computer Computer Assisted Instruction. Note thatScience, Engineering or Business studies I have not mentioned computer experts,depa. tments toeffectively monopolize programmers and system analysts. AU- hardware and Computer Centre staff. The THOR was specifically designed to be used student interface had to be simple and G Using Computers Intelligently 437 robust to deal with the average apprentice 1. Plan the lesson or test using the built-in who was likely to have no experience of word processor. computers at all. 2. Use the other editors in the system to At this stage Microcraft had produced create required graphics and/or anima- some stand-alone educational software for tions or recorded playback of external CP/M microcomputers and Data General software. minicomputers. Looking for a new project we discussed two possibilities, a wine data-3. Expand your V q into the lesson using base for the discerning connoisseur and a the appropriate commands and ques- generalized authoring system for the tion types. teacher. We decided to go ahead with AUTHOR. 4. Test the lesson to see how it will appear to the student. At that time I was working in the Computer Centre at Richmond TAFE.Microcraft 5. Repeat steps 3 and 4 until you are satis- agreed to provide the software free to the fied with the result. college in exchange for a large test site. 6. Enter the name of the lesson and a short AUTHOR went through five versions at description into the Lesson register. It Richmond TAFE before Version 6 was re- may now be run by students. leased commercially for the IBM PC. Ver- sion 6 supported 5 question types and col- To alter or update a lesson our graphics yet required an external text editor and graphics package for lesson 1.Call up the word processor make any creation.Limited student management alterations and test the lesson. and results reporting were provided. 2. Use the other editors to change graphics Version 7 was released in late 1984 and or animations. incorporated a text editor and graphics design system and also some advanced features such as videodisc and speech syn- AUTHOR Language thesizer support.Version 8 (late 1986) extended what had become the AUTHOR The format of an AUTHOR lesson is quite procedural language and featured a com- simple. Any text that is not a command or pletely redesigned user interface that had a comment will be presented to the student. both function key and/or pull downQuestions are restricted to 8 distinct types menus. Animation and networking were that each have their own format. Various added and the ability to use all the colour commands may be used to extensively graphics and text modes of the IBM PC. I customize the lesson. Branching is usually will not enumerate all the additions that performed using IF.. THEN.. ELSE state- occurred over the years but simply referments but simple branching between les- you to our 750 page manual. I hasten to add sons and randomization may also be used. that the tutorial section of the manual is a lot more digestible at around 80 pages. Tne AUTHOR language currently has over 80 commands. Each command must be Lesson development preceded by a backslash ( \ ).This makes commands easy to distinguish from text in Using AUTHOR is better demonstrated the lesson and also makes the parser's job than described but I will attempt to cover (and the programming) much easier. The the basic method. 447 438 Using Computers Intelligently structure may become clearer with the fol-The question text is finished with a row of lowing example underlines called an End Line within the AUTHOR system. The End Line is used by A multiple choice question which willa number of AUTHOR commands to de- provide feedback if answered incorrectlylineate the end of variable amounts of infor- and then loop until the correct answer ismation. given. Feedback is given to the student for incor- \ -ANSWERS rectly answering the question using a mul- \Beginning. tiple line \IF.. THEN statement. Within \ MULTIPLE this block, text is specified to be displayed 4,1,C on the screen and the lesson will WAIT until a key is pressed. The lesson will then If you specify a command in an AUTHOR return to the Beginning label ( \ GOTO) and lesson but omit the backslash ( \ ) AUTHORask the question again. This process will will repeat until the correct answer is given. a. Give up When we were designing the language we b. Act on the command did go through some mental torment as to c.Place the command on the screen as text whether to include a \GOTO statement. d. Crash the system However, although we fully support the concept of structured programming we \IF WRONG THEN found that the simple GOTO was more understandable than the other structures of I'm afraid not! If you omit the backslash the procedures and \ WHILE... \ WENT loops. command will be treated as text and dis-In the final analysis we were more inter- played on the screen. ested in subject experts quickly producing effective CAI than producing structured \WAIT programmers - even at the risk that the \GOTO Beginning structure of some lessons might resemble spaghetti. \ ENDIF Resources Explanation Graphics, animations and colour text The first few lines are comme nts and will be screens are considered to be resouces ignored by AUTHOR. Comments beginwithin the AUTHOR system and as such with a single quote. are distinct from the lesson. This allows different lessons to use the same material The first command ( \ -ANSWERS) turns without wasteful duplication. These re- off the automatic display of the correct sources are created using the appropriate answer. The next ( \ Beginning:) is simply a editors. label that marks this point in the lesson. The graphic editor has all the features of the Next a multiple choice question is specifiedcommon PC paint programs and is most ( \ MULTIPLE). The next line specifies 4 effective when used with a mouse, how- choices and 1 answer required of the stu- ever all functions may also be u sed from the dent, the answer C. You may, of course, keyboard. The editor currently supports have multiple answers. all CGA and EGA graphic resolutions and Using Computers Intelligently 439 the new VGA 256 colour mode will bestage, that it would become the default released in early 1988. standard for hardware and software. Education and training departments have The animation editor allows you to ani- never been endowed with large budgets so mate up to 4 objects on the screen at once specialized hardware and expensive soft- with internal animation of each object. ware could only appeal to a limited number of people. We did however, require a col- Design considerations and decisions our graphics adapter and monitor as we believe them to be essential to the process of An integrated approach vs stand-aloneCAL AUTHOR does support more expen- lessons sive peripherals such as videodisc players and video digitizers, but effective lessons An authoring system should be integrated can still be written without the aid of these with built-in management of students anddevices. results. Use of the function keys and/or pull A number of authoring systems are similar down menu to general language compilers - i.e. they Functions within the various editors produce standalone lessons that may beshould be rationalised and assigned to distributed. This approach means that les- common keys. Users should have a choice sons cannot be altered and usually exist on of interfaces including use of the function floppy disk. The process of integrating keys with the aid of a template and a mouse these lessons into an easy to use structure is or keyboard driven point-and-choose generally left to the user. This is beneficialmenu system. The interface should not to the creators of the lesson but not veryclutter the user's work area. Novice users useful for educators. AUTHOR provides a and those with mice tend to prefer to use standard user interface and management the pull down menu. The menu can be of students, lessons and resuliz.Othercalled at any stage. As soon as the user software may also be defined as lessonsselects the required function the menu dis- within the system and accessed in the same appears from the work area. Use of the way as AUTHOR lessons. menu is slightly slower than use of direct keystrokes, especially when the user Interactive lessons vs production ofknows the keystrokes without reference to printed tests the template. This mode is preferred by experienced users. In the early 70's hardware was expensive (especially colour graphics hardware) and Independence of the user a number of systems concentrated on pro-The system should be directly used by the ducing printed tests for students whichsubject expert with no need for computer could be subsequently corrected when the experts. studeni returnedto enter his answers into the system. These systems emphasized Great care was taken to ensure that lesson statistics and had little interactivity. AU- and resource creation was straight forward THOR is more concerned with educationand that the system could be used at a than testing. simple or complex level. Use of standard hardware Graphic creation We deliberately chose IBM PC compatibil- Graphics should be drawn rather than ity because it was obvious, even at an earlyprogrammed. 449 440 Using Computers Intelligently The use of rubber band functions and stan- dard graphic shapes such as circles, boxes, Tony Mc Sherry, is one of the directors ellipses and lines meant that graphiccrea- of AUTHOR, Computer Manager at tion is interactive, fast and possible for Richmond College of TAPE and co- those who feel they lack drawing skills. creator of AUTHOR. Tony has been working in the area of CAL for ap- Networking and remote delivery proximately eight years. He is also qualified and experienced in the fields The system should be able to be networked of psychology, literature, media stud- and make remote delivery of lessonspos- ies and video production. sible. AUTHOR will use any network which operates under DOS 3.+ allowing central storage of student records and lessons. Any workstation on the networkmay also be used for authoring by instructors with password access. Remote del very of lessons is possible in two ways.Standard AUTHOR lessons may be sent on floppy disk or via the .tele- phone system for use with a remote AU- THOR system. Alternatively lessonsmay be distributed independently of the AU- THOR system using the Lesson System. This system is available to commercial developers. Conclusion AUTHOR is currently used in over 140 organizations across Australia and over- seas. T -ganizations range from univer- sities a illegeo to banks and insurance companies. The people using the system have a variety of backgrounds and many have limited experience in using comput- ers.All have used AUTHOR to create effective Computer Assisted Instruction. 5 0 Diverse CBT Applications using AUTHOR John D. Rundle, Director Microcraft Pty Ltd 41wwwsamos.m TheAUTHOR usercommunity has responded The lessons are in everyday use, and have enthusiastically toAUTHOR'sflexibility. Five been developed without help from com- specific lessons are to be demonstrated as devel- oped by AUTHOR users.AUTHORis used in a puter specialists, and in most cases without wide variety of subject areas. attending specific courses in the use of AUTHOR. When we originally designedAUTHOR,our main design criteria were to maximise flexibil- ity and user control of lesson presentation. Our I would also like to thank the authors for users have responded enthusiastically to this giving me permission to demonstrate the flexibility and have outstripped us in discover- following lessons. ing innovative ways to useAUTHOR.When they have been unable to achieve specific goals, we have been badgered into further developing Basic Electrical Circuits. Developed by: AUTHOR,making it even more flexible. Mavis Bird at Richmond College of TAFE. This lesson provides trainee As part of our contact with our user com- auto electricians with an introduction munity, we often view lessons or presenta- to electrical circuits. tions developed with AUTHOR. Our users have provided us with some excellent ideas Sheep Dentition. Developed by: John and suggestions, some of which are beyond Kent at the Melbourne College of Tex- the capability of computers today, but tiles. A lesson covering sheep dentition which we will implement when suiable as related :n wool types and the age of hardware becomes available. sheep. The Applications Key Facts, A series of lessons on Diabe- Developed by: Dr Matt Cohen, Today, I would like to show you five les- Director of the Lions International II,- sons selected from a wide range of avail- betes Institute Education in self-man- able subject areas to illustrate AUTHOR's agement for diabetics. versatility. Chemistry Familiarisation. Developed I must stress that these lessons have been by:J A Robinson and S Kobot of developed by our users and while I might Queensland Institute of Technology. suggest ways to improve them, I am pre- Familiarisation of students with AU- senting them as they have been developed. THOR and CBT. None of the developers do not have any Basic Indonesian. Developed by: Jo special training in CBT, but they are experts Fischer, Sacred Heart College, Ballarat. in their own subject areas. An introduction to basic Indonesian vocabulary. 4 442 Using Computers Intelligently The Demonstration Our approach, embodied in AUTHOR, is that the subject experts can develop CBT if I will work through parts of the five lessonsgiven the right tool. as a student, making some deliberate mis- takes (and some not so deliberate) so that you can see the lessons as the students see John D. Rundle is one of the directors them. of Microcraft Pty Ltd, and co-creator of AUTHOR. John's previous experi- I will also comment on specific features, ence includes working for Norwich plus ways of presenting information and Life Insurance (PC Coordinator), BHP asking questions. (Information Systems Training Offi- cer) and the Victorian Education De- Conclusion partment (Maths and Computer Stud- ies teacher). John is particularly inter- Some of the major subject areas covered by ested in the application of PC's in AUTHOR users include the following: providing solutions under the direct control of the user. Accountancy Banking Chemistry Computer Studies Computer Training Diabetes Drawing Electronics Engineering Internal medicine Insurance Languages Law Library and information science Mathematics Physics Scier cc Wool classing AUTHOR lessons are easily updated and modified and may be transported to any AUTHOR system. Lessons may be sold, exchanged or donated at the author's dis- cretion. Microcraft Pty Ltd publishes the "AU- THOR Directory: A guide to AUTHOR users and lessons" based on results of sur- veys of our users. Please contact us if you would like a copy. 2 The Advanced Features of AUT!!OR Liz McArthur Director, Microcroft Pty Ltd The use of videodisc, robots, gr_2hics and Using videodisc with AUTHOR speech ,Yroduction with AUTHOR will be d scribed and demonstrated. The \ VIDEO command is used in the les- son whenever access to the videodisc is Videodisc required.This is followed by the com- mands used by the specific player. For Videodisc technology allows fast, directexample, the following command for an access to video and audio information.LD-V4200 player will cause the player to This makes it very suitable for use withsearch for frame 5000 and then stop. microcomputers for computer based train- It complements the use of the cora- \VIDEO 1 2r by offering high quality sound and FR5000SE speech and "real-life" images, both still and moving. The videodisc may also be used to hold over 50,000 still frames making it ideal Menus to the video material may be cre- for picture and graphics storage. ated, questions may be asked and se- quences reviewed if the student answers a Practical application question incorrectly. For example, if the student answers the following question Until recently videodisc technology hasincorrectly the section of the videodisc primarily been used by large organisationswhich deals with this topic is played. such as IBM and GMH which could justify the high cost of production, firstly to create \ TRUE/FALSE a master and the.t duplication of the vide- T odiscs. An adequate fuel supply is one factor that However, smaller organisations can snakecontributes to efficient combustion. cost effective use of videodisc technology by re-authoring existing videodiscs to suit \ IF WRONG THEN their needs. Unfortunately you are wrong. Strisce a key AUTHOR offers specific support for theto review efficient combustion. Pioneer LD-V4200 wad can be used with any player which can be controlled via the \WAIT computer's serial port. This means that any \ VIDEO CAV (Constant Angular Velocity) disc may FR25443SE be used as a source of still frames and 26080PL motion picture sequences for severs? les- sons. \ ENDIF 453 ,-- 444 Using Computers Intelligently Some features: The combination of microcomputer and videodisc offers a powerful educational Customisation tool.Its use is currently restricted by the cost of production of videodisc material, The \ DEFINE VIDEO command is usedtohowever, the material already produced assign the functions of a particular player canbe used in a variety of ways. Tworecent to any of the function keys on the keyboard. developments in the field suggest that the For example the following commandas- use of videodiscs may increase dramati- signs the play function to the F4 key, stillcally over the next few years. Theseare the frame reverse to F5, still frame forwardtoability to store an hour of video materialon F6, fast speed reverse to F7 and fast speeda Compact Disc and the demonstration of forward to F8. different systems for producing erasable videodiscs. \DEFINE 1/ ID E 0 F4=PL Robots F5=SR F6=SF F7=180SPMR AUTHOR lessons may incorporate the control of external devices such F8=180SPMF as robots for industrial training.Devices may be controlled through the RS-232 serialport Text overlay from an AUTHOR lesson using the \ PORT and \ TALK commands. Text may be overlord on the video image using AUTHOR if your player (e.g. T.DThe \ PORT command is used to setup the V4200) has th±e. facility. Text overlaymayprotocol between the rc bot and computer. be used to create titles, arrows and otherThe \ TALK command is used to send the markers to point out areas of interest and robots control language. display menus. Example: controlling the Sal robot Direct student control \ TALK The student may be given direct control ofM 0,200 the player from the keyboard by including the \ KEYS command in your lesson. You may choose to provide ale student with a This l-l-/ um end will move the robots shoul- table of contents on the AUTHORscreen -der from position 0 to position 200. listing frame numbers with subjects, al- lowing them to view sections at theirown Graphics pace and analyse some parts in detail. Functions include: AUTHOR supports a range of graphic screens and special effect screen wipes for Play changes between screens. Graphics may be Reverse simply displayed, overlayed with ques- Still frame forward tions or text or presented as a series of Still frame reverse slides.The following modes of graphic Fast forward and reverse screens are available: Frame forward and reverse Scan Text character graphics (created with CED, Show frame numbers the Colour Text editor) 4 4 Using Computers Intelligently 445 40 X 25 character 16 colour CGA or can resolve and the number of shades of EGA grey it provides - grey scales. The more 80 X 25 character 16 colour CGA or EGA grey scales the more life-like the image will appear. Pixel based graphics (created with GED, the Graphic editor) Using the PC-EYE 320 X 200 pixel 4 colour CGA or The image may be captured using a stan- EGA dard camera for input to the graphics edi- 640 X 200 pixel 2 colour CGA or tor. The displayed examples use only 4 EGA 640 X 200 pixel 16 colour EGA grey scales at a resolution of 640 x 350 640 X 350 pixel 16 colour EGA pixels. A video digitiser can saves substan- tial amount of time in creation of graphics. Special effects Practical application Special effects are used to add interest and continuity to the graphics being displayedBest results are achieved with black and and to create the impression of animation.white line drawings. Twenty four special effects may be usedVoice Production with GED screens ranging from vertical or horizontal wipes to a dissolve. no speed There are two ways of creating speech - of the effect can be controlled. digitisation and synthesis. Speech digitisa- tion samples real speech and converts it to Fx Description digital information. This gives high quality speech but requires a large amount of disk 0 Instant load to screen space for storage of the information. 1 Wipe from top to bottom Speech synthesis builds up words from 2 Wipe from bottom to top phonemes. 3 Wipe from left to right 4 Wipe from right to left Using the VOTALKER IB speech synthe- 5 Crush from top & bottom sizer 6 Crush from left & right 7 Expand to top & bottom The VOTALKER comes with a standard set 8 Expand to left & right of words and the user can also build a set of 9 Crush words which may be modified using the 10 Expand phonetic editor. The rate of speech, ampli- 11-20 As in 0-10, but with edges tude, inflection and filter can be controlled. 21 Fade 22 Interleave from left & right The \VOTALKER command is used to 23 Diagonal wipe from bottom. left send text to the speech synthesiser. 24 Diagonal wipe from bottom.right For example. Video Digitisation using the PCEYE card. \VOTALKER Video digitisation is the process of convert- 2 ing visual images to digitalformation. Hell[I=5] o.[i= 0]This is the VOTALKER The sophistication of a digitiser is ex- speech board. pressed in the number of dots or pixels it 455 446 Using Computers Intelligently The VOTALKER will say this sentence with an upward inflection on the o in Hello after an initial delay of 2 seconds. Practical application Realistic speech can only be obtained by patient use of the phonetic editor. Conclusion AUTHOR may use a variety of perhipher- als as lesson resources. Any device that has an RS232 serial port may be controlled from within an AUTHOR lesson. We are com- mitted to the continuing evolution of the AUTHOR system. As new devices which promise to enhance the lez.ming process appear on the market, they will be incorpo- rated into the AUTHOR system. Liz McArthur, is one of the directors of Microcraft and co-creator of AUTHOR. Liz is also qualified and experienced in the areas of psychology, history and philosophy of science, librarianship and production of audiovisual media. 4 r Object-Oriented Control for Intelligent Computer Assisted Learning Systems Tom Richards, Dept of Computer Science, La Trobe University Geoff Webb, Computing & Information Studies, Griffith University Noel Craske, EDP Depeartment, Chisholm College ABSTRACT This paper investigates an approach to providing a general-purpose authoring/tutoring shell for intelligent computer assisted learning systems.The approach is tooutline an object-oriented representation of task/goal hierarchies, then to consider the ways in which domain expertise, student information and teacher expertise can be made to interact with such hierarchies. The result is a skeleton in terms of which some exploratory lessons are being constructed; and in terms of which further research on the domain expert system, student modelling, educational expertise modelling, and user interfacing can more concretely be developed.1 Introduction Most current work on the construction of Intelligent Computer Assisted Learning (ICAL) systems has concentrated on one-topic systems: diagnosing faults in a particular electrical circuitthe SOPHIE system (Brown, Burton & DeKleer, 1981), solving the WUMPUS game (Goldstein, 1981), and ACE, which discusses the analysis of nuclear magnetic resonance spectra (Sleeman & Hendley, 1981).These systems and others like them were not intended as vehicles for widespread educational use; rather, they are vehicles for investigating and solving important research problems: natural-language communication, modelling the evolution of procedural knowledge in a learner, and dialogues to monitor and guide student reasoning, respectively. Those are 1This research is supported by grants from the Apple Educational Foundanon, Australian Telecommunications and Electronics Research Board, La Lobe University School of Mathematical and Information Sciences, and the Victorian Department of Industry, Technology and Resources. 457 448 Using Computers Intelligently problems that need solving first ifwe are to use artificial intelligence effectively in education. In the EXCALIBUR Project. byway of contrast, the authors and others are investigating what is involvedin providing a general-purpose ICAL system, one which providesan authoring and knowledge engineering shell, analogouslyto a standard expert system shell, anda delivery system for students (Richards,1986a). Such a system is intended to be general-purpose in that the architecture andknowledge representation methods used should involveno specific assumptions about the content or nature of the subject domainto be taught, nor about the modes of learning or presentation(instructional, student-driven, exploratory, problem-oriented, etc)to be used. To use such a system, aperson whom we will call the author will interact with the system tocreate what for lack of a less peijorauve word we will call a tutorial a program which can then berun by a student with the aim of learning from hisor her interaction with it. Westress that it is up to the author to decide theextent to which the student's interaction with the tutorial is to be constrainedby (for instance) predetermined delivery sequence and studentassessment results, in which case the author will need to specify instructionalmodes or rules to the system;or to which the student may treat the tutorialas an educational information kit and tool bag, in which case the author willneed to specify whatresources and modes of explorationare available to the student. The pointis that a general-purpose ICAL system should placelittle constraint on the educational philosophy the authorwishes to adopt. The question is: how mightone build such a general-purpose system? The main problemswe have identified, for which solutionsmust be found that do notcompromise the sought-after generality,are: (a) How to store data on the subject domainof the tutorials, and how to perform analyseson that material in ways relevant to the learning process at hand (designingthe Domain Information System facility, whichin many cases will amount to an Expert System or Knowledge BasedSystem); (b) How to construct a knowledge baseabout the student, including what the student believes andhas learnt in the domain of the tutorial, and why and how.Tnis facility is called the Student Model. (c) How v) incorporate rules about studentknowledge and performance criteria that can beused to tailor or guide the presentation of the student sessionin the light of information Using Computers Intelligently 449 gathered about the studentso far. This is the Mentor facility (Sussex, 1987). (d) How to unify these undera System Driver which handles authoring sessions as wellas student sessions. The Domain Information System isnot necessarily the same as the material which the student may learn. Often,it may be much more extensive, for the samereason as a teacher usually needs to know far more about a subject than s/he will teach, just inorder to handle the teaching process well. In the case of an ICAL system, the DomainInformation System may often need to contain intelligentsubsystems enabling the computer to sulve problems whose solution is neededfor the tutorial interaction, but whose solution method iseither far too advanced for the student or is a computer-orientedprocess that can be quite confusing to a human. Crudely put, machines and peopledo not think alike. Examples are: The parsing of a given Frenchsentence in a French grammar tutorial, in order to generate the specificgrammatical and pragmatic information the student needsfor the interaction about thZsentence.Machine parsing methods are usually totally different from those used ina French language tutorial class, even if the end-products (parts of speech,agreement relations, etc.) are much the same. The checking or production ofa proof step in the derivation of a proof in a tutorial on mathematicsor formal logic. A student may, for example, be learningone particular method of doing formal logic, such as natural deduction;yet the tutorial system will use a pattern-matching procedure which isnot in the least educationally relevant, to check the student's proofsteps. We cannot easily do without thespecial expertise of the Domain Information System. If we didaway with the French parser, thenan author could provide the parsing informationneeded by the student for each stored sentence; but then thatinformation could not be generalized to cover sentences typed in by the student. Inthe logic example thingsare worse, for proof checking informationcannot in principle be provided by the author: for non-trivial derivationsthere can be an infinity ofcorrect "next steps". Only a special proof-checkingprocedure will do the job. We turn now from considerationsabout the material to be learned. to questions of how to learn it. In Frenchgrammar, logic. and indeed courses on most topics there may bemany modes of learning. The student might be, for example: .4 5 9 450 Using Computers Intelligently Working through a set system-controiledsequence c.,t. actions, :'reely exploring an environment containingknowledge and techniques relevant to the subject domain, Trying to devise his/herown techniques for reachinga goal, and testing them. The more of such learning modesas these that are made availableto the student, the more demands are putonto the content and problem solving power of the underlying Domain Information System.But even more stress can be put onto the other three facilities, the StudentModel. the Mentor, and the System Driver. In the rest of this paper we will lookat a method of designing and unifying these four facilities. The resultis the specification ofan architecture for a general-purpose ICALsystem. Goal/Task hierarchies One place to start in tryingto analyze this problem is the idea ofGoal/Task hierarc.zies. See Figure 1. A tree-structured Goal Task hierarchyhas at its root the overall goal of the current tutorial. and the (immediate)dependants of any nodeare the subgoals that need to be achieved inorder to achieve the node'sown goal. On each node is a task, whichto be carried out requires that thethe tasks on the dependant nodes have been executed.The terminals of thetree are initial goals which have no prior goal:tasks which can be executed without performing prior tasks. Formally.the links from a node Nto its dependant nodes N1, N2,... mean that the tasks on the dependants are individually necessary and jointly sufficientfor the execution of the task on N; so a task hierarchy is an and-tree. anda link from N to a dependant node N1 means that achieving the goalon N1 is a necessary condition of achieving that on N. That isto say, in order to achieve the goalon N it is (logically, pedagogically, epistemically,or ...) necessary to achieve.or to have achieved. the goals on N1, N.) and when they have beenachieved it is (logically, pedagogically, ...)possible to achieve the goalon N by doing the task on N. The relation betweenthe goal and the taskon a node is simply that the goal is what hasbeen achieved when that task has been executed. given that the subgoalson N1, N2. ... have been achieved. Any tighter definition of the linkage relationin a goal /task hierarchy is not really a matter for the system designer.but for the author of a tutorial. 4C () Using Computers Intelligently 451 who will wish to exploit itas a control structure thatmay be adapted to his or her ideas about the flow of the tutorial. In Figure 2 we have a small example of a Goal/Task hierarchyfor detecting subject-verb agreement in number, which illustratesthe points just made. The knowledge neededto determine agreement of subjectand verb (the goal on N) is the gender and numberof the subject (left-hand sub-goal) and the gender and number ofthe verb (other sub-goal).When they are known, a comparison of themto see if they are thesame (task on N) suffices to reach the goalon N. Suppose now a student is attemptingto use an ICAL program based on this Goal/Task hierarchy to learn aboutsubject-verb agreement. We imagine that the student is presented with a series of examplesFrench sentences that may or may not obey therules about agreement a.,,d is required to move through the hierarchywith each example. (The pedagogical effectiveness of thisapproach is not up for discussionat present: we simply need zn exampleto motivate this description ofour system.) Then on each node we need to record a method ofdetermining if the student is correct in his/hertask solution. Oneway to do this might be to look up the answer as entered by thetutorial author ona record containing the current sentence example (S1 of Figure 3 isa record containing the needed data). Moreintelligently, sucha frame might be created by a parse procedure belongingtc the underlying Domain Information System (Figure 4). Figure 3 contains (sketchily) three types of control (or, ifyou prefer, information). 1) Domain information,as in the record Si. 2) Tutorial driving procedures,as in the Student Test slot of N 1. 3) Student modelling services.as in the Student Record slot of the frame Ni. Each of these three controlregimes may embodya great deal of informauon, procedures. and thelike: but as Figure 3suggests. each regime intersects witha Goal/Task record. suchas Nl. in a simple way. The architectural design ofthe system is basedon this. brings us to the central design issues of thesystem. 452 Using Computers Intelligently Object hierarchies and Type-systems In Object-Oriented ProgrammingLanguages (OOPL's:see Byte, 1986; OOPSLA, 1986) the entry that does the workis not a procedure,as in procedural languages, butan Object (with a capital 0 to avoid ambiguity) containi.ig local data (like Pascal records) andprocedures (called Methods), which are accessed and activatedby Messages (like command lines) sent by one Object to another.and returning values. Objectsare similar to the sets and individuals ofset theory, in that one Object can bea member (usually called an instance) ofother Objects; or it can bea subset (subtype) of other Objects. For instance,an Object containing information about the behaviour of the Frenchverb voir will be an instance of another Object containinggeneric information about the behaviour of transitive verbs. That Objectin turn will be a subtype of another Object containing generic informationabout the behaviour of verbs in general. Note the differencebetween an instance anda subtype: voir is an instance of a transitive verb.and of a verb. It is nota subtype of either. And a transitive verbs,as a type. are not an instance of verbs, they are a type of verb. Subtype relations form a hierarchy (see Figure5), with instances as the terminal nodes of the hierarchytree. Objects in the tree normally inheritproperties belonging toany higher Object; for instance if Verb in Figure5 recorded that all verbs have subjects, that would hold of transitiveverbs and intransitive verbs,and all instances etre and voir as well. So if a query were addressedto this tree: 'Do transitive verbs take subjects?''he Transitive Verb Object would receive this query message andanswer 'Yes.' If the lesson designer wished to treat the verb etreas an exception to the rule about subjects. s/he records on the Etre object that itdoes not take a subject. and the inheritance of the contradictoryproperty from Verb is blocked. The query 'Does etre take a subject?' wouldgo to the Etre Object. which would directly answer 'No,' withoutconsulting further up the hierarchy. In the same way procedures can be stored on the appropriate Object.and get consulted by lower objects unless overridenby a lower procedure. The procedure for findinga verb's subject belongs on Verb. andthe procedure for finding a verb's objectbelongs on Transitive Verb. Objects that are instances typicallyrecord different sort of information from objects thatare types.Instance Objects record information specific to the thingthat Object represents. buta type Object 4 6 1 Using Computers Intelligently 453 will record generic information true of all instances of that type. The type Object Student Model would record the pass mark needed for any student to succeed in the current tutorial; but an instance of that type. such as Student36, would record that student's name and mark achieved. In our ICAL design, the Goal/Task hierarchies are not Object hierarchies as described above, even though the nodes (see Figure 2 and NI of Figure 3) are Objects. So from now on we will call them Goal/Task Objects. The hierarchy shown in Figure 2 is plainly not one of inheritance, for Ni does not inherit anything from N. The link as remarked earlier, is that the lowerGoal/Task Object contains a goal that is a necessary condition for attaining that in the higher Object. A goal. filling a slot in a Goal/Task Object such as N1, is however an instance of the type Object called Goal, which specifies the structure of any of its instances, which are of course individual goals. (See Figure 6.) Figure 6a shows a type Object called Goal containing a number of structured Instance Slots, which describe generically the slots occurring on each token of this Object that may get created. i.e. on any individual goal Object. Goal5 is a token of Goal; and what the Goal slot of N1 in Figure 3 really contains is a pointer to Goa15. To create Goal5. the Instance Slots of Goal are called upon to define the structure of Goa15. and so is any Object of which Goal is a subtypein this case Universe, which is the unique upper bound of the inheritance lattice. This is the method used to create any new Object. The Own Slots contain data or Methods belonging to that Object itself; these slots are the ones defined by the correspondinggenenc InstanceSlot data on the higher Objects. One Object can access the data in another Object's Own Slots, or activate Methods stored there. by sending it a Message, containing arguments for the Method's parameters if needed. For example, since the Goal slot of a Goal Task Object pointsto some instance of Goal, Goal/Task Objects are designed so that theycan send messages to Goal. For example, the pointers to Goal/Task nodes making up the Goal/Task hierarchy (as seen m Figure 2) are determined by a Goal/Task Object sending a message to the instance of Goal that its Goal slot points to. In this way, a Message from N1 to its Goal instance G5.to find G5's supergoal. will return N's goal. Then a Messageto that coal to report its subgoals will return G5's "sibling subgoals" (just one, in the example) which must also be achieved by the student before the goal of N can be attempted. Messages can now be sent from N1 to each of those subgoals. accessing the Goal/Task objects they belong to (just N2) and 483 454 Using Computers Intelligently checking which of them the student has alreadyachieved. This helps to determine the flow of tutorial control. Where does the code for thesemessages reside? In OwnSlots in the type Objects Goal/Task and Goal, to be inherited by theirinstances. For example, the Method to find a Goal/Task Object'ssiblings resides on the type Object Goal/Task, and looks like this, in pseudocode: Get Siblings: supergoal := ask Goal for its supergoal Object; sib-goals:= ask supergoal for list of its subgoal Objects; siblings := collect into a list: for each sib in sib-goals: ask sib for its Goal/Task Object; return (siblings less self). Now if Ni is sent the messageGet Siblings itwill return the list (N2). This example of how a slot in an Objectcan point to another Object. which belongs to a particular hierarchy of Objects.illustrates an important aspect of the architecture ofour design. We have identified a number of Object hierarchies, or clusters thereof,that are needed in our ICAL structure2. Major ones are: Goal type-systems, discussed in detail above. Task type-systems, defining activities for thestudent. A type-system to describe the Student Modelsfor individual students and associated diagnostic procedures. Logic type-system to provide rule frameworksfor a logic- programming reasoning service whereneeded. Sets of these rules compnse the Domain Expert System,reasoners about the student's belief base, and the analysersfor a natural language interface. Object networks We remarked earlier that the hierarchy ofGoal/Task Ot jects wasnot linked up by inheritance. so in present terminologyit is not a type-system. Type-systems are used to define Objectsas types. subtypes, andultimately instances, and to provide through inheritance powerfulbuilt-in reasoning services. But the hierarchy of Goal/Task Objectsis effectively a simple semantic network: data nodes joined by labelled links. 2A rype systemhere is a sublatnce of the one overall Object latnce.and it has a single upper bound which is some very significant type Object suchas Goal or Student or LogicRule. G 4 Using Computers Intelligently 455 Overall, the system we are designing can be viewed as comprising several connected semantic networks or conceptual graphs (Sowa, 1984), which we will call Object networks. Major Object networks in a tutorial program are: The underlying Domain Information System The Student Model facility The Mentor facility The System Driver The Goal/Task network Human/machine interfaces Natural language processor Reasoning systems, primarily for the Domain Information System if it is knowledge-based, and for conducting rational dialogues with system users. Concentrating attention on these Object networks as the architectural framework of an ICAL system makes the type-systems disappear into the background. On the other hand, if you concentrate on the type-systems, then you see the Object networks as largely derivative structures, being created by the pointers in slots of the Objects as created. maintained, and manipulated by the type-systems. Look at how we found the siblings of a particular Goa /Task Object, which illustrates how the Goal type-system defined the Goalfrask Object network. Figure 7 tries to depict, on two-dimensional paper, the multi- dimensional structure of the resulting system. It would be better to visualise slots (in Target of Figure 7. for instanc;.., not as containing pointers, but containing the Objects poi-ted to. and to visualise each type- system as lying in a different plat,r dimension. The Object networks cannot be entirely determined by the background type- systems and slot pointers; but to the extent that they are. the authonng task for a given tutorial is simplified and automated. The authoris presented with ready-made prototypes for type- systems. e.g. the type Objects GoaL'Task and Goal. and needs to specify instances and whattoes into the Own Slots of those instances. That can be more likea data-entry 4 6 5 456 Using Computers Intelligently task than a programming task. To the extent that Methods need to be developed by the author of a tutorial, typically in an intelligent Domain Information System, then to that extent programming or knowledge engineering is needed. At its simplest, then, the tutorial construction procedure for an author is: Construct a Goal/Task network for the tutorial domain. Construct a Domain Information System which will provide the domain data and processing that a student (or the system) may need at each node of the Goal/Task network. From the point of view of the system designer, who has to think about how to make the authoring process simple yet powerful. three important research goals that we are exploring are: To devise skeleton Goal/Task node objects that put few constraints on the type, content, or style of tutorial to be developed around those tasks, yet which allow the author to express the cognitive structure of the tutorial with a minimum of special programming or modification. To provide a clean interface to the Domain Information System that allows domain information to be entered securely; and in a way that permits reasoning on that information and creates suitable output for student users. To develop an authorisystem interface that requires as little learning as possible by authors. while giving them as much power and control as possible over the system and the tutorial they are creating. To do this very high-level editors and browsers are needed. such as network displays and network editors. To illustrate the problem of designing authoring facilities for a general-purpose ICAL, and the way in which an Object-based architecture helps, we will now look at the queson of providing various tutoring strategies or methods for any given tutorial material. Provision of Tutorial Strategies The object-oriented approach allows several useful features and procedures to be developed. virtually free, as aspects of the OOPL methodology. 4G6 Using Computers Intelligently 457 The Goal/Task hierarchy can be generalised to an and/or dir.anh, in the manner of the Feature Networks of (Webb, 1986). "1i's now allows us to represent any structure of necessary and sufficient conditions between nodes. which give great freedom in representing the epistemological structure of the concepts, skills. etc. being tutored. Do First and (inversely) Do Next slots on a Goal/Task Object can contain pointers to other such Goa/Task Objects, thus controlling lesson sequencing in a rigid manner. Alternatively, the GoaVTask hierarchy can be used to provide default tutorial sequencing.If the type Object for these nodes has an instruction to do first the dependants of a node (if any), then a bottom-up tutorial regime is established as the default. Or, an instruction there to do the parent first establishes a top-down regime. allowing the student to avoid doing explicitly the subcaslcs of a task s/he can do in one hit. A Tutorial Strategy slot on a Goal/Task Object can contain code for an explicit tutorial st.ategy, to override the contents of Do First and Do Next slots, but accessin_ them if needed typically under certain conditions laid down by the author or inferred from the student model. Do First and Do Next slots will in general create a pointer universe for the Goal/Task Objects that organizes them as a directed lattice or even a general digraph, with lower bounds and upper be nds representing the first and last things to do. This ca.n be even more complicated if these slots contain disjunctions or conjunctions of pointers. A system-provided graph search algorithm will act as a tutorial planner for the student, finding feasible paths to the root task/goal node(s). In general. a student will have some task to do at a Goal/Task node, even if it is only to read something. Slots on the node can specify different tasks for different tutorial strategies. or for different conditions that the student is in: slog such as Remediate. CleckKno ledge. Test Knowledge. TellStudent. Access to these will typically be determined by the ode on the Tutorial Strategy slot. Thus. except in rather primitive lessons. the GoaVTask network need not of itself define any sort of sequencing of student tasks. The task(s) at a GoailTask node can recursively be defined in terms of another "lower-level" GoaUTask hierarchy. 1 6 7 458 Using Computers Intelligently effectively providing different levels of abstractionor detail at which a student might proceed with the tutorial. Costs can be statically or dynamically associatedwith nodes on the graph, reflecting difficulty of task, desirability of doing the task, etc. The graph-search algorithmmentioned above will then seek minimal-cost pathsfor the student. Any slot on a Goal/Task node can havea User Access? facet on it, indicating whether the user may access the information or execute the procedure in that slot. This providesa control over the openness of the tutorial material to the student, and the extent to which they cai. browse, sele-ttheir own learning method. etc. Sucha facet may of course contain code, not just Yes or No, determining theaccess conditions for that slot, and distinguishing betweentypes of user, such as programmer, tutorial author, and student. Authoring a tutorial can be treatedas one mode of use of the tutorial. This allows a studentaccess to the authoring tools. and a high level of access to slots in the Goal/Task network. as well as to the Domain Information System and the Student Model system. This has the advantages of providinga uniform user interface, and notone for the author a-1 another for the student; and of openingup a powert learning-by-making tutorial strategy. The disadvanta,is that the student could damage or destroy the tutorial by his/her modifications; so a student-as-authoraccess class needs to be distinguished froma true-author access class. Student Modelling Here too, the OOPL approach to system design allowsa lot of student information to be collected and usedmore or less automatically. Demons on a Goal/Task node can collectany type of monitoring data on student behavior when the goal is visited. The data can then be dispatched to appropriate nodesin a student-modelling structure. Demonson nodes in that structure can then update student statistical data. revisc the profile of the student. and despatch informationto the blackboard that will be needed for thenCustages of user interaction Collected student data. either on a Goal/Task nodeor in the Student Model database or on the blackboard.can be accessed 4, I zS, yawasatairmaam Using Computers Intelligently 459 by message-passing to help control the tutorial or to determine how to respond to the user. Overlay models of student beliefs (Goldstein, 1981) are effectivet generated automatically by storing student beliefs about 7. Goal/Task node on that node, and similarly by storing student beliefs about some item or concept in the Domain Information System database on the Object encoding that concept. This provides the raw data for analysing bugs in student beliefs (Stevens, Collins & Goldin. 1981), which are important in directing further tutorial interaction. Comparison with other systems The idea of Goal/Task nodes is found in a simplified form in the ECCLES system (Richards & Webb, 1985); although there they are arranged strictly in an or-tree whereas here any and-or graph is constructible. The point of the or-tree was that a path through the tree representedan appropriate series of tasks for the student when faced witha particular problem in the domain. Analysis of the grammar ofone type of French sentence. for example, would prok.,....ti down one path. determined by the grammatical features of that type, and analysis of sentences of another type would proceed down a second path. taking alternatives appropnate to the different features of the second type. Eachsentence to be analysed is flagged by a tip node in the decision tree, thus determining the path of tasks to take. See Figure 8. Also. in ECCLES. the actual tutorial interactionsare determined by questiun-response material stored on each node. On thepresent model, many tutorial modes are possible, and a student is not constrained to any path through the Goal/Task network (thoughcan be). Moreover, in the present system, if a student is "at" a particular Goal/Task node. the interaction there can be of many sorts. and be determined byprocesses in the Student Model. Mentor. and Domain Information System. andas outlined in the previous section of this paper. Teaching materialon a Goal/Task node generally just consists ofpointers to the relevant objects in the Domain Information System. or. inmore complex cases. messages to the Mentor who organizes appropriate maten....1 from the Domain Information System. It is important to realize that the Goa 'Task Objects in the present system are mainly thereas points of intersection of Objects from many type-systems, thereby actingas nodes for an Object network 4 6 9 460 Using Computers Intelligently (the Goal/Task network) that describes epistemological relationships in the subject domain to be taught. The Goal/Task network provides a description of the operations that students must perform to complete learning tasks from a subject domain. Often this network will map directly onto the Domain Information Ssytemthe network that describes the knowledge that the student is to acquire through the performance of those tasks. For example, in Figure 2, N1 should map directly onto a node in the Domain Information System specifying that the number of the subject of a sentence must be singular or plural. Ni should map onto a similar node relating to verbs. Finally, N should map onto a Domain Information System node specifying that there is subject/verb number agreement in a phrase if and only if the number of the subject is the same as the number of the verb. A Goal/Task network will only map directly in this manner onto a Domain Information System network if each element of knowledge that the student is to learn is to be taught by a single operation that the student is to perform and if each operation in turn relates only to a single element of domain knowledge. There will be many domains for which these conditions do not hold. The idea of basing lessons on explicit descriptions of the knowledge to be taught has been explored through the DABIS system (Webb, 1986). DABIS utilises a form of and/or digraph, the feature network. to represent the epistemological structure of the domain to be taught. Thus. DABIS lessons are based onelicit descriptions of the knowledge they are to teach (the Domain Inrormation System) rather than on explicit descriptions of the teaching strategy that is to be pursued (the Goal/Task Hierarchy), as is the case with ECCLES. Figure 9 illustrates the difference. This allows DABIS to construct far better student models than ECCLES in domains for which there is not a one to one correspondence between the Goal/Task hierarchy and the epistemological structure of the domain. However. DABIS hasthe shortcoming that it is difficult to specify instructional flow that differs substantially from the structure of the Domain Information System. Thus. like ECCLES. and for the converse of the same reason. DAB IS operates best in domains in which the Goal/Task hierarchy maps directly onto the Domain Information System. As a result, tutorials for both systems tend to be selected where these conditions hold. This. unfortunately, tends to obscure the fundamental .4 r i Using Computers Intelligently 461 differences between the semantics of the underlying structures utilised by the two systems. However, as described in (Webb, 1986), these differences are both real and significant. The present Project, EXCALIBUR, can be seen as unifying the ECCLES and DABIS approaches to allowing the explicit description of both a Goal/Task network and a Domain Information System within the one system. What EXCALIBUR adds to the other systems is: explicit Goal/Task networks (missing from DABIS), explicit knowledge structures (missing from ECCLES), the allowance that student paths through the Goal/Task network need not be constrained by the necessary-condition vectors of that network. but by pointers m Do First and No Next slots in the Goal/Task Objects, the provision an Object-oriented architecture for the construction of ICAL's around the Goal/Task networks. The reason for treating the Domain Information System separately is just that in general a lot more information and reasoning ability is needed in a teacher than the curriculum they teach (see the Introduction, above.) However, the Domain Information System can, in the Object- structured architecture of the present system, be coupled as closely as needed with the GoalrI'ask network. In some cases it could be dispensed with entirely; as where the tutorial author provides the necessary parse data for the problem French sentences. Conclusion:further research The design of a suitable architecture is just the starting pony: of the EXCALIBUR Project. Current active research by the authors consists of: Building small tutorial systems along the above lines, as different in subject area and tutorial approach as possible. a. testbeds for the design of Goal/Task hierarchies and nodes. and of the various Object networks and their interaction (Richards, 1986b) Abstracting from those cases to an adequate common Object formalism. Other current EXCALIT,UR work comprises: Design of Student Mc del systems that incorporate different theories of cognitive ,tructure and learning; with the aim of 4rya 462 Using Computers Intelligently using EXCALIBUR as empirical tests of those theories (Cumming, 1986). The use of ICAL systems as testbeds for cognitive theories has been explored by (Anderson et al, 1984). The problem of incorporating current work on "Instructional Design"the study of how to design good teaching vehicles and learning environments-- into EXCALIBUR as features available by default or on the author's choice (Barrett, 1986). Natural language processing for the system (Sussex, 1987). User-computer discourse management. including answer negotiation, logic of dialogue, and modelling student beliefs 1986.)3 REFERENCES Anderson, J.R., Boyle, C.F., Farrell. R., & Reiser. B.J. (1984). Cognitive Principles in the Design of Computer Tutors. Proceedings of the Sixth Annual Conference of the Cognitive Science Society.Boulder, Colorado. Barrett. J. (1986).Intelligent CALsome drwelopment considerations. Proceedings, 1986 CAL1TE conference.University of Adelaide. 23-37. Brown, J.S., Burton, R.R., DeKleer, J. (1981). Pedagogical. natural language and knowledge engineering techniques in SOPHIE I, II and III. In ( Sleeman & Brown.1982), Ch.11. Byte, (1986). [Special issue on OOPL's] Byte, 11:8, August. Cumming, G. (1986). The Mentor Component in EXCAL. In Proceedings of the First Round Table Conference, La Trobe University, EXCALIBUR Project, Technical Report 3. Girle, R.A. (1986). Dialogue and discourse. Proceedings, 1986 CALITE conference. University of Adelaide, 123-135 Goldstein, I.F. (1981). The genetic graph, a representation for the evolution )f procedural knowledge. In (Sleeman & Brown. 19821, Ch.3. 2 The authors wisn to tnank Jan Reed. Dennis Hill ant Roly Sussex for valuaLle discussions. Using Computers Intelligently 463 OOPSLA (1986). Proceedings of the OOPSLA '86 Conference. Constituting SIGPLAN Notices,September 1986. Richards, T.J. (1986a). A Description cf the Educational Expert Systems Project. La Trobe University, EXCALIBUR Project, Technical Report 1. Richards, T.J. (1936b). Knowledge representation in expert system based computer assisted learning. Proceedings, 1st Australian Artificial Intelligence Congress,Deakin University, section H. Richards, T.J., Webb, G.I. (1985). ECCLES- An 'Expert System' for CAL. Proceedings of the 1985 Western Educational Computing Conference Oakland. California. California Educational Computing Consortium, 151-157. Sleeman, D., Brown, J.S. (eds) (1982). Intelligent Tutoring Systems. London, Academic Press. Sleeman, D.H., Hendley, R.J. (1981). ACE: a system which an2..yses complex explanations. In (Sleeman & Brown. 1982), Ch. 5. Sowa, J.F. (1984). Conceptual St; uctures.Reading, MA, Addison- Wesley. Stevens, A., Collins, A., Goldin, S.E. (1981). Misconceptions in students' understanding. In (Sleeman & Brown, 1982), Chapter 1. Sussex. R. (1987). Natural Language Interfaces and Representations. La Trobe University, EXCALIBUR Project, Technical Report 4. Webb. G.I. (1986). Knowledge representation in computer-aided learning. PhD. Thesis, La Trobe University, Department of Computer Science 7 464 Using Computers Intelligently N Goal Determine Agreement Task EsIabksh match of number of subject and verb NI F/12 Goat Delemehe numberor,orb Goal Determine number of sublect task Task f A tali; /goal hierarchyIf ifcultteu yeti, :pc:merit ,r.k Inehtit by Acc-se utlinctl Goal/ I Si ISENTENCEI OwnSlot I aSentence InstanceSlot 1 aSentence Sentence The guts were running Sentence Parse parse data Parse parsing function returning parse data Subject-number Plural Subject- number (subject-number Parse) Verb-number Plural Verb-number (verb-number Parse) Subject The girls Subject (Subject Parse) Verb were running Vero (Verb Parse) figure 4 Goal Determine number of subject A type Object for sentence data Note that SI of Figure 3 is an instance of this Object. StudentTest (= (response-to 'What is the number of the subject of this sentence') (subject-number CurrentSentence)) StudentRecord (if (= StudentTest true) then (put CJrrentStudent Goal 'passed) Verb else (put CurrentStudent Goal 'tailed)) Diqu___ N A Record or Object SI holding an example sentence and Instance Subtype ,Zibtype grammatical information about it, and a corresponding Goal/Task Object Nl. kalic text is a description of data, Etre TransitiveVerb IntransitiveVerb not die data itself. Parenthesised text is Method code Instance Voir 4 5 Fleury 5 An inhentance hierarchy of Objects (a) Instance Of Subtype Of Universe Instance Slot theGoal I Key function returning unique keys IDescnption instances put English descnption here InstanceStot GoaVTask instances put a pointer to a Label Coe-I-ask Object here InstanceSlot Subgoats instances put pointers to subgoals here InstanceSlot Supergoals instances put a pointer to argot the supergoal here InstanceSlot GoalData instances put Methods or pointers here to obtain data about the goal (b) InstanceOf Goal SuotypeOf OwnSlot tnoGoal Key j Goa15 Descnp ion Determine number of subject OwnSlot GoaVTask N1 OwnSlot Subgoals () figure 7 OwnSlot Supergoals Goa149 Type-systems and Object networks OwnSlot GoalData () Four type-systems are Chown. linked by plain lines showing inheritance. One Object. called Target, shows its slot pointers (thin arrows) to other Objects. Inheritance and slot pointers (those for other Objects are not shown) allow Methods in Objects to determine the conceptual graph relations (heavy arrows) that define an Object system (a) Thetype Object "Goal"(simplified): (b) An instance. Goa15. of Goal. Using Computers Intelligently 467 I(ROOT) 3' Aux is ETRE 4 Aux. Is Pronominal 1_2___Ai4 Is AVM 5 Direct Oh)ect pre- 6 00 follows PP cedes Pas! Pattie le SO Free 00 Prec. 9' Nee DO is I Preceding 00 Is Is tem pi masculine singular Is masa plural feminine sing PP not FS PP is FP PP not FP "PP Is MS PP not MP PP is FS Ungramma Gramma- Ungramma Gramma Gramme. Ungramma- Gramma. Ungramma Ilea! tical Steal Ilea! heal deal deal Iea! Terminal 6 !ermine( 7 terminal 8 Terminal 1 Terminal 2 Terminal 3 Terminal 5 E(itaut1 Part of an ECCLES decision tree for anal) zing the (*catch reach insti CCM '47044 cons/moon Nodes show the information gained to Them: terminal nodes (rounded boxes) also show the pathnumber to that node. (Entry) jrctPrece E r None Unr;rkeO\ mg /Avoir `,4 0 0co ® 0 CD 55 figure 9 Part of a DABS feature network for the Frenchpas si compose This covers much the same .eatures as the ECCI.ES tree inFigure 8. The ,00p intim?, the links coming down from int top nodeindicate; that it is an AND-node: all four subtrees oeinw it mustbe visaed. not Justone. 117 7