An Enhanced Standard Computer Keyboard System for Single-Finger and Typing-Stick Typing

Veterans Administration

Journal of Rehabilitation Research and Development Vol . 25 No . 4 Pages 17—24

An enhanced standard computer keyboard system for single-finger and typing-stick typing

ROBERT A. CHUBON, PhD, and MISCHELE R. HESTER, BS Rehabilitation Counseling Program, Department of Educational Psychology, University of South Carolina, Columbia, SC 29208

Abstract—A prototype keyboard system was developed, The personal computer has evolved into one of using off-the-shelf hardware and software, as an inexpen- the most useful high technology tools for persons sive keyboard-based system to facilitate data entry for with disabilities. It may now be considered a form single-finger and typing-stick typists. Evaluation estab- of megaprosthesis. New uses are emerging daily in lished that the system can increase entry rate by 50 treatment, work, educational, and home settings, percent or more . The underlying concepts may provide a and considerable effort is being directed toward basis for developing other configurations that accelerate development of interfaces that enable persons with and simplify computer keyboard use for persons with a variety of hand impairments. the spectrum of severe disabilities to benefit from these uses (3,4,8,17) . However, the specially-de- Key words: computer keyboard system, hand impair- signed and relatively costly input systems for per- ments, macros, single finger and typing-stick typing. sons with disabilities appear to be considerably less efficient than the standard computer keyboard being used by the typical nonimpaired, 2-handed touch INTRODUCTION typist. Thus, in word processing and other applications in which input speed is critical, the person with The high cost of many potentially useful high hand impairment may be functional, but is still left technology developments results in their being avail- with a substantial disadvantage. able to a very limited number of persons . Although The importance of this problem is further under- many developments enable persons with disabilities scored by a report published by the Office for to perform critical functions heretofore not possible Technology Assessment (13) which indicates that (e.g., create written communication), their efficiency there are more than 2.5 million people in the United remains limited compared to the performance of States with non-paralytic, upper-extremity impair- individuals without disabilities. Consequently, where ment. When the number of people with impaired productivity is concerned, persons with disabilities hand function stemming from high-level spinal cord are still left with a substantial handicap which may injuries and other disabling paralysis is added to render them non-competitive. Thus, it is imperative that group, there is an indication of the substantial that cost and efficiency be given priority in research number of people who are impeded in keyboard use. and development activities. Because of cost, availability, and related problems with special input devices, the most frequent approach to utilization of personal computers by Address all correspondence and reprint requests to Dr . Robert Chubon, Rehabilitation Counseling Program, Department of Educational Psy- hand-impaired individuals involves using a typing chology, University of South Carolina, Columbia, SC 29208 . stick (either hand- or mouth-held), or single-finger

17 18

Journal of Rehabilitation Research and Development Vol . 25 No . 4 Fall 1988 entry on a standard keyboard . There is an obvious others arranged in order of decreasing frequency inefficiency in these approaches, although for most toward the periphery (3) . The logic of this approach of those who are able to utilize the techniques, entry rests on the concept of a "minimal distance" rates are probably competitive with those attainable keyboard or scanning board configuration, which with available alternative input systems . These prob- emanated from efforts to develop more efficient lems prompted initiation of a project directed alternative input devices by limiting the distance the toward reconfiguring the standard keyboard input input medium must travel (4,10,11,17). system to make it more efficient for single-finger Additional effort was made to keep letters which and typing-stick typists, while containing cost within frequently occur together in close proximity to each more affordable bounds. other in accordance with research findings (17) . The This became feasible with the development of work of Solso and Juel (15), and Solso and King memory resident keyboard utility software (e .g., (16) provided the usage frequency data required for SuperKey' ), which enables key reassignment with- reassignment of key position according to the out special costly hardware or hardware modifica- minimal distance concept . Their data were derived tion for IBM-PC's and compatible computers . Ad- from the Kucera and Francis assemblage of words, ditionally, the widespread availability of low cost which comprise a corpus of more than a million IBM-type computer keyboards contributes to the words selected from written English prose (7) . This practicality of this approach to keyboard adapta- is the largest sample constructed for language tion. analysis and is relatively recent (2,14) . The data derived from these works, and used in rearranging the keys, are contained in Table 1 and Figure 1. KEYBOARD DESIGN The key caps were physically removed from the keyboard and relocated to keep their character Review of keyboard design-related research and labels consistent with the changes . Based on the observation of single-finger and typing-stick typists data, the most frequently occurring letter is E, and it revealed 5 significant factors affecting keyboard use: was used as the center of the keyboard arrangement. a) the distance the finger or typing stick must travel; Although it has been suggested that the space bar b) the speed at which the stick or finger can be and ENTER key are actually the most frequently moved; c) the time spent locating the keys to be depressed (17), they were not reassigned to other pressed; d) interruptions evoked by the need to locations . Because the space bar runs across almost figure out how to expedite simultaneous keypress the entire keyboard, and is in close proximity to combinations ; and, e) the number of key strokes several keys, there would be little accomplished by required for some functions (6) . These 5 factors moving it. This assumption is supported by the work provided the focus for reconfiguration efforts. of others (6) . The RETURN key was not included An IBM PC-type keyboard and keyboard utility among those considered for movement, because software (SuperKey), having a combined cost of appropriate frequency-of-use data was not found, approximately $100, provided the basis for the and because the use of the RETURN or ENTER key system. SuperKey was most suitable for meeting the is diminished by most word processing programs software needs of the project because of its low cost which now include "word wrapping" or automatic and its capacity to support a variety of keyboard returns when the end of a line is reached. input functions, including complete keyboard layout Ultimately, the layout in Figure 2 evolved. It can reassignment and alteration of simultaneous key- be seen that the letter E is actually located nearer the press commands to a sequential mode . A flattened bottom center of the keyboard . Because there are 4 "target-like" arrangement was used as the primary alphanumeric key rows, the alternative would have guideline for the key reorganization . Within this been to locate it somewhere in the upper center framework, previous research suggested clustering quadrant of the arrangement . This placement was the most frequently used letters in the center, with tried, but most users preferred the lower position, and in fact, 2 suggested that near reaches in the 1 SuperKey is a product of Borland International, 4585 Scotts Valley bottom 2 rows were easier than far reaches . Consid- Drive, Scotts Valley, CA 95066 . eration of this preference eventually led to the

CHUBON and HESTER : Enhanced Keyboard System

I E 1 T 1 0 1 A I I 1 N I S I R I H I L I D I C I U1 El 157636 13861 126780 119057 177325 175885 11393751122849150017 172459 137796 15593 TI 156058 170665 182986 134716 151937 126698 113881218130 1127 115493 124658 I 01 12935 128342 175865 124377 171722 120040 126677 113805 132567 135591 I Al 121352 182189 143063 161307 137057 154323 121577 134575 18378 I 11 1100018160166 135926 131964 139218 127844 132212 14151 I NI 117103 16327 11377 12870 148547 113841 117952 I 115281 113453 17205 14713 15866 126849 I RI 13832 13989 110420 18943 122696 ► HI 1592 1214 121256 12945 I LI 112440 15631 117035 I DI 188 17723 I Cl 110730 1

1 M 1 F 1 P 1 G I W I Y 1 B 1 V I K I X 1 J 1 Q 1 Z I EI42256 113895 122970 117358 118450 110152 122920 139318 110976 17477 11694 11638 12127 I T11121 13350 13106 1713 12862 17883 1605 16 167 11457 ii I 1147 I D132778 157534 119894 18414 121801 17302 111415 19272 13290 1593 12304 140 1386 1 A130368 18694 117573 112983 120897 110069 113941 111914 14899 11587 11269 159 11377 I 1123651 117124 18090 115087 114837 11270 16822 117947 16022 11690 1150 1434 12759 I NI1169 11944 1228 139395 13598 14555 1176 11603 14484 1101 1374 1191 1102 I S15925 1642 18681 11997 12286 15018 11687 178 13528 13 13 1 120 I R17058 18849 116628 110627 11659 18786 14999 12218 13561 117 180 I 134 I HI429 183 12999 19791 114883 11701 1209 19 1165 1131 I1 118 118 I 141276 14339 110176 12404 11044 117679 18914 11125 11574 120 1 115 1123 I D1614 172 132 11132 1488 12366 1142 1612 142 1 1227 167 1 ► C150 Ill 111 I 123 11478 124 11 16003 1896 1 I 115 I U19138 14224 18832 18098 179 1262 111193 1170 1159 1292 12436 14913 1142 ► MI 1160 18289 1269 173 13036 13544 14 167 I1 1 1 13 1 Fl 161 162 165 1315 114 I 1114 119 12 I Ii I p t 125 194 11020 126 13 161 12395 14 I I GI 119 1632 138 Ii 179 I 1 I 110 1 WI 1340 174 I 1151 113 I1 12 16 Yl 16048 1204 1418 184 12 I 1163 I BI 1214 142 12 1626 I 17 VI 12 11 I 11 19 1 KI I I 117 111 I xl I 17 I JI I I I 41 1 ZI

Figure 1. Combined forward and backward bigram frequencies.

placement of more alphabet keys at the bottom of Ion, hyphen, and question mark seem to fall in the arrangement. between . This cursory study provided the basis for No usage data were available for the numeric, placement as pictured . It is noted that the numeric punctuation, and utility keys, but a review of several keys were located around the periphery of the layout sample essays and other examples of general written to reflect their low frequency, and the miscellaneous communication in an English textbook suggested keys were used to fill in the remaining unused that numbers are used relatively infrequently and positions and appropriately paired to maintain that the period, comma, quotation marks, and logical combinations. apostrophe are most common . The semicolon, co- A final consideration in determining key place-

Journal of Rehabilitation Research and Development Vol . 25 No . 4 Fall 1988

Table 1. Rankings of letters by overall frequency of use and first position use.

Rank Ordered Overall Use Rank by First Letter Frequencies Percentage Position Occurrence E 577583 12.62 12 T 427179 9.33 7 0 350121 7.65 18 A 348411 7.61 4 I 336166 7.34 14 N 325652 7.11 17 S 297531 6.5 1 R 281881 6.15 9 H 252191 5.51 11 L 188261 4.11 13 D 181054 3.95 8 C 142336 3.11 2 U 124736 2.72 20 M 116574 2.54 5* F 107219 2.34 10 P 93040 2.03 3 G 89499 1 .95 15 W 86563 1 .89 16 Y 78749 1 .72 24 B 70714 1 .54 5* V 45707 .99 19 K 29838 .65 21 X 9032 .19 26 J 7296 .15 22 Q 4936 .10 23 Z 4316 .09 25

*Estimates for these two letters were essentially the same and they were both ranked fifth.

ment was the likelihood that a letter would be the beginning with each was estimated from the respec- first of a word. Since words are usually preceded by tive number of pages in Webster's New World a space, it was concluded that there would be an Dictionary (5). advantage to placing letters most likely to begin Another consideration in the enhancement of the words near the space bar if consistent with the other keyboard was to facilitate the visual scanning or placement criteria. To determine which letters were key-locating process . Observation of typing-stick candidates for this placement, the numbers of words users revealed that their mode of keyboard entry 21

CHUBON and HESTER : Enhanced Keyboard System

C 321 V U P 6 7 8+ 4 Q M I T S C K Z 9: 5 J G N R E H B Y X 0 / ' F 0 A D L W, ------

Figure 2. Reassigned key positions. was heavily dependent upon visual scanning, unlike highlight the overall pattern, and to provide a touch-typing, which is predominantly dependent discriminable characteristic in addition to the labels upon kinesthetic feedback (1). Measures were imple- on the keys. The "bull's eye" letter E was covered mented to assist the user in developing a scanning with a red transparency to make it stand out. The pattern consistent with the target-like arrangement. numbers were made blue, punctuation marks green, Generally, English reading/writing people learn to and the miscellaneous letters yellow . The tilde (— ) scan from left to right . However, with this target key, which serves as a special function key, was arrangement, it was felt that it would be advanta- relocated in the upper left corner and made orange. geous if users maintained focus on the center of the Color coding was accomplished using commercially- keyboard and learned to use this as their reference produced, transparent, stick-on key top covers, point, since the layout was designed to concentrate which allowed the character labels to show through. the keyboard activity in this area. To this end, the The resultant target-like appearance can be seen in numbers were arranged to curve around from near Figure 3. the top center of the keyboard in order of occur- Two approaches were taken to eliminating key- rence. Thus, as can be seen in Figure 2, the numbers strokes. First, the use of macros, that is, the 1 through 5 run counterclockwise, and 6 through 0 assignment of words, phrases or other multiple are arranged in clockwise fashion. character strings to a defined and relatively limited The second measure taken to facilitate scanning keying sequence, was incorporated into the scheme. and location of the keys was to color code them to The software provides for the creation of these aids.

Figure 3. Reconfigured keyboard . 22 Journal of Rehabilitation Research and Development Vol . 25 No . 4 Fall 1988

Usually macro strings are derived from frequently- groups, each comprised of 3 adults having little or typed material, such as a person's signature, which no typing experience, engaged in two 90-minute can be assigned to a keying combination like Alt S, single-finger typing sessions . Subjects who were and thereafter typed in entirety by entering this inexperienced in typing were used to eliminate the limited keystroke combination . Because this project intrusion of negative transfer as a confounding was directed at more general usage, a set of macro variable (12). The control group was instructed to words was derived from the Kucera and Francis practice single-finger typing, using a standard IBM- word corpus analysis (7) . The frequencies of the 200 PC keyboard and a word processing program . The most-used words were multiplied by their length to experimental group was oriented to the features of obtain products that reflect the relative number of the reconfigured keyboard system, and instructed to strokes required by their use when typing substantial practice single-finger typing with the altered key- amounts of ordinary written English . The 5 words board and the same word processing program used yielding the largest products were utilized for mac- by the control group . Additionally, all participants ros . Thus, the following words were assigned a were provided with written instructions for their keying sequence consisting of the Alt key and an systems to use as needed. alphabet character (or the Ctrl key and the alphabet The sessions, scheduled a week apart, involved character to create capitalized macros) : that, from, typing practice exercises contained in a typing with, this, and which. instruction textbook (9). During the last 5 minutes A second means of eliminating keystrokes was of each session, 5-minute proficiency tests were derived from the macro concept . Because spacing is administered to the participants . Following the usually required after punctuation marks, the macro second session and proficiency test, a debriefing concept was used to configure the system to auto- interview was conducted with each participant, and matically insert 1 space following the comma and each member of the experimental group was also semicolon. Upon pressing the appropriate punctua- asked to complete a rating scale developed to obtain tion key, 2 spaces are automatically inserted follow- their opinions about the value of the enhancements. ing the period, question mark, and colon . The apostrophe and quotation mark keypress functions were not changed because of their frequent use RESULTS without trailing spaces. If it becomes necessary to use these punctuation The data from the typing tests conducted at the marks without trailing spaces, one possible solution end of the second practice sessions were subjected to is to backspace to the desired position after entering statistical analysis . The number of errors was sub- the reconfigured punctuation marks ; however, the tracted from each subject's total number of keypress software creates an "undo" key (the orange-colored entries to obtain proficiency scores . The mean tilde key mentioned above) which returns the key- calculated from the scores of the standard keyboard board to its nonmacro state for 1 keypress, and this group was 264.7 (sd = 27.4), and the mean for the provides an effective solution. reconfigured keyboard group was 409 .7 (sd = 29 .7). The final enhancement to the keyboard input Review of individual scores revealed that all 3 system was to utilize the software's "single-finger" subjects in the reconfigured keyboard group scored mode . This enables entering keyboard commands, higher than the highest score in the standard which normally would require simultaneous, multi- keyboard group . A t test supports the conclusion ple keypresses (e.g ., shift and a letter key to type that the difference between the means was signifi- upper case), to be entered as sequential commands. cant (p< .005, t = 5 .69, df = 4). From another per- spective, the experimental group rate was approximately 55 percent faster than that of the control or EVALUATION standard keyboard group. Responses from the debriefing interviews also The reconfigured keyboard input system was revealed a notable difference . All 3 subjects in the evaluated in a research study that followed a formal control group (standard QWERTY-arranged key- experimental paradigm . Control and experimental board) reported experiencing finger fatigue during 23

CHUBON and HESTER : Enhanced Keyboard System practice, while none of the experimental group their unique usage patterns (including word and members cited this as a problem . Examination of letter frequencies), would likely provide a better the rating scale data obtained from the reconfigured basis for keyboard rearrangement than general keyboard group suggested that all of the enhance- usage material . It is noted that some of these issues ments contributed positively to their performance, are being addressed at Tufts University (11). including both speed and accuracy, except the macro Attempts to develop analytical algorithms must be words . They attributed problems in using the mac- pursued carefully, however . This study, as well as ros to their inability to quickly learn and recall the others cited, suggest that to be most effective, keystroke sequences, and suggested that the se- reconfiguration must take into consideration a num- quences were "unnatural" or not compatible with ber of human factors that can vary substantially the flow of thought and concentration commanded from individual to individual . Included are matters by the typing activity. such as visual scanning rate, extremity/digit strength, stamina, dexterity, alternating extremity/digit response time and sequential response time, and DISCUSSION cognitive factors such as memory and attention capabilities . Consequently, to be effective, analytical This limited study suggests that the standard algorithms will have to be able to produce much computer keyboard entry system can be enhanced to more than arithmetically-determined key arrange- improve the proficiency of hand-impaired typists. ments based on typing or other entry material Although the study does not indicate the maximum content and the number of fingers or typing aids entry rates which may be attainable by single-finger used. Each additional critical factor may cause the and typing-stick typists using this reconfigured complexity of the required algorithm to increase keyboard system, substantial gains may be realized. geometrically. The most feasible approach may use Moreover, the study confirms that relatively inex- a moderately encompassing design algorithm, user pensive avenues to enhancement do exist. trial, and an algorithm to analyze system perfor- It is also noted that this keyboard arrangement mance. may not be of benefit to persons using other than a In sum, the experience gained from this project single-finger or typing-stick entry technique. For indicates that there are numerous relatively inexpen- example, persons using 2 typing sticks or 2 fingers sive, off-the-shelf productivity enhancements that may actually find the arrangement to be a hin- hold potential benefit for persons with disabilities. drance. Because the activity is concentrated in the However, their optimal use may rest with the middle of the keyboard, 2 or more entry media may development of new assessment and design tools. collide and interfere with each other . On the other hand, the project outcome supports the notion that it may be possible to discern and create an optimal keyboard arrangement to accommodate any hand- REFERENCES impaired individual and his or her unique residual function . The low cost of keyboards and keyboard 1. Alden DG, Daniels RW, Kanarick AF: Keyboard design utility software, and the flexibility inherent in the and operation: A review of major issues . Human Factors software approach to reconfiguration, make the idea 14:275-293, 1972. plausible. 2. Cress CJ, Vanderheiden GC : Applications of multiple wordstudy analysis in development of vocabularies for The key to widespread customizing of keyboard augmentative communication systems . In Proceedings of entry systems may lie in the development of analyti- the 9th Annual Conference on Rehabilitation Technology cal algorithms that enable rapid evaluation of 6:351-53 . Washington, DC : Association for the Advance- various configurations as they are being tried by ment of Rehabilitation Technology, 1986. potential users . Similarly, software designed to 3. Getschow CO, Rosen CJ, Goodenough-Trepagnier C : A analyze unique kinds of material that users will be systematic approach to design of a minimum distance alphabetical keyboard . In Proceedings of the 9th Annual typing may provide a basis for optimal, individually- Conference on Rehabilitation Technology 6:396-398. tailored key arrangements and macros . Programs Washington, DC : Association for the Advancement of that could analyze an individual's text files to reveal Rehabilitation Technology, 1986 .

Journal of Rehabilitation Research and Development Vol . 25 No. 4 Fall 1988

4. Goodenough-Trepagnier C, Levine S: An optimal, inex- Association for the Advancement of Rehabilitation Tech- pensive text entry system for the orthopedically and nology, 1986. neurologically disabled . Rehabilitation R & D Progress 11. Minneman SL: Keyboard optimization technique to im- Reports 22(4) :134. Washington, DC : Office of Technol- prove output rate of disabled individuals . In Proceedings ogy Transfer, Veterans Administration, 1985. of the 9th Annual Conference on Rehabilitation Technol- 5. Guralnik DB (Ed.): Webster's New World Dictionary, ogy 6 :402-404 . Washington, DC : Association for the Second College Edition, New York: World Publishing Advancement of Rehabilitation Technology, 1986. Co., 1970. 12. Noyes J : The QWERTY keyboard : A review . Interna- 6. Kinkead R : Typing speed, keying rates, and optimal tional Journal for Man-Machine Studies 18:265-281, 1983. keyboard layouts . In Proceedings of the Human Factors 13. Office of Technology Assessment . Technology and Hand- Society 19th Annual Meeting, 159-161 . Santa Monica, icapped People, Washington, DC : U.S . Government CA: Human Factors Society, 1975. Printing Office, 1982. 7. Kucera J, Francis WN: Computational Analysis of 14. Shapiro SI, Gordon GP: Contemporary norms of word Present-Day American English, Providence : Brown Uni- and phonetic frequencies . Journal of Verbal Learning and versity Press, 1967. Verbal Behavior 10:92-94, 1971. 8. Lee CC, Vanderheiden GC, Rasmussen A : One finger 15. Solso RL, Juel CL : Positional frequency and versatility of operation of the IBM family of personal computers . In bigrams for two- through nine-letter words . Behavioral Proceedings of the 9th Annual Conference on Rehabilita- Research Methods and Instrumentation 12:297-343, 1980. tion Technology 6:265-266 . Washington, DC : Association 16. Solso RL, King JF : Frequency and versatility of letters in for the Advancement of Rehabilitation Technology, 1986. the English language . Behavior Research Methods & 9. Lloyd AC, Rowe JG, Winger FE: Typewriting for Instrumentation 8:283-286, 1976. Colleges, New York : McGraw-Hill, 1964. 17. VerVoort S, Winningham D, Swarts A, Krouskop T: A 10. Levine SH, Goodenough-Trepagnier C, Rosen MJ, computer keyboard design for people who use typing-stick Getschow CO: Adaptive technique for customized inter- typing. In Proceedings of the 9th Annual Conference on face design with applications to nonverbal communica- Rehabilitation Technology 6:394-395 . Washington, DC: tion. In Proceedings of the 9th Annual Conference on Association for the Advancement of Rehabilitation Tech- Rehabilitation Technology 6:399-401 . Washington, DC : nology, 1986.