Direct Instruction Revisited: A Key Model for Instructional Technology D] Susan G. Magliaro Barbara B. Lockee John K, Burton Rooted in behavioral theory, particularlythe [0 Rooted in behavioral theory, particularly radical or selectivist behaviorism of B.F. what Skinner labeled the radical or selectionist Skinner (1953, 1954, 1966, 1968, 1974), the behaviorism (see, e.g., Skinner, 1953, 1966), the direct instruction (DI) approach to teaching is direct instruction (DI) of Siegfried Engelmann now well into its third decade of influencing (Bereiter & Engelmann, 1966) is now well into its curriculum, instruction,and research.It is third decade of influencing curriculum, instruc- also in its third decade of controversy. Our tion, and research. It is also in its third decade of purpose is to present the DI model with the controversy (c.f., Gersten, Baker, Pugach, notion that the designer can and should use Scanlon, & Chard, 2001). the model eff-ectively based on appropriate To begin, we offer a definition assessment of the learners,content, context, and our stance related to DI-which has become the whipping and task at hand. To accomplish our goal, we post in some pedagogical camps, while the pan- begin with a general discussion of the basic DI framework,followed by a summary of the acea in others. For clarity, DI is not a lecture approach (e.g., Freiberg major DI models that have been used in live & Driscoll, 2000). It is an instructionalcontexts. We then shft to a instructional model that focuses on the interac- review of how DI has been used in tion between teachers and students. Key compo- technology-based learningenvironments. nents of DI include "modeling, reinforcement, Finally, we conclude with a look into the feedback, and successive approximations" future of DI. (Joyce, Weil, & Calhoun, 2000, p. 337). Joyce and colleagues specified the instructional design principles, which include the framing of learner performance into goals and tasks, breaking these tasks into smaller component tasks, designing training activities for mastery, and arranging the learning events into sequences that promote transfer and achievement of pre- requisite learning before moving to more advance learning. Essentially, DI is "modeling with reinforced guided performance" (Joyce et al., p. 337). Our intent in this article is to explicate the genesis, components, and permutations of DI as it has evolved in practice, and describe how it is being used in instructional technology. Three purposes undergird this article. (a) First, we believe that DI is a viable, time-tested instruc- tional model that plays an important role in a ETR&D, Vol. 53, No. 4, 2005, pp. 41-55 ISSN 1042-1629 41 42 ETR&D, Vol. 53, No. 4 comprehensive educational program. The DI was found to be effective and superior to research indicates its usefulness in maintaining other models in everything from learning time on task, the learning of skilled perfor- engagement to achievement to student affect. mance, and high rates of success when designed As a selectionist model, DI is underpinned by correctly (e.g., Fisher et al., 1980; Slavin, Mad- the basic notion that behavior, like physical den, Dolan, & Wasik, 1996). Therefore, we characteristics, evolves or is selected by the envi- believe that instructional designers, software ronment. Those behaviors that work are selected designers, teachers and the like ought to know by the consequences that follow the behavior. its foundation, essential components, historical Since there are different consequences for the and current uses, and potential for designing same behavior in different environments, in contexts. (It is impor- instruction that promotes student success for behaviors are situated of a behavior particular instructional objectives. (b) Second, tant to note however that the cause in and related to the first, our experience with lay is not the context but rather the consequence, leaves do not cause a faculty (and some instructional technology prac- the same sense that high giraffe's neck to grow. Rather the consequence titioners) who design instruction, especially of longer neck mutations is to be able to eat online education, indicates a dearth of knowl- that few other animals can reach.) edge regarding the research and application of leaves models such as DI. Over the past two decades, DI has been over- Further, in behavioral-based is assumed that learners must be active used by some, maligned by others, and fre- DI, it In The Technology of Teaching, quently been wrongly equated with a pure (behaving) to learn. Skinner (1968) stated, lecture approach. DI is not for all uses, objec- tives, or learners; no approach is. DI is a useful tool for the appropriate purpose, objectives, and It is important to emphasize that a student does not passively absorb knowledge from the world around context, and the appropriate learners. (c) Finally, him but must play an active role, and also that action is over while DI has maintained its core principles not simply talking. To know is to act effectively, both time, it has evolved in response to new under- verbally and nonverbally. (p. 5) standings about learners and learning. We will these variations (e.g., expository elaborate on Moreover, in such models it is assumed that teaching) and the research that indicates their learning is universal, in the sense that the same utility. selectionist principles are involved in learning The DI model was created by Engelmann and from planeria to people; from shoe tying to tying his colleagues in the 1960s at the University of off the last suture in brain surgery (and every- Illinois at Champagne-Urbana under a Project thing in between). Selectionists might agree that Follow Through grant. The research first what we call higher level or higher order activi- appeared in 1966 (Bereiter & Engelmann). Sci- ties may separate us from the rest of the animal ence Research Associates published the first kingdom, but they believe that the way we learn implementation of the model known as Direct such things does not. Instruction System for Teaching And Remedia- Further, behaviorally based models reject tion (DISTAR), programs that addressed begin- logical positivism, mentalisms such as mind ning reading, language, and math (Engelmann (although not mental activity), and free will. As & Bruner, 1969; Engelmann & Carnine, 1969; might be expected, rejection of such concepts Engelmann & Osborn, 1969). Few models have causes passionate reactions even today. For been as researched as DI, including the largest example, an editorial in Early Childhood Educa- educational evaluation ever conducted compar- tion Journal (Jalongo, 1999) reported the author's ing it with 12 other models, across nearly 30 (and her classmate's) first reaction to a movie that years, and involving nearly 75,000 students at showed the DI curriculum, DISTAR, as a "harsh, 180 sites. In that large evaluation (Bock, inflexible, and depersonalizing approach" (p. Stebbins, & Proper, 1977; Watkins, 1997), as in 139) that she worried could resurface today. She numerous studies (e.g., Madaus, Airasian, & said that she would "like to see a stake driven in Kellaghan, 1980; Rosenshine, 1970, 1971, 1985), the heart of DISTAR" (p. 139). Yet, in the same DIRECT INSTRUCTION 43 editorial Jalongo conceded that DI "does have a Nowhere is DI more evident than in com- place"--that "it is the method of choice for low- puter-mediated learning environments. From level tasks such as learning to cut with scissors computer-aided instruction to distance learning or tying shoes" (p. 139). She also saw it as useful experiences, the basic tenets of DI are infused- for special needs children. with greater and lesser fidelity. And, although DI is no longer the most prominent instructional The issue, then, is not whether selection accounts for at least some behaviors or whether framework for the overall design of computer- behavioral approaches work with humans, the mediated applications (c.f., Cognition and Tech- issue is whether some other type of learning nology Group at Vanderbilt, 1996), DI is the evolved and "kicks in" that is unique to some strategy of choice when the learning objective requires higher level behavior in humans. Indeed that the learners have direct practice in what must be done, or said, or written (Cazden, although many critics would argue against DI as 1992). a model for higher level learning or perfor- mance, the model does work well in situations Consequently, our purpose here is not to pro- where motor skills or prerequisite intellectual mote DI as the only instructional framework to skills are involved (e.g., Gagnd, 1985). Such pre- promote learning, either in live or computer- requisite skills might include learning such mediated learning environments. Our purpose things as "mathematical procedures, grammati- is to present the DI model along with the notion cal rules, the states of New England, alphabetiz- that the designer can and should use the model ing, carburetor overhaul, scientific equations, effectively based on appropriate assessment of and the periodic table of elements to name a few" the learners, content, context, and task at hand (Gunter, Estes, & Schwab, 1999, p. 79). Moreover, (Shambaugh & Magliaro, 1997). To accomplish as Gunter et al. put it, "every teacher, in every our goal, we begin with a general discussion of subject, at every level of schooling has some the basic DI framework, followed by a historical learning objectives related to basic skills that trace of exemplar DI models that have been must be mastered before the learner can move to studied and used in live instructional contexts other levels of thinking and learning" (p. 79). over the past 30 years. We then shift to a review of how DI has been used in technology-based In fact, DI has re-emerged in recent years as a learning environments. Finally, we conclude viable instructional strategy that can be situated with a look into the future of DI.