Ode to Zig (and the Bard): In Support of an Incomplete Logical-Empirical Model of Direct Instruction

Abstract

In this article, I offer my perspective on several elements of Engelmann’s Direct Instruction. I hypothesize Engelmann’s thinking about the schooling environment that arguably provoked his theoretical, philosophical, and conceptual insights into the design of Direct Instruction. I also examine the research on Direct Instruction as a national educational model, but only as an extension of Engelmann’s commitment to falsifying his own thinking. In addition, I survey the research on the design of instruction to highlight how greatly different disciplines can find common ground around “faultless communications.” Along the way, I offer examples and descriptive analyses of selected design of instruction elements of Direct Instruction. Finally, I conclude with a brief ode to Engelmann.

The singular and arguably unprecedented focus of this special section is on Siegfried Engelmann’s Direct Instruction that guest editors, Heward and Twyman, describe as “a powerful teaching system that combines logical selection and sequencing of examples and high rates of student responding to mastery” (Call for Papers: Direct Instruction). An important objective of this singular focus is ostensibly to ascertain the salient and irreducible features of Engelmann’s Direct Instruction, including the irreverent Siegfried Engelmann himself (Barbash, this issue) and his fierce and unflinching commitment to the learner as the ethical imperative that drives the “faultless design” (Twyman, this issue) of every one of his teaching sequences, as well as every one of his more than 100 commercial curriculum programs and their critical but often elusive technical and pedagogical intricacies.

As a collection, the articles in this special issue make it conspicuous that Engelmann’s Direct Instruction (DI) is indeed a powerful teaching system. However, selected features and elements important to the fabric of Direct Instruction are arguably Shakespearean in nature because they unwittingly conceal more than they reveal (Kame’enui, 1996). Unless, of course, as teacher, school administrator, curriculum specialist, researcher, teacher aide, instructional designer, local school board member, parent, or observer, you know what elements to pay particular attention to, at the right time, and for the right reasons, as in the Bard’s plays.

In Engelmann’s Direct Instruction, not unlike Shakespeare’s plays and sonnets, everything matters—but not equally so—at least not in its impact on student learning. In Engelmann’s Direct Instruction, as in Shakespeare’s work, conceptual and analytical hooks are threaded throughout its fabric for the benefit of the teacher, yet purposefully designed to privilege the learner’s success. These conceptual and analytical hooks are likely derived from the “How To Manual” for Direct Instruction—that is, Engelmann and Carnine’s (1982) Theory of Instruction: Principles and Applications. Other principal sources, however, are also implicated historically and instrumentally (e.g., Becker et al., 1971, 1975; Engelmann & Bereiter, 1966; Engelmann & Carnine, 2011; Engelmann & Engelmann, 1966). These sources offer detailed blueprints for the design of instruction across a range of disciplinary content (e.g., reading, math, science, social studies, history). It is important to note that these hooks are also strategically calibrated to communicate from teacher to learner—clearly and unambiguously—the essential content of the target discipline into the what that gets taught, when and at what points in time, with what levels of scaffolded support, to what criterion-level of student performance, how, how fast, how often, and, of course, in deference to the Bard, why said content is important to teach.

The conceptual and analytical hooks, the associated technical details, and the overarching architectural design of Engelmann’s Direct Instruction are consistent with Smith and Ragan’s (1999) definition of the design of instruction as “a systematic process of translating principles of learning and instruction into plans for instructional materials and activities” (p. 3). Likewise, they also honor Tennyson and Christensen’s (1986) assertion that the design of instruction requires “initially preparing instruction that has a high probability of preventing learner errors and/or misconceptions and misrules” (p. 4; emphasis added). As such, the design of instruction relies on a probabilistic model (i.e., increasing the probability of preventing learner errors) and on the prescription of instructional procedures “to achieve particular changes in learner behaviour” (Moore, 1986, p. 202).

Thus, to achieve changes in the learner’s behavior, the “design” of Engelmann’s Direct Instruction is comprised of a range of carefully crafted individual instructional elements, which are detailed in Table 1.

Table 1.

Direct instruction elements

Design Element	Example or Function
1. Prescribed delivery for teaching a lesson	Scripts as in Shakespeare’s plays that specify what the teacher says and does including student responses.
2. Explicitness of the instruction	Instruction that employs a Model-Lead-Test routine (Carnine & Silbert, 1979) in which the teacher “models” the student response, “leads” or guides the student in the response, and “tests” or assesses all students, as well as each student’s independent response. Archer (2020) calls it the “I do, We do, You do” routine.
3. Scaffolding	Guides teachers in direct support of learners negotiating new and unfamiliar conceptual terrain before releasing the learner to full independent learning.
4. Correction procedures: Immediate and delayed	Prescribes for teachers how to address students’ errors, predictable or otherwise.
5. Group and Individual responses	Responses are solicited, staged, and juxtaposed throughout a lesson.
6. High number of opportunities for students to respond	Responses are prompted at a high rate during instruction as a group or individually.
7. Pacing of the instruction	Instruction that systematically schedules how much content is covered and how fast throughout a lesson and across lessons in a program.
8. Cumulative review	Review of previously taught content strategically scheduled within lessons but also between lessons and throughout a curriculum program.
9. Mastery of student performance	Student performance requirements of lessons include highly specified criterion levels of student performance to demonstrate proficiency.
10. Practice: Distributed or chunked	Practice is distributed or “chunked” but systematically allotted within and between lessons.
11. Part–whole discrimination tasks	Tasks that provide learners with “punctuated” and timely opportunities to practice a new part of a task/problem to reinforce and strengthen a response before opportunities are provided to complete the whole or entire task/problem within a lesson.
12. Initial teaching sequences	Designed and employed for the introduction of new concepts and skills in which example selection and sequencing are tightly controlled.
13. Expanded teaching sequences	Designed and employed for familiar and previously taught concepts, rules and skills that shifts the burden of acquisition and transfer to the learner to demonstrate independence.
14. Placement tests	Designed and employed for initial grouping of students by skill level within and between programs.
15. Ongoing assessments	Designed and employed to monitor student learning and progress and regroup students for instruction when necessary.
16. Scope and sequence	Designed for “mastery” that reflects an analysis of disciplinary content (e.g., reading, math, social studies, science) and “knowledge forms” aligned with Engelmann and Carnine’s (1982) Theory of Instruction but always in direct service to ensuring the range of content, however complex, is accessible to the learner.

Engelmann’s Direct Instruction programs intentionally privilege, at a minimum, these 16 pedagogical and design of instruction elements. In practice, these design of instruction elements are embedded and woven into the very fabric of DI programs, such that in total, a DI program is plainly “more than the sum of its parts.” That is, the program disguises—and in cases where the content to be taught is complex in its demands on the learner, perhaps even conceals—how specific tasks and teaching sequences are imagined, designed and employed within a lesson; how they are conceptualized, staged, and woven between lessons such that the essential elements of lessons are connected from one to the next; and finally, how all lessons within and across a curriculum program are threaded together structurally, thematically, logically, and coherently. It is important that these design elements within a program are not fungible but distinct and separable and thus, retain their independent integrity for individual testing, and continued revision, and improvement.

Some of the elements are considered conspicuous signature elements of Engelmann’s Direct Instruction—that is, Big D and Big I, Direct Instruction—including, for example, the explicitness of the instruction, or the prescribed delivery and teacher scripting of lessons that specify exactly what the teacher says and does. But even these signature elements are not wholly transparent. For example, in prescribing the teacher wording, Engelmann asserts that the wording be “clear and concise,” “truthful and not misleading,” and that the vocabulary used is “limited to what is immediately necessary for the discrimination or operation being taught” (Engelmann & Colvin, 2006, p. 16).

In addition, the so-called signature elements may arguably detract from other elements that are equally (or more) critical to the efficacy of Engelmann’s Direct Instruction, such as the actual number of opportunities to respond that a learner is provided to gain the necessary practice on a new concept or operation through the prescribed distribution of the individual teacher-guided prompts within a lesson or across lessons in a program. Another elusive example is the level of difficulty of a sequentially ordered group of lessons that Engelmann argues should “not change drastically from one group of lessons to the next even though more complex content is being introduced” (Engelmann & Colvin, 2006, p. 7).

Like apparitions in the Bard’s plays, the apparent signature elements may also serve to confuse the nominal yet critical distinction between Big DI and little di (little d, little i) (e.g., Baumann, 1983; Berliner, 1981; Duffy & Roehler, 1982) approaches and programs even though the “design elements” of the DI programs differ powerfully in their “pedagogical DNA” from little di (Kame’enui et al., 2013, p. 492). If this concern strikes you as trifling, Engelmann himself thought it sufficiently worrisome that he penned and published a text with his good friend and colleague, Geoff Colvin, “Rubric for Identifying Authentic Direct Instruction Programs” (Engelmann & Colvin, 2006) in which he describes the “axioms of Direct Instruction” for identifying the essential features of DI. Don't be fooled, however: even that professional disclosure arguably conceals more than it reveals.

In this article, I offer my perspective on several elements of Engelmann’s Direct Instruction. In doing so, I hypothesize about Engelmann’s own thinking on the schooling environment that provoked his theoretical, philosophical, and conceptual insights into the design of Direct Instruction. I also examine the research on Direct Instruction as a national educational model but do so only as an extension of Engelmann’s commitment to falsifying his own thinking. In addition, I survey the research on the design of instruction to highlight how greatly different disciplines can unwittingly find common ground around “faultless communications.” Along the way, I offer examples and descriptive analyses of selected design of instruction elements of Direct Instruction. Finally, as an apprentice of Engelmann’s Direct Instruction, I conclude with a brief ode to Engelmann.

Begin at the Beginning: A Two-Question Conceit

“‘Begin at the beginning,’ the King said, very gravely, ‘and go on till you come to the end: then stop’” (Carroll, 1865). It’s not always obvious where the beginning is on some topics—that clearly identifiable thread to pull that releases and exposes the hidden contents in a clear and magical sequence. Engelmann’s Direct Instruction is such a topic. The archetypal starting point to understanding Engelmann’s Direct Instruction is likely Engelmann, the actual “thread” himself, as evidenced in this special issue (Barbash, this issue; Kame’enui, 2015). Another equally logical, but less compelling starting point is Engelmann’s own thinking about teaching and learning, which includes a set of important and uncompromising assumptions about the learner (e.g., If the learner fails to learn, the fault is not with the learner), the teacher (e.g., If the learner fails to learn, the fault lies with the teacher), the content (e.g., If the learner fails to learn, examine what the learner has failed to learn), and by logical extension, the schooling process (e.g., If the learner fails to learn, the schooling process is responsible for the performance of the learner).

In deference to the king’s “Begin at the beginning” petition in Lewis Carrol’s Alice in Wonderland, I offer a peculiar conceit. In particular, I tender two questions below that I imagined Engelmann posed and answered for himself in his original analysis of the learner, the teacher, the content, and the schooling process, and presumably before he ever developed the audacious assumptions enumerated previously and in his Direct Instruction approach. The two-question conceit required Engelmann to engage in a Shakespearean “process of dubitation,” not unlike Hamlet’s celebrated fourth soliloquy—“To be or not to be”—in which he posed a doubt (or two), then in benign submission to the soliloquy, talked himself through the “doubt” in a thought experiment to an insight and resolution. No doubt, Engelmann would appreciate this process; after all, he once asserted, “The goal of thinking is to reduce or eliminate the doubt and replace it with certainty” (Engelmann & Carnine, 2011, p. 81). Thus, the two-question conceit I imagined Engelmann posed to himself:

• First Question: What in the school environment is the learner, as an inexperienced and guileless observer of the world, required to negotiate on a moment-to-moment basis that sets the stage for his or her success or failure in school?

• Second Question: What in the province of the learner’s school environment can a teacher control, engineer, and take full and direct responsibility for to increase the odds—significantly so—that the learner will be successful?

The hypothetical answer to the two-question conceit served as the critical “Begin at the beginning point” to Engelmann’s systematic and life-long conceptualization of Direct Instruction. I hypothesize that the answer to both questions is the symbolic system, that is, the system of symbols found in the content of schooling materials that learners must negotiate successfully to access directly, and on a moment-to-moment basis in order to engage in schooling tasks successfully. Such schooling tasks include, for example, reading (i.e., to negotiate an alphabetic writing system, at least in the United States), mathematics (i.e., to negotiate the Arabic or Hindu-Arabic number system and mathematical notations and operations), communicating—speaking or writing (i.e., to negotiate a primary [English] or secondary [Spanish, Russian] spoken and written language), and thinking and learning (i.e., to engage in ideation that permits further individual growth and development) for which the learning environment, including the teacher, is charged with an implied (or in some cases, explicit) “public trust” to ensure that learners are indeed successful.

To appreciate the importance of any symbolic system, more than 4 decades ago, Engelmann and colleagues (Becker et al., 1971) conceptualized a set of fundamentally basic distinctions between the features of a symbolic system (cognitive tasks) and that of a nonsymbolic system (physical tasks). To appreciate Engelmann’s logical analysis in the design of Direct Instruction, including the creation of highly specified tasks, teaching sequences, lessons and curriculum programs, it is essential to understand these fundamental distinctions, because they were in direct service to both his philosophical and technical commitments to the learner as the ethical imperative.

Symbolic versus Nonsymbolic Systems

More than 40 years ago, Becker et al. (1971, 1975) identified four important distinctions between symbolic systems and nonsymbolic systems. As noted in Table 2, nonsymbolic systems involve physical tasks that do not necessarily require reliance on negotiating symbols per se, such as opening a door, riding a bike, or dribbling a basketball. These physical behaviors cannot be hidden at any time but are necessarily public and always remain public for everyone to observe, evaluate, and manipulate, as necessary.

Table 2.

Distinction between symbolic and nonsymbolic systems

Nonsymbolic Systems and Operations (Physical Tasks—Examples: Running, shooting a basketball)	Symbolic Systems and Operations (Cognitive Tasks—Examples: Reading, writing, solving math problems)
1. All component behaviors are overt and remain overt.	1. The component processes are covert and only the final product is overt.
2. Feedback from the environment is typically instant and provided on every trial.	2. Feedback from the environment is either delayed or not given.
3. In general, component behaviors and skills are clear and identifiable.	3. Component skills or processes are not clear and typically difficult to identify.
4. In general, the execution of the component skills and behaviors in a particular sequence is clear.	4. The execution of the component skills or processes in a particular sequence is not clear.

A nonsymbolic, physical task is in direct contrast to a symbolic task that involves cognitive processes and require negotiating symbols, such as the words, punctuation, and numbers on this page written in English. Such a symbolic task requires the reader to negotiate an alphabetic writing system (i.e., or a syllabary system as in Japanese; or a logographic system as in Chinese; Rayner & Pollatsek, 1989) that implicates and triggers a unique set of cognitive processes, all of which take place covertly and privately. These cognitive processes provoke, for example, the visual system (e.g., retinal processing), the speech system (e.g., phonological activation and processing; Broca’s area), multiple neural networks (e.g., occipito-temporal cortex; parieto-temporal cortex), and more (Dehaene, 2009; Shaywitz, 2003; Wolf, 2007, 2018). Such cognitive processes cannot be directly observed or manipulated physically, like that of dribbling a basketball. Only the final outcome or product of the cognitive process—that of reading the words in a narrative passage aloud—is overt. If the reader fails to read a word correctly, there is no direct observation of the misfire, only the misread word as the public outcome.

In addition, in a nonsymbolic physical task like dribbling a basketball, the environment including the basketball, the gymnasium floor, the learner’s eye–hand coordination, and other factors (e.g., presence of a coach or audience, the basket, backboard) either concurrently or sequentially provide the learner with instant feedback about the learner’s dribbling performance. If the learner fails to bounce the ball with sufficient force and rapidity, the ball will cease to bounce, and the learner will gain instant feedback that his or her basketball dribbling behavior must change in order to be successful. In contrast, when reading a narrative passage silently, the only opportunity for feedback is if the teacher or a peer asks the learner to answer passage-specific comprehension questions. If the reading is out loud, the only feedback provided—instant, delayed, or otherwise—is from a teacher, peer, adult, or a programmed machine-designed feedback system. Thus, feedback to the learner in a cognitive task must be planned, intentional and scheduled. It is not a natural feature of the task provided instantaneously as part of the learner’s environment, as is dribbling a basketball.

The last two features that distinguish a physical, nonsymbolic task from a cognitive, symbolic task is the identification of the component skills or processes, and the proper sequence of the component behaviors or constituent skills required to execute and complete the target task. These features are the most complex and difficult to ascertain, especially for a cognitive task, because it requires both a logical analysis of the components of the target task, as well as solid, scientific empirical evidence to guide (i.e., confirm or reject) the logical analysis. In a physical task such as dribbling a basketball to make an unguarded layup, the proper sequence of critical component behaviors is rather straightforward, predictable, and logical in executing the layup for a bucket. The component behavior and skills for a right-handed player, for example, include the following six steps (https://www.basketballforcoaches.com/how-to-do-a-layup/):

• Step 1—Get your eyes up to focus on the “hoop” and the defense;

• Step 2—Control your body and use your outside foot to take a long step toward the basket as you pick up the ball, and position it for the layup;

• Step 3—Use your inside foot (i.e., left-foot for right handers) to jump high and gain momentum;

• Step 4—Use your nonshooting hand to protect the ball from defenders;

• Step 5—Follow through by bringing the ball above your head, extending your arm and flicking your wrist to guide the ball to the backboard and hoop;

• Step 6—Practice at different angles and speeds and on both sides of the basket to perfect your skills.

In a cognitive task, the component skills and associated behaviors, as well as the proper sequence for the execution of them required to produce the target outcome successfully are often difficult to identify. In fact, these features have been and continue to be the very topic of turbulent (and in some disciplines, toxic) argumentation over the adoption and implementation of the school curriculum and teaching approaches across a range of disciplines, including, for example, the teaching of beginning reading (Adams, 1990; Castles et al., 2018; Goodman, 1992), the teaching of math (Schoenfeld, 2004), and the teaching of social studies (Evans, 2004).

As an illustration of this disciplinary tribalism, for more than 100 years there has been fierce argumentation characterized as the “Reading Wars” (Lehman, 1997; Liberman & Liberman, 1990; Matthews, 2020) about how children learn to read in a “writing system” (e.g., alphabetic, logographic, syllabary), how this complex system of reading in a writing system should be taught, and whether teaching reading requires the identification and actual “teaching” of specific component skills and operations. Unfortunately, the legitimate argumentation of how to teach such a complex process as beginning reading has been shamelessly reduced to a “vulgar dichotomy” (Kame’enui, 2002) of phonics versus whole language, as if a highly complex cognitive process could be captured succinctly and robustly in a one or two-word approach to teaching beginning reading. Such a characterization notwithstanding, in brief, a “whole language” approach privileges meaning and considers learning to read as natural as learning to speak. As such, it does not rely upon or employ explicit, systematic instruction, nor does it dare to reduce reading into component skills and processes. Such a deconstruction of reading into component skills to be taught is viewed as “trivializing” the very process of authentic reading for meaning (Goodman, 1987).

In stark contrast, a “phonics approach” privileges learning the component skills of reading in an alphabetic writing system, a system of writing invented more than 5,000 years ago (Wolf, 2007). Thus, the act of reading doesn’t occur naturally, like the act and process of speaking, because speech is ontogenetically prior to reading any form of print. Almost all children learn to speak naturally, and without formal instruction, in part, because humans have a biological predisposition to develop oral language (Pinker, 1994). In stark contrast to that of learning to speak, learning to read in a writing system does not come naturally for most children but requires formal instruction (Dehaene, 2009; Shaywitz, 2003; Wolf, 2007). Because human beings created this unique symbolic system—the writing systems—it must be learned and taught, including the following component skills: phonemic awareness, the alphabetic principle, vocabulary development, fluency, and reading comprehension (Adams, 1990; Carnine et al., 2017; Dehaene, 2009; National Reading Panel, 2001; Seidenberg, 2017; Snow et al., 1998; Wolf, 2007, 2018).

The primary point in understanding the distinction between symbolic and nonsymbolic systems is that the orthogonal contrast between these two systems likely prompted an important conceptual insight for Engelmann: Given the unconcealed and highly public nature of nonsymbolic, physical tasks—which are entirely overt, consists of component parts that are reasonably easy to deconstruct, and provide the learner with instant feedback, why not “model” covert symbolic tasks after nonsymbolic, physical tasks, especially for teaching and learning?

In other words, why not design the teaching of symbolic systems in ways that require the learner to respond overtly? In doing so, why not also break down complex cognitive tasks into manageable, public pieces that can be directly observed, taught explicitly, and made accessible? Finally, why not provide instant feedback to the learner about their performance on each overt piece along the way? This reverse engineering analysis likely served as the conceptual hook for Engelmann’s philosophical and technical approach to teaching and learning, as well as his analysis of the design of instruction principles that served as the foundation for Direct Instruction. Of course, in keeping with Shakespearean convention, it conveniently answers the two-question conceit posed at the beginning of this section.

Engelmann’s Direct Instruction Philosophy

Understanding the reverse engineering of symbolic, cognitive tasks and Engelmann’s analytical commitment to impersonate the basic features of nonsymbolic, physical tasks so that the component parts are overt and directly accessible to the teaching process is critical to also understanding Engelmann’s uncompromising approach to the learner’s success. Armed with this conceptual insight, as well as his analysis of the content to be taught (Engelmann & Carnine, 1982), Engelmann was unsurprisingly confident in his ability to analyze the full range of disciplinary knowledge and content, however elusive, into teaching sequences, tasks, operations, and routines and to make them accessible to the teacher and in particular, the learner (Engelmann & Carnine, 1982). Thus, that confidence permitted him to make the following bold assumptions about the learner, teacher and the schooling content:

1. The premise from which all the procedures derive—either directly or indirectly—is that the teacher is responsible for the learning and performance of the children (Engelmann, 1969, p. 39).

2. The first and most important step in cause finding is to discover what the child has failed to learn (Engelmann, 1969, p. 8; emphasis in original).

3. We must assume that the learner’s behavior is lawful, which means the learner who possesses the assumed mechanism will learn what the communication demonstrates or teaches (Engelmann & Carnine, 1982, p. 4).

4. A further teaching assumption is that the more carefully skills are taught, the greater the possibility that the child will learn them (Engelmann, 1969, p. 25).

These are bold assertions. In fact, other than Engelmann, I don’t know of another prominent or popular figure in education who unconditionally (and publicly) embraced the full and unfettered burden for student learning without demurring, or placing the burden, either partially or entirely, on the back (or in the brain) of the learner.

For some, these assertions may seem naïve, perhaps even reckless because they place the full burden of teaching and learning on a single individual—the teacher—while failing to recognize the unforgiving complexities of moving a child from a “state of unknowing or partial knowing” about the world, to a state of more “complete knowing” about the world (Carroll, 1963), especially in the complex host environments known as schools (Kame’enui et al., 2000). Of course, Engelmann was not naïve to the demands placed on learners by schools and the unforgiving content they were required to learn in a typical 180-day school year. Nor was he naïve in his thinking about the full-scale systems (e.g., teacher professional development, teacher preparation, design and development of curriculum programs, professional organizations, administrative leadership) required to support teachers in the daily demands placed on them, especially when teaching the full-range of diverse learners, from children identified as gifted to children identified with disabilities (Engelmann, 2020; National Institute for Direct Instruction, https://www.nifdi.org/).

Instead, Engelmann’s bold assertion that the “teacher is responsible for the learning and performance of the children” reflects his analysis of symbolic systems as described previously and summarized in Table 2. It is important to note that the distinction and analysis permitted him to make such bold assertions, in part, because he also knew there was an empirical path to “exposing” complex cognitive tasks that take place in the brain, and in doing so, to give learners, teachers, and curriculum developers immediate access to them, as well as a fighting chance to be successful. It is also important that he recognized that it provided educators with a systematic, logical, and scientific basis for designing instruction and curricula that gave the learner and teacher a high probability of succeeding. And it is important that the assertion reflected Engelmann’s deep and uncompromising commitment to the success and welfare of children—the ethical imperative. As evidence of this commitment, Evan Haney, coauthor of Direct Instruction programs with Engelmann, offers the following poignant observation:

Program development isn’t about tinkering with formats for the sake of artificial precision. Fundamentally, it’s about ethics. We sweat and fuss and sometimes torture ourselves for hours to find just the right form of presentation that will disburden students in their pursuit of knowledge. We should carry that burden. Students should not. . . . I don’t think you get Zig if you don’t see that everything he did, especially the painstaking work of program development, was driven by a passionate ethical concern for the welfare of children. (Haney, 2019)

For Engelmann and his colleagues, the learner should not carry the “burden” for learning. Instead, the burden should always be on the program or curriculum developers, the teacher, and the teaching process. This tenet was simply not negotiable, and Engelmann was intensely consistent in his commitment to this ethical imperative.

The Falsifiability of Engelmann’s Direct Instruction

Engelmann’s fervid ethical commitment to the welfare of children was predicated on a set of principles that governed how an individual or group should conduct itself. It is interesting that it was important for Engelmann that the ethical commitment was not just ethical, but also logical and empirical, because it was anchored to how he as an educator would conduct his professional work, including his thinking, writing, and curriculum development. Engelmann and Carnine (2011) explain their two-part logical and empirical analysis:

In broad terms, the first and primary analysis is logical. Questions of clarity are approached first from a logical perspective, then from an empirical perspective. Is the presentation clear in terms of what we show and the discriminations we teach? In practice, the answer is never definitively yes, but rather, apparently, yes, until the empirical analysis renders the final decision of clarity. (p. 125)

Given this commitment to a “logico-empirical” model, Engelmann made what could be characterized as an “ethical” decision to expose Direct Instruction as a comprehensive educational approach to the scientific method and a longitudinal program of evaluation and research. An important tenet of the scientific method is that claims to knowledge require evidence and must be falsifiable; that is, claims must be empirically tested and potentially disproven (Kuhn, 1962; O'Hear, 1980). Falsification in science insists that proponents of a particular approach, irrespective of discipline, must make every effort to “falsify” or strive to invalidate claims about their approach, ideas or hypotheses.

The University of Oregon Direct Instruction Model and the National Follow Through Project

To that end, in the late 1960s, Engelmann, along with his colleagues at the University of Oregon, whose contributions to the development of Direct Instruction cannot be overstated, Wes Becker and Doug Carnine, agreed to participate in a national, federally sponsored research study. This study, known formally as the National Follow Through Project, was designed to examine multiple educational approaches to increasing student achievement for disadvantaged children and was considered at the time to be the “largest and most expensive social experiment ever launched” (McDaniels, cited in Becker, 1977, p. 520). Much has been written about the quasi-experimental research design that the National Follow Through Project employed to compare the effectiveness of the University of Oregon Direct Instruction Model with eight other educational models and approaches (e.g., Piagetian Cognitively Oriented Curriculum Model, the Bank Street College Model, the University of Kansas Behavior Analysis Model, the Tucson Early Education Model), as well as non-Follow Through comparisons (Becker & Carnine, 1980).

The results of Project Follow Through have been debated vigorously for more than 2 decades (1980s and 1990s), subjected to a range of different statistical analyses, and reported broadly (Becker, 1977; Becker & Carnine, 1980; Guthrie, 1977; House et al., 1978; Stebbins et al., 1977). The U.S. Department of Education charged two independent contractors (Stanford Research Institute and Abt Associates) with analyses of the data and found that the Direct Instruction model was the most effective model for raising the basic skills achievement of low-income students who participated in Project Follow Through for 4 years (Stebbins et al., 1977). In summary, the Direct Instruction Model produced greater gains in basic skills (e.g., word reading, math computation, and spelling), cognitive problem solving (e.g., reading comprehension, mathematical reasoning), and affective behavior (e.g., self-esteem and locus of control) than the other eight major educational models. It is interesting and unexpected that several of the other educational models (e.g., Open Classroom Model, Piagetian Cognitively Oriented Model, Bank Street College) failed to increase student achievement.

The Direct Instruction Model: Strongest Evidence of Effectiveness

As further evidence of the effectiveness of Engelmann’s Direct Instruction, Borman et al. (2003) conducted an exhaustive meta-analysis on comprehensive school reform models and student achievement. The meta-analysis reviewed the research on the achievement effects of comprehensive school reform models and summarized the effects of 29 widely implemented educational models. The researchers grouped the 29 models into four categories ranging from Strongest Evidence of Effectiveness to Greatest Need for Additional Research (p. 154). Models with the strongest evidence of effectiveness included “those that had 10 or more studies of schools and students across the United States,” and those that demonstrated “statistically significant and positive achievement effects in studies using comparison groups or third-party comparison designs and have accumulated evidence from at least 5 third-party comparison studies” (p. 161).

Direct Instruction was one of only three reform models that was found to have the “strongest evidence of effectiveness” with an overall effect size of d=.21 (Z = 11.61, p < .01), and a 95% confidence interval of d =.17 to d = .25 effect size. A total of 49 studies were included with 38 of those as third-party comparison studies. Overall, DI was considered the most studied program and demonstrated the largest effect sizes for student achievement. It is worth noting that the DI model had (1) a strong focus on early literacy, (2) explicit instruction on the five essential elements of beginning reading (National Reading Panel, 2001), (3) on-going professional development throughout the school year, (4) high-quality implementation support, and (5) strong research support (American Institutes for Research, 1999). The DI model also utilized placement tests to conduct initial grouping of students by skill level and ongoing assessments to regroup students for instruction when necessary. According to Borman et al. (2003), the Direct Instruction model represented a class of school reform models that held significant promise for improving student achievement.

In addition, the American Institutes for Research (1999) produced an independent report that profiled 24 models of school reform in which each were rated for their effectiveness on raising student achievement. Direct Instruction was one of only two models that received a rating of strong research support. In summary, it appears that Direct Instruction, as a “powerful teaching system” (Heward & Twyman, this issue), is unique in the amount of research support it has garnered demonstrating positive effects on student achievement.

Carnine’s Program of Research on the Components of Direct Instruction

The longitudinal, large-scale research on Engelmann’s Direct Instruction as an educational model in Project Follow Through represents but one major strand of research that provides evidence about the model’s effectiveness as a “powerful teaching system” (Heward & Twyman, this issue). Another equally important strand of research that was part of the National Follow Through Project at the University of Oregon but not widely known was an applied program of research that Doug Carnine developed and directed for the lifecycle of the Follow Through Project. This program of research examined many of the “design of instruction” components of Direct Instruction. As noted previously, this research had a particular focus—the development and preparation of instructional materials that have a “high probability of preventing learner errors and/or misconceptions and misrules” (Tennyson & Christensen, 1986). In addition, this program of applied research gave Engelmann and Carnine an experimental “sandbox” to explore and empirically test their ideas concurrent with authorship of the Theory of Instruction (1982), which took 10 years to write. Although they used this sandbox to pilot and field test many ideas as a precursor to conducting formal studies, there remains a great deal of sand still to be sifted in the Direct Instruction sandbox.

Carnine’s program of applied research employed both single-case and group design experiments to examine a wide-range of issues related to designing effective instruction. Research studies focused, for example, on the manipulation of textual variables (e.g., pronoun constructions, proximity of critical information to target vocabulary words) in narrative and expository text to determine what inhibited or disrupted reading comprehension, as well as the selection and sequencing of examples for teaching a range of content in reading and math. Selected studies from this program of applied research for a selected time period (1977–1980) are found in Table 3.

Table 3.

Selected studies of direct instruction applied research (1977–1980)

Authors & Year of Publication	Topic of Study
Carnine (1980)	Three procedures for presenting minimally different positive and negative instances
Kame’enui et al. (1980)	Explicit teaching of reversible passive-voice and clause constructions
Kame’enui et al. (1982)	The role of redundant information in vocabulary learning on vocabulary-specific and general comprehension questions
Carnine, Stevens, Clements, & Kame’enui, 1982c	A facilitative questioning strategy for identifying character motives
Carnine, Kame’enui, and Woolfson (1982a)	The effects of a range of textual variables on text-based inferences in reading comprehension
Carnine, Kame’enui, and Maggs (1982b)	The components of analytic assistance including statement saying, concept training, and strategy training
Patching et al. (1983)	The direct instruction of critical reading skills
Carnine, Kame’enui, & Coyle (1984)	The utilization of contextual information in determining the meaning of unfamiliar words
Kame’enui and Carnine (1982)	The ecological validity of fourth grader’s comprehension of pronoun constructions in narrative and expository texts
Kame’enui and Carnine (1986)	The preteaching and concurrent teaching of the component skills of a subtraction algorithm
Kame’enui et al. (1986)	Two approaches to the development phase of mathematics instruction

As an interesting aside, Adams and Engelmann (1996) incorporated the effects of many of these studies on the components of Direct Instruction in their own meta-analysis of 350 research studies on Direct Instruction conducted over a 25-year period (see Engelmann & Carnine, 2011, Appendices A & B, for list of studies; also see Stockard, this issue).

The Design of Instruction Research

Carnine’s program of applied research on selected design of instruction elements was not a local artifact of the National Follow Project at the University of Oregon. In fact, it represented an uncommon slice of an otherwise wide-ranging, esoteric, and provocative research literature that has its theoretical and historical roots in multiple disciplines, including, for example, experimental psychology, applied behavior analysis, computer science, mathematics, communication, and theory of information.

This research literature embraces prominent design of instruction theorists and architects in concept and rule learning (Gagne et al., 1992; Markle, 1978; Markle & Tiemann, 1969, 1970; Tennyson & Christensen, 1986; Tennyson & Cocchiarella, 1986; Tennyson & Park, 1984), as well as widely respected design of instruction disciples, such as Jerome Bruner. Bruner’s (1966) design of instruction model included four critical features: (1) a description of the experiences necessary for learning, such as the “prerequisite skills” a learner must know as a precondition for instruction and for achieving a change in learning behavior; (2) an analysis of the structure and forms of knowledge; (3) a specification of teaching sequences in the materials to be learned; and (4) a system for monitoring and rewarding student performance during the instructional process (cited in Moore, 1986). Engelmann’s Direct Instruction coincidentally incorporates all of Bruner’s design of instruction features and more (see Engelmann & Carnine, 2011, for a critique of Bruner’s analysis).

In addition, the design of instruction research shares theoretical and conceptual linkages to the “instruction set architecture” invoked in the computer science, machine-mediated learning, and the theory of information research literature (Goldstine, 1972). This research literature includes Feynman’s computation lectures on coding and information theory (See Chapter 4 in Hey & Allen, 1996). It is interesting that information theory is not traced to the 1965 Nobel laureate in physics, Richard Feynman, but to Claude Shannon of Bell Telephone Laboratories, who is considered the “father of information theory” (Campbell, 1982). Shannon specified a set of mathematical theorems for sending messages from one place to another “quickly, economically and efficiently” (Campbell, 1982, p. 17) and was known for his design of early-warning radar systems, “color television” transmission, and the recovery of intact messages from distant spacecraft. What Shannon offered the field of information theory was a formal logic and mathematical framework for thinking about information as communication, including a mathematical formula (i.e., Shannon’s entropy) for calculating the “minimum capacity required” to reliably transmit “information” contained in a message (Verma, 2020).

In principle, Shannon’s logic about “information as communication” is not unlike Engelmann and Carnine’s (1982, 2011) “logico-empirical approach to creating “faultless communications” (Twyman, this issue)—communications in which, from Shannon’s perspective, entropy is low (uncertainty is reduced) and the probability of receiving a clear signal/message is high. For both Engelmann and Shannon, the primary test of the efficacy of their systems of communication was an empirical one that would “render the final decision of clarity” (Engelmann & Carnine, 2011, p. 125): Was the message successfully received—by either a spacecraft, or a learner?

An Ode to Glerm1

The scientific evidence for Engelmann’s logical and empirical analysis of Direct Instruction as an educational model and its components is extraordinary, and with few peers (American Institutes for Research, 1999). I have studied Engelmann’s Direct Instruction, as well as its pedagogical inflections, such as little direct instruction, for 45 years. I have also had the rare privilege of working directly with Zig Engelmann, Wes Becker and more intimately with Doug Carnine at the University of Oregon for almost 3 decades. That noted, from my perspective, however, the two-dimensional logico-empirical approach that Engelmann, Carnine, and colleagues have offered the field for more than a half-century is necessarily incomplete. Why? It fails to incorporate in its analysis at least one other critical dimension—one that is understandably elusive and especially difficult to capture, quantify, replicate, and represent clearly and unambiguously. It is important to note that this missing dimension is, from my perspective, the most important to Engelmann’s Direct Instruction. However, it is also the most ineffable—ideation.

Shakespeare was a master at employing literary devices (e.g., repetition, allusion, personification, anaphora, monologues, soliloquy, anagnorisis) in his plays and poetry. The primary purpose of these “tricks” was to convey the power of his message. However, these literary devices were no substitute for his ideas and words, which wielded the power of his shrewd and enduring message.

Likewise, Engelmann and colleagues employed numerous “devices” in Direct Instruction programs that served as conceptual and analytical hooks. These design of instruction “devices” described in Engelmann and Carnine’s (1982) Theory of Instruction are many, and include, for example, a “taxonomy of knowledge forms” from “basic” to “complex” (pp. 19–33)—forms that covary with learner knowledge and development. It also includes a set of “juxtaposition principles” (pp. 38–43) designed to guide the systematic selection of the teaching materials (i.e., the “set up”) and to convey “sameness” that requires the highly specified selection, sequencing, and presentation of positive examples designed to communicate how the examples are the same. Likewise, it includes the “difference” principle that requires the highly specified selection, sequencing, and presentation of nonexamples to communicate the essential “difference” (pp. 11–12) between examples and nonexamples of the target concept/content. The placement of the “minimally different” examples (or as Markle & Tiemann, 1969, called them, “close in” examples) in a teaching sequence is prescribed logically, empirically, and exactingly.

Engelmann and Carnine (1982) also provide concrete details and examples of how these devices can be used to develop and teach single dimension concepts (e.g., under, wider, balanced precariously), multiple dimension concepts (e.g., furniture, animal; pp. 37–78), if–then rule relationships (e.g., “If two things happened together, it doesn’t necessarily mean that one causes the other”; pp. 93–106), and “cognitive routines” (e.g., factoring any number from an expression; pp. 194–213). Of course, Engelmann reminds us that “A faultless sequence of examples is one that admits to only one inference” (p. 38). I am rather confident he meant to say, “one and only one inference!”

It is important to note that these devices were not mere tricks for “causing learning.” Instead, they fundamentally represented insightful ideas, and it is these very ideas and the resulting ideation of Engelmann, Carnine, and their coauthors that is missing from the two-dimensional logico-empirical analysis. From my perspective, no amount of logic and empiricism would likely produce these ideas and so many others. Thus, their logico-empirical model is necessarily incomplete.

The unique ideas communicated through the design of instruction devices are expansive and impressive. Moreover, they cut across a range of different disciplinary content to demonstrate the power of sameness in the communication of the selected message. These ideas, for example, coupled with the design elements previously noted, are found in the teaching of (1) reading (e.g., Carnine et al., 2017; Coyne et al., 2011); (2) morphographic spelling (e.g., Dixon, 1992); (3) basic math (e.g., Stein et al., 2006); (4) writing (Coyne et al., 2011); (5) geometry (Carnine, 1992); (6) complex math problem-solving (Engelmann et al., 1992); (7) secondary science instruction (Woodward & Noell, 1992); (8) social studies and history (Kinder & Bursuck, 1992); (9) the fundamental skills of higher order thinking (Carnine & Kame’enui, 1992; Grossen, 1992); (10) problem solving (Neidelman, 1992); (11) classroom management (Kame’enui & Darch, 1995); and (12) the management of the full-range of social and behavioral challenges (Colvin, 2020).

The ideas are unique because they reduce complex (and often “messy”) content into parsimonious and pithy bites (and “bits”) of information that capture and represent a “grammar” or a Big Idea (Kame’enui & Carnine, 1998) for inducing what is the “same” (sameness principle) across a range of examples that are greatly different (difference principle). For example, in geometry, Engelmann and Carnine (2011) reduced seven complex formulas for calculating the volume of seven 3-dimensional figures into a variation of a single basic formula (base x height). Likewise, in the study of American history, Carnine et al. (1998) created a “Problem-Solution-Effects” grammar for conceptually linking and recalling a full range of greatly different social and historical events. Finally, in beginning reading, Carnine et al. (2017) developed a systematic strategy for decoding a range of words types through a chaining task transformation sequence—a sequence for teaching a range of word types, beginning with vowel-consonant (VC) and consonant-vowel-consonant (CVC) word types with continuous sounds (e.g., at, sam) first, followed by introducing CVCC word types that begin with continuous sounds (e.g., runs, lamp), then CVC word types that begin with stop sounds (e.g., hot, cap), then CVCC word types that begin with stop sounds (e.g., cast, hand), etc. (see Carnine et al., 2017, pp. 80–81).

These “Big Ideas” represent a provocative and transformative device for gaining efficiencies in a curriculum designed for the explicit purpose of maximizing precious instructional time. As Engelmann and Colvin (2006) assert, “DI programs attempt to control all the variables that influence student performance within the context of a packaged program. . . . If variables are adequately controlled, more is taught in less time, and student learning is accelerated” (p. 6). Of course, time—allocated or engaged—is a critical factor in a fixed 180-day school schedule, especially for students who are assessed as not keeping pace in their academic performance with grade-level peers. Students who are “behind” face the “tyranny of time” (Kame’enui, 1993) and being taught “more in less time” is absolutely essential. Anything less places their future in enormous academic and economic jeopardy.

In the complete absence of implementing Engelmann’s Direct Instruction programs at a high level of fidelity coupled with a commitment to monitor students’ academic progress closely employing progress monitoring measures that are independent of DI programs (as a means of falsifying DI program bias), teaching more in less time is arguably problematic. Why? Because it sets the stage for significantly complex instructional (e.g., managing student absences, scheduling “double dose” lessons) and teacher personal teaching efficacy issues (e.g., teaching fatigue associated with teaching more in less time, teaching to a high criterion level of student performance, managing small-group instruction; Smylie, 1988). Traditional commercial curriculum programs are primarily designed to cover a wide range of content and topics. Thus, they are ostensibly designed for “exposure” not mastery, with an architecture that is more “horizontal” (i.e., “a mile wide and an inch deep”) than “vertical.” Unless these traditional commercial curriculum programs are redesigned to teach to mastery, or designed to “teach less more thoroughly,” in which the “less” that is taught includes the most critical Big Ideas and content specified in state content standards across disciplines, then as the Bard in Hamlet reminds us, we “defy augury.”

It is asserted that the Bard changed the English language with the minting of thousands of words and phrases, not including nonce words, as in Lewis Carroll’s first stanza of his poem “Jabberwocky,” or Engelmann’s famous teaching sequence, “glerm,” in which he used a nonsense word to teach a basic single dimension concept, such as “over” (Engelmann & Carnine, 2011, p. 9). There is, however, little argument that the “Immortal Bard” has had a sempiternal influence and impact on the written and spoken language, and perhaps on our big and little ideas about human nature.

It’s too early to determine if Engelmann has changed the field of education, at least in the way the Bard changed the English language—perhaps it’s an untenable standard for anyone, but especially one who was a factory worker and investment advisor (Barbash, this issue), and stumbled unwittingly into the field of education. Education is arguably an immature “science” (Education Sciences Reform Act, 2002) that is markedly susceptible to “false and fashionable” ideas (Kame’enui, 1991, p. 17). The heroic and underfunded efforts of the Institute of Education Sciences (IES)—the research, evaluation and statistical arm of the U.S. Department of Education—notwithstanding, education as a discipline is more like English literature and the study of Shakespeare than it is a mature science such as chemistry or even the concept and rule learning research of experimental psychology. However, those in education who, like Engelmann, hold to the same ethical imperative—faultless communications, a logical-empirical (and ideational) analysis of the “type of learning that is caused by teaching” (Engelmann & Carnine, 2011, p. 11)—will no doubt consider his contributions prodigious and immortal,2 which is entirely and predictably Shakespearean.

Lest you think that immortality loomed large in Engelmann’s world, he would likely invoke the king’s words in admonishment—Shakespeare’s King Lear, that is—to remind us that the real challenge is in the “smell of mortality” itself that warrants our full, unsullied attention and dogged efforts. In doing so, he would also insist, nay, demand, that we continue to keep our focus on the ethical imperative and challenge ourselves to continue to falsify, revise, improve, and “perfect” the design of instruction so that all children can successfully negotiate the symbolic system that the Bard saw as a stage where “. . . all men and women are merely players; They have their exits and their entrances, And one man in his time plays many parts. . .” (Shakespeare, As You Like It) (Steevens, n.d.).