Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13

Three Innovative Teaching Systems: Common Sources of Effectiveness

by George L. Geis - 1970

Sections of Dr. Geis' paper have been adapted from an address he delivered to the 22nd Annual Meeting of the American Conference of Academic Deans. His concern is to examine three teaching environments—the talking typewriter, the coursewriter, and the SAID system (a speech auto-instructional device). After explaining what each involves and demands, he goes on to explicate the instructional engineering principles common to all three. Evoking the Skinnerians, the writer stresses the effectiveness of the three systems and argues for an informed approach to "planning for, and effective production of, human learning" He wishes to express his appreciation to Drs. Fred S. Keller and Harlan L. Lane for criticism and suggestions. Dr. Geis is now with the Center for Research on Language and Language Behavior at the University of Michigan.

Over the years teachers have suggested, and psy­chologists (with an embarrassing lack of modesty) have agreed, that education can be markedly improved by communication of the findings from psycho­logical laboratories to those who are engaged in instruction. The fact that until recently almost no changes in education have resulted from such dis-semination could be attributed to the psychologist's feeble powers of persuasion or to the perverse rigidity of the teacher. A more likely explanation is that the gap be­tween laboratory and classroom cannot be bridged by talk. Instead, a technol­ogy of behavioral modification is needed, analogous in many ways to engineer­ing and medicine. The birthdate of such a technology, spanning the distance between the art of teaching and the science of learning, might be 1954, the year in which B. F. Skinner's pronunciamento was published.1 Slowly devel­oping since that time has been a set of principles more specific than, but of course not incompatible with, general principles of learning. These engineer­ing principles, if the guidelines for practitioners can be so characterized, are emerging from the applications themselves. To distinguish them from more general and theoretical principles of learning, they might be called principles of teaching. However, teaching is usually equated with the activities that teachers traditionally engage in. Those activities represent only a few of the many possible conditions for learning, often the most ineffective ones. The job of those interested in student learning is not to engage in "teaching behaviors" but to develop and manage those sets of conditions which optimize student learning.

It is the purpose of this paper to explicate some instructional engineering principles which emerge from an examination of three teaching or learning environments. These systems will be described and common characteristics extracted which seem to be basic requirements for effective and efficient teaching.

The Talking Typewriter

 For several years the psychologist Omar Moore has been interested in formally arranging the en­vironment, its opportunities and its consequences, to produce learning in young children who are average, ultra-rapid, or ultra-slow learners.2

The part of Moore's "responsive environment" to be described is the labo­ratory, only one element of the whole school environment. The responsive environment, Moore insists, is a social and cultural, as well as mechanical, sys­tem—an intricate and carefully planned set of experiences. A child sits in one of the laboratory booths at what is popularly called the "talking typewriter." A booth assistant, at a control panel mounted outside of the booth, controls the presentations of the complicated typewriter output. The mechanical con­trol and recording system involves a computer with multiple memory systems, an audio recording, a visual, and a logic and control system.

The laboratory buildings are windowless and air-conditioned. The rooms or booths allow immediate observation through one-way mirrors; permanent rec­ords may be obtained through photographic recordings with cameras placed at special camera ports designed not to intrude upon the child at work.

The new child at school is introduced to the laboratory by another child rather than by an adult. The child-guide explains rules relevant to the labo­ratory: for example, that the child need not come to the laboratory unless he wants to, that he can leave when he wants to, that he need not explain why he is going, that if he leaves he is allowed to come back the next day. At a child's appointment time each day, the classroom teacher informs him of his oppor­tunity to go a few yards across the playground to the laboratory. The child may accept or reject his turn. Moore remarks that "day-in and day-out, chil­dren elect to come to the 'laboratories.' However, it frequently happens that a child does not want to leave when his time is up."

On his first trip to the laboratory after the guided tour, the child is directed to a chair in front of the apparatus by the booth assistant, who leaves after telling the child to enjoy himself and raise his hand if he wants anything. This first phase Moore calls "free exploration." The environment is designed to take the brunt of some aspects of exploration; e.g., the keyboard is able to re­sist pounding. It responds immediately and sensitively to some other behaviors; e.g., when a key is struck, the typewriter types the letter in large type and the apparatus pronounces the name of the character. On the other hand, the type­writer will not respond to certain kinds of behavior; e.g., no two keys can be depressed simultaneously. What do these youngsters, many of them below three years of age, do in this environment and in the absence of a human teacher? As the name of the phase suggests, they explore; some of them pains­takingly go over the keyboard, pressing each key in turn; others pound ran­domly on the instrument. After this half-hour session, the child is usually eager to come back.

In the exploring phase, as in all others, the computer records the child's per­formance in the laboratory; thus, the length of time the child spends in the booth and the number of times he presses each key are recorded. In addition, the booth assistant, who has been observing the child, keeps notes on other behaviors. At daily conferences teachers, attendants, and other professional staff review each child's laboratory performance. The laboratory supervisor decides when the child is ready to move on to the next phase. An important factor in that decision is the child's demonstrated interest; if his interest seems to wane, whether it be after dozens of sessions or only two or three, he is usually advanced to the next level.

Moore's writings describe in detail the sequence of learning that moves the child slowly through "search and match," word construction, and reading and composition. Reports of other applications of this general system are of equal interest and might well be consulted.3 It is sufficient to point out here a few salient features of all phases. For example, the child not only paces the instruc­tion but also contributes to it: his own vocabulary is used to provide interest­ing, relevant content and to individualize the subject matter as well as the process of instruction.

The almost complete record of the child's daily laboratory activities is sup­plemented by a comprehensive history of the child, including information on his physical condition and his development history, which has been collected along with intelligence test data before the child enters the laboratory." Daily evaluations of each child can be made on the basis of extensive data. "Evalua­tion" here is not a means of screening and discarding students. It is a way continually to monitor the student in order to adjust and optimize this learning environment.

Moore has described many individual cases which demonstrate the power and effectiveness of this comprehensive plan to educate the child—a plan based in part upon a complex set of instruments and in part upon a careful analysis of the child, of his educational needs, and of the physical and human resources necessary to meet certain instructional goals.

As a result of being in the learning environment which includes the talking typewriter, first-grade children develop remarkable performances. Many, for instance, achieve at the sixth-grade level on standardized reading tests.

Moore's own definition of the responsive environment summarizes the fea­tures being stressed here: "(a) It permits the learner to explore freely, (b) it informs the learner immediately about the consequences of his actions, (c) it is self pacing, (d) it permits the learner to make full use of his capacity for dis­covering relations of various kinds, and (e) it is so structured that the learner is likely to make a series of interconnected discoveries about the physical, cul­tural or social world" (p. 2). Some of the features of this learning system characterize another and quite different environment—that of a computer-based instructional system for college level students.


The   University   of   Michigan's Center for Research on Learning and Teaching has been the site of a student station for computer-based instruction.4Roughly 15 such installations are in operation on as many campuses. The apparatus consists, again, of a typewriter at which the student sits and auxiliary equipment (such as a tape recorder and a slide projector and screen) under the control of the computer. The com­puter keyboard and additional apparatus are connected directly with a con­trolling system at the I.B.M. Watson Research Center in Yorktown Heights, New York. (The system is called Coursewriter by I.B.M.) Typically, a stu­dent identifies by code on the typewriter his selection of a particular piece of teaching material. A wide variety of sample units, developed by members of the faculty and the Center, are available at the University. In addition, the station at Michigan has available materials developed at other universities and at the Watson Center.

Text material is presented, usually in the form of a bit of information typed out by the typewriter, followed by a question which requires the student to type an answer. The answer can be as simple as a choice from a list of alterna­tives or as complicated as a whole paragraph of text. The sequence, then, is normally a problem followed by an appropriate, probing question requiring active participation by the student before he can advance to the next problem or question. Depending on the student's answer, there are differential consequences. The student may be told that he was right. If wrong, he may be told merely that he was wrong; usually, however, he receives more complicated feedback. For example, he may be told what part of his complex constructed answer was correct. Thus, in a French language program, a student's answer: le plume de man aunt may be returned as: l. . plume de m . . . .nt. ., with the computer retyping the correct part of the student's answer. If over a period of time a student has shown a serious lack of understanding, he may be switched to a remedial branch; on the other hand, a student demonstrating ad­vanced skills may be skipped to a later sequence.

Attachments such as tape recorders and slide projectors offer additional means of input to the student. Other accompaniments are possible, and a vari­ety are already in use: for example, in a course in Otology, a model of a human skull is used, and in a course in Physics, an optics laboratory bench.

As in Moore's learning environment, the student is required to respond. In addition, there is provision for feedback to the student about his performance. A third characteristic is the production of a permanent record of that per­formance. The student goes through a particular instructional sequence, leav­ing, as it were, a trail of footprints describing his route of learning. These data can be used as more traditional test data are (i.e., to judge the student). They also can be the basis for diagnosing student difficulties and prescribing appropriate remedial work, or for diagnosing the effectiveness of the teaching material itself, with examination of the error data leading to revision of the material. With such a computer system, it is possible to reduce and treat stu­dent data so that average error rates, average latencies of response, and the like can readily be obtained after a group of students has gone through an instruc­tional sequence, allowing precise and immediate revision of materials.

The development and operation of such installations (i.e., Coursewriter, the talking typewriter) require not an individual but a team. In this instance, the team includes the faculty author, engineers and experts at the Watson Research Center, and Dr. Karl Zinn and a staff of assistants at CRLT; "team teaching" has taken on another meaning here.

It is with this system and similar installations that we most clearly see the new function that a faculty member assumes when he becomes an author of material to be used in a controlled learning environment. The material is pro­grammed. Programmed instruction, whether in the form of a simple text or in this more elaborate computer-based format, requires the author to do a careful and exacting analysis and description of so-called terminal behaviors—the be­havioral changes which he plans to produce in the student. The later design and development of the material itself is a difficult task, but the prerequisite specification of instructional objectives is an even more onerous job.

It is true that teachers, at least in higher education, have traditionally been part-time authors, but their writing has been primarily reportorial, expository, rather than truly pedagogical. The writer of a program, the designer and builder of a learning environment, is in many ways more like the engineer of a bridge than the author of a textbook.

For example, the Coursewriter system demands that the instructional engi­neer or program author specify precisely the limit of acceptability of any answer. The computer, it is often said, is a slave and, indeed, it "will do only what it is told to do. It will recognize as correct only those student responses which it has been told to call correct. The author must consider all possible alternatives (synonyms and the like) for complex as well as simple answers and instruct the computer to accept them as he would. He has the further exciting but exacting option of describing to the student the degree of accuracy of a response which misses the bullseye but still lands somewhere on the target.

The I.B.M. Coursewriter system is in some ways quite unlike the talking typewriter and unlike the speech-shaping device to be described later. Its de­signers did not have a specific instructional aim in mind, nor were they devel­oping a research tool. Rather, a general-purpose system was designed for the widest possible set of users; it is a practical machine for instruction. In some ways it resembles the more traditional ways of teaching: it is limited, in good part, to the textual behavior of student and author; it might be said to in­corporate, unlike the other environments discussed here, only the minimal characteristics which psychological research in learning suggests are desirable in such a system. It has, however, proven to be an effective teacher and, not incidentally, a training device for program authors.

The SAID System     

Another teaching system at  the University of Michigan is part of the instrumentation for the Center for Research on Language and Language Behavior. This is the speech auto-instruc­tional device, acronymically: SAID.5Included on the experimenter's side of the system is a computer with a teleprinter output, tape reader input, monitor­ing equipment, parameter extractors and tape recorder. At present, the equip­ment has been used solely in studies of prosodic features of speech, such as the stress and duration of speech sounds. Yet it is the model of a device which would allow us to teach various dimensions of spoken language to first and second language learners as well as to speakers in need of remedial work such as stutterers or aphasics.

The equipment provides three significant features which both psychological research and practical experience in second language teaching suggest are necessary in a system designed to teach production of accurate prosody in language. First, the system presents tape recorded pattern sentences or sounds which the student is instructed to imitate. The system then "processes the stu­dent's imitation, and instantaneously evaluates its acceptably on the basis of three distinct prosodic features: pitch, loudness, and tempo." This is the an­swer-comparator function mentioned earlier in relation to Coursewriter and which is implicit in the locked keyboard phase of the talking typewriter. Fi­nally, as in the other learning environments examined, the student sees not only whether his imitation is acceptable but, in this particular system, in what direction he should modify an unacceptable imitation in order to meet the requirements.

In the student's cubicle there are loudspeakers for presentation of the pat­tern sequence and a microphone into which the student speaks his imitation when a light on the left-hand side of the panel is turned on. Lights on the right-hand side indicate the particular mode (i.e., pitch, loudness or tempo) which the student is to attempt to mimic; a dial in the center of the student panel feeds back to the student the degree to which he has undershot or over­shot the target sound.

A sample sequence would be this one: The student is seated in the sound insulated booth and listens to a tape-recording pattern. The illuminated pitch light tells him to imitate the pitch of the pattern he will hear. As he mimics the sound, the meter needle swings to one side or the other of the zero center, in­forming him of the degree and direction of his error, indicating that his pitch was too high or too low to match the sample sound. The consequences which follow can be any of a great variety. Programs can be developed which make the computer respond differentially to correct and incorrect utterances in terms of the sequence of material on the tape, of the criteria which are set for acceptability of a mimicked utterance, etc. For example, the student may be asked to repeat until he has achieved three successful imitations of a given sound or ten acceptable imitations. Review sequences can be set in periodically. Criteria of acceptability can be made more rigid on successive reviews. In short, the instructional designer can explore a great variety of instructional sequences and strategies. The equipment can be extremely sensitive to indi­vidual students as well as to different instructors. It can tailor each successive step to the student's momentary competence as indicated by his previous performance.

Several features of the SAID system should be reviewed. First, there is im­mediate and direct monitoring of the student's performance via the teleprinter output. Secondly, the machine has superhuman discriminative and processing capacity. It can make discriminations and minute analyses of particular dimen­sions of the speech sound which are almost impossible for the human teacher to make. Even when the "teacher" is not required to make such fine discrimina­tions, the capacity of the machine to process the student input immediately and reliably and to do something about it (in terms of feeding back specific error) is well beyond that of the human. Truly effective teaching of second languages or effective correction of first language errors may require such a microscopic analysis of the student's production, and systems like SAID are needed for such a task.

The SAID system is unlike the others described in one important respect: it can be used to create entirely new behaviors. The other two systems primarily produce new discriminative controls, bringing already established responses under the control of new stimuli. The SAID system is designed to shape, out of the clay of the existing behavioral repertoire, entirely unique responses. Thus, the student may never have emitted a certain intonation or stress pattern before. By careful, patient guidance, it is possible for the SAID system to es­tablish such articulation in the student's vocal repertoire. SAID allows for automated skill building.

Critical Features

The most immediately ob­vious common feature of these three instructional systems is the sophisticated instrumentation which each of the environments contains. The science and technology of computers is awe inspiring. But a subtler and more impressive science and technology lies behind these learning environments, almost lit­erally behind them: that is the science of behavior and its translation, as in these examples, into a technology of education.

The fact that the three systems discussed all involve computers should be viewed as perhaps indicative but not exclusive. If one were to design optimal environments for learning, it is true that he would probably not produce a classroom. Yet a number of recent non-computer efforts at maximizing the instructional effectiveness of that environment exemplify as well some of the principles enunciated below.6

In each of the three cases cited, several important characteristics obtain which might be termed emerging principles of an educational technology.

(1) In each case an environment for learning has been consciously and carefully designed using a system or set of psychological principles which when compounded should produce effective learning. One or two may serve as examples. The SAID system illustrates the principle of developing new responses by the reinforcement of successive approximations to the criterion be­havior. This procedure of moving toward the behavioral goal in small steps is a direct application from laboratory work on both human and infra-human subjects. In addition, all of the systems incorporate the provision for immedi­ate reinforcement, a principle so crucial to learning that it will be elaborated as a separate point later. The environments are not collections of hardware, not the newest models coming off the audio-visual production line; they are sys­tematic applications of basic principles of learning.

(2) In each case, the environment is designed to be sensitive to the stu­dent's behavior. This is most obvious in the case of the SAID system, where the equipment is able to make extremely fine discriminations. But even in the case of Coursewriter, a well-constructed instructional program could antici­pate a great variety of student responses, the programmer having determined, with more thought and perhaps more skill than most teachers, what the limits of allowable answer variation should be in any instance.

(3) Each of the environments provides for continuing reinforcement of behavior. The environment is so arranged that a correct response produces a change in the environment which strengthens the behavior that produced the change. The reinforcer may merely be a signal to show the student how he did. In the case of the SAID system, for example, a flickering light on the panel tells the student that he is right. If necessary, that system as well as the others can be adapted to provide other kinds of reinforcers such as trinkets and tokens, including money. In any case, the correct behavior of the student im­mediately produces reinforcing consequences.

(4) Each of these systems is adjustive. It was pointed out that these en­vironments are sensitive to the student. They respond differently to differing inputs; their "receptors" are acute. They also have differing outputs: their "effectors" produce a variety of responses. What happens to a student is de­termined, for the most part, by the interaction among: what he has just done, where the teacher wants him to go next, and the path that has been planned to get him there. This extensive flexibility, which allows these learning en­vironments to adjust almost infinitely to human variation, challenges the de­signers of the teaching procedures and materials used in the environments as teachers have never previously been challenged.

(5) Each of these environments provides information to both the student and the teacher. Again, taking the SAID system as an example, the student has immediate feedback to tell him not only whether he is right or wrong but also, at least in gross terms, how far right or wrong he is. (As noted earlier, in many cases such feedback may be considered reinforcing.) The teacher has a permanent record of the student's performance; the permanency and detail of that record is an important feature to note. Not only can the teacher locate a particular student somewhere along the path of learning after each response or after each session but also he is provided with invaluable information about the success or failure of the particular learning sequence. Revision of the in­structional materials or programs in each of these environments is not merely possible; it is an integral part of these systems. They have the potential for continued growth and improvement.

(6) In none of the three cases described is the student in a classroom or an environment that resembles a classroom—even the teacher seems to be miss­ing. The student is connected to the teaching program and materials—to the teacher—by a thin electrical umbilical cord. The environments, therefore, have the promise of mobility. It is possible or even likely that in a very few years a telephone subscriber will become a student merely by dialing the correct ex­change. The symbolically thick, ivy-encrusted walls of the college and uni­versity will not be able to contain, literally contain, the knowledge that has traditionally resided therein. Educators have held a banker's view. The college or school has knowledge on deposit and the means of transmitting knowledge and producing skills. The student in order to make a withdrawal must appear in person and remain in residence for a long time. The possibility of opening the vaults, of moving out to the student, comes just at the time when such a change is sorely needed. Continuing education—education as a continuing part of everyday life throughout one's lifetime—is possible on a large scale only when the rigid geographical and temporal limits of current education are removed. A way is suggested in the foregoing examples. Mechanization allows adjustment to individual variations in terms of both time and teaching strategy. In addition, the environments can be mobile, allowing education to travel to the learner, reducing the need for time and geographic constraints.

(7) In each of the environments examined, a team of people was involved in the design and the monitoring of the learning environment: teachers, psy­chologists, engineers. The basic unit of education has traditionally been the teacher located in the traditional environment, the classroom. The untraditional computer-based environment for learning plus a team of specialists and man­agers represent a new basic unit.

(8) One general point should have emerged from the present discussion: probably the best contribution science can make to education is a model or method rather than a specific solution. It is a simple model in which the designer, experimenter, teacher or teaching machine adjusts future actions on the basis of the effects of previous behavior. The laboratory experimenter de­cides upon future research strategies and upon specific studies to be done as a result of observing the outcome of previous studies he has conducted. This closed-loop model also characterizes each of the learning environments dis­cussed in this paper. In this model, the teacher is placed primarily under the control of the student. The teacher's behavior adjusts moment by moment as a function of the success or failure of his actions in producing the desired changes in the student. This is the crucial aspect of these environments. To the extent that the feedback loop is incomplete, educational changes will con­tinue to be whimsical.

In Conclusion

Data are available on each of these systems which indicate their teaching effectiveness. Indeed, it is almost gratuitous to add that they are effective, for, as was just pointed out, each con­tains provision for self-correction; elements of the learning environment can be continually changed until the environment does what it is designed to do.

Reviewing the attributes of these systems, it is clear that what they have in common are functional, not formal characteristics. Each aims at a different population, provides different modes of instruction, is not unique to a par­ticular content nor to a particular set of behaviors; they do not necessarily ex­clude nor require a traditional teacher (although it is interesting to note how many human jobs are involved in designing, developing and maintaining these "automated" systems). They provide appropriate conditions not only for the student to learn but also for the instructional engineering team to learn about the student, the instructional sequence, and the system.

Designing, maintaining, and revising such a system—the hardware, the con­tingencies, the instructional materials—is an infinitely more complex and dif­ficult job than "teaching." Yet so it might be, for when we engage success­fully in the task of arranging optimal instructional environments, we are dem­onstrating one of the few uniquely human abilities: conscious planning for, and effective production of, human learning.


1    Skinner, B. F. "The Science of Learning and the Art of Teaching," Harvard Edu­cational Review, 24, 1954, pp. 86-97.

2   Omar Khayyam Moore. Autotelic Responsive Environments and Exceptional Children. Publication of the Responsive Environments Foundation, Inc., 20 Augur Street, Hamden, Connecticut, September, 1963.

3   See, for example, Lassar G. Gotkin, "The Machine and the Child," A-V Communica­tion Review, 14, Summer 1966, pp. 221-241.

4   K. L. Zinn, "An Introduction to IBM's Experimental 7010 Coursewriter System as Used via IBM 1050 Remote Terminals at the University of Michigan" (mimeo), Center for Research on Learning and Teaching, Ann Arbor, July, 1966.

5   R. Buiten and H. L. Lane, "A Self-Instructional Device for Conditioning Accurate Prosody." 1RAL, 3, 3,1965, pp. 205-219.

6   For example, see L. E. Homme, P. C. de Baca, J. V. Devine, R. Steinhorst, and E. J. Rickert, "Use of the Premack Principle in Controlling the Behavior of Nursery School Children," Journal of Experiment, Analysis, Behavior, 6, 1963, p. 544; F. S. Keller, "Goodbye Teacher . . ." (mimeo), Invited Address to American Psychological Associa­tion, 1967, Washington, D. C.; D. E. P. Smith and J. W. Kelingos, Manual: A Program for Teachers (Michigan Successive Discrimination Language Program), 1964, Ann Arbor Publishers, especially pp. 7-33.

Cite This Article as: Teachers College Record Volume 71 Number 3, 1970, p. 379-390
https://www.tcrecord.org ID Number: 1756, Date Accessed: 1/19/2022 5:00:43 AM

Purchase Reprint Rights for this article or review
Member Center
In Print
This Month's Issue