3(2) p. 68

PRINTOUT: Two Books on CAI

Ken Autry

Olsen, Solveig. (Ed.), (1985). Computer-aided instruction in the humanities. New York: MLA.

O'Shea, T. and Self, J. (1983). Learning and teaching with computers: Artificial intelligence in education. Englewood Cliffs, NJ: Prentice-Hall.

The earliest hopes for computers in education were pinned on computer-aided instruction (CAI) of the sort now often referred to as "drill-and-practice." Some of the learning sequences touted just ten or twelve years ago, including some primitive efforts of my own, now seem ludicrously quaint and serve as reminders of how far we've come since then. The unforeseen pervasiveness of the microcomputer and the concurrent preoccupation with word processing in composition instruction have shifted the focus away from "drill-and-practice"

p. 69

and toward more sophisticated forms of CAI. At the same time, advancements in composition research have heightened our suspicions about any sort of programmed learning in writing classrooms.

Each of the two books reviewed here in its own way assesses the state of CAI in the humanities and suggests possible future directions. Solveig Olsen has edited one of these, the second volume in the MLA "Technology in the Humanities" series. Part I consists of twelve essays, while Part II, occupying about a third of the book, contains a list of journals and other resources, a list of 124 universities with CAI projects, and a 55-page bibliography.

Each of the first four selections describes a CAI project in a particular discipline: Western Civilization at Notre Dame, German at BYU, French at Western Ontario, and logic at Carnegie-Mellon. Candidly acknowledging faculty reservations and organizational problems, Robert Burns explains how he and his colleagues at Notre Dame wrote 600 tutorials at an average of four double-spaced pages each. Randall L. Jones describes how the BYU German grammar course was developed using TICCIT (Time-shared Interactive Computer-Controlled Information Television), a large-scale learning system started in the 1970s with National Science Foundation funding. Glyn Holmes and Preston K. Covey review projects in French and logic, respectively.

The scope of these projects and the energy invested in them is impressive, and I have no doubt that these are exemplary projects, perhaps models for others within their disciplines. Yet I cannot imagine that the CAI principles illustrated here would be of much use in the

p. 70

teaching of writing. Furthermore, the effectiveness of the projects described may be open to question even in the context of their own disciplines. The Burns essay, for example, describes a Western Civilization course in which writing plays a role only in testing, and the notion of "writing to learn" seems foreign to this course, which is structured around faculty lectures and hundreds of question-and-answer routines.

CAI would seem to be made to order for learning the principles of a foreign language quickly and methodically. Yet, as Randall L. Jones himself acknowledges, this method flies the face of some of the most prominent current language acquisition theories. Also, though the TICCIT approach has proven popular with students, he notes that there is "no reliable empirical evidence" of success.

Two items in the Olsen collection relate directly to CAI in writing. Kathleen E. Kiefer's contribution surveys the most prominent aids to writing, subdividing in what has come to be the accepted taxonomy: prewriting, writing, revising and editing, and comprehensive (those that encompass the other three). Actually, most of the packages Kiefer describes are not CAI in the usual sense of that acronym. Rather, they are helps or adjuncts to the writing process.

The article by Donald Ross, Jr., and Lillian S. Bridwell concurs with Kiefer's contention that drill-and-practice routines are not much help in learning to write (although there are some, such as the computerized COMP-LAB modules developed by Michael Southwell, that have received much attention). They believe that the most appealing software developments are aids

p. 71

to invention and to analysis of text. They argue that teachers and programmers must work together in designing software, and they provide an enticing glimpse of their development of an authoring system at the University of Minnesota. Here, and at several other places in this book, I wished for more detail, particularly more illustrations of a sample program at work.

Two other essays in this collection seem worthy of note: one by Joan Sustik Huntley that anticipates the future importance of videodiscs in CAI and one by Gerald R. Culley that revealingly justifies the seemingly inefficient practice of developing software on mainframes before converting it for microcomputers.

I find the final four essays in Part I the least substantial in the book, partly for their brevity, and partly for their duplication of information on such matters as software evaluation, which is covered in countless other sources.

It is clear that CAI is alive and well but has not yet produced the results that would mark it as a major evolutionary step in education. The authors of Learning and Teaching with Computers explain why CAI has not lived up to expectations and go on to suggest that artificial intelligence (AI) will create a breakthrough.

Tim O'Shea and John Self, both British educators, include among the characteristics of AI the ability to remember and access knowledge, the ability to extend knowledge, and the capacity to draw analogies. The authors illustrate important aspects of AI using examples from two types of program. There is

p. 72

LOGO, a programming language often taught in elementary schools that allows systematic expansion of programming power. And there is MYCIN, a sophisticated medical diagnosis program.

In a chapter reviewing the history of computer use in education, the authors outline eleven types of CAI and point out why the success of each is qualified. They express reservations even about sophisticated CAI networks such as PLATO and TICCIT. They favor computer systems that allow learners' prerogatives and abilities to shape their search for knowledge. And they believe AI to be ideal in fostering this flexibility.

The heart of the book is chapters four and five: "The Computer as Teacher" and "The Computer as Tool." The first argues (as do several contributors to the Olsen book) that most current CAI authoring systems are simply not sophisticated enough to meet student needs and that the time required to program effectively is not available to the typical teacher. Chapter 5 emphasizes student programming rather than programmed learning. "Discovery learning," as the authors call it, puts the "computational means of education. . . directly in the hands of the students." This outlook will be familiar to any who have read Seymour Papert's Mindstorms.

The book's final two chapters look ahead to future possibilities for Al in education. The authors assert that using the computer to maximum effect implies "a re-thinking of the goals of the educational system itself." They are not even hopeful about the possibilities of videodiscs; in their view even these are

p. 73

pre-packaged, forcing students into a lockstep learning sequence of the sort that has proven detrimental to CAI in the past. In the end, the book is a refreshing manifesto for computerized instruction that is adaptable and that leads to genuine intellectual discovery rather than question-and-answer sequences. O'Shea and Self acknowledge that applying artificial intelligence principles to instructional computing takes time, energy, and money, but in their opinion it's the only way to go. The authors do not assess the appropriateness of CAI or AI for particular disciplines. But those interested in the suitability of CAI for writing instruction would do well to ponder the provocative guiding principles suggested here.

Ken Autry teaches at the University of South Carolina.