Dialogue with a dumb terminal

Terry Mayes describes how higher education students can help to construct a learning resource for their peers

It is a measure of the increasing sophistication of the debate about the impact of learning technology in higher education that fewer academics talk of replacing lectures with courseware. As a teaching technique the lecture is cost effective, even though the quality can vary between charismatic and mind-numbing. Nevertheless, it is often argued that lectures are rather pointless, as the content can readily be learned from handouts or books. Empirical research demonstrates that lectures are ineffective for the transfer of knowledge. Personally, I have to admit that this point was forced home early in my lecturing career. Enthused by a youthful spirit of enquiry, I was keen to measure how much of what I was saying was understood and remembered during the lectures themselves. On a number of occasions I interrupted a lecture and asked the students to answer some questions, all of which were easily answered by the material I had just delivered. The results were dismal. Clearly my students were not listening properly.

I tried forbidding the students to take notes during my lectures, except in designated pauses which I inserted every ten minutes or so. These pauses were not long enough for anything except a summary of the main points. Now, the unanticipated tests were better answered, and I felt that I had discovered something important. By imposing this constraint, students were required to listen for meaning, and not simply take dictation. By this time the students had decided that I was a hopeless eccentric, and started voting with their feet. Only when I gave them my word that there would be no more funny business did they return in numbers.

Learning occurs as a by-product of understanding; understanding comes through solving problems and thinking. Any task which forces people to do this makes an effective learning task, and this fact underpins the approach known as constructivism. The assumptions have been well described by Seymour Papert:

"We understand 'constructionism' as including, but going beyond what Piaget would call 'constructivism'. The word with the 'v' expresses the theory that knowledge is built by the learner not supplied by the teacher. The word with the 'n' expresses the further idea that this happens especially felicitously when the learner is engaged in the construction of something external or at least shareable . . . a sand castle, a machine, a computer program, a book. This leads us to a model using a cycle of internalisation of what is outside, then externalisation of what is inside and so on."

It is not surprising, then, that being required to summarise ten minutes worth of exposition is a more effective learning task than writing verbatim notes, where the objective is simply to record as much as possible of what is said. Constructivism emphasises the learner's search for meaning, and the production of a summary is an effective construction for the achievement of meaning. Despite this, however, the lecture remains an effective pedagogical event because it is more accurately viewed as a social gathering than as useful knowledge-transfer.

Lectures are occasions where the individual is confirmed as a member of a learning community. The whispered asides, the shared jokes, the passing around of handouts are more meaningful interactions than the engagement (or lack of it) with the subject matter. It is also an occasion for the detailed specification of the real task facing the learners: dealing with assessments. One reason why detailed handouts are seen by the students as inadequate substitutes for lectures is that they do not convey the same quality of emphasis. The lecture will convey information on what the lecturer feels most enthusiastic about, and may give good clues about exam questions. Lecturers will also mark out the effective boundaries of a topic, sometimes unintentionally.

Sensitivity to such information may well be as important a determinant of academic success as hours spent studying. The subtlety and richness of this seems associated only with live lectures: most attempts to provide televised lectures have been abandoned in response to student antipathy. It is easy to see that courseware aimed at providing this primary contact with the subject matter is also unlikely to act as an effective substitute. Courseware offers another resource, like books, which the learners can effectively draw on in implementing their learning strategies. Of course some courseware will be effective in allowing the learner to gain a better understanding of the subject matter than is achieved through reading text, or listening to a spoken account. Simulations, virtual worlds, hypermedia, even drill and practice problem solving can all be successful in allowing the learner to achieve understanding. We have called this primary courseware: it provides computer-based access to subject matter content. Nevertheless, the constructivist view of learning is that mere exposure to content is not enough. To achieve real understanding the learner must use the information in the performance of some task. This is, of course, achieved through such conventional teaching methods as the essay and the lab report. There are many ways of using technology to support learners in selecting appropriate content, linking this in new ways and building new constructs. The tools for, and products of these tasks we have called secondary courseware.

To describe the essence of learning in higher education, the above account is necessary but not sufficient. The missing ingredient is dialogue. The learner's understanding must be put to the test of other people's judgement, through discussion. Of course, a kind of internal dialogue is possible: reflection. But a learner's thinking advances primarily through dialogue with others. For most of the history of the university, dialogue has been at the heart of the learning process, primarily in a setting where there was ample opportunity for informal conversation, but also through the more formal device of the tutorial. Over the past 30 years, however, the lecture, rather than the individual or small group tutorial, has become the cornerstone. Associated with this is a creeping assumption that the main job of the university teacher is to present knowledge, in as structured, vivid and palatable a form as possible, but essentially the lecturer's role is to be a delivery agent rather than a respondent and discussant. Recent rapid deterioration of staff/student ratios means that it is no longer possible to offer every student the opportunity of personal dialogue. Few lecturers who maintain an open door policy to students will feature as "research active" on the Research Assessment Exercise return.

A lot is now riding on the multimedia wave. Policymakers recite the expected benefits of new learning technology as a kind of mantra. Yet I believe the central question is rarely addressed: can technology offer an effective substitute for the loss of real dialogue? This is being partially tackled by the collaborative research programme between the Centre for Learning and Teaching Innovation in Glasgow Caledonian University, my own institution, and the Human Communication Research Centre in the University of Edinburgh. The central question is the extent to which learners can benefit from dialogues in which they have not participated directly. The assumption is that much real learning occurs through observation of other learners engaged in active dialogues. Vicarious learning is not a new construct in learning theory; it has been an influential idea in social and developmental psychology, and has been discussed by Robert Gagne in the context of education. It has, though, been somewhat neglected through the emphasis on active task-based participation. Vicarious learning is important in classroom experience, where the questions asked, and comments offered, by some learners can often articulate and expose aspects of conceptual difficulty experienced by others. These dialogues can involve discourse between learner and teacher, discussion between peers, or even direct interaction with courseware. To the extent that such dialogues reveal aspects of learning that are relevant and illuminating to the individual learner, it will be beneficial to make them "re-usable", as a new kind of technology-based learning resource. This we have termed tertiary courseware.

Providing the experience of dialogue without the simultaneous presence of a tutor has been the main goal of intelligent tutoring systems. Although a quarter of a century has passed since Jaime Carbonell demonstrated that it was possible to simulate a tutorial dialogue through machine representation of domain knowledge, the full achievement of that goal has remained largely out of reach. In a recent overview of his group's achievements at Carnegie-Mellon over 15 years of the building of intelligent tutors, John Anderson explicitly abandoned the aim of emulating human tutor-learner dialogues. Only a few well-publicised examples in computer programming and mathematics have met the conditions necessary for modelling the subject matter, the learner and a tutorial strategy in a computationally and pedagogically effective way.

If we cannot yet construct dialogue-providing tutors in situations where discussion and reflection are important components, with programs which can "understand" the conceptual difficulties a learner is experiencing and which can offer direct help, how else can advanced learning technology provide learners with an experience of dialogue? There are at least two strong candidates. One is to design supportive environments for peer tutoring, in a way that will capitalise on the global availability of peers across the Internet to locate suitable one-to-one dialogues. A second, the subject of our research, is to take real discussion between learners and tutors as the raw material from which a new kind of courseware can be built.

Our research is trying to identify those questions and answers, fragments of real dialogue, or even parts of "think-aloud protocols", that will prove to be genuinely "re-usable", that is, will be demonstrably beneficial to other learners. When we understand this, in some kind of principled way, the challenge will be to design a system for capturing dialogues and structuring them for optimal access. A possibility is to record dialogues of previous learners in equivalent situations and to make these accessible for new learners at the right moment in the conceptualisation cycle. The idea is enticing, not least because it avoids apparent weaknesses of mainstream approaches to computers in education and training, the huge investment of time and money to develop effective content-based courseware. There is also a satisfying user-centred basis to the notion of evolving courseware out of real teaching and learning experiences. In a modest way, we can already see this idea beginning to emerge in the form of lists of FAQ's (frequently asked questions). One can readily see that these could be elaborated into a more effective resource. An approach based on learners asking questions, and using these to retrieve answers already available and to identify questions which still need to be answered, is one of the possible implementations already prototyped. The main inspiration for this was a system, called the Answer Garden, developed as part of the Athena project at the Massachusetts Institute of Technology. This demonstrated the potential for integrating a database of FAQ's with email, where questions that could not be answered by querying the database would be emailed to the "expert" who had ownership of the relevant part of the database. The answer would then be added to the database. Attempts to use the idea in teaching have been made at Heriot-Watt's Institute for Computer-Based Learning, and Nottingham's life sciences department. There is a well-established cognitive science approach to instruction which involves an analysis of the errors learners make, so that we may understand the common misconceptions.

However, rather than placing the burden on the technology to detect the misconception, as intelligent tutoring systems try to do, here the responsibility is shifted to the learner to decide whether being exposed to a dialogue about someone else's learning misconception may illuminate conceptual difficulties. Essentially, the learner is told: "Here are some problems previously experienced by other learners, see if you can find one that is similar to your own difficulty". Sometimes misconceptions might be revealed that would have remained hidden. Placing the burden of matching on the student acknowledges that the same question asked by different students may not mean the same thing. Nevertheless, it is common in small group teaching to recognise the question asked by a fellow student as one that you yourself were struggling to articulate. Simply being exposed to other learners' questions appears sometimes to facilitate conceptual advance. There is a danger, however, that too early an exposure to this "discussion level" material might actually inhibit individual reflection.

Even a brief consideration of the basic idea of "re-usable" dialogues raises some fundamental issues. Asking one's own question, and having it answered by a tutor sensitive to the context in which it has been asked, is valuable. How valuable, though, will that same answer be to a future questioner whose learning context may be subtly different? More important, how useful would a database of questions and answers be to a learner who cannot yet pose a question at all?

Of course, it is not necessary always to seek further understanding from asking questions. It may be enough to identify an area in which clarification is sought, and to explore detailed expositions in that area. In which case, one may suppose that primary courseware structured into several levels of detail might provide most of what is required. Our assumption, however, is that there is a crucial advantage in using material generated by real questions and discussion from real learners. This will provide a learner's perspective on a topic that cannot be "authored" by subject experts. It is not a substitute for anything, but an additional resource for learning.

Terry Mayes is professor and head of the Centre for Learning and Teaching Innovation, Glasgow Caledonian University.