John Lee, Jean McKendree, Finbar Dineen and Richard Cox
Dialogue is an important aspect of a rich learning experience, and central to the learner's ``enculturation" into the patterns of language and thought characteristic of an academic discipline. Also, we believe, dialogue is often the most effective way for a learner to overcome conceptual impasses during problem-solving. We have proposed, further, that learning can occur not only through participation in dialogue, but also through observing others participating in it -- vicarious learning. Discussing Schober and Clark's (1989) claim that dialogue is only really useful to those who directly participate, we argue here that their study really shows people to have a range of effective interpretation strategies usable in different situations, which can be exploited to develop a coherent model of the ``common ground" in even an observed dialogue. We argue that external representation, especially multimodal e.g. graphical re-representation, is important to many of these strategies. We consider the nature and educational implications of dialogues that facilitate such vicarious re-use.
Dialogue is clearly an important aspect of a rich learning experience, of the sort one hopes to provide in higher education for example. We have argued elsewhere (McKendree et al 1997) that dialogue is central to the learner's ``enculturation" into the patterns of language and thought, discussion and criticism, that are characteristic of an academic discipline; we have noted also that, on a more detailed and local level, dialogue is often the most effective way for a learner to overcome e.g. a particular impasse during problem-solving, or to resolve a difficult conceptual issue. We have proposed, further, that learning can occur not only through participation in dialogue, but also through observing others participating in it. We call this vicarious learning, and we believe that in some manifestations it is very common, as when silent students in a tutorial group learn from discussions between the tutor and others. We seek to understand this process more clearly, and to investigate the ways in which we can take advantage of it to create databases of re-usable dialogues as a helpful resource for distance learners and others.
One challenge to our conception of re-use is presented by Schober and Clark (1989), who claim that in fact dialogue is only really useful to those who directly participate. They describe a study in which ``overhearers" hear recordings of dialogues between pairs where one participant is describing arrangements of Tangram figures to the other, whose task, repeated over 6 trials, is to copy them. Some overhearers (``early") hear all 6 trials, others (``late") only the last three.
The participants tend to use less words over the trials, as they are able to refine the ways they refer to the figures. At the same time, they become more accurate. Overhearers also become more accurate through the trials.
Schober and Clark's objective is to establish that collaboration is an essential part of effective communication in dialogue; they conclude that ``the social role of interacting in conversation plays a central role in the cognitive process of understanding" (p. 228), and that ``[u]nderstanding can only be guaranteed for listeners who actively participate" (p. 230). However, it seems clear to us that these conclusions at least demand further inspection. It is striking how well the overhearers do in fact perform. Our view of these experiments is that they show people to have a range of interpretative strategies that they are capable of bringing to bear, and which can compensate remarkably well for the inevitable disadvantage of not receiving direct, tailored feedback from a conversational partner.
Schober and Clark's analysis depends on the concept, now standard in theories of dialogue, of common ground. The common ground (CG) is essentially the information which is shared by the participants in a dialogue and which forms the basis on which they understand each other. It evolves continuously, though in a task such as the Tangram matching much work in establishing it is done in the earlier trials. The argument is that the overhearer, lacking the attunement that arises between participants, is highly disadvantaged in developing and maintaining a coherent view of the CG. The position is naturally much worse still for late overhearers.
We do not challenge the basis of this analysis, but we note again that the overhearers are actually quite successful. Their descriptions of what is going on, recorded by Schober and Clark, suggest that they are very good at using what cues they can find to conjecture at the CG, and to correct their notion of it in the light of later evidence. Like the participants, they improve in their success at the task, despite the fact that words become fewer. Even the late overhearers improve at a remarkable rate. They certainly achieve significant understanding, and indeed learning.
We accordingly want to modify Schober and Clark's conclusion in that understanding for the overhearer can be at least quite typically expected. Their conclusion about the social role of dialogue we do not dispute: rather, we embrace it. We believe that the social role of interaction is indeed critical in achieving full understanding, through enculturation, but we urge that the relation between the cognitive and the social is more complex than Schober and Clark suggest. Even as a dialogue pair's language converges on a shared understanding, Garrod and Doherty (1994) have shown how pairwise interactions among a population result in the emergence of shared conventions among that population -- enculturation under laboratory conditions. That overhearers are able to participate in this sharing may be of great value e.g. to otherwise isolated distance learners.
Critical factors in vicarious learning will be the extent to which the learner already shares a good deal of common ground with the participants in the dialogue which is being ``overheard", and the explicitness of cues to its establishment and further development. The dialogue will be immediately understood in as much as CG is shared, but on the other hand it will result in an expansion of the CG (and hence at least one kind of learning) in so far as the overhearer can somehow infer those aspects not already shared.
Schober and Clark are concerned with the level of nominal grounding, definition and identification. However, important uses of language in education go wider than this, covering e.g. the relationship of terminology to conceptual distinctions. When listening or participating in a learning dialogue, we must be constantly monitoring our own orientation toward the words being used. Vicarious learning may actually allow us to do this more consciously than those involved in `the heat of the moment', having to generate the discussion. It is at the higher processes of meaning, function, conceptual integration that the overhearing of discourse may provide the means for grounding which would otherwise be missed.
How can we develop this wider picture of the importance of vicarious learning? We believe that such dialogues can provide a model for learners to analyse and emulate. Many of the best learning dialogues involve the making explicit of inferences which would normally remain unspoken. The central role of inference is further emphasised by our ``logic model" of dialogue (cf. McKendree at al. 1997), in which we note that educational dialogues, in particular, are commonly concerned with derivation rather than exposition. The student learns, not only about the subject matter, but also about the process of argumentation, which is to an important degree essentially explicit; it's a procedure that can be observed. The development of the CG being thus relatively out in the open, derivational dialogues are well suited to vicarious following of the process by overhearers.
There is much in educational dialogue that cannot be characterised simply as the transmission, or even the collaborative construction, of informational structures. If this were the case, we would hypothesise that the delivery of exposition would be more effective than dialogue in promoting learning. We believe that dialogue often gains from affective factors and from its association with a particular episode in a student's struggle to develop a new conceptual structure. In some sense, the latter might be seen as a new bit of common ground that the student has to come to share with the tutor; but to us it is more, because of its power to reorient the student's thinking and provide insights that are difficult to get otherwise. Though direct evidence of this is hard to acquire, we see plausible examples in the dialogues reported by Pilkington and Mallen (1996) and in recent investigations of our own (Cox et al., in preparation), including early evidence that students may sometimes learn more from watching dialogues in which poor students struggle through a problem than watching those where good students perform well (Monaghan and Stenning, in preparation).
The following example conveys a flavour of what we have in mind. We take, first, a snippet from a tutorial exposition (in which a tutor is explaining how to construct a syntax tree for a sentence):
...for the noun phrase, I want a rule that's going to result in a determiner and a noun. Every is the determiner and the noun is cat ...We compare this with a snippet from a dialogue in which a student is struggling to construct a parse tree, with assistance where needed from the tutor. Here, the student is speaking:
Every cat ...I would guess that's a ...determiner so we've got ...Every cat and now the bird ...ummm would be pretty much like the others were ...er ...let's see a bird would again be a ...determiner and a noun, let's see a ...bird
Among the differences here, we note that the student, unlike the tutor, indicates in a potentially student-friendly manner the similarity of the current example to earlier examples, and makes decisions explicit in ways that might be helpful to another student struggling with a similar example.
We are concerned with methods for allowing students better access to learning dialogues in situations where they might not otherwise have such opportunities, most obviously distance learners who have no contact with peers and little with tutors. We are interested in investigating the power of technology to perhaps augment current practice and allow vicarious learning opportunities for all students. This is where multimodality begins to emerge as important. This is a point in which we are particularly interested: the educational effectiveness of multiple representations, including dialogue.
It seems clear that the activity of external representation, in an individual's development of a concept or conceptual system, can stimulate various kinds of reorganisation of the system. It seems likely that there are two things that contribute to this; on the one hand, the revelation of ``emergent" aspects through restructuring of the representation; on the other, the very activity of creating an external representation itself perhaps driving a cognitive process that helps to develop a system of concepts. This latter possibility may be especially effective where the external representation is being generated for the benefit of another individual.
Evidence is accumulating for the power of re-representation, or the use of multiple representations, in learning and problem-solving (cf. Cox 1997), perhaps because the transition from one representation to another may make available interpretations which are emergent with respect to the first and more explicit in the second. It seems plausible that an important aspect of the value of collaboration and dialogue is that it forces externalisation and re-representation of ideas and conceptual schemes, intensifying the inherent value of doing this by exposing representations to multiple interpreters.
This argument presupposes the value of multiplying perspectives on a set of concepts. It emphasises a synthetic, breadth-first style over an analytic, depth-first approach, cf. the particular importance of dialogue in learning the kinds of synthetic skills -- design, diagnosis, etc. -- discussed e.g. by Donald Schön (1987). But these interlink; it is often after a period of analysis leading to a dead end that a change of representation is most effective.
Of course, the switch of modality has another value from our point of view: it is typically highly observable. We conjecture that multimodal dialogues will be some of the most effective in terms of promoting vicarious learning. At present we have little direct evidence for this idea, though what we do have clearly indicates that the ``cognitive style" of the learner, in interaction with the type of representation involved, may be a critical parameter. Where diagrams are used directly in teaching, some learners learn better than others, who may indeed learn better where diagrams are not used. Therefore, the particular modality in which dialogue participants elect to discuss a problem may affect the effectiveness of the dialogue for learners both direct and vicarious -- and potentially, we suppose, in opposite directions.
This research is supported by the UK ESRC Cognitive Engineering programme and UK EPSRC Multimedia and Networking Applications programme. HCRC is an Interdisciplinary Research Centre funded by ESRC. We are grateful to Terry Mayes, Keith Stenning, Richard Tobin and Padraic Monaghan for many helpful discussions.
Cox, R. (1996) Analytical reasoning with multiple external representations. PhD thesis, Department of Artifical Intelligence, University of Edinburgh.
Cox, R., Lee, J., Mayes, T., McKendree, J., Stenning, K. and Tobin, R. (in preparation) Learning from dialogues versus learning from discourse: a controlled comparison. Human Communication Research Centre, University of Edinburgh.
Garrod, S. and Doherty, G (1994). Conversation, coordination and convention: An empirical investigation of how groups establish linguistic conventions. Cognition, 53, 181-215.
McKendree, J., Stenning, K., Mayes, T., Lee J., and Cox, R. (1997). Why Observing a Dialogue may Benefit Learning: The Vicarious Learner. Proceedings of PEGpi97, and submitted to Journal of Computer Assisted Learning.
Monaghan, P. and Stenning, K. (in preparation). Teaching Euler's circles. Human Communication Research Centre, University of Edinburgh.
Pilkington, R. and Mallen, C. (1996). Dialogue Games to Support Reasoning and Reflection in Diagnostic Tasks, Proceedings of EuroAIED, Lisbon: Edies Colibri, pp. 213-219.
Schober, M.F. and Clark, H.H. (1989). Understanding by addressees and observers. Cognitive Psychology, 21, 211-232.
Schön, D. (1987). Educating the Reflective Practitioner: toward a new design for teaching and learning in the professions. Jossey-Bass.
This document was generated using the LaTeX2HTML translator Version 97.1 (release) (July 13th, 1997)
Copyright © 1993, 1994, 1995, 1996, 1997, Nikos Drakos, Computer Based Learning Unit, University of Leeds.
The command line arguments were:
latex2html -split 0 CMC98.tex.
The translation was initiated by John Lee (EdCaad) on 7/1/1998