We suggest that topics and replies are highly constrained during casual conversation. Topics must be problematic, and replies have to strengthen or to reduce this problematic aspect. However, things may be slightly more complex as far as replies are concerned, mainly because replies call for further replies. In fact, there is little to be added : reinforcements, banalizations, invalidations and antagonistic reactions may follow one another, as we can verify on the excerpts given in this paper. For instance in [ex_doors p.13], the antagonistic reaction A4 follows the invalidation B3. But there are several other, logically minor, possibilities to give relevant replies, and we go through rapidly now.
We did not mention agreements. We had some in the same excerpt [ex_doors] (B4 and B6). However, the importance and frequency of agreements, at the logical level, should not be overestimated. For instance from a logical point of view, replies like B1 or B2 in [ex_thirst p.18] are not agreements but banalizations, as we saw. We should also mention replies that make the complex explicit, or ask for an explicitation. We had such utterances after every breakdown we gave as example, but also in [ex_Goffman p.6] A6, or below reply B3. Lastly, they may be more complex patterns, when interlocutors introduce " meta-arguments ". Consider the following excerpt.
[ex_ping-pong]
context: A is amazed at the strange technique of a Chinese student who is playing table tennis.
A1- T'as vu comment il tient sa raquette ?
B1- C'est un Chinois, les Chinois ils tiennent tous leur raquette comme a
A2- C'est pas naturel ! Comment ils arrivent à jouer ?
B2- Oh ! C'est pas un sport a !
A3- Enfin, tu transpires !
B3- Un sport o tu cours pas, c'est pas un sport.
[É]
A5- Mais comment ils font pour smasher, les Chinois ?
A1- Did you see how he holds his paddle?
B1- He is Chinese. They all do the same
A2- It isn't natural! How do they manage to play?
B2- Oh ! you can't call it a sport
A3- Well, you do sweat
B3- If you don't run, it isn't a sport
[É]
A5- But how do the Chinese succeed in smashing?
Here B2 is a meta-argument. First A1 is intended as a paradox (as A2 and again A5 clearly show). But B take it as an improbable event, and banalizes it with B1, which cannot of course make A satisfied. With B2, something else happens : if table tennis is not a sport, then the previous topic about the holding of the paddle is no longer interesting. The ensuing discussion is a meta-conversation only because the interestingness of the initial topic is at stake. To all other aspects, it is a normal conversation (notice the invalidation A3, and the effect of B3 which is to make a clause of the logical context explicit). Only in A5 do interlocutors come back to the initial topic.
The analysis of this excerpt illustrates how our model can be used to understand the internal structure of conversations. Not in terms of acts or intentions, but in terms of logic. Usual theories of discourse or of conversation pay more attention to the interlocutors themselves and to their intentions than to the content of their utterances. This is perfectly legitimate, and allows to make interesting predictions (e.g. conditions of satisfaction of verbal acts, typology of negotiation, etc.). We tried to show however that some constraints may exist between arguments, regardless of who uttered them[7]. This point of view gives a very different perspective. For instance, when B in [ex_doors p.13] suggests to use a wire brush, his intention is certainly to be helpful. However, the effect of his argument is to invalidate A's reasoning. If A uses a wire brush, then the inference from burning off the old paint to repainting is tough work is no longer valid. Cooperation, intention, conversational goal, all these concepts commonly used in the study of conversation have no meaning at the level of the logical organization of arguments. Arguments have " their own life ", which obeys strong constraints. Interlocutors are for the major part unaware of the logical constraints they put on the other's arguments or on their own. We believe that the study of their intentions and of their discourse strategies should take advantage of a previous analysis of the argument logical structure the way we propose here.
The model presented in this text has given rise to some technical developments. We are currently developing a computer program that tries to generate arguments that have been uttered in real conversations, using a knowledge base containing knowledge about the subject [Dessalles 1990]. We also designed a Computer Assisted Learning program that sustains conversations in the paradoxical mode with students [Dessalles 1991]. We believe that besides its potential value as theoretical reference for technical applications in human-machine verbal interaction, this model of the logical structure of conversation may generate several research programs, that we already begin to explore. More work is needed to delimit the scope and validity of the model, to establish quantitative experimental evidence, to study developmental aspects (how logical constraints are learnt, if they are learnt at all), cultural variability, disorders[8], etc.
We are by nature reluctant to accept that our freedom is not as wide as what we believed, especially during free, spontaneous conversation. Anyone making a close examination of the content of what we say in everyday interactions must however admit that the rules of the conversational game are very strict, and that they often dictate our next move, depending on what we know of the subject. There are often not many ways to be logically relevant. Such constraints on our verbal behavior, like other constraints observed at other levels (linguistics or social), can no longer be ignored.
Acknowledgments: I am grateful to Lamia Choukair, Alain Grumbach, Georges Sallé and an anonymous reader for having read early versions of this paper, and for their valuable criticisms.
References
Bateson Gregory, Jackson Don D., Haley, Weakland (1956). Toward a Theory of Schizophrenia. Behavioral Science, vol. I, N[exclamdown] 4, 1956
Bunt H.C. (1989). Information Dialogues as Communicative Action in Relation to Partner Modelling and Information Processing. In Taylor M.M., Néel F., Bouwhuis D.G. , The Structure of Multimodal Dialogue, North Holland, Amsterdam 1989, pp. 47-73
Clark Herbert H., Schaeffer Edward F. (1989). Contributing to Discourse. Cognitive Science 13, 1989, pp. 259-294
Clark Herbert H. (1993). Managing problems in speaking. In Shirai K., Kobayashi T., Harada Y., International Symposium on spoken dialogue, , Tokyo 1993, pp. 10-12
Coulthard Malcom (1977). An Introduction to Discourse Analysis. Longman, London 1985
Dessalles Jean-Louis (1990). The simulation of conversations. In Kohonen Teuvo, Fogelman-Soulié Franoise, COGNITIVA 90 - Proceedings of the Third Cognitiva Symposium (Madrid), North Holland, Amsterdam 1991, pp. 483-492
Dessalles Jean-Louis (1991). Conversation Assisted Learning: The SAVANT3 Dialog Module. In Forte Eddy N., Proceedings of Calisce'91, Presses Polytechniques et Universitaires Romandes, Lausanne 1991, pp. 159-165
Dessalles jean louis (1992a). L'incidence logique de l'interaction dans la communication d'informations. Technologies Idéologies Pratiques Vol. X n[exclamdown] 2-4, Aix-en-Provence 1992
Dessalles jean louis (1992b). Model-Based Surprise and Explanation: a way to negotiate concepts. In Brezillon Patrick, Proceedings of the ECAI-92 Workshop on Improving the Use of KBS with explanations, Rapp. LAFORIA 92/21 Univ. Paris VI, Paris 1992, pp. 107-113
Dessalles jean louis (1993). Modèle cognitif de la communication spontanée, appliqué à l'apprentissage des concepts. Thèse de doctorat, TELECOM-Paris 93E022, Paris 1993
Drew Paul, Holt Elisabeth (1990). Complainable Matters: the use of idiomatic expressions in making complaints. In Conein Bernard, De Fornel Michel, Quéré L., Les formes de la conversation, CNET, Paris 1990, pp. 108-143 (T.1)
Falzon Pierre (1991). Cooperative Dialogues. in Distributed Decision Making: Cognitive Models for Cooperative Work,J.Rasmussen & al. ed., J.Wiley, London 1991, pp. 145-189
Fox Barbara A. (1987). Interactional Reconstruction in Real-Time Language Processing. Cognitive Science 11, 1987, pp. 365-387
Goodwin C., Heritage J. (1990). Conversation analysis. Annual Review of Anthropology 19, 1990, pp. 283-307
Grice H.Paul (1975). Logic and Conversation. In Cole Peter, Morgan Jerry L., Syntax and Semantics volume 3: Speech Acts, Academic Press, New York 1975, pp. 41-58
Grosz Barbara J., Pollack Martha E., Sidner Candace (1989). Discourse. In Posner M.I., Foundations of Cognitive Science, Bradford Book, the MIT Press, Londres 1989, pp. 437-468
Heritage John (1990). Interactional Accountability: a Conversation Analytic Perspective. In Conein Bernard, De Fornel Michel, Quéré L., Les formes de la conversation, CNET, Paris 1990, pp. 23-49 (T1)
McTear Michael F. (1985). Children's Conversation. Basil Blackwell, Oxford 1985
Moeschler Jacques (1990). Conversation: cohérence et pertinence. In Conein Bernard, De Fornel Michel, Quéré L., Les formes de la conversation, CNET, Paris 1990, pp. 79-104
Norman M.A., Thomas P.J. (1991). Informing HCI design through conversation analysis. International Journal of Man-Machine Studies 35, 1991, pp. 235-250
Reichman Rachel (1985). Getting Computers to talk like you and me. MIT Press, Harvard 1985
Reichman Rachel (1989). Convention versus Intention. In Taylor M.M., Néel F., Bouwhuis D.G. , The Structure of Multimodal Dialogue, North Holland, Amsterdam 1989, pp. 121-134
Schank Roger C., Lehnert Wendy (1979). The Conceptual Content of Conversation. International Joint Conference on Artificial Intelligence, Tokyo 1979, pp. 769-771
Scholtens Anneke (1991). Planning in ordinary conversation. Journal of Pragmatics 16, 1991, pp. 31-58
Searle John R. (1969). Les actes de langage - Essai de philosophie du langage. Hermann (Speech Acts, Cambridge University Press), Paris 1972
Shannon Claude E. (1948). Mathematical Theory of Communication. Bell Systems Technical Journal 27, 1948, pp. 379-423,623-656
Sperber Dan, Wilson Deirdre (1986). La pertinence. Les Editions de Minuit, Paris 1989
Stubbs Michael (1983). Discourse Analysis. Basil Blackwell, Oxford 1983
Tannen Deborah (1984). Conversational Style - Analyzing Talk Among Friends. Ablex Publishing Corporation, Norwood 1984
Taylor Jasper A., Carletta Jean C. (1994). Limited nested beliefs in cooperative dialogue. In Ram Ashwin, Eiselt Kurt, Proceedings of the Sixteenth Annual Conference of the Cognitive Science Society, Lawrence Erlbaum Associates, Hillsdale, N.J. 1994, pp. 858-863
Trognon Alain, Brassac Christian (1988). Actes de langage et conversation. In Intellectica n1/46, A.R.C., Paris 1988, pp. 211-232
Vuchinich Samuel (1980). Logical Relations and Comprehension in Conversation. Journal of Psycholinguistic Research, Vol.9, N[exclamdown].5, 1980
Watzlawick P., Helmick Beavin J., Jackson D.D. (1967). Une logique de la communication. Seuil, Paris 1972
Young R. Michael, Moore Johanna D., Pollack M.E. (1994). Towards a principled representation of discourse plans. In Ram Ashwin, Eiselt Kurt, Proceedings of the Sixteenth Annual Conference of the Cognitive Science Society, Lawrence Erlbaum Associates, Hillsdale, N.J. 1994, pp. 946-951
[7] We can even observe people replying to themselves.
[8] Some conversational disorders have been shown to be related to logic [Watzlawick & al. 1977], [Bateson & al. 1956].