Review Article: Does Conversation Analysis Have a Role in Computational Linguistics

Computational linguists are a vicious warrior tribe. Unconstrained by traditional disciplinary boundaries, they invade and plunder neighboring disciplines for attractive theories to incorporate into their own. The latest victim of these attacks is sociology-in particular, a branch of ethnomethodological sociolinguistics called Conversation Analysis (hereafter, CA). The book reviewed here, Computersand Conversation, is the first significant result of this. It is based on a symposium held at the University of Surrey in September 1989. Its purpose is to show what CA has to offer to the study of human-computer interaction (hereafter, HCI), especially interaction in natural language. In this review article, I want to use the book to explore at some length the question of whether CA has indeed anything to say to computational linguistics. Up to now, almost all research in computational linguistics on dialogs between humans and machines has taken the approach of Discourse Analysis. Discourse Analysis concerns itself with talk at the level of linguistic constituents (Levinson 1983, pp. 286ff; Stubbs 1983, pp. 1, 9-10). The disfluencies of spoken language are ignored. Utterances are treated as meaningful units of language, and are analyzed with regard to their semantic content. So the human and the machine are both seen as rational agents whose utterances are always perfectly formed and fully in accordance with Gricean rules. These interlocutors like to spend their time seeking information from each other, testing each other's abilities in plan inference. Typical interchanges are the following: