Preliminaries to a collaborative model of dialogue

Many writers have argued that dialogue should be regarded as a joint activity (see for example (Clark and Wilkes-Gibbs, 1986; Grosz and Sidner, 1990; Schegloff, 1981; Suchman, 1987)), something that agents do together, rather than simply as a product of the interaction of plan generators and recognizers working in synchrony and harmony, as plan-based theories propose. Such plan-based approaches do not explain why addresses ask clarification questions, why they confirm, or even, why they do not walk away. Rather, the joint action model claims that both parties to a dialogue are responsible for sustaining it. Participating in a dialogue requires the conversants to have at least a joint commitment to understand one another. The key questions to be answered include how to formalize such general commitments precisely, and to show how they predict the fine-grained synchrony so apparent in ordinary conversation. To begin to answer these questions, we sketch here how a formal theory of joint action explains confirmations that arise in task-oriented telephone dialogues. A more formal account is given in (Cohen and Levesque, 1991a). Then we argue that extensions of this analysis to dialogue more generally will be difficult. In particular, it will force us to give up our simplistic analyses of propositional content and literal meaning.