In this study, I propose and defend a new theory of speech acts predicated on the assumption that speech act theory, if it is to be of genuine empirical and theoretical significance, must be embedded within a general theory of conversational competence capable of accounting for how we do things with words in naturally occurring conversation. The theory I shall propose, Dynamic Speech Act Theory (DSAT), can usefully be seen as a synthesis of traditional speech act theory (cf., especially, Searle 1969, 1975, 1979), conversation analysis (cf., especially, the references to Schegloff, Sacks, and Levinson in the bibliography), and artificial intelligence research in natural language processing (cf., especially, the citations in the bibliography to the work of Allen, Cohen, Litman, and Perrault, also Patten, Geis, and Becker (1992)). Important additional influences are the work of Brown and Levinson (1987) on politeness and Halliday and Hasan (1989) on register.

I shall take the position here that the goal of a theory of conversational competence should be specification of the properties of devices capable of engaging in conversational interactions – devices that we might call “conversation machines.” As such, the theory would be a theory of the conversational competence that underlies our ability to engage in goal-achievement and goal-recognition in conversation and our ability to produce and understand utterances (and nonverbal behaviors) appropriate to the context. I shall argue in these pages that correctly conceived, speech-act-theoretic structures will play a critical role in accounting for these abilities.