Speech acts and pragmatics in sentence generation

Abstract

A fundamental advance in recent theories about natural language pragmatics involves the realization that people use language not just to describe propositions, but also to perform actions. This idea can be taken as a given starting point for investigating the questions: how are linguistic actions, or speech acts, performed and understood? How far can descriptions, as locutions, be used as speech acts What role does inference play in the performance and understanding of speech acts?

Many previous theories of speech acts have taken speech acts to be independent and primitive units of communication, implicit in, but separate from, description and inference. In this thesis, I argue for an alternative model of speech acts. I propose that speech acts can be explained by a combination of description and inference, without the requirement of separate conventions. This explanation relies instead on an account of explicit linguistic units, including clausal moods and performative verbs, in addition to the inferential mechanism provided by Gricean conversational implicature.

In addition to outlining a model for the description and comparison of speech acts, I present a small sentence generator that partially implements the model, and discuss how it can be enhanced in future. Also, I illustrate the relevance of this model for a current computational theory of syntactic style. My model of speech acts suggests how this computational stylistic theory can be extended into the areas of semantic style and lexical style.