Communicative Action as Bargaining: Utility, Relevance, Elementary Social Acts 1. Some Problems of Communicative Action Theories

I outline an approach to communicative action (CA). Its bases are (i) a bargaining game approach to Elementary Social Acts (ESAs) and (ii) a Decision-Theoretic Semantics (DTS) in which utility and, as a special case, evidential relevance play a central role. The act analysis is extended to a variety of act types. Examples of various kinds of`conventional' and`conversational' impli-cature illustrate how act-and relevance-structure feed into one another. More speciic issues addressed include: implications of modelling communication as a coordination game, empirical problems of act-imputation by discourse participants , and mining utterance structure for cues to ostensible utterer's intention. I should like to submit for evaluation an approach to some of the phenomena CA theories habitually deal us. I believe the approach makes inroads on well-known problems in the theory of meaning for natural languages , and I shall illustrate some, giving references to other cases studies. 1 I shall also attempt to engage some points of discussion raised by symposium con-venors, some with reference to responses in discussion drafts. The exposition aims to provide tools for addressing two speciic problems raised explicitly or implicitly by participants to the symposium: People can be shown to vary widely in assignment of speech act categories to utterances (Di Eugenio et al 1997). Many working systems analyzing and synthesizing aspects of CA very consciously opt for a `template' approach to utterance meaning (e.g. Thomason and Hobbs 1997). Problems akin to those of the rst kind have prompted many to dismiss speech act type as a piece of super-1 Several technical reports are available online at uous baggage (Gazdar 1981). Similarly, the inadequacy of traditional speech act theory to deal with intonation (Cutler 1977) may warrant suspicions that its categorizations were inadequate not only to this purpose (Merin and Bartels 1997; cp. Pierrehumbert and Hirschberg 1990) but also to others. On the other hand, lives do occasionally hang in the balance of explicit assignment of traditional speech act categories. 2 This problem is related to the one singled out by the second problem. Explicit performatives are rare birds. Even when they do occur, there is often no guarantee that what they appear to commit the speaker to will not be subverted by telltale clues in surrounding parts of his discourse. So we should ask: to what extent can a semantics { not necessarily truthconditional { of parts of the closed-class lexicon help to get a …