Implicatures and the recognition of speaker intentions in a game theoretic model

In this paper we study the role of the recognition of the speaker’s intentions in utterance interpretation. We are interested in the question whether the recognition has to be explicit and whether it precedes or follows interpretation. We present a game theoretic model of pragmatic interpretation and study it in the context of iterated best response models as these models make the interlocutor’s reasoning about each other explicit. We claim that for normal communication backward induction is sufficient for interpretation, and hence that the recognition of the speaker’s intentions follows rather than precedes interpretation. Relevance implicatures are generally considered to be part of the communicated meaning. By definition, it seems that the hearer has to take the speaker into account when calculating these implicatures. We will therefore pay special attention to them.