Grammar Formalisms and Explanations of Dialogue

Contrary to what for years has been a widespread belief that conversational data are riddled with mistakes, false starts, and simply ungrammatical data, Pickering and Garrod (2004) extended a challenge primarily to psycholinguists, but also to theoretical linguists that their models should be evaluatable by how well they provide a basis for characterizing the patterns systematically displayed in conversational dialogue, this being the core language data. In this paper we argue that by adopting a Dynamic Syntax perspective, with the time-linear dynamics of building up structured semantic representations relative to context built in to the architecture of the grammar formalism, these patterns are directly predicted, while still retaining the assumption of grammar as a system which underpins language use but does not fully dictate it, i.e. without writing conversational rules or a grammar of conversation.