Repository | Book | Chapter

This chapter argues that the occurrence of jointly constructed utterances (split utterances) in conversation has wide implications for current linguistic theories. Firstly, we show that standard formal syntactic and semantic/pragmatic theories are unable to cope with such conversational data due to the widely assumed competence/performance distinction. We then present some recent developments in the domain of formal models of dialogue in order to assess whether they meet the design features that a general analysis of dialogue, and of the split-utterance phenomenon in particular demand. We argue that what is crucial for such an account is incorporating both the physical and social situatedness of language use, combined with modelling the incrementality of linguistic processing, within the grammar formalism employed. In previous work, we have argued that the grammatical framework of Dynamic Syntax (DS) augmented with the flexible semantic representations made available by Type Theory with Records (TTR) meets these requirements. Accordingly, through the phenomenon of split utterances, we illustrate how the grammar itself needs to be seen as a holistic, action-based model of language use incorporating incremental interaction with context and flexible mechanisms of processing. These requirements are needed in order to deal not only with what have been traditionally thought of as indexicals but also with the representation of fine-grained sub-sentential utterance events, speech-act information, roles assigned to participants, etc. This stepwise interaction is necessary for a general account of how a speaker-change in mid-utterance affects the form and interpretation of linguistic elements. As a result, the incremental stance allows a natural characterisation of split utterances as continuations/interruptions, whereas, without it, the only recourse is the assumption of widespread ellipsis, mind-reading and multiple ambiguity of sub-sentential fragments. We then take a wider view of the data characterised as the Turn Taking Puzzle (Ginzburg 2012) by combining the phenomenon of split utterances with an account of the function of why? fragments ((Split-)Turn-Taking Puzzle, STTP). On the basis of the STTP data, we argue that it is crucial for syntactic specifications and interpretation to interact with the modelling of the sub-sentential dynamics of the discourse-situation updates. From these interactions, we draw conclusions as to the significance of the STTP data for the design of grammar formalisms and dialogue models, as well as for the general conception of linguistic knowledge.

DOI: 10.1007/978-3-319-12616-6_28

Full citation: