CCNCS Seminar Details
Interactive Misalignment: Repair-driven Co-ordination in Dialogue.
|Date/Time:||Wednesday 11 March 2009, 4.15pm|
|Location:||SB110B, Computing Laboratory, Cornwallis|
Convergence --the tendency for people to align with each other's speech style, body movement and language use in conversation-- is widely seen as characteristic of successful human communication (see Giles, Coupland and Coupland, 1991 for a review). Recently, it has been proposed that convergence is the result of a basic priming mechanism that underpins all successful dialogue (Pickering and Garrod 2004, Garrod and Pickering 2006). I will argue against this view. Conceptually, the claim that priming is the central mechanism of co-ordination in dialogue is vulnerable to a reductio ad absurdum; it implies that conversations should become locked into cycles of verbatim repetition. More importantly, it also presupposes that conversational participants have the same lexical, syntactic and semantic repertoires and that co-ordination consists in selection between these shared alternatives. Radical nativism notwithstanding, it seems unlikely that any two people ever satisfy this ideal. Priming also cannot account for innovations in language use. Using data from corpus analyses I will show that a) repetition in dialogue is in fact rare and that b) repair sequences (in which problems with mutual-intelligibility are directly encountered and addressed) are common. These observations are used to motivate the claim that misalignment is the norm and that it is the processes for dealing with this that are the central mechanisms of dialogue co-ordination. This claim is tested by two chat-tool experiments which selectively interfere with people's ability to engage in repair. The results show that this manipulation has a marked effect on people's ability to co-ordinate their language use. This leads to a view of dialogue as a contingent, repair-driven, co-ordination process in which languages are continuously adapted to the evolving needs of participants.