Going Up a Blind Alley: Conflating Conversation Analysis and Computational Modelling

Publisher Summary This chapter discusses the desirability of developing computational models of conversational phenomena, and the supportive role given to conversation analysis (CA) in the development of such models. The arguments presented in this chapter are not an attempt to restrict the range of creative resources that software designers might turn to for inspiration. In particular, it is implicitly endorsed in the attempts to develop descriptively adequate models of conversation for use in computer systems, and explicitly endorsed when it is argued that by providing a simulacrum of conversation one has naturally occurring conversation between computers and humans. The attraction of CA for people who want to develop rules of conversational organization that can be used to program computers is two-fold: (1) CA might seem to provide a ready-made package of conversational rules that they can use or adapt for their purposes; and (2) their models may be authorized by appealing to CA. However, CA is used to authorize computational models of conversation that misrepresent the details of how conversation works.