Both the collaborative and interactive alignment models of communication treat communicative feedback as fundamental to co-ordination in dialogue. However, they differ in their treatment of negative feedback: clarification requests such as "What?", "Where?", or a partial repeat of the problematic utterance vary in strength at locating the source of the problem in the preceding dialogue, and are incorporated by the grounding model to index understanding within a four-level framework. By contrast, the interactive alignment model associates this locatability with the extent to which interactants take each other's perspective into account. This talk presents a "Maze-game" experiment that introduces artificially generated clarification requests into participants' dialogue in order to investigate the effects of locatability on the development of semantic coordination. The results indicate that both models do not fully explain the patterns of co-ordination observed for different kinds of clarification request, suggesting the importance of an account that is sensitive to semantic differences between different forms of co-ordination.

Return to Conference Programme