The Common Code: Diagnosis, a treatment and its prognosis

It is commonly, often tacitly, assumed that the mutual-intelligibility of natural language dialogue is underwritten by a common semantic code. Ultimately, this assumption is both emprically and conceptually unsustainable and adequate models of dialogue need to provide for ways in which semantic con icts can be addressed. One possiblity for resolving such con icts is by the use of some form of matching strategy to accomodate situations in which interlocuters do not share the same semantic model. The most clearly articulated proposal along these lines is the Input-Output coordination model. This paper examines this model and, on the basis of data obtained from a taskoriented dialogue, argues that it does not provide a su cient mechanism for the resolution of semantic con icts.