In this paper we sketch the general lines of a model of an elementary
interchange in a dialogue, and concentrate on the analysis of the dial
ogue failures and exploitations the model can account for, as well as
on deceitful uses of communication and non-expressive uses of language
. We show that our model of standard communication can explain non-sta
ndard uses of language without special adaptation. In this way we can
account for a variety of linguistic phenomena which are often neglecte
d in formal treatments of dialogue, and provide a first validation of
the knowledge structures and inference processes assumed in the model.