This paper calls for an analysis of the notion of “shared knowledge” through its role in inference. First of all, selection restrictions have to be excluded from the domain of shared knowledge because they are not inferred; rather, they are the ground of every possible inference. Secondly, in the field of shared knowledge, real cognitive models must be kept apart from idiomatic prejudices: the former produce inferences which can be defined as “free”, while the latter produce inferences which can be defined as “frozen” or “fixed”. This separation raises a problem on the bounds of the notion of stereotype.
Français (France)
English