The distributive ignorance puzzle

Abstract

We observe that verbs like wonder do not just imply that their subject does not know the answer to the embedded question, but a stronger form of ignorance, which we call distributive ignorance. This is not predicted by existing work on the semantics of wonder, and we argue that it cannot be straightforwardly derived as a pragmatic inference either. We consider two possible semantic accounts, and conclude in favor of one on which the lexical semantics of wonder involves exhaustification w.r.t. structural alternatives as well as sub-domain alternatives of its complement.