Predicting polar question embedding

This paper shows that it is fully predictable whether a polar interrogative clause can appear under a declarative embedding predicate. The idea in a nutshell is as follows: whenever a predicate appears to not embed polar interrogatives, the interpretative component independently derives a trivial meaning for the sentence. Such trivial meanings manifest themselves in unacceptability. A crucial property of this proposal is that interrogative embedding is polaritysensitive, which is shown to be empirically supported. As a consequence, one must not stipulate in the lexical entry of a given predicate whether or not it embeds polar interrogative clauses.