Friday, March 6 in Silver 509. Reception to follow in the linguistics department.
Title: The projection problem for pragmatic inferences
Abstract
I will present a system which accounts for the behavior under embedding of a variety of pragmatic inferences: scalar implicatures (1), free choice inferences (2) and presuppositions (3).
(1) John read most of the books. => He didn't read them all.
(2) You may have an apple or a banana. => 'You' may choose which fruit 'you' have.
(3) Mary doesn't know that it's raining. => It's raining.
I will focus on a potential problem for the resulting system: presuppositions of quantified sentences. At first sight, the predictions of the present system are at odds with other dominant positions taken on that topic: e.g., presuppositions are predicted to project neither existentially à la Beaver (5a) nor universally à la Heim (5b) but somewhat intermediately (5c):
(5) Few of these students know that they are lucky.
a. Existential presupposition: at least one of them is lucky.
b. Universal presupposition: all of them are lucky.
c. Present prediction: most of them are lucky.
I will present experimental evidence which support the predictions of the present system.
On a more conceptual side, I will discuss the consequences of the very existence of this type of unified system for the architecture of a theory of pragmatic inferences and discuss in particular the following question: when does common ground come into play?