Language and logic: the cases of negative polarity items and scalar implicatures

Jury :

David Barner (University of California San Diego, rapporteur)
Gennaro Chierchia (Harvard University, rapporteur)
Angelika Kratzer (University of Massachusetts, Amherst)
Lyn Tieu (Western Sydney University)
Emmanuel Chemla (Ecole normale supérieure, directeur de thèse)
Benjamin Spector (Ecole normale supérieure, directeur de thèse)

Numerous linguistic phenomena have been shown to correlate with some logical properties of the sentence with which they occur. Examples of such properties are logical entailments supported by the sentence, or the logical relation between it and some alternative sentence. In this thesis, we explore (A) whether these logical correlates play a causal role in the linguistic phenomena in question, and (B) at what level these computations of logical correlates are performed. There are two broad possibilities concerning (B): (i) these logical correlates could be computed in a formal system that does not have access to contextual knowledge, call this system grammar (Fox and Hackl 2006, a.o.), or (ii) they could be computed post-grammatically, and therefore have access to contextual knowledge. We investigate these questions in relation to the licensing of negative polarity items (NPIs) and the derivation of scalar implicatures. Our findings add to a growing body of evidence that logical correlates of NPI licensing and scalar implicatures play a causal role in these phenomena. Furthermore, our results suggest that logical correlates to NPI licensing are calculated post-grammatically, while logical correlates to scalar implicatures are calculated grammar-internally.