Meetings this semester are:
Fridays from 2:10-3:20 (by default) or Thursdays from 9:30-11 (when the default time does not work) in 32-D831 Please see schedule for details.
Andreea Nicolae (Harvard)
An alternative account of the distribution of NPIs in interrogatives
Despite much research on the NPI front, their behavior in questions remains puzzling. Many recent theories, building on Ladusaw’s original insight, have developed an alternative–based approach whereby their distribution, a requirement to be in a downward entailing (DE) context, follows from the way their alternatives are “exhaustified,” without having to stipulate a licensing-by-DE condition. In fact, Guerzoni & Sharvit (2007) show that when it comes to the distribution of NPIs in questions, DE-ness cannot be a factor, and claim instead that the crucial factor is strength. While NPIs are always acceptable in direct questions (modulo some intervention facts), matters are more complicated in embedded questions: wonder verbs always allow NPIs, surprise verbs never allow NPIs, and know verbs have an intermediate status.
(1) a. Mary wonders which students brought anything to the party.
b. %Mary knows which students brought anything to the party.
c. *It surprised Mary which students brought anything to the party.
Noting that the NPI’s acceptability correlates with whether the embedded question is interpreted as weakly (WE) or strongly (SE) exhaustive — wonder embeds SE questions, surprise embeds only WE questions, while know arguably admits both — G&S draw the generalizations that NPIs are only admissible in embedded questions that receive a SE interpretation. Summing up, the situation is the following. We have a promising theory of NPIs, a good generalization about their distribution in questions, and a theory of WE versus SE questions, but we don’t know how these come together, and in particular how the generalization may follow given what we know about the distribution of NPIs in non–interrogative contexts. In this talk I am going to propose a new way to derive the WE/SE distinction in embedded questions and show how this new approach allows us to tackle the puzzle of NPIs in questions.
The emerging picture is that not only is it the case that DE–ness is not a factor in non–interrogatives and strength not a factor in interrogatives, but that in fact all occurrences of NPIs can be accounted for in an arguably elegant way by simply looking at the interaction between their alternatives and how the grammar uses up these alternatives across different environments. I will also suggest ways to analyze the intervention facts observed with NPIs in questions.