Locality of wh-movement and the individuation of events

I propose a novel characterisation of the counterexamples to the generalisation that wh-movement out of an adjunct is impossible. The thesis discusses three classes of exceptions, namely extraction from in order clauses (la), prepositional participial adjuncts (lb), and bare present participial adjuncts (lc). (1) a. What are you working so hard in order to achieve? / b. Who did John go home after talking to? f c. What did John drive Mary crazy whistling? r These contrast with cases where extraction is impossible (2). (2) a. * What did you get upset because Mary said? r b. * Who have you been really happy since talking to? / c. * What does John work whistling? / These two groups do not form natural syntactic classes, but are distinguishable in event-structural terms. The minimal constituent containing the head and foot of the chain in (1) describes a single event, on an appropriate definition of event, but does not in (2). Accordingly, I propose the following condition: (3) W/z-questions carry a presupposition that the minimal constituent containing the head and the foot of the chain describes a single event. Wz-movement is permitted only if the denotation of that minimal constituent can be construed accordingly. Chapter 2 of the thesis develops a formal and cognitively well-motivated model of the internal structure of events. Key to this model is the recursively defined notion of extended events, corresponding to plan formation. This recursion crucially allows (3) to capture long-distance A'-dependencies. Chapter 3 applies this model, assuming (3), to locality data, deriving the contrast be tween (1) and (2), as well as other data such as the absolute prohibition on extraction from tensed adjuncts (2a), and the distinction between bridge verbs and factive islands (4). (4) a. What did John think that Mary did? r b. * What did John regret that Mary did? t l.