Covert Movement in Multiple-Wh Questions: Experimental and Theoretical Investigations

Results from a new grammaticality-judgment experiment in French confirm the published finding in English that sentences containing a Superiority violation involving a bare extracted element and a lexically restricted intervener (e.g., ‘What did which student buy?’), a configuration termed inverse inclusion, are more acceptable than those involving a lexically restricted extracted element and a bare intervener (e.g., ‘Which book did who buy’), a configuration termed inclusion. To account for this pattern, we adopt an explicit implementation of covert movement and propose some modifications in the characterization of the class of interveners. Interestingly, experimental findings on extraction from wh islands attest the opposite pattern: there, inclusion is more acceptable than inverse inclusion. We argue that whereas (overt) extraction from wh islands is sensitive to the feature content of the extractee and the intervener (i.e., whether or not they are lexically restricted), the degree of (un)acceptability of Superiority violations hinges on the different landing-site options that the features of the extractee and the intervener permit.