Retour vers liste

Détail de la contribution

Auteur: Shevaun LEWIS

Co-Auteur(s): Dave KUSH, University of Maryland, United States Bradley LARSON, University of Maryland, United States

No filled gap effects in coordinated wh-questions

Abstract/Résumé: In previous work we have argued on empirical and theoretical grounds that the first wh-word in a coordinated-wh question like (1) enters into a purely semantic variable-binding dependency, which has no instantiation at the syntactic level. Here, we present psycholinguistic evidence that this dependency is also processed differently from standard wh-dependencies. Specifically, in a self-paced reading study, we did not observe a cost for filled gaps in sentences like (2)—an apparent illusion of grammaticality. (1) What and when did Ivy eat? (2) *The man wondered what and when Georgina would eat something. We presented participants with declarative sentences containing embedded CWh questions and either optionally- (3) or obligatorily-transitive (4) verbs. We manipulated Wh Type (‘what’/‘when’/‘what and when’), as well as whether the gap was filled with an overt argument, the word ‘something’. Reaction times (RTs) were analyzed for the 3 words following the verb. (3) The diplomat had to make a schedule of {what/when/what and when} his lazy assistant would translate (something)... (4) The busy executive was especially worried about {what/when/what and when} his lazy assistant would overlook (something)… With both verb types, we observed a filled gap effect in ‘what’ sentences: RTs in all post-verb regions were significantly longer than in ‘when’ sentences. With optionally-transitive verbs (3), RTs for ‘what and when’ sentences patterned with ‘when’ sentences in all post-verb regions: there was no filled gap effect. Under our account, this is due to the lack of a movement dependency between ‘what’ and the verb. With obligatorily-transitive verbs (4), RTs in the first post-verb region in ‘what and when’ sentences were significantly longer than in ‘when’ sentences. In subsequent regions RTs patterned with ‘when’ sentences. The initial slow-down may reflect the unacceptability of the obligatorily-transitive verb. The filled gap does not register as an additional violation, since there is no movement dependency between ‘what’ and the verb. The lack of filled gap effects supports our hypothesis that the non-final conjuncts in coordinated wh-questions participate in non-syntactic wh-dependencies.