Young children find (some) object relatives much harder to understand than subject relatives. The main finding of this article is that not all object relatives are difficult. The difficulty with object relatives (and object Wh-questions) is selective: it depends on the structural similarity between the A'-moved element and the intervening subject. We interpret this selective effect in terms of a proper extension of Relativized Minimality, the principle of syntactic theory which expresses locality effects linked to intervention, and whose psycholinguistic relevance has been highlighted in Grillo's work on agrammatism. Six experiments have been conducted with 22 Hebrew-speaking children aged 3;7-5;0 to substantiate our claims empirically. Experiments 1 and 2 tested the comprehension of headed subject and object relatives with and without a resumptive pronoun, in sentences with lexically restricted (D NP) subjects. Subject relatives were comprehended well, but the performance on object relatives was at chance. The addition of resumptive pronouns did not improve comprehension. Experiments 3 and 4 manipulated the lexical restriction of the moved element and the intervening subject, using free relatives and impersonal pro subjects, respectively. When the moved constituent and the intervening subject were structurally dissimilar in terms of lexical-NP restriction, the performance significantly improved. Experiment 5 showed a similar comprehension pattern in another type of A' movement: Wh-questions. The comprehension of who and which subject and object questions was difficult only when both the crossing element and the intervener included a lexical NP restriction. Finally, a similar pattern emerged in production: in an elicited production study, children showed a tendency to avoid producing structures in which both the moved element and the intervener are lexically restricted. We conclude suggesting a line of analysis of the difference between children and adults in dealing with object A'-dependencies.
- Relative clauses
- Relativized minimality