Retour vers liste

Détail de la contribution

Auteur: Anamaria BENTEA

Co-Auteur(s): Stephanie DURRLEMAN, Université de Genève, Suisse

Atoms of intervention in the acquisition of A'-dependencies

Abstract/Résumé: Research on the acquisition of A'-dependencies has shown that not all movement configurations are problematic for children. Grammatical/intervention accounts explain such selective difficulties in terms of sensitivity to the structural similarity between the moved object and the intervening lexical subject. One question remains: what are the exact atoms of intervention? Here we address this by investigating French-speaking children's comprehension of object wh-questions and relative clauses (RCs) and by focusing on the nature of the moved object: +NP vs -NP; +Animate vs -Animate. 61 children aged 5 to 8 years old were prompted, in a character-selection task, to choose the correct character as identified by a wh-question (1-2) or RC (3-4). Each test sentence was associated with two pictures depicting the same action with reversed Agent-Patient roles. The results show that the crucial factor for comprehension of A'-dependencies is the presence of similar features on the A'-moved object and the intervening subject: children perform worst with structures in which the +NP+Animate features on the intervener are included in the set of features on the A'-object (F(1,61)=7.97,p=.006). That intervention effects do not disappear in 'free' +Animate RCs (t=-7.35,p<.001) follows from the internal structure of 'celui/celle' in French whose morphosyntactic features match those of the intervener (i.e. number and gender). A'-dependencies headed by a +NP-Animate element yield better performance in older children only (all ps<.001). That younger systems cannot exploit the mismatch in animacy with the intervener (p>.05) stems from animacy being a subfeature of NP, i.e. it is too deeply embedded to be taken into account for computation by immature systems. Across all age groups, a mismatch in animacy did not significantly improve comprehension in -NP questions (p>.05). We take this as evidence that the animacy effect depends on the locus where the feature is expressed (whether it surfaces with a +NP or -NP feature). Featural intervention across the board increases cognitive load, as shown by the link between accuracy of performance and memory scores (r=.660,p<.001), suggesting that limitations of computational resources impact the processing of A'-dependencies in children. 1. Quel chameau/Qui est-ce que le zèbre suit? 'Which camel/Who ESK the zebra is following?' 2. Quelle balançoire/Qu'est-ce que la fille frappe? 'Which swing/What ESK the girl is hitting?' 3. Montre-moi le chameau/celui que le zèbre suit. 'Show me the camel/the one the zebra is following.' 4. Montre-moi la balançoire/ce que la fille frappe. 'Show me the swing/what the girl is hitting.'