Retour vers liste

Détail de la contribution

Auteur: Romy LASSOTTA

Co-Auteur(s): Akira Omaki, Daniele Panizza, Sandra Villata, Julie Franck

Garden-path recovery and cognitive control in children and adults: Evidence from French wh-questions with filled-gaps

Abstract/Résumé: Research on processing syntactic garden-paths found that children fail to revise initial parsing commitments. It has been argued that such failure may result from weak cognitive control, but there is little empirical support for this hypothesis to-date. Our study presents evidence of a link between sentence revision in wh-questions and cognitive control abilities measured by Dimensional Change Card Sorting (DCCST) and N-back tasks. Children (age=7yrs, N=48) and adults (age=23yrs, N=48) took part in a Question-after-Story comprehension task involving temporarily ambiguous wh-questions like “A qui Marie a dit [à sa maman]PP qu’elle avait donné du chocolat?” Listeners were expected to initially attach the wh-element to the main verb (MV), due to active gap filling, and subsequently revise this attachment to the embedded verb (EV), due to the presence of an overt PP (filled-gap) blocking MV attachment. A globally ambiguous condition without PP served to establish the default interpretation preference. Questions followed cartoons that made MV and EV attachment equally plausible. Offline responses and eye-tracking data were recorded. The DCCST measured accuracy and RT while participants sorted objects according to one out of two dimensions (shape or color). Critically, the sorting rule changed during the task to force participants to inhibit the previously relevant dimension. Switch cost was used as index of cognitive control, calculated by the difference between trials where the sorting rule persisted and trials where the rule changed. An N-back task was designed for participants to decide whether the current stimulus in a sequence of fruits matches the one situated two items before. It assesses working memory updating and inhibition of irrelevant lures. Cognitive control was indexed by overall accuracy and hit latency. Adults demonstrated a clear MV preference in the ambiguous condition (86%), but not children (45%). Both populations gave significantly less MV responses in the filled-gap condition (29% and 10% respectively, p<.05). Eye-data revealed that upon encountering the filled-gap, only participants who provided EV responses redirected their fixations from the MV to the EV location, attesting of revision. Mixed model analyses yielded significant interactions between the type of question and switch cost, N-back accuracy and hit latency (ps<.05), indicating that participants with stronger cognitive control provided more EV responses and increased looks to the EV location in the filled-gap condition only, as compared to participants with weaker cognitive control. These results suggest that cognitive control plays a role in syntactic revision in children and adults.