Mémoire de Travail et détection d'erreurs d'accord verbal : Etude chez le novice et l'expert

Résumé En

Alexandra DEDEYAN, Pierre LARGY et Isabelle NEGRO: Mémoire de Travail et détection d’erreurs d’accord verbal: Étude chez le novice et l’expert The revision of the verbal agreement is an activity requiring at first "to detect” an error or a risk of error, before operating possible modifications. Two procedures of detection are differentiated (Largy & Dédéyan, 2002): an algorithmic novice procedure and an automatized expert procedure based on the treatment of cooccurrences between inflectional morphemes (e. g., case of two successive words respectively bent by -s and -nt). The first seems sensitive to semantic constraints, such as the plausibility between the local noun and the verb (Largy, Dédéyan & Hupet, 2004); by contrast, the second is influenced by factors of surface perceptive, such as the police of writing (i. e., bold, degraded, Dédéyan & Largy, submitted) of the marks of the local noun and the verb. To test the hypothesis according to which these factors are of visual nature, this research studies the impact of the visuo-spatial sketchpad (Baddeley, 1986, 1992, 2003) on the novice and expert detection. To do it, children, teenagers and adults have to detect verbal agreement errors within sentences "Noun1 Noun2 Verb” of which half contains a proximity concord (e. g., The dog of the neighbours arrive). A temporal constraint is imposed. This task is realized in isolated condition at first, then simultaneously in a secondary task implying the visuo-spatial sketchpad. The results show differentiated profiles according to the level of expertise. It indeed seems that the experts make more errors relative to the secondary task than the novices. The expert detection thus seems to lean on visual cues.