Séminaire Octogone Lordat : "Semantic neighbours in the lexicon: how does their number influence picture naming performance?"

Publié le 24 août 2015 Mis à jour le 15 mars 2016
le 23 septembre 2015
MdR, salle E323

par Solène Hameau (Macquarie University Sydney)

It is a common claim that in the course of spoken word production, several semantically related representations are activated along with the target word. It remains unclear to date, what type of semantic relationship between these representations exists, and how this co-activation of semantic neighbours can influence word retrieval. We investigated the influence of semantic neighbourhood density (SND: number of words that are similar in meaning to the target word) on picture naming performance of (i) unimpaired speakers (n=50) and (ii) speakers with aphasia (n=193). We first carried out a principal component analysis that allowed us to distinguish four conceptually different types of SND. In a second step, a multiple regressions analysis was applied to our “unimpaired” data and did not reveal any influence of SND on response time or accuracy, consistent with previous findings (e.g. Bormann, 2011). Third, a linear mixed effects modelling approach was used analysing the aphasic data:  we found facilitating effects of high SND on accuracy for the group, contrary to previous findings (e.g. Mirman, 2011). Further investigation of the influence of SND on error types and the interaction with the patients’ type of impairment was carried out and discussed within psycholinguistic models of speech production.