The role of semantic features in word production.

According to Levelt's language production model, to name an object, speakers must first conceptualize and lexicalize the object before it can be named. Conceptualization is conducted through the semantic network, with concepts activating lexical items at the lemma level, i.e., lexicalization. So far, research has focused on the roles of semantic categories (i.e., semantic category interference) and single semantic features (i.e., semantic feature interference) but less so on the number of overlapping features. To investigate the role of the number of overlapping features in language production, we conducted a picture-word interference study in Mandarin Chinese, varying the semantic category and shape congruency whilst controlling for classifier congruency. We also recorded behavioural and electrophysiological responses. We observed a main effect of semantic category that was stronger than a main effect of shape (i.e., a semantic feature). That is, the reduction in naming accuracies, the increase in naming latencies, and the increase in ERP amplitudes between 275 - 575 ms post-stimulus onset (N400 effect) for congruent vs. incongruent conditions were larger for semantic category than for shape. In addition, we found an interaction effect between semantic category and the semantic feature 'shape' regarding naming accuracies and also at the electrophysiological level. We conclude that, with increasing feature overlap between word pairs, there may be more spreading of activation between such pairs in word production.