Determiner selection in speech production Evidence from Norwegian

This thesis reports on two picture-word interference experiments where I investigate the processes involved in selection of grammatical gender in Norwegian Noun Phrase production. The respondents, consisting of subjects with a two gender system (masculine and neuter) and subjects with a three gender system (masculine, feminine and neuter), produced a gender-marked demonstrative + noun NP in response to a picture shown with either a distractor word of the same grammatical gender or one with a different grammatical gender. Though there are three different genders in Norwegian, there are only two demonstratives; den for feminine and masculine nouns, and det for neuter nouns. Significant longer naming latencies were found when the picture was shown with a distractor word of different gender than if the word and picture had identical gender. This effect however, was only obtained for subjects with a three gender system. This finding suggests that the interference does not occur between the independent word forms, the demonstratives, as proposed by Schiller and Caramazza (2003), but occur when selecting the noun's grammatical gender feature as originally suggested by Schriefers (1993). This implies that selection of grammatical features, at least gender, is based on competition, and does not come as an automatic consequence of selecting the lemma. A computational model based on the findings was tested in WEAVER++ (Roelofs, 1992). The simulations showed that WEAVER++ accounts for the some of the empirical findings