Loerts, H. , Stowe, L. A. & Schmid, M. S.
University of Groningen
The purpose of this project is to examine the online processing of grammatical gender by natives and late L2-learners of Dutch who either have a gender system in the L1 (Polish) or lack any form of gender concord in the L1 (Turkish). To be able to examine the fast processing of language in real time, this study recorded event-related brain potentials (ERPs) in response to spoken Dutch sentences. The participants, 28 adult native speakers of Dutch (age range: 30-60; mean: 48) with varying educational backgrounds, listened to correct sentences and sentences containing local gender violations (e.g. het[neu]/*de[com] boek; the[neu]/*the[com] book). In addition, the sentences varied in terms of semantic expectancy/cloze probability. In line with previous findings, our results reveal a clear P600 for gender violations, which is thought of to indicate a repair or re-analysis process in response to the gender mismatch. Sentences in which the target noun had low semantic expectancy elicited an increased N400, reflecting semantic integration difficulty. The N400 was independent of the gender mismatch variable. Interestingly, an interaction was found between gender mismatch and cloze probability in the P600 time-window, i.e. the P600 started significantly later in the low-cloze conditions as compared to the high-cloze conditions. A similar interaction was found by Gunter et al., (2000), who reported a P600 only for high-cloze nouns and not for low-cloze nouns. The results suggest that semantic expectancy facilitates the repair process as reflected in an earlier P600 for high-cloze items. Planned experiments will show whether L2-learners process gender and semantic expectancy similarly to natives and whether the presence or absence of a gender system in the L1 affects these processes. The poster presents the results from the native control group and will elaborate on the hypotheses concerning Polish and Turkish learners of Dutch.