Grammatical gender in the sphere of minimal context: An ERP investigation

Hofmann, J., Friederici, A. D., and Kotz, S. A.
Max Plank Institute of Cognitive Neuroscience, Leipzig, Germany
E-mail: hofmannj@cns.mpg.de

The nature of gender information, whether it is more syntactic or semantic, is still in debate. A previous ERP sentence processing study reported a left anterior negativity (LAN) reflecting syntactic processing and a P600 reflecting integration and reanalysis processes for gender violations (Gunter and Friederici, 2000). However, in a word level ERP experiment no LAN, but an N400 effect indicating lexical-semantic processing was observed for gender violations (Hofmann, Friederici & Kotz, 2003). A superordinate analysis revealed a LAN for nouns with phonologically and derivational-morphologically marked endings in contrast to semantically marked words. As gender marked words were presented out of context the question arose, whether syntactic processing is triggered by context only.

The current experiment investigated the influence of visually presented minimal context information (noun phrases) and no context information on gender processing. Additionally nouns were cross-balanced over three gender information categories (phonological, derivational-morphological and semantic) and three gender types (feminine, masculine and neuter).

All three gender information categories elicited an N400 effect for incongruent versus congruent trials. This effect is more pronounced in the context condition. Also an early frontal effect beginning at around 200 ms post-stimulus onset occurs only in the context condition and is more pronounced for the phonologically marked gender categories (phonological and derivational-morphological). This ERP response might reflect a syntactically driven process followed by a lexical-semantic process as indicated by the N400. Thus, minimal context seems to be a necessary condition to initiate syntactic processing of gender information.