Native and L2 processing of homonyms in sentential context

We compare native and non-native processing of homonyms in sentence context whose two most frequent meanings are nouns (e.g., sentence) or a noun and a verb (e.g., trip). With both participant groups, we conducted a combined reaction time (RT)/event-related brain potential (ERP) lexical decision experiment with two stimulus-onset asynchronies (SOAs) of 200 and 500 ms. At the 200 ms SOA, overall RT and ERP N400 priming was observed in both natives and non-natives, indicating multiple access for both homonym types. At the 500 ms SOA, RTs revealed that contextually inappropriate meanings were no longer active for both groups. In contrast, the ERP data showed that activation of inappropriate meanings had decayed for natives, but not for non-natives. Results suggest that non-natives show native-like multiple access at an early processing stage, but differ from the natives later in processing when sentence context information is used to disambiguate meanings. (C) 2004 Elsevier Inc. All rights reserved.