To revise or not to revise? What the P600 can tell us about semantic processing in both native and non-native sentence comprehension

Frenck-Mestre, C. 1 , Carrasco, H. 2 & Sneed, E. 2

1 Centre National de Recherche Scientifique
2 Université de Provence

Debate has long waged over the underlying linguistic processes revealed by the various ERP components. In particular, the significance of the P600 has been a polemic topic. Recently, several studies have shown that the violation of thematic constraints can give rise to a P600 effect. In the current study we provide clear evidence that purely semantic anomalies can elicit a clear P600 effect, in addition to the classical N400 effect. The P600 was observed however, only in native speakers. For non-native speakers, whereas a classical N400 was found to semantic violations no subsequent P600 appeared. This pattern was found in three independent experiments involving independent participant populations (3 independent groups of French native speakers, two groups of English-French learners and a group of Spanish-French learners) as well as various semantic manipulations, attesting to the robust nature of the finding. These data imply that the P600 is indeed a hallmark of revision processes but is in no way limited to the computation of syntactic dependencies. Moreover, the difference in ERP patterns for native and non-native processing provides novel information about how the two groups may differ as concerns their capacity to deploy attention during on-line processing.