Number specification in L2 processing of Subject-Verb agreement: An ERP study

Abstract

This is part of an ongoing study that investigates the processing of subject-verb agreement violation in Norwegian adult second language (L2) learners of English with advanced proficiency, using ERP responses. Unlike English, Norwegian lacks overt S(ubject)-V(erb) agreement marking. We manipulated number marking on the subject by using determiners and added S-V violations to see the influence of determiner-specification on non-local verbal agreement processing: 1. The window of the house is really clean. 2. *The windows of the house is really clean. 3. That window of the house is really clean. 4. *Those windows of the house is really clean. ERP data were acquired while participants (N=11, F= 7, mean age=23) read sentences on the screen and each sentence was followed by a grammaticality judgment, yielding a behavioral response. Preliminary results from repeated measure ANOVAs on the behavioral and ERP data suggest that participants detected non-local agreement violation, indexed by a P600 effect (F = 0.36, P = 0.56). Contrary to the previous findings (e.g., Cheng et.al, 2020) the manipulation in the determiner-number specification did not have a facilitation effect in the P600 time window and there was not a significant difference in the amplitude of the P600 for specified vs unspecified sentences. Instead, the effect of the number-specified determiners in interaction with grammaticality is reflected in the N400 time window, with a higher amplitude for the specified sentences (F = 13.82, P = 0.003). The findings will provide further insights into the underlying mechanisms of agreement processing when there is no structural overlap between L1 and L2 and thus one of the most debated questions in the field: whether native and non-native processing mechanisms are similar or not.