A Morphological Processing of Determiner-noun Agreement by Korean Learners of English: An ERP Study

Abstract

The current study attempted to investigate number agreement processing by advanced Korean learners of English, with distance and working memory capacity as modulating factors. Studies have reported that agreement is acquired at a late stage of second language acquisition, and that L2 learners are insensitive to agreement violations in real-time language processing even at an advanced level. English number agreement is not an exception, especially for L2 learners whose L1 does not have the same agreement system (Chen et al., 2007

Jiang, 2004

Tanner et al., 2012). Nonetheless, previous studies have focused on subject-verb agreement and how L2 learners process number agreement between English determiner and noun is yet to be investigated. A related question is whether L2 learners with higher working memory capacity show better sensitivity to agreement violations as a function of distance between agreeing constituents. While working memory capacity is argued to be an important element of language aptitude, its role for high proficiency L2 learners is controversial. Granted, the current study examines the online processing of number agreement between English determiner and noun with ERP (event-related potentials) methodology for advanced Korean learners of English with high, middle, and low working memory capacity.

Twelve English native speakers and eighteen Korean learners of English with different levels of working memory capacity (high, middle, low) read English sentences, half of which contained agreement violations. While they were working on the sentences, their EEG data were recorded. The sentences also varied in terms of the distance between determiner and noun (i.e. short distance and long distance). While no LAN effect was observed in both groups in the time window of 300-450 ms, the native group showed a P600 effect (500-600 ms) in response to agreement violations. LAN is an early indicator of morphological violations while the P600 signals repair or reanalysis of syntactic violations. The P600 was not elicited in the learner group in the same time window, for both short distance and long distance conditions. While the short distance conditions overall elicited more positive waveforms than the long distance conditions for both groups, the distance did not affect the size of P600. Finally, L2 learners with high working memory capacity showed a small P600 effect for short distance conditions at a later time window (700-800 ms).

The result demonstrates that L2 learners do not process number agreement in a native-like way when the equivalent grammatical feature is absent in their mother language. The ERP response of the native group and the learner group to varying distance conditions highlights their difference even more. The native group showed a consistent P600 effect even for long distance conditions, whereas the learner group did not show sensitivity to agreement violations even when the distance was short and therefore the cognitive load was less taxing. Nonetheless, working memory capacity was found to be an influential factor in processing agreement for L2 learners.

The current study has the following implications in regard to second language acquisition. First, L1 effect is crucial in L2 processing. Second, high working memory capacity can in part compensate for L1 effect.