CLS Speaker Series - Patricia Schempp (Penn State University)
L2 Learners’ Processing of Grammatical Gender Varies According to Cognate Status and Proficiency: An ERP study
Previous studies highlight the difficulty of mastering L2 gender (e.g., Hopp 2013), however recent ERP evidence suggests that after training, L1 English speakers can exhibit native-like ERPs to gender violations (e.g., Morgan-Short et al., 2012). This study extends Morgan-Short et al.'s (2012) work on artificial language learning to natural language, investigating whether classroom-based L2 German learners are sensitive to gender violations in L2 German, before and after training, and whether advanced L2 German learners are sensitive to these same violations. Additionally, it investigates L2 noun-gender mappings for cognates versus non-cognates, broadening research showing that L2 speakers process cognates faster than non-cognates (see Schwartz & van Hell, 2012). In an ERP task, twenty-three intermediate L2 German learners and 19 advanced L2 German learners read target questions and then answered the question by choosing the appropriate picture on screen. The target questions were grammatically correct or incorrect, and contained cognate or noncognate nouns. Next, intermediate participants were trained offline to high accuracy with the nouns and their gender using picture naming. One week later they were tested on their vocabulary and gender accuracy offline, retrained, and then repeated the ERP task. ERP analyses timelocked to the critical noun phrase revealed that before training, intermediate learners exhibited no significant ERP responses to grammatical gender violations. After training, intermediate participants exhibited an N400 effect for gender violations with cognates only. At 500-900ms there was a frontal positivity for gender violations with cognates, and at 900-1100ms there was a frontal positivity for gender violations with cognates and non-cognates. Analysis of the picture naming test prior to retraining reveals that cognates were easier to recall, but gender accuracy did not vary with cognate status. These results suggest that for intermediate learners cognate effects extend to the processing of L2 morphosyntactic features, whereby cognates facilitate the online processing and retrieval of nouns and associated grammatical information. In contrast, advanced learners exhibited sensitivity to gender violations for noncognates only, suggesting that there is a fundamental difference in the processing of gender for cognates and noncognates among L2 learners, that may vary as a function of proficiency level.