Online processing of syntactic constructions unique to the second language: An eye-movement study of subject–verb agreement with Swedish advanced learners of English

Abstract: It is a hotly debated issue whether second language (L2) learners can achieve target-like online syntactic processing of relations not instantiated in their first language (L1). The Shallow Structure Hypothesis (SSH) predicts that L2 learners will not process configurations with structural distance in a target-like online fashion (Clahsen & Felser, 2006 a,b,c), whereas some studies find that target-like processing is possible (e.g. Foucart & Frenck-Mestre, 2012). In view of the divergent findings, the present study asks how Swedish advanced learners of English (n=18) process complex subject–verb agreement (e.g. the editor of the newspaper/s was/'were admired by the staff) compared to native English speakers (n=15) as measured in an eye-tracking paradigm. Importantly, Swedish lacks subject-verb agreement whereas English marks it. Six conditions are investigated, in which number of the head noun (the editor of the newspaper, singular/plural), number match between head and modifying noun (the editor of the newspaper), and grammaticality (head noun/verb agreement in number) are manipulated. The results indicate that native English speakers are sensitive to manipulations of complex subject–verb agreement, with reading times varying depending on head/modifying noun match and grammaticality. In contrast, Swedish L2 learners show no difference in reading times across conditions. Moreover, more detailed analyses reveal that native speakers are more likely to skip copulas during reading than L2 readers, suggesting a greater predictive capacity and a subtle timing advantage in processing. The results are in line with the SSH prediction and previous studies of this structure (e.g. Chen et al., 2007; Jiang, 2004). Implications for theories of syntactic processing in the L2 are discussed.