NEUROCOGNITION AND NON-NATIVE GRAMMATICAL PROCESSING: THE VIEW FROM ANAPHORA IN SUCCESSIVE-CYCLIC WH-MOVEMENT IN L1-MANDARIN/L2-ENGLISH

Abstract

Adult learners of a new language (L2) tend not to achieve target-like knowledge and performance, so the cognitive status of L2s has long been debated. While monolingual native (L1) speakers robustly apply their grammatical knowledge to input as it unfolds, there are competing claims about whether L2 speakers also do so reflexively (Dekydtspotter, Schwartz & Sprouse, 2006) or whether they cannot do so (Clahsen & Felser, 2006). In line with the claim from Dekydtspotter, Schwartz and Sprouse, findings in research on this topic have often pointed towards an epiphenomenal frailty in the application of L2 grammatical knowledge, a result of a proportionally greater expenditure of resources in L2 for upstream processes like lexical access and L1 suppression. While the “wetware” of the brain (Berwick & Chomsky, 2017) ultimately gives rise to language and its processing, most neurocognitive research on non-native languages has been paradigmatically divorced from these issues. The current study thus asks to what extent L2 speakers compute real-time syntactic representations, to what extent those computations take place as robustly as in L1 speakers, and to what extent the neurocognitive activity for them is similar to the neurocognitive activity for grammatical computations in L1 speakers. L1-English speakers and L1-Mandarin L2-English speakers completed a rapid serial visual presentation task while EEG was recorded as well as a self-paced reading task. Stimuli involved anaphora in successive-cyclic wh-movement like Which picture of X did Ben say that Amy deleted by accident? Crucially, the Binding Theory constrains when and how X is licensed on the basis of whether it is a singleton reflexive (himself), singleton pronoun (him) or conjoined pronominal (both himself/him and Sara). On one hand, L1 and L2 speakers yielded ERPs and reading time asymmetries implicating the real-time application of grammatical knowledge. These measures further suggested that computations started in similar time windows in L1 and L2. On the other hand, the topographical distribution of ERPs differed in L1 and L2, suggesting distinct neurocognitive action at work, even for the same computations. Intriguingly, the specific nature of L1-L2 differences in topography and duration of ERPs changed across positions in the sentence, suggesting a dynamism that must be accounted for in experimentation and theorizing.