José Alemán Bañón, Stockholm University

A fundamental question in second language (L2) acquisition research concerns whether adult L2 learners recruit processing mechanisms similar to those used by native speakers during real-time comprehension (e.g., Clahsen & Felser, 2006; Hopp, 2022). For example, it is well established that native speakers rely on prediction, which broadly refers to the use of context to facilitate the processing of upcoming input (Kuperberg & Jaeger, 2016). L2 processing is also predictive, although learners recruit predictive mechanisms to a lesser extent than native speakers. Nevertheless, Kaan (2014) has argued that predictive processing is qualitatively similar in L1 and L2 speakers, and that differences between the two populations can be explained by the same linguistic and cognitive factors that account for variability among L1 speakers, such as the quality of lexical representations and individual differences in cognitive abilities, in addition to factors unique to the L2, such as L1-L2 dissimilarity. More recently, Grüter and colleagues have argued that differences in predictive behavior between L1 and L2 speakers arise because prediction has reduced utility in the L2 (Grüter & Rohde, 2021; Kaan & Grüter, 2021). According to this proposal, predictive cues tend to be less reliable in the L2 than in the L1 due to differences between L1 and L2 acquisition and cross-linguistic differences, among other factors. For example, when a predictive cue is absent in the L1 or realized differently, learners may weigh its predictive strength differently from L1 speakers.

Against this background, I will discuss two studies that investigate predictive processing in L1 and L2 speakers. Study 1 examines whether Swedish-speaking learners of English generate lexicosemantic predictions to the same extent as L1 speakers when changes in the linguistic environment increase the reliability (and utility) of the predictive cues. Study 2 investigates whether L2 learners of Swedish whose L1 is Finnish, a language that lacks grammatical gender, use gender cues in Swedish to facilitate lexical retrieval, and how this process is affected by properties of the L2 grammar (e.g., markedness). Overall, the findings from these studies suggest that prediction is qualitatively similar in the L1 and L2 and is modulated by similar factors, at least at these levels of linguistic representation.

Part 2 of 3 on a series of talks on the neurological underpinnings of language, organised by the TEAM research programme (Transdisciplinary Approaches to Learning, Acquisition, Multilingualism)