Research has shown neural changes following second language (L2) acquisition after weeks or months of instruction. But are such changes detectable even earlier than previously shown? The present study examines the electrophysiological changes underlying the earliest stages of second language vocabulary acquisition by recording event-related potentials (ERPs) within the first week of learning. Adult native English speakers with no previous Spanish experience completed less than four hours of Spanish vocabulary training, with pre- and post-training ERPs recorded to a backward translation task. Results indicate that beginning L2 learners show rapid neural changes following learning, manifested in changes to the N400 — an ERP component sensitive to lexicosemantic processing and degree of L2 proficiency. Specifically, learners in early stages of L2 acquisition show growth in N400 amplitude to L2 words following learning as well as a backward translation N400 priming effect that was absent pre-training. These results were shown within days of minimal L2 training, suggesting that the neural changes captured during adult second language acquisition are more rapid than previously shown. Such findings are consistent with models of early stages of bilingualism in adult learners of L2 (e.g. Kroll and Stewart’s RHM) and reinforce the use of ERP measures to assess L2 learning.

Event-related potentials were recorded during the visual presentation of words in the three languages of French-English-Spanish trilinguals. Participants monitored a mixed list of unrelated non-cognate words in the three languages while performing a semantic categorization task. Words in L1 generated earlier N400 peak amplitudes than both L2 and L3 words, which peaked together. On the other hand, L2 and L3 words did differ significantly in terms of N400 amplitude, with L3 words generating greater mean amplitudes compared with L2 words. We interpret the effects of peak N400 latency as reflecting the special status of the L1 relative to later acquired languages, rather than proficiency in that language per se. On the other hand, the mean amplitude difference between L2 and L3 is thought to reflect different levels of fluency in these two languages.

Research in the past few decades generally supported a nonselective view of bilingual lexical access, where a bilingual’s two languages are both active during monolingual processing. However, recent work by Costa et al. (2017) brought this into question by reinterpreting evidence for nonselectivity in a selective manner. We manipulated the factor of first language (L1) attrition in an event-related potential (ERP) experiment to disentangle Costa and colleagues’ selective processing proposal versus the traditional nonselective processing view of bilingual lexical access. Spanish–English bilinguals demonstrated an N400 effect of L1 attrition during implicit L1 processing in a second language (L2) semantic judgment task, indicating the contribution of variable L1 lexical access during L2 processing. These results are incompatible with Costa and colleagues’ selective model, adding to the literature supporting a nonselective view of bilingual lexical access.