Je Voudrais Parler Français: Brain Activity During Sleep Supports Second Language Learning

Abstract

Language learning depends on a variety of cognitive abilities, including long-term memory. Sleep is important for the enhancement of memory for newly acquired information and skills. Much of what we know about the relationship between sleep and memory has come from the investigation of two distinct long-term memory systems: declarative (memory for e.g., facts, figures and events) and procedural (memory for e.g., strategy, rules and motor skills). Several sleep-specific electrophysiological markers of memory processing have been identified. More specifically, sleep spindles (bursts of neural oscillatory activity which characterize non-rapid eye movement (NREM) sleep) may be a marker of consolidation for declarative memory (e.g., semantics, facts, figures, events), while rapid eye movements may serve as a marker for cognitive aspects (e.g., grammatical rule-learning) of procedural memory. In adults, second language acquisition (SLA) is thought to depend at first on declarative memory for grammar and linguistic rules (i.e., “early SLA”), and then shifts to procedural memory as the learner gains experience (i.e., “late SLA”). Given the unique roles of spindles and rapid eye movements in declarative and procedural memory consolidation, it was hypothesized that sleep spindles would correlate with language improvement during early SLA, whereas rapid eye movements would correlate with language improvement during late SLA. Eight Anglophone University students (six females, one left-handed, ages: 17-28) completed polysomnographic recordings four times throughout an intensive 6-week French course. Electrode sites W1 and W2 (Wernicke’s area- left and right hemispheres) were used to extract sleep spindle data, as well as examine event-related EEG spectral power time-locked to rapid eye movements. A correlational analysis indicated a significant positive relationship between improvements in French proficiency (post-course minus pre-course) and the change in spindle density (early minus late SLA) at electrode W1 for NREM stage 2 and SWS. A nonparametric cluster-based permutation analysis examining event-related power time-locked to REMs, indicated that there was a significant effect of SLA stage (early versus late), whereby an increase in power during late SLA was observed in the theta band. Importantly, this increase in theta power (late minus early) positively correlated with improvements in French proficiency. This follow-up study reveals an important role for sleep spindles in early SLA, and suggests that cortical theta activity during rapid eye movements is a potential marker of learning for cognitive procedural memory during late SLA.