The bilingual brain considers non-target language during speech comprehension: Evidence from event-related potentials

Converging evidence suggests that there is language non-selective access to the bilingual lexicon in the domain of written word comprehension (for a review see: Dijkstra, 2005). However, for auditory comprehension the situation is less clear-cut. Whereas some studies show that interlingual phonemic overlap can lead to nonselective activation there are indications that lexical candidates from a non-target language are nevertheless unavailable for semantic integration. The current study further investigated the availability of first language (L1) semantic features in second language (L2) speech processing. The first experiment identified electrophysiological correlates of perceiving congruent and incongruent language switches. The second experiment investigated whether interlingual homophones activate L1 and L2 semantics in a language non-selective manner. In both experiments we recorded EEG from 30 Dutch-English bilinguals who listened to spoken sentences in their L2 (English). Experiment 1 used a spoken L2 sentence context (e.g., Pork is a kind of ) in which the critical word was (a) an L2 word that was congruent with the sentence context (e.g., meat), (b) an L2 word that was incongruent with the sentence context (e.g., noise), (c) a congruent L1 word (e.g., vlees ‘meat’), or (d) an incongruent L1 word (e.g., schoen ‘shoe’). Incongruent L1 and L2 words elicited a canonical N400 effect compared to congruent L2 words. Furthermore, results revealed a transient negativity to congruent L1 words that offset earlier than the N400 to inconruent L1 and L2 words. Lastly, both congruent and incongruent language switches elicited a late positive component (LPC). In Experiment 2 the critical word in the sentence was an L1-L2 interlingual homophone (e.g., pet ‘hat’). In separate conditions: (a) the L2 meaning was congruent (e.g., “His cat was his favourite pet …”), (b) the L1 meaning was congruent (e.g., “The policeman wore a pet …”, or (c) neither meaning was congruent with the sentence context (e.g., “Jeremy drove the pet …”). These conditions were compared to a semantically fully congruent control condition (e.g., “We went to the Vatican to see the pope ..”). Analogous to results from Experiment 1, whenever the L1 meaning of the homophone was congruent with the sentence context we observed a transient negativity compared to congruent L2 items with an earlier offest than the N400 effect to incongruent homophones. When the L2 meaning of the homophone was congruent a transient negativity emerged, but substantially later than the N400 effect to incongruent homophones. Taken together these results firstly show that interlingual homophones activate semantic features from L1 and L2, however L1 semantic features become available later in time, resulting in a decreased or increased incongruity response, respectively. Secondly, while L1 words are not initially available for semantic integration, congruent L1 words are nevertheless eventually successfully integrated. Finally, the LPC may reflect the perception of language switches. References: Dijkstra, T. (2005). Bilingual Visual Word Recognition and Lexical Access. In J. Kroll & A. M. B. de Groot (Eds.), Handbook of bilingualism: Psycholinguistic approaches (pp. 179-201). New York: Oxford University Press