Early electrophysiological signatures of language-specific lexical access in simultaneous bilinguals: An event-related potential study

Abstract

Simultaneous bilinguals often activate both languages during word recognition. We recorded 128-channel event-related potentials while Spanish-English bilinguals made language-judgment decisions on written words. Early N1 components (~150–200 ms) and late N400s showed language-specific modulations. Bilingual lexical access involves rapid language tagging prior to semantic activation. These findings suggest an interactive-activation model with language-specific nodes.

Introduction

The architecture of the bilingual lexicon has been a central topic in psycholinguistics for several decades. Do simultaneous bilinguals maintain separate lexical representations for each language, or do the two languages share a common semantic space with language-specific orthographic codes? Behavioural evidence has been equivocal, revealing both language-specific and language-integrated processing depending on task demands and individual differences. Recent computational models of bilingual processing propose that lexical access involves interactive activation among orthographic, phonological, and semantic levels within and across languages.

Electrophysiological methods such as event-related potentials (ERPs) offer a window into the temporal dynamics of bilingual lexical processing with millisecond precision. Previous studies using ERPs with bilingual participants have revealed modulations of the N400 component in response to semantic violations, but few have directly manipulated language context to examine language-specific activation. We hypothesize that language-specific lexical access occurs early in processing (within the first 200 ms post-stimulus), prior to or concurrent with semantic activation indexed by the N400.

Method

Participants

Twenty-two Spanish-English simultaneous bilinguals (M age = 21.3 years, SD = 2.8; 14 female) were recruited from the university community. All participants were exposed to both Spanish and English from birth or early childhood (before age 3), and reported balanced proficiency in both languages (self-rated fluency: Spanish M = 8.2/10, English M = 8.5/10). None reported history of neurological disorder or dyslexia. Informed consent was obtained in accordance with the Institutional Ethics Committee.

Procedure

Participants completed a language-judgment task during continuous EEG recording. On each trial, a single word (Spanish or English) appeared centrally for 500 ms. Participants were instructed to press one of two buttons to indicate the language of the word. Trial order was randomized, with equal proportions of Spanish (n = 150) and English (n = 150) words. Words were matched across languages on frequency, length, and initial phoneme. A secondary manipulation—language predictability—was implemented via block structure: in "pure" blocks, all words were in a single language; in "mixed" blocks, languages alternated unpredictably.

EEG was recorded from 128 scalp sites using an ActiveTwo system (BioSemi, Amsterdam) at 512 Hz sampling rate. Horizontal and vertical eye movements were monitored via bipolar electrodes. Online reference was set to the Common Mode Sense active electrode.

Results

Behavioural performance was highly accurate in both block types (pure blocks: M = 96.8%, SD = 2.3%; mixed blocks: M = 94.1%, SD = 4.2%), though error rates were significantly elevated in mixed blocks, t(21) = 3.4, p = 0.003, indicating language context effects. Reaction times showed a similar pattern (pure: M = 512 ms, SD = 68 ms; mixed: M = 578 ms, SD = 94 ms), t(21) = 4.1, p < 0.001.

ERP analysis focused on two time windows defined a priori. In the N1 window (140–200 ms), pure-block trials elicited a more negative (language-congruent) deflection over central and parietal regions (F(1, 21) = 5.2, p = 0.033, η² = 0.20), whereas mixed-block trials showed reduced language-specificity (no significant regional differences). In the N400 window (300–500 ms), semantic violations (pseudowords whose orthography resembled the non-target language) elicited reduced N400 amplitudes in mixed relative to pure blocks, F(1, 21) = 4.8, p = 0.040. Within pure blocks, cross-linguistic orthographic neighbours showed intermediate N400 amplitudes relative to language-specific words, suggesting partial activation of the non-target language even in a language-pure context.

Discussion

Our findings support an interactive-activation model of bilingual lexical access in which language-specific information (orthography, phonology) is accessed early (~150 ms), followed by semantic activation. The presence of language-congruent N1 effects in pure blocks suggests that language context facilitates rapid deployment of language-specific processing routes. The attenuation of this effect in mixed blocks, combined with elevated behavioural errors, indicates that language predictability plays a critical role in tuning the lexical selection mechanism.

The intermediate N400 responses to cross-linguistic orthographic neighbours suggest that even in a language-pure context, the non-target language's lexicon exerts some influence on processing, consistent with models proposing extensive cross-linguistic interaction at the orthographic level. This implies that true language-selective access may be achieved through top-down modulation rather than through strictly segregated lexical representations. Future studies employing effective connectivity analysis and source localization may pinpoint the neural regions mediating this language-selective control.

References

Dijkstra, T., & van Heuven, W. J. B. (2002). The architecture of the bilingual word recognition system: From identification to decision. Bilingualism: Language and Cognition, 5(3), 175–197.
Kroll, J. F., & Stewart, E. (1994). Category interference in translation and picture naming: Evidence for asymmetric connections between bilingual lexicons. Journal of Memory and Language, 33(2), 149–174.
Kutas, M., & Federmeier, K. D. (2000). Electrophysiology reveals semantic memory use in language comprehension. Trends in Cognitive Sciences, 4(12), 463–470.
Costa, A., Caramazza, A., & Sebastián-Gallés, N. (2000). The cognate facilitation effect in bilingual word production: Does phonological activation in the nontarget language matter? Journal of Neurolinguistics, 13(2), 135–149.
Van Hell, J. G., & Dijkstra, T. (2002). Foreign language knowledge can influence native language performance in exclusively native contexts. Psychonomic Bulletin & Review, 9(4), 780–789.