Brain potentials reveal semantic priming in both the ‘active’ and the ‘non-attended’ language of early bilinguals Clara D. Martin a,b,c, ⁎, Benjamin Dering b,c , Enlli M. Thomas c,d , Guillaume Thierry b,c a Departament de Tecnologias, Universitat Pompeu Fabra, C. Roc Boronat,138, 08018 Barcelona Spain b School of Psychology, Bangor University, UK c ESRC Centre for Research on Bilingualism in Theory and Practice, Bangor University, UK d School of Education, Bangor University, UK abstract article info Article history: Received 14 May 2008 Revised 2 April 2009 Accepted 8 April 2009 Available online 15 April 2009 Keywords: Bilingualism ERPs Language Semantics A key question in the study of bilingual functioning is whether both the languages known are active at all times or whether one language can be selectively inactivated when bilingual individuals are tuned to the other language. Psycholinguistic and neuroscientific investigations have provided inconsistent data regarding the level of semantic activation of the two languages, even in the case of highly proficient bilinguals. In the present study, highly proficient, early Welsh/English bilinguals were presented with words in both their languages and were required to make word length decisions on words in one language while disregarding words in the other. Participants were not explicitly told about the organization of the word stream in pairs manipulating (a) semantic relatedness, (b) language of the prime and (c) language of the target in a fully counterbalanced two-by-two-by-two design. We observed significant semantic priming for both English and Welsh target words, irrespective of the active language, and independent of performance in the low-level letter counting task. We conclude that accessing the meaning of a written word is automatic in the two languages even when fluent bilingual adults are instructed to disregard words in one of their languages. © 2009 Elsevier Inc. All rights reserved. Introduction Most prominent theories of bilingualism assume that the mental representation of languages can be divided into a lexical (word form) and a conceptual (word meaning) level (Potter et al., 1984; Snodgrass, 1984). In the so-called hierarchical model (Kroll and Stewart, 1994), words in the two languages (L1 and L2) generally have separate representations at the lexical level, but at least partially shared representations at the conceptual level. The Bilingual Interactive Activation (BIA+) model of bilingual word recognition developed by Dijkstra and Van Heuven (Grainger and Dijkstra, 1992; Dijkstra and Van Heuven, 1998, 2002) posits that the bilingual word recognition system is fundamentally nonselective and involves a parallel lexical search within an integrated L1/L2 lexicon. Based on these two major theories, a number of experiments on bilingual word recognition have been conducted in experimental psychology, neuroimaging and electrophysiology. Behavioral studies have suggested that bilingual individuals who have acquired their languages in parallel from birth (referred to below as early bilinguals) develop independent routes for accessing meaning in their two languages (Scarborough et al., 1984; Gerard and Scarborough, 1989). However it remains debated whether the lexicon of one language is inactive during processing in the other (Macnamara and Kushnir, 1971; Scarborough et al., 1984; Soares and Grosjean, 1984; Gerard and Scarborough, 1989; Rodriguez-Fornells et al., 2002; but see also Grainger and Dijkstra, 1992; Dijkstra and Van Heuven, 1998; Green, 1998) or whether both L1 and L2 lexicons are co-activated, with influences of one language on the other (Kirsner et al., 1984; De Groot et al., 2000; Dijkstra et al., 2000; De Bruijn et al., 2001; Hernandez and Reyes, 2002). In order to gain further insight into the interplay of L1 and L2 in bilingual individuals, neuroimaging studies have explored the neural repre- sentation of the two languages. Such studies have shown that processing of L1 and L2 involves non-overlapping brain areas in late bilinguals, i.e., individuals who have acquired their second language after the age of 12 (Perani et al., 1996; Dehaene et al., 1997; Kim et al., 1997). In highly proficient bilinguals, however, regions activated by L1 and L2 processing tend to overlap to a large extent (Kim et al., 1997; Perani et al., 1998; Chee et al., 1999a,b; Perani and Abutalebi, 2005). Event-related potentials (ERPs, i.e., averaged electrical signals derived from brain activity), have provided insight into the time- course of neural events taking place during word recognition. They have suggested that access to L1 and L2 can be selective as early as the orthographic stage (Proverbio et al., 2002; Rodriguez-Fornells et al., 2002) or/and at the semantic level in late bilinguals (Alvarez NeuroImage 47 (2009) 326–333 ⁎ Corresponding author. Departament de Tecnologias, Universitat Pompeu Fabra, C. Roc Boronat,138, 08018 Barcelona Spain. Fax: +34 93 542 25 17. E-mail address: clara.martin@isc.cnrs.fr (C.D. Martin). 1053-8119/$ – see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2009.04.025 Contents lists available at ScienceDirect NeuroImage journal homepage: www.elsevier.com/locate/ynimg