Elsevier

Brain and Language

Volume 203, April 2020, 104739
Brain and Language

Co-activation of the L2 during L1 auditory processing: An ERP cross-modal priming study

https://doi.org/10.1016/j.bandl.2019.104739Get rights and content

Highlights

  • Non-dominant language accessed automatically when only dominant language in use.

  • Co-activation occurs in absence of overt sound overlap with dominant language.

  • Results suggest highly interconnected bilingual mental lexicon.

Abstract

Several studies have shown that unbalanced bilinguals activate both of their languages simultaneously during L2 processing; however, evidence for L2 activation while participants are tested exclusively in their L1 has been more tenuous. Here, we investigate whether bilingual participants implicitly activate the label for a picture in their two languages, and whether labels activated in L2 can prime activation of cross-linguistically related L1 lexical targets. We tested highly proficient early Spanish-Basque bilinguals on an ERP cross-modal priming task conducted only in their L1, Spanish. Participants activated prime picture labels in both Spanish and Basque. More importantly, participants activated Basque translations of Spanish auditory targets, even in a Spanish experimental environment with no reference to Basque. Results provide strong evidence for non-selective bilingual lexical access, showing co-activation extending to lexical levels beyond phonological overlap. Our results add to the growing body of evidence for the interconnective nature of bilingual language activation.

Introduction

A central question in the study of bilingualism has been whether there is co-activation of a bilingual’s languages when hearing, reading, or speaking in one language alone. So, when a Spanish-Basque bilingual hears the Spanish word perro “dog,” will the Basque word for dog, txakur, also be activated? Early accounts of bilingualism proposed the equivalent of a mental switch that allowed the bilingual to turn off the irrelevant language during speech processing or production (Penfield & Roberts, 1959). It is by now a recognized phenomenon, however, that bilinguals display concurrent activation of both languages, regardless of whether they are reading, speaking, or listening to one language alone (e.g., Colomé, 2001, Costa et al., 1999, Hermans, 2000, Macizo et al., 2010, Martín et al., 2010, Spivey and Marian, 1999).

Based on these results, recent accounts of bilingualism argue that the cognitive architecture of the bilingual is fundamentally nonselective, but under certain circumstances operates selectively (Kroll et al., 2006, Martin et al., 2016, Molnar et al., 2015). The question now is no longer whether co-activation can occur, but under what (experimental and language) conditions it does or does not, and what levels of language processing are typically affected. In the current study, we present evidence from an ERP experiment with highly proficient Spanish-Basque bilinguals for language co-activation at the phono-lexical level. We show that Basque language representations are co-activated at both the phonological and the lexical level in a Spanish-only experimental environment, especially with regard to Basque translations and rhyme words.

Recent calls in the literature (e.g., Wu & Thierry, 2010a) have emphasized that new work must pay particular attention to the context of language processing, and especially situations that may inadvertently engage both languages due to experimental requirements or broader social context. In particular, one criticism of studies showing coactivation to the lexical level is that the methodological paradigm frequently requires the use of both languages (Costa et al., 1999, Guo and Peng, 2006, Hermans et al., 1998). Recent studies have addressed this concern through experimental designs using one language alone (e.g., Thierry & Wu, 2007; Wu & Thierry, 2010b), but even these studies are open to criticism, as they have often tested participants in their non-dominant language, which is thought to be mediated through the dominant language until high levels of proficiency have been reached (Kroll & Stewart, 1994). In other words, it is less surprising that the L1 would be active during L2 language use if the L2 requires the L1 for processing in the first place. We argue that a stronger case for language coactivation at the lexical level would be to show coactivation of the L2 in highly proficient bilinguals in a situation that only requires the L1 to be used.

Indeed, several past studies suggest that the non-dominant language may be active in an L1-only context (e.g., Marian and Spivey, 2003a, Spivey and Marian, 1999, Von Holzen and Mani, 2014), but crucially only under conditions that induce bottom-up activation of the L2 (but see Villameriel, Dias, Costello, & Carreiras, 2016, for evidence from hearing bimodal bilinguals). Specifically, these previous experimental paradigms still required overt auditory input from the L1 to overlap with the non-target L2 phonology. In the visual-world paradigm studies by Marian and colleagues, participants saw four objects and heard instructions to pick up one of the objects. The spoken L1 name of the target object (e.g., marka, Russian for “stamp”) overlapped in phonology with the L2 label of one of the four objects presented in the display (e.g., English “marker”). In Von Holzen and Mani (2014), participants saw a picture prime, followed by an auditorily presented target word in the L1. In the critical condition, the auditory L1 label (e.g., Kleid, German for “dress”) rhymed with the L2 label of the picture prime (e.g., English “slide”). In both of these paradigms that showed activation of the non-dominant L2, participants physically heard an L1 target word that overlapped in phonology with the L2 picture label. It is possible that activation of the non-dominant L2 label for the picture prime is triggered by the overt presentation of the phonologically related L1 auditory target. These findings, therefore, limit the scope of language co-activation to bottom-up acoustic/phonetic input from the L1 target word and cannot speak to the automaticity of L2 activation in the absence of overt L1 phonological overlap, as is the case in most real-world situations of L1 language processing (see Von Holzen & Mani, 2012, for similar investigations in children). Connectionist models such as the Bilingual Language Interaction Network for Comprehension of Speech (BLINCS; Shook & Marian, 2013) allow for extensive cross-language interactions within the bilingual lexicon, and yet experimental evidence to-date has not been able to substantiate the extent of these possibilities (see also the Bilingual Interactive Activation Model of Lexical Access, BIMOLA, Léwy & Grosjean, 2008, which allows for both bottom-up and top-down spreading activation).

To address this question, we expand the scope of inquiry to further investigate lexical-level L2 co-activation in the absence of overt phonological overlap between L1 and L2. In the present experiment, we use the intermodal priming paradigm (Desroches et al., 2009, Von Holzen and Mani, 2014) to further test the limits of cross-language activation and cascaded activation in a balanced bilingual population. Similar to other recent EEG studies, we exploit a rhyming relationship between primes and targets to investigate language co-activation (cf., Wu and Thierry, 2011, Desroches et al., 2009, Von Holzen and Mani, 2014). However, to examine whether both languages are active at the lexical level, we ask whether there is bilingual language co-activation in a condition where there is no overt phonological overlap between prime and target.

In the present study, participants were presented with unlabeled prime pictures, followed by an auditory target in the L1 Spanish while they completed an unrelated picture matching task. We recorded participants’ ERP response time-locked to the onset of the L1 Spanish auditory word targets. To investigate the level of language co-activation, we manipulated the phonological overlap between the picture prime label and the auditory target. Specifically, we created rhyming relationships between the prime label in Spanish (L1) or Basque (L2) and the auditory Spanish target or its Basque translation. The relationship between primes and targets was manipulated in five critical conditions (see Fig. 1 for an illustration). To determine whether participants were sensitive to the phonological overlap between prime label and auditory target, we focused on the N400 component. The N400 is a negative deflection in the ERP peaking 400 ms after the onset of a stimulus sensitive to context or reduced processing of the stimulus. We anticipated that if participants were sensitive to our manipulations of the relationship between primes and targets, the N400 component would be reduced, indicating facilitated auditory recognition. Indeed, previous monolingual and bilingual studies have shown a reduced N400 for rhyme words in reading, picture naming, and cross-modal picture-auditory word paradigms (Barrett and Rugg, 1990, Desroches et al., 2009, Grossi et al., 2001, Praamstra and Stegeman, 1993, Von Holzen and Mani, 2014).

Of particular interest to us was the condition in which the Basque label for the silently presented prime image rhymed with the Basque translation of the Spanish auditory target. In other words, after seeing a picture prime of a needle (Basque orratz), when balanced bilinguals hear the Spanish auditory target lápiz (“pencil”), do they co-activate its L2 Basque translation arkatz? We believed this condition would provide the strongest test of the hypothesis that the L2 Basque was co-activated during L1 Spanish processing as it would require co-activation of the L2 label for the silently presented picture prime in the absence of any overt phonological overlap with the auditory presented Spanish target. If the Basque prime image label influences the recognition of the target, we anticipated a reduced mean N400 amplitude for this cross-language prime-target rhyme pair relative to the unrelated prime-target pair.

Section snippets

Participants

28 Spanish-Basque bilingual participants from the Basque region of Spain (7 male, 21 female) participated in the study (M = 23.55 years, SD = 5.04, Range = 19–36). Four additional participants were tested, but were removed from analysis because they had too much noise in the signal. All participants were right-handed and had no history of hearing loss or neurological impairments. We were careful to recruit participants without referring to their bilingual status (e.g., Marian & Spivey, 2003b).

Omnibus ANOVA

Fig. 4 plots the mean distribution of differences for the condition comparisons for both the early (250–400 ms) and late (400–700 ms) time windows. In the 250–400 ms time window, a repeated measures ANOVA with the factors hemisphere, region, and condition revealed significant interactions between condition and hemisphere, F(4,108) = 8.97, p < .001, np2 = 0.25, condition and region, F(12,324) = 3.35, p = .009, np2 = 0.11, and condition, hemisphere, and region, F(12,324) = 2.97, p = .014, np2

Discussion

In the present study, we considered the extent to which the non-dominant language is active while processing the dominant language alone. Using an intermodal priming paradigm, participants saw picture primes followed by an auditory target in their L1 Spanish. We manipulated the extent of phonological overlap between the prime labels and auditory target or target translation. Consistent with the idea that the target is processed easier when there is a complete match between the label of the

Acknowledgments

We thank Eri Takahashi, Clara Furio, Xabier Urizar, Oihana Vadillo, Rebecca Mattli, and Nicole Altvater-Mackensen. Funding: SCB, KV, and NM were supported by the German Excellence Initiative Award to Georg-August-Universität Göttingen (Third funding line: Institutional Strategy). MC was supported by the Spanish Ministry of Economy and Competitiveness (MINECO) [RTI2018-093547-B-I00], the European Research Council [ERC-2011-ADG-295362], and the award “Centro de Excelencia Severo Ochoa

References (47)

  • K. Von Holzen et al.

    Language nonselective lexical access in bilingual toddlers

    Journal of Experimental Child Psychology

    (2012)
  • P. Boersma

    Praat, a system for doing phonetics by computer

    Glot International

    (2002)
  • S. Chabal et al.

    In the mind’s eye: Eye-tracking and multi-modal integration during bilingual spoken-language processing

  • K. Chauncey et al.

    Primed picture naming within and across languages: An ERP investigation

    Cognitive, Affective, & Behavioral Neuroscience

    (2009)
  • A. De Bruin et al.

    The BEST dataset of language proficiency

    Frontiers in Psychology

    (2017)
  • C. De Cat et al.

    Representational deficit or processing effect? An electrophysiological study of noun-noun compound processing by very advanced L2 speakers of English

    Frontiers in Psychology

    (2015)
  • A.S. Desroches et al.

    Investigating the time course of spoken word recognition: Electrophysiological evidence for the influences of phonological similarity

    Journal of Cognitive Neuroscience

    (2009)
  • P.E. Dussias et al.

    Re-learning to parse a first language: The role of experience in sentence comprehension

    Proceedings of the 27th CUNY conference on human sentence processing

    (2014)
  • P.E. Dussias et al.

    The effect of exposure on syntactic parsing in Spanish-English bilinguals

    Bilingualism: Language and Cognition

    (2007)
  • G. Grossi et al.

    Phonological processing in visual rhyming: A developmental ERP study

    Journal of Cognitive Neuroscience

    (2001)
  • T. Guo et al.

    Event-related potential evidence for parallel activation of two languages in bilingual speech production

    NeuroReport

    (2006)
  • D. Hermans

    Word production in a foreign language

    The Netherlands: Unpublished doctoral dissertation

    (2000)
  • D. Hermans et al.

    Producing words in a foreign language: Can speakers prevent interference from their first language?

    Bilingualism: Language and Cognition

    (1998)
  • Cited by (13)

    • Ultra-rapid and automatic interplay between L1 and L2 semantics in late bilinguals: EEG evidence

      2022, Cortex
      Citation Excerpt :

      Increasing globalization and migration flows have transformed our world into a predominantly bilingual one (Bialystok, Craik, & Luk, 2012). Despite the majority of the bilingual population learning their second language (L2) after childhood, and to a lesser degree of proficiency than their native language (L1), most scientific research on cognitive underpinnings of bilingualism has traditionally focused on samples of perfectly balanced and highly proficient bilinguals; in these individuals the two languages are learned simultaneously and to a comparable degree in early life, as a consequence of upbringing in a bilingual household or, even more frequently, in a traditionally bilingual society (e.g., Bobb, Von Holzen, Mayor, Mani, & Carreiras, 2020; Duñabeitia, Dimitropoulou, Uribe-Etxebarria, Laka, & Carreiras, 2010; Gámez & Vasilyeva, 2020; Lallier, Acha, & Carreiras, 2016). Recently, bilingualism research has been shifting towards a more ecological approach, starting to consider the more widespread case of late, sequential, and unbalanced bilingualism, studying bilinguals who learn their L2 after childhood (i.e., after full attainment of their L1) and to a lower degree of proficiency in comparison to L1 (e.g., Gallo, Novitskiy, Myachykov, & Shtyrov, 2021; Novitskiy, Myachykov, & Shtyrov, 2019; Peeters, Dijkstra, & Grainger, 2013; Vega-Mendoza, West, Sorace, & Bak, 2015).

    • Cross-linguistic semantic preview benefit in Basque-Spanish bilingual readers: Evidence from fixation-related potentials

      2021, Brain and Language
      Citation Excerpt :

      A possible approach to overcome such limitations is to use translated words from a second language, since translations hold one of the strongest semantic relationships between two words (Altarriba, 1992). Multiple studies have shown that bilinguals co-activate both of their languages during reading (e.g., Bobb, Von Holzen, Mayor, Mani, & Carreiras, 2020; Dimitropoulou, Duñabeitia, & Carreiras, 2011; Macizo, Bajo, & Martín, 2010; Martín, Macizo, & Bajo, 2010; Perea, Dunabeitia, & Carreiras, 2008). Interactive activation models, such as the BIA (Dijkstra & Van Heuven, 1998; Grainger & Dijkstra, 1992; Van Heuven, Dijkstra, & Grainger, 1998) and BIA+ (Dijkstra & Van Heuven, 2002) models, endorse a non-selective access to a lexicon that is integrated across languages.

    View all citing articles on Scopus
    View full text