Bilingual cross-language activation has been the subject of substantial research over the last two decades. Evidence supporting the idea that when a bilingual processes speech in one language, the other language is actived has been fundamental in shaping our understanding of whether these languages are discrete systems, or represented jointly in an integrated lexicon.
A growing body of evidence for bilingual language co-activation has used electrophysiological methods (ERPs) to reveal implicit activation of the native (L1) language in the absence of conscious awareness and behavioural effects. Here, I investigate how two aspects of spoken language phonology, namely accent and lexical stress, influence activation of the native language when highly-proficient bilinguals process second language speech. Furthermore, based on a review of key aspects of recent literature on bilingual language co-activation and phonological processing, I propose that a review of our current understanding of the ERP correlates of phonological mapping may be required.
In studies 1 and 2 we explored how accent influences activation of L1 phonological and semantic representations. In Study 1, Welsh-English bilinguals made relatedness decisions on English word pairs, unaware that some concealed a phonological overlap via Welsh translation. We found a facilitatory effect of accent, such that priming between these otherwise unrelated word pairs was facilitated by the presence of a Welsh accent. Results suggest that for these Welsh-English bilinguals, activation of the L1 was elicited by the presence of a Welsh accent. In Study 2, we went on to investigate whether this same facilitatory effect occurred for translation links at the semantic level. German-English bilinguals listened to English sentences produced by a native English or native German speaker that ended in a visually presented target word. In the critical condition, target words that were unexpected in English translated to polysemous German words. The alternate meaning of these words was congruent in the context of the English sentence. We found no effect of accent, nor a difference in responses to correct-via-polysemy targets as compared to a control condition. Results are discussed both in terms of study design and models of bilingual language activation.
In studies 3 and 4 we investigated how suprasegmental lexical stress influences L1 activation. Languages that feature stress are generally divided into those in which stress is variable, that is, occuring on any given syllable, or fixed according to certain linguistic rules. This difference in stress systems is thought to affect the mechanisms by which bilinguals process lexical stress. In Study 3 we tested native English and Welsh-English bilingual participants to determine whether the presence of Welsh stress patterns heightened the activation of L1 representations. Our findings were inconclusive. Based on limitations in the design of Study 3, we designed and conducted Study 4, in which we tested how stress influences L1 activation, and incorporated an accent manipulation to explore how the two factors may differentially influence this effect. We found that whilst accent heightened L1 activation, lexical stress did not. However, we found a significant main effect of stress compatible with prior findings, suggesting that stress-deaf individuals process lexical stress on the basis of stored pre-lexical templates.
Considered together, our results suggest that suprasegmental cues such as accent and lexical stress play a role in bilingual language processing, and that this role may be underrepresented in current models of bilingual lexical access. Furthermore, our results highlight that the differential influence these factors have may depend on language background, proficiency and exposure, suggesting that bilingual word recognition mechanisms may differ substantially across different bilingual populations.