Is there adaptation of speech production after speech perception in bilingual interaction?

In dialogue, speakers tend to adapt their speech to the speech of their interlocutor. Adapting speech production to preceding speech input may be particularly relevant for second language (L2) speakers interacting with native (L1) speakers, as adaptation may facilitate L2 learning. Here we asked whether Dutch-English bilinguals adapt pronunciation of the English phonemes /æ/ and coda /b/ when reading aloud sentences after exposure to native English speech. Additionally, we tested whether social context (presence or absence of a native English confederate) and time lag between perception and production of the phoneme affected adaptation. Participants produced more English-like target words that ended in word-final /b/ after exposure to target phonemes produced by a native speaker, but the participants did not change their production of the phoneme /æ/ after exposure to native /æ/. The native English speaking confederate did not show consistent changes in speech production after exposure to target phonemes produced by L2 speakers. These findings are in line with Gambi and Pickering's simulation theory of phonetic imitation (Gambi & Pickering, 2013). (PsycINFO Database Record (c) 2019 APA, all rights reserved).

Do semantic sentence constraint and L2 proficiency influence language selectivity of lexical access in native language listening?

We investigated whether language nonselective lexical access in bilingual auditory word recognition when listening in the native language (L1) is modulated by (a) the semantic constraint of the sentence and (b) the second language (L2) proficiency level. We report 2 experiments in which Dutch-English bilinguals with different proficiency levels completed an L1 auditory lexical-decision task on the last word of low- and high-constraining sentences. The critical stimuli were interlingual homophones (e.g., lief [sweet] - leaf /li:f/). Participants recognized homophones significantly slower than matched control words. Importantly, neither the semantic constraint of the sentence, nor the proficiency level of the bilinguals interacted with this interlingual homophone effect. However, when we compared the slow and fast reaction times (RTs), we observed a reduction in the homophone interference effect when listening to high-constraining sentences in L1 for the slow RTs, but not for the fast RTs. Taken together, this provides strong evidence for a language-nonselective account of lexical access when listening in L1, and suggests that even when low-proficient bilinguals are listening to high-constraint sentences in L1, both languages of a bilingual are still activated.

Interlingual lexical competition in a spoken sentence context: evidence from the visual world paradigm.

We used the visual world paradigm to examine interlingual lexical competition when Dutch-English bilinguals listened to low-constraining sentences in their nonnative (L2; Experiment 1) and native (L1; Experiment 2) languages. Additionally, we investigated the influence of the degree of cross-lingual phonological similarity. When listening in L2, participants fixated more on competitor pictures of which the onset of the name was phonologically related to the onset of the name of the target in the nontarget language (e.g., fles, "bottle", given target flower) than on phonologically unrelated distractor pictures. Even when they listened in L1, this effect was also observed when the onsets of the names of the target picture (in L1) and the competitor picture (in L2) were phonologically very similar. These findings provide evidence for interlingual competition during the comprehension of spoken sentences, both in L2 and in L1.

Knowledge of a second language influences auditory word recognition in the native language.

Many studies in bilingual visual word recognition have demonstrated that lexical access is not language selective. However, research on bilingual word recognition in the auditory modality has been scarce, and it has yielded mixed results with regard to the degree of this language nonselectivity. In the present study, we investigated whether listening to a second language (L2) is influenced by knowledge of the native language (L1) and, more important, whether listening to the L1 is also influenced by knowledge of an L2. Additionally, we investigated whether the listener's selectivity of lexical access is influenced by the speaker's L1 (and thus his or her accent). With this aim, Dutch-English bilinguals completed an English (Experiment 1) and a Dutch (Experiment 3) auditory lexical decision task. As a control, the English auditory lexical decision task was also completed by English monolinguals (Experiment 2). Targets were pronounced by a native Dutch speaker with English as the L2 (Experiments 1A, 2A, and 3A) or by a native English speaker with Dutch as the L2 (Experiments 1B, 2B, and 3B). In all experiments, Dutch-English bilinguals recognized interlingual homophones (e.g., lief [sweet]-leaf /li:f/) significantly slower than matched control words, whereas the English monolinguals showed no effect. These results indicate that (a) lexical access in bilingual auditory word recognition is not language selective in L2, nor in L1, and (b) language-specific subphonological cues do not annul cross-lingual interactions.

Roman digit naming: evidence for a semantic route.

Earlier research with monolinguals and bilinguals showed that numbers may be named through both a semantic and a phonological route, depending on the number's language and format (Arabic or verbal), task demands, and naming language. The present study investigated the importance of the semantic route for the processing of a third representation of magnitude, namely Roman digits. Using an interference paradigm, we showed that the processing of Roman target digits is influenced by Arabic digit distractors, both in a naming task and a parity judgment task. Roman digits were processed faster if the target and distractor were of the same magnitude. If this was not the case, processing speed slowed down as the numerical distance between target and distractor increased. This strongly suggests that semantic access is mandatory when naming Roman digits. Implications are discussed for the number processing domain and for models of translation in bilinguals.