| Popis |
First language (L1) phonotactic knowledge shapes perception and production in the L1; illicit sequences are perceptually repaired and show low contrast sensitivity5. Exposure to a second language (L2) with more permissive phonotactics can restructure the L1 system3,4. However, no previous studies have investigated whether L1 phonolexical representations may be updated (i.e., made more/less precise) as a result of this restructuring. Studies on the lexical representation of BP mid-vowels (/e~?, o~?/) find variability in their encoding precision: even L1 BP speakers sometimes accept variation in mid-vowels, even when it results in non-words1. The source of this variation can be twofold: (Hypothesis A) overall leniency to alternate forms due to exposure to English-accented Portuguese (i.e., forms with common English-based mispronunciations), or (Hypothesis B), acceptance of alternate forms only occurs under specific phonotactic conditions. However, this phonotactic effect may be reduced with exposure to English (phonotactics). To determine the potential role of phonotactics on the lexical representation of BP mid-vowels, we examined stressed mid-vowels in three contexts2: (a) an obligatory phonotactic constraint: only close mid-vowels precede a nasal consonant (“N” context; e.g., ‘ônibus’ /'o.ni.bus/); (b) a phonotactic tendency: open mid-vowels are more likely in antepenultimate-stressed syllables (“Ante”; e.g., ‘ótimo’ /'?.t??i.mu/); and (c) no clear phonotactic tendency (“Else”, e.g., ‘neto’ /'n?.tu/). Importantly, English has no constraint on pre-nasal mid-vowels, nor a tendency regarding antepenultimate mid-vowels. Participants (n=60 L1 BP adults with sequential L2 English) completed auditory lexical decision (LDT) and preference judgement (PJT) tasks. Stimuli consisted of real target words containing a mid-vowel in the stressed syllable (TW; e.g., ‘ônibus’ /'o.ni.bus/), target non-words created from the TWs by switching the mid-vowel (TNW; e.g., /'?.ni.bus/), and additional control (non-)words. Participants also completed an AX discrimination task, a hearing test, and a background questionnaire to assess exposure to English, and to English-accented Portuguese. Hypothesis A predicts a generalized effect of exposure to English-accented Portuguese, with more exposure leading to lower accuracy in rejecting non-words regardless of phonotactic context. Yet, participants’ preference would still favor the real words (TW) over TNW. Hypothesis B (phonotactics-based effects) predicts a context effect (not overall leniency) due to perceptual repairs. Here, participants would accept alternate forms (i.e., be less accurate for TNW) more in the N vs. other contexts, and (perhaps) in the Ante vs. Else context. Additionally, Hypothesis B also predicts greater selection of ‘Both are equal’ in the PJT in the N context. In both tasks we would expect an interaction with English exposure: greater exposure could reduce perceptual repairs and increase contrast sensitivity, leading to higher accuracy in the N context. We found no effect of exposure to English-accented Portuguese in either task. A clear context effect emerged: in the LDT, accuracy in the N context was much lower than in other contexts, and there was an interaction effect with English exposure: accuracy improved in the N context with increased English exposure (Figure 1). In the PJT, the ‘Both are equal’ option was chosen much more in the N context than the other contexts, but the likelihood of preferring the TW increased with English exposure (Figure 2). We interpret the results as supporting hypothesis B: phonotactics affected lexical encoding. Additionally, with increasing English exposure estructuring of the L1 phonotactic grammar occurs, leading to (slowly) increasing precision of the L1 phonolexical representations, due to the new pressure to encode contrasting mid-vowels.
|