You are viewing the SNL 2018 Archive Website. For the latest information, see the Current Website.


Poster C59, Friday, August 17, 10:30 am – 12:15 pm, Room 2000AB

Neural discrimination of non-native vowel contrasts by late Spanish-English bilinguals

Daniela Castillo1,2, Eve Higby3, Sarah Kresh1, Nancy Vidal-Finnerty4, Jason A Rosas1, Valerie L Shafer1;1The Graduate Center of the City University of New York, 2Queens College, 3University of California, Riverside, 4Iona College

In this study, we used the mismatch negativity (MMN), an event-related potential (ERP), to address whether native Spanish speakers who learned American English after 14 years of age rely more on spectral or durational information to distinguish English vowels that are non-contrastive in Spanish. Spanish vowels are highly distinct in spectral information but do not differ from each other in duration (Bradlow, 1995). Spanish speakers who learn English late in life have particular difficulty with the American English /ɑ/ (as in “hot”), /ʌ/ (as in “hut”), and /æ/ (as in “hat”) and typically assimilate these three vowels into a single category (Flege et al., 1997). Second language (L2) speech perception can improve with experience, but may be less automatic than perception in the first language (L1). Strange (2011) hypothesized that under difficult task circumstances or when attention is directed elsewhere, L2 learners fall back on their L1 perceptual routines. Even so, Burnham (1986) argues that durational cues are more robust than spectral cues. Thus, we asked whether Spanish L2 learners of English show good discrimination of American-English vowels that differ in duration. ERPs were recorded from 64 electrode sites using an Electrical Geodesic Net from 10 Spanish L2 learners of English and 12 American-English monolingual controls. Speech stimuli were presented in an oddball discrimination task; in one condition /ɑpə/ was the standard (80% of the time) and /ʌpə/ and /æpə/ were deviants (10% each), and in another condition /ʌpə/ was the standard and /ɑpə/ and /æpə/ were deviants. Three natural speech tokens of each stimulus were used that varied in non-target acoustic cues so that listeners would have to rely on phonological information to discriminate the stimuli. Approximately 150 deviants of each stimulus type were presented. Participants performed a visual oddball task to draw attention away from the auditory stimuli. The amplitude and latency of the MMN at fronto-central sites served as dependent measures. Participants were also tested on behavioral discrimination of the vowels via a button press whenever they heard a deviant stimulus, following the ERP study. Spanish L2 listeners were better at behaviorally discriminating /æpə/ and /ɑpə/ than /ʌpə/ and /ɑpə/ (85% and 68% accuracy, respectively). ERPs supported these results with a greater amplitude MMN to the /æpə/ versus /ɑpə/ than to the /ʌpə/ versus /ɑpə/ contrast, but only when /ʌpə/ was the standard and not when /ɑpə/ was the standard. Spanish listeners showed worse behavioral categorization than the American-English controls; in addition, their MMN was smaller in amplitude to /ʌpə/ versus /ɑpə/ than observed for the American English listeners. These findings indicate that Spanish listeners relied more on spectral than duration differences for American-English vowel discrimination. Our future direction will be to examine whether English exposure and use versus English proficiency level lead to improved discrimination of these vowels and increased reliance on spectral and/or durational information at the automatic level of processing indexed by the MMN.

Topic Area: Multilingualism

Back