ISCA Archive PSP 2005
ISCA Archive PSP 2005

Vowel discrimination in early bilinguals: how plastic?

Anders Højen

Recent neuroscience research has revealed far more plasticity (e.g., in motor, sensory, and auditory functions) than previously thought possible. This has spurred renewed interest in whether the systems underlying human speech perception remain plastic. Recent L2 speech perception research has shown that even early bilinguals may differ from native speakers. This might suggest a lack of plasticity in the perceptual system following L1 acquisition. However, the finding that some early bilinguals show native-like performance raises the question of 'to what extent' L2 perceptual learning is constrained by the L1 perceptual system. To evaluate this question, two reference points are needed. Bilinguals' perception of the L2 should be compared to that of monolingual speakers of the target L2, to reveal how much the bilinguals have yet to learn, and they should be compared to monolingual speakers of the L1, to determine how much they have already learned. This study compared the discrimination of English vowels by English monolinguals, Spanish monolinguals, and early Spanish-English bilinguals (n=20 each). The vowels in one 'easy' contrast were heard as two distinct Spanish vowels by Spanish monolinguals whereas vowels in the three 'difficult' contrasts were heard as a single Spanish vowel. To avoid the ceiling effects often seen in cross-language vowel discrimination research, within-trial F0 variation was introduced into the categorial AXB test used here, and the 64 trials testing all four contrasts were presented in two randomized blocks that differed in inter-stimulus interval (0 vs. 1000 ms). This design was intended to yield low scores on difficult contrasts for Spanish monolinguals, but high scores for English monolinguals. As expected, all three groups obtained high percent correct scores for the easy contrast (means=88-98%). As intended by the design, the Spanish monolinguals obtained near-chance scores for the three difficult contrasts (mean=58%) whereas the English monolinguals obtained very high scores (mean=96%). An ANOVA revealed that the bilinguals obtained significantly higher scores (mean=89%) than the Spanish monolinguals did for all three difficult contrasts in both ISI conditions. They obtained significantly lower scores than the English monolinguals for two difficult contrasts in the 0-ms condition, but no contrast in the 1000-ms condition. Sixty-five percent of individual bilinguals obtained scores falling within 2 SDs of the English monolinguals' mean scores, as against just 2% of individual Spanish monolinguals. The results will be discussed with respect to theories of plasticity.