Individual Differences in Non-Native Phonological Contrast Learning: The Role of Perceptual Sensitivity to Sub-Phonemic Variation in Native Categories

Abstract

The current study explored individual differences in sensitivity to sub-phonemic variation of acoustic cues in the perception of a native language (L1) category in order to test the hypothesis that second language (L2) learners’ different sensitivity to the L2-relevant acoustic dimension in L1 perception could explain individual variability in nonnative phonological contrast learning. In addition, this study also investigated whether the modified High Variability Phonetic Training (HVPT) paradigm could aid in nonnative phonological contrast learning. The cue-attention switching training was added to the typical HVPT paradigm with multiple talkers, expecting to reallocate learners’ attention away from the less relevant acoustic dimension to the more informative acoustic dimension in the perception of the target nonnative contrast(s). The present study targeted two groups of learners with different L1 backgrounds: naïve adult English learners of Korean and intermediate adult Korean learners of English. The multiple HVPT sessions trained English learners of Korean on a Korean three-way laryngeal contrast in stop (/p’/-/p/-/ph/) and trained Korean learners of English on three English vowel contrasts, /i/-/ɪ/, /ɛ/-/æ/, and /ʊ/ -/u/.The Visual Analogue Scaling (VAS) task measured English adult listeners’ sensitivity to sub-phonemic acoustic details in the perception of English stop voicing contrast with a stimuli continuum of English voiced and voiceless stops (/b/-/p/) varying in VOT and f0 at vowel onset. For Korean adult listeners, the AXB oddity task quantified learners’ sensitivity to within-category differences induced by spectral and duration cue changes, using a set of stimuli belonging to the Korean /i/ vowel but with different spectral and duration properties. The results of the HVPT training in experiments 1 and 3 revealed that in both groups, L2 learners with higher sensitivity to L2-relevant acoustic cues in L1 perception had an initial advantage in L2 contrast learning and showed more nativelike cue utilization during and after the HVPT. On the other hand, learners with less sensitivity to the “right” acoustic cues failed to systematically use those cues in perceiving the target L2 contrast(s). Learners who received the modified HVPT with the cue-attention switching training with L1 stimuli in experiments 2 and 4 demonstrated more native-like use of acoustic cues in L2 perception than learners who received only the typical HVPT with multiple talkers. English learners of Korean with relatively less sensitivity to f0 cues in the perception of English voicing contrast performed similarly to those with relatively high sensitivity to f0 cues. For Korean learners of English, the benefit of the cue-attention switching training was observed in learning the English /i/-/ɪ/ contrast, but not in more challenging /ɛ/-/æ/ and /ʊ/ -/u/ contrasts. Korean learners of English with the cue-attention switching training showed more reliance on spectral than duration cues like English native listeners. This study showed the relation between individual differences in sensitivity to sub-phonemic acoustic details in L1 and the nonnative novel phonological contrast learning and a possible type of training to overcome disadvantages due to the individual differences. The results suggest the transfer of L1 cue sensitivity to L2 cue utilization. That is, how successfully L2 learners progress to become more nativelike listeners can be predicted in terms of to what degree they have sensitivity to the L2 informative acoustic cue in L1 speech perception. This implies that individual differences in the L2-relevant cue sensitivity may determine the initial stage of learning and to what extent learners can benefit from L2 training. Moreover, this study emphasizes the importance of considering individual differences to predict L2 learners’ learning outcomes and provide appropriate L2 training to learners whose perceptual abilities may place them at a disadvantage. The VAS and AXB oddity tasks showed possibilities as pretraining assessments to predict the acquisition of L2 phonological contrasts and L2 cue-weighting strategies.