In two different longitudinal studies, I am examining factors that influence learning progress at the very beginning stages of late-onset L2 phonological development.
In the first study on L1 English adult learners of Korean, I am investigating change in learners’ L2 production and perception during the first six weeks of language classes. Data from three experiments (imitation, elicited production, forced-choice identification) are being compared in order to probe learners’ knowledge of the phonetic space of novel L2 categories (the lenis, fortis, and aspirated stop consonants of Korean), to test their implicit knowledge of obligatory phonological alternations they have not been explicitly taught (e.g., obstruent nasalization), and to examine the effectiveness of certain non-linguistic variables (e.g., basic psychoacoustic perceptual acuity, attitudinal factors) in predicting acquisition outcomes for individual learners. So far, findings from elicited production experiments have shown that learners vary widely with respect to the phonetic spaces they construct for the new L2 laryngeal categories, that the patterns are not fully predicted by cross-linguistic perception results, and that some learners continue to perseverate the same non-native patterns based upon erroneous ideas about the nature of the L2 contrast (Chang, 2010a, b).
In the second study on L1 English adults being exposed to Mandarin, I am collaborating with Anita Bowles and others to investigate the development of L2 tone contrasts during a short word learning paradigm. We have found that pitch-specific perceptual abilities are the best predictors of tone learning (more powerful than measures related to musical aptitude, musical experience, general foreign language aptitude, and general cognitive ability), which argues in favor of a feature-specific approach to language aptitude (Bowles, Chang & Karuzis, 2016). We have also found that coarticulatory variability in tone contour has a significant effect on the learning of tone contrasts in different contexts (Chang & Bowles, 2015).