In joint work with Yao Yao, Erin Haynes, and Russell Rhodes, I have investigated how American heritage speakers of Mandarin Chinese differ from late-onset second language (L2) learners of Mandarin and English in both their Mandarin and English phonologies. Our findings have shown that heritage speakers generally outperform late-onset L2 learners in the production of both language-internal phonological contrast as well as cross-linguistic phonetic contrast (Chang, Haynes, Yao & Rhodes, 2009, 2010; Chang, Yao, Haynes & Rhodes, 2011).

In our most recent work, we have examined the acoustic and perceptual properties of heritage speakers’ tone production, as well as the demographic classification of heritage speakers on the basis of their speech. Our results (reported in Chang & Yao, 2016, 2019) suggest that heritage speakers approximate native-like production more closely than L2 speakers in some, but not all, aspects of tone (namely, the pitch contour of Tone 3, durational shortening in connected speech, and rates of Tone 3 reduction in non-phrase-final contexts) and show higher levels of tonal variability than native or L2 speakers. Further, heritage speakers’ tones differ from both native and L2 speakers’ in terms of intelligibility and perceived goodness, and heritage speakers are more difficult to classify demographically than native or L2 speakers. In certain cases, such as production of ‘neutral’ tone in non-obligatory contexts, heritage speakers are actually less intelligible to L1 (mainland) listeners than are L2 speakers, which may be due to differences in dialectal and educational experience. In short, the findings suggest that early heritage language experience can, but does not necessarily, result in a phonological advantage over L2 learners with respect to tone, and add support to the view that heritage speakers are language users distinct from both native and L2 speakers.

In a different strand of research with Sunyoung Ahn, Robert DeKeyser, and Sunyoung Lee-Ellis, I have examined how American heritage speakers of Korean (L1 Korean-L2 English bilinguals) differ from Korean speakers living in Korea (native controls) with respect to perception of their L1 (i.e., their heritage language). Our findings, published in Language Learning (Ahn, Chang, DeKeyser & Lee-Ellis, 2017), showed that bilinguals were less accurate than control listeners on L1-specific contrasts, and their accuracy was significantly correlated with their age of reduced contact with the L1, an effect most pronounced for the L1 contrast most dissimilar to L2. These findings suggest that the earlier bilinguals are extensively exposed to L2, the less likely they are to perceive L1 sounds accurately; however, this relationship is modulated by L1-L2 similarity. Additionally, a turning point in both L2 acquisition and L1 attrition of phonology appears to occur at around 12 years of age.

Recently, I wrote a review chapter synthesizing the research in the area of heritage language phonetics and phonology, which appears in the new Cambridge Handbook of Heritage Languages and Linguistics (Chang, 2021). Yao Yao and I also wrote a chapter for the forthcoming volume The Phonetics and Phonology of Heritage Languages, edited by Rajiv Rao, describing an approach to analyzing individual differences among heritage speakers using multiple factor analysis (Chang & Yao, 2024). (Comments welcome!)