I am conducting a cross-linguistic investigation into the perception of unreleased stops by speakers of languages in which final stops are obligatorily unreleased (i.e., canonical), variably unreleased, or never unreleased. The project aims to shed light on the role of frequency information vs. canonicity information in shaping perceptual biases, as well as the manner in which perception of a second language is influenced by these biases interacting with structural constraints of the native language.
One set of findings with non-native listeners of English from a native Korean background suggests that canonicity of a particular form can result in a “non-native advantage” in speech perception (Chang & Mishler, 2012). Moreover, this native-language transfer benefit perseveres even when exposure to the language is attenuated in adolescence and adulthood, as in the non-canonical acquisition profile of heritage speakers (Chang, 2016).
Recently, I have been developing the notion of perceptual attention to the auditory cue, guided by the cue’s relative functional load across the language, as the locus of native-language transfer to non-native perception (Chang, 2018).