While heritage research has progressed across various domains few studies have explored the perceptual patterns of heritage learners. This study investigates how changes in voice onset time (VOT) in Korean intervocalic stops (i.e. /k p t/) affect the perception of heritage language learners (i.e. Korean Americans) compared to that of native Korean listeners and second-language (L2) learners of Korean. Fourteen adult native speakers each of Korean, English, Mandarin and Heritage Korean participated (i.e. 33 females and 23 males 56 in total). All Chinese, English and Heritage participants were students enrolled in a beginning-level Korean language course at the university when the experiment was conducted. A discrimination task was created of ten synthetic /C1V1C2V2/ stimuli that differ in VOT of C2 by 20ms steps (ranging from -90ms to +90ms) for three stops to test the differences among language groups (C1 = /k p t/ C2 = VOT manipulated stops of C1; V1 and V2 = [u]). Three test syllables of C1V1C2V2 were uttered by a female Korean speaker. Each participant heard two repetitions of each stimulus for a total of 60 tokens (10 continuum * 3 stop * 2 repetition) in a randomized order. A two-second interval was included between each stimulus. The participants were instructed to listen to the stimuli and they were provided with an answer sheet and asked to circle “O” if the pronunciations of the two consonants in the C1V1C2V2 sequence were the same or to circle “X” if the two consonants were different to their ears. The results showed that L2 Chinese and English listeners displayed the typical categorical perception in VOT variations of Korean intervocalic stops with a sharp rising curve at + 30ms (Chinese) and +50ms (English) and a peak at +90ms of VOT. Heritage learners and Korean native listeners however did not demonstrate this categorical perception and perceived the two consonants in C1V1C2V2 as the same for most stimuli regardless of the VOT length variation of C2. Heritage learners were predicted to be accurate in discriminating the VOT variations like native speakers of English which is their dominant language. However their perception is found to align largely with that of native Korean speakers. The unexpected insensitivity to the VOT variation of Heritage listeners suggests that the linguistic behavior of Heritage listeners resembles that of native listeners in their abstraction away from phonologically irrelevant phonetic details and their perception has been optimized to signal to only those aspects that are relevant to the phonemic cues in the target language (voicing is phonetic but not phonemic in Korean). This can be interpreted as an indication that the heritage language phonology and phonemic system is already built in their perception in a similar way to that of the native listeners while L2 learners in the same beginning Korean class primarily focused on the phonetic variations. In addition, it is suggested that the heritage language phonology supersedes more than the phonetic or phonemic experience of their dominant language in their heritage language perception.

View slides here.