It is well established that Korean listeners tend to perceptually repair illicit consonant clusters by inserting an epenthetic vowel. This study investigates how Korean listeners with differing exposure to English perceive initial consonant clusters in English words, focusing on their temporal threshold for detecting an inserted vowel. Using a computational model of speech production, we generated an 11-step continuum of an epenthetic vowel between the consonants of four English words containing voiced stop + liquid sequences (bleed, breed, glass, grass) by varying the duration between stop release to liquid onset from 40 to 130 ms. Participants included seven Korean listeners living in the U.S. for over 2 years, six Korean listeners residing in Korea with no overseas experience, and a control group of native English listeners. In a forced-choice task, participants detected whether a vowel was present between the two consonants. All listener groups reliably detected the vowel when its duration exceeded 70–80 ms, except for the word glass. Both Korean groups consistently failed to perceive the vowel in glass even at durations around 100 ms. These results suggest that Korean listeners require longer vowel durations than native English speakers to perceive epenthesis, especially in velar + lateral clusters.
Kim et al. (Wed,) studied this question.