Phonetic patterns of Koreans producing English vowels, /i, I, u, U/, in words and in sustained phonation

Ten native-Korean male speakers and ten native-American-English male speakers produced the target vowels /i,I,u,U/ in both words and sustained phonation. These recorded speech samples were judged by a panel of three expert judges for vowel identification. They also were analyzed by the experimenter to determine the formant fi-equencies (F1 and F2) and the vowel duration. All measurements had a high degree of reliability. Korean speakers produced the phonemic vowels, /i/ and /u/ with a high accuracy similar to that of native English speakers. Korean speakers produced the non-phonemic vowels, /I/ and /U/ with very low accuracy (high number of errors); these errors were not observed in the Americans. Korean errors for non-phonemic /I/ and /U/ were predictable errors, i.e., /i/ for /I/, and /u/ for /U/. In addition, Korean speakers had more errors for sustained vowel phonation than words. The first formant (F1) was a better predictor for both the Koreans' phonemic and non-phonemic vowel identification. The second formant (F2) and vowel duration were inconsistent predictors of vowel identification. The Korean phonetic patterns provided information for ESL teaching strategies.