Some Phonetic Characteristics of Mid-vocalic Lax Stops and Pre/Post-stop Vowels in Korean

다국어 초록 (Multilingual Abstract)

It has been claimed that Korean mid-vocalic voiceless unaspirated lax stops are phonetically realized with voicing throughout the oral closure phase. Acoustic measurements were undertaken to examine the claim with four Korean native speakers using /V_1CV_2/ words where the vowel (V_1 = V_2) was /i,a,u/ and the C was voiceless unaspirated lax stops /p,t,k/. Findings: (1) During mid-vocalic stops /k/ and /p/ the vowel /u/ was accompanied generally by a significant increase in voice cessation time as percentage of the oral closure interval (PCT) than the vowel /a/, regardless of subject-dependent, (2) The effcts of vowels on PCT were significantly greater in mid-vocalic /k/ than /p/, regardless of subjects, (3) The mean PCT, averaged across six tokens, ranged from 17 % to 100 %, giving overall mean 61 % in which the standard deviation was ±30, and (4) Overall 67 % of the total of mid-vocalic unaspirated lax stops were produced with a substantial period of devocing and voicing lag. Considering these results, it is difficult to agree with the existing claims that Korean voiceless unaspirated lax stops are phonetically realized with voicing throughout the oral closure phase. Other phonetic variables, including the durations of pre/post-stop vowels, voice onset time, voice cessation time, and the duration of oral closure, were measured.
Keywords: voicing, Korean intervocalic lax stop, Korean vowel, duration
The duration of V_1 and V_2. Across six tokens, the mean durations of V_1 ranged from 52 ms to 153 ms, giving overall mean 20 ms, averaged across six tokens, four subjects, three vowel types, and three stop types. The range of the mean (across six tokens) duraion of V_2 was from 196 ms to 292 ms., giving overall mean 252 ms. Overall (across four subjects, three vowel types, three stop types and six tokens), the mean ratio of the duration of pre-stop vowels and post-stop vowels was 1 (V_1) : 2.5 (V_2). In general, the inconsistent resutls of V_1 and V_2 resulted genrally from between-subject differences and ocassional differences.
The duartion of oral closure (DOC). Across six tokens the mean DOC of mid-vocalic unaspirated lax stops in unstressed position ranged from 50 ms to 97 ms depending upon subjects, with overall mean 70 ms. The effects of the place of articulation for stops and vowel types on the DOC was generally inconsistent depending upon subjects. S_1 produced significantly greater DOC for /p/ than /t/ and /k/, mean while S_4 yielded significantly less DOC for /p/ than /t/ and /k/. S_2 and S_3 showed the insignificant effects of the place of articulation on DOC.
Voice onset time (VOT). VOT was generally dependent of subjects, and there were considerable effects of vowel types on VOT. Across three stops, three vowels and six tokens, the range of mean VOT was from 4 ms (S_2) to 17 ms (S_1), with overall mean 10 ms (SD = ± 10). Voice cessation time as percentage of the oral closure interval (PCT). During mid-vocalic stops /k/ and /p/ the vowel /u/ was accompanied generally by a significant increase in PCT than the vowel /a/, regardless of subjects, whereas in mid-vocalic alveolar stop /t/ the effects of vowels on PCT were subject-dependent. The effects of vowels on PCT were significantly greater in mid-vocalic /k/ than /p/, regardless of subjects. The mean PCT, averaged across six tokens, ranged from 17 % to 100 %, giving overall mean 61 % which the standard deviation was ±30. In mid-vocalic unaspirated lax stops where the vowel was /a/, overall (across four subjects, three stops, and 6 tokens) 86 % of the toal of mid-vocalic stops were produced with a substantial period of devoicing and voicing lag (see Table I, II and III). This is agreeable with Kim's findings (1987) in which 98 % of the total of mid-vocalic stops were manifested with a substantial period of devoicing and voicing lag. However, overall 67 % of the total of mid-vocalic unaspirated lax stops were produced with a substantial period of devocing and voicing lag. Considering these results and Kim's findings (1987), it is difficult to agree with the earlier claims that Korean voiceless unaspirated lax stops are phonetically realized with voicing throughtout the oral closure phase (e.g., Zong, 1973; Kagaya, 1974; Han and Witzman, 1970; Kim, 1965; Lee, 1969; Heo, 1972; Ladefoged, 1973). The voicing during intervocalic lax stops appears not to be allophonic, i.e., rule-governed, but it is rather one of potential phonetic correlates of the intervocalic lax stops. The range of the mean values for timing variables, overall mean values, and the mean ratios may be useful in programming a communication system.