This research is to analyze the contemporary tendency of intervocalic deaspiration phenomenon in four strongly aspirated Korean stops and affricate, /ph, th , ch, kh/.
The sound value of /ph, th , ch, kh/ is fully represented in the word-initial posi ...
This research is to analyze the contemporary tendency of intervocalic deaspiration phenomenon in four strongly aspirated Korean stops and affricate, /ph, th , ch, kh/.
The sound value of /ph, th , ch, kh/ is fully represented in the word-initial position. They are neutralized as unreleased [p, t, t, k] in the syllable-final position. In the intervocalic position also they are supposed to be pronounced as strongly aspirated. The contemporary tendency is that the strongly aspirated sounds are deaspirated even in the intervocalic position. Moreover, the coalition phenomenon that a plain stop or affricate, /p, t, c, k/, to produce [ph, th , ch, kh] does not occur in many speakers. The coalition even after a complex coda of a verb stem which ends with /h/ as in /kkinh+ta/ and /kkulh+ta/ to [kkintha] and [kkultha] does not occur.
For this research, 12 different groups of 10 people each were involved in the experiment. Forty people from three different Korean dialect speaking areas were in the experiment. Each group had 10 subjects, who were males and females of in their 20’s and 50’s. Each subject read 52 sentences and each sentence had one token of aspiration/deaspiration information. Thus the total number of the tokens was 3 (dialects) x 2 (male/female) x 2 (generations) x 10 (people each) x 52 (sentences) = 6240.
The result was as follows. First, the deaspiration rate among the four aspiration consonant cases in nouns was: /kh/ >> /ph/ >> /ch/ >> /th/ (88 >> 57 >> 23 >> 12%). This was the same in the three dialect groups, males and females, and the 20’s and the 50’s.
Second, the rate of the blockage of the coalition, the change of a plain stop or affricate plus /h/ to an aspirated sound (i.e., k+h, h+k --> kh), is also considerably high in the three dialect groups, males/females, and in younger/older generations. The average rate was 31% in nouns, 25% in verbs, and 29% in nouns of the redundant position (i.e., kh+h --> kh).
Third, the rate of deaspiration in verb conjugation was not very significant. Only 1% of the tokens was found for a verb followed by a vowel, and 6% for a verb preceded by /h/ for coalition.
Fourth, the coalition rate of /nh/ and /lh/ of the verb stem followed by a plain obstruent was 41, 44, 28% and 9, 58, 22%, respectively in three different dialect groups. A notable finding is that Kyungsang speakers lost aspiration in the /lh+C/ position much more than the other two dialect speakers did. This was the same in the total rate of deaspiration phenomenon as it is Kyungsang Speakers 38%>> Cholla Speakers 33%>> Seoul Speakers 30%.
Fifth, there was a significant difference between the two generations. The young generation produced 837 deaspiration sounds out of 3120 cases (27%), while the older generation produced 1260 deaspiration sounds (40%).
Sixth, there was no significant difference in deaspiration between males and females. The male subjects produced 1082 deaspiration sounds out of 3120 cases (35%) while female subjects produced 1017 sounds (33%). A notable point is that Seoul female speakers significantly lower rate of deaspiration cases (26% vs. 35%).
The phonological issues to deal with from the result of the experiment are various.
First, it is found that Kyungsang speakers produce more deaspirated (less aspirated) sounds than Cholla or Seoul speakers. This is an interesting point in that it has been a general impression that Kyungsang speakers’ utterances use more aspirated sounds, as in /hatəla/ --> [khadəra] ‘I heard that…’, than any other dialect speakers, while Cholla speakers lose many aspirated sounds, as in /kilih+co/ to [kirico ] not to [kiricho] ‘Yes, sure’.
Second, the strongly aspirated obstruents not only lost their aspiration but also reduced to voiced sounds in the intervocalic position. The change from /Ch/ --> [C] --> [Cvd] is unlicensed in standard Korean phonology, but this did occur in historical stages or in a series of changes. This can be interpreted as losing strong aspiration triggers intervocalic voicing, a regular phenomenon formerly limited to unaspirated lax (plain) obstruents. This trend causes to lower the frequency of the occurrence of strongly aspirated sounds prevocalically and in coalition, i.e., very few kh-ending words are left.
Third, OT analysis does not have explanatory simplicity in accounting for various optput forms from the same input. For instance, a coronal consonant produced many different phonetic output forms such an alveopalatal stop, alveolar stop, alveolar affricate, or dental affricate, voiced stop, voiced affricate, and even a fricative. The input form /salkach+il/ showed its output forms as [salkachil], [salkathil], [salkacil], salkasil], [salkazil], and so on. This causes many violations on Ident-IO or Faithfulness constraints. Serial derivation, however, has no problem in giving an explanation for the deaspiration phenomenon. Deaspiration feeds intervocalic voicing to produce the correct phonetic representations. The deaspiration case of /muliph+i/ --> [muriphi] --> [muripi] --> [muribi] ‘knee-nom’. This phenomenon is supported by some words with coronal deaspiration in the historical change like /kacʰ/ --> [kacuk] --> [kajuk] ‘animal skin/hide’.
Fourth, velar and labial tends to be deaspirated much more frequently than coronal consonants and labial consonants. This conforms to the hypothesis that peripheral elements (labials and velars) are more apt to change their laryngeal feature than default elements (coronals) as labials and velars are more characteristic even without the laryngeal feature. The more neutral sounds, [th] and [ch], might have to keep their laryngeal feature to carry functional load.
Fifth, the Voice Onset Time (VOT) is not responsible for the difference in the degree of deaspiration. In various research on the VOT of Korean obstruents, the length is in the order of [ch >> kh >> th >> ph]. The rate of deaspiration is not correlated with the length of the VOT.