Vowel duration changes depending on voicing feature of following consonants by
Korean learners of English
Dong빲Jin Shin
*Abstract
The current study investigates whether the vowel length is affected by the voicing feature of the following consonants, and also how this phenomenon is differently demonstrated by three speaker groups (English native speakers, English bilinguals, and Korean L2 speakers).
Each group of subjects produced given target words embedded in a carrier sentence with three repetitions. The results showed a significant group difference as well as a significant interaction between group and voicing feature. English native speakers and bilingual speakers showed a significant difference in the length of the preceding vowel depending on the voicing feature of the following consonants. Korean L2 speakers also showed a significant difference as a function of the following consonant voicing feature, however, the degree of difference was not as large as the other two groups. As a result, the study showed that the non빲native빲 likeness of the vowel lengthening effect as a function of the voicing feature of the following consonant by Korean learners might be because L2 learners failed to produce voiced consonants in a native빲like way, resulting in foreign accented speech.
Key words : L2 speech, vowel duration, voicing feature, coarticulation
34)
* Jeonju University Assistant Professor
1. Introduction
Learning a new language can be challenging for second language learners.
Especially, for adult L2 learners, it is difficult to obtain native빲like pronunciation of the second language compared to people who started to learn the target language at younger age. Among other abilities, learning pronunciation in a native빲like way is especially challenging for L2 speakers, and in order to acquire the sounds that do not exist in their native language, they have to be exposed lots of time to the target language. Cross빲linguistically, vowel duration can be changed by many phonetic environments such as vowel height, vowel tenseness, and the voicing feature of the following consonants. In addition, suprasegmental features such as speech rate, lexical stress, focus, and rhythm structure can also affect vowel duration. Specifically, low vowels have longer duration than high vowels and tense vowels have longer duration than lax vowels. Also, slower speakers produce a vowel with longer duration than fast speakers, and vowels in stressed position have longer duration than in unstressed position.
Previously, a number of studies have investigated the effect of the progressive/regressive coarticulation. In particular, the correlation between the length of vowels and the voicing feature of following consonants have been studied by many scholars (House & Fairbanks, 1953; Denes, 1955; Peterson &
Lehiste, 1960; House, 1961; Lehiste, 1972a; Klatt, 1976; van Santen, 1992).
The difference of the length of the preceding vowels due to the contrastive voicing feature in English is greater than Korean, since Korean has voiced consonants only as allophones in intervocalic position. If the vocalic difference between following voiced and voiceless consonants is a characteristic of English, rather than in Korean, then, can this phenomenon in English be acquired by Korean learners of English? If the vocalic difference due to the
voicing feature of the following consonant is language specific, will English빲 Korean bilingual speakers show a similar pattern as native English speakers, given the fact that the amount of exposure to English by bilingual speakers is smaller than native English speakers? Current study examined how Korean L2 learners of English show this difference in their production.
Present study investigated how the vowel duration of monosyllabic word is differed according to the voicing feature of following consonants, especially with fricatives and affricates. Furthermore, this study will also examine how the differences in vowel lengths are changed according to the different speaker groups.
2. Methods
1) Subject
A total of ten speakers participated in this study. In native English group, there were four native English speakers, among which three were from North America (Toronto, Missouri, and Colorado) and one from Sidney, Australia. In Korean빲English bilingual group, three speakers, who lived in English빲speaking environment more than 10 years, participated in this study. In this group, a male speaker lived in New York for 20 years and another male speaker lived in the same region for 10 years. A female speaker lived in Philippine more than 10 years and studied in international school. Four subjects (three females, one male), whose major is in English, participated as the group of Korean learners of English. The reason for designating the subjects as English majors is to minimize the problems due to pronunciation errors as some words have quite unfamiliar pronunciation. None of these subjects reported any articulatory or hearing problems.
2) Procedure
Monosyllabic minimal pair words containing voiced and voiceless sibilants in coda position were prepared for the production test. Subjects inserted all target words in a given carrier sentence, ‘Say "______" again' with three repetitions.
The target words are as follows in Table 1.
<Table 1> minimal pair words
Target sounds Minimal Pairs
Fricatives
/f빲v/ duff 빲 dove
leaf 빲 leave
/ɵ빲ð/ teeth 빲 teethe
/s빲z/ bus 빲 buzz
kiss 빲 kids
/ʃ빲ʒ/ cash 빲 cas(ual)
me.sher – mea.sure
Affricates /ʧ빲ʤ/ rich 빲 ridge
search 빲 surge
Among the target words, the pair of cash and cas(ual) were chosen because there was no counterpart of ‘cash’ in English so we used part of word casual [kæʒuəl]. In the case of fricative, ‘mesher’ is not a common word in English.
However, we choose this words to make a minimal pair to ‘measure’ (van Santen, 1992). Furthermore, as ‘Mesher’ is a proper name in English, we considered that the participants would be able to produce this word correctly. In addition, target words like ‘search’ and ‘surge’ were measured from the onset of the rhotic vowel to the beginning of the sibilants, since separating vowel from liquid sound is almost impossible. In the case of rhotic, although the consonant /r/ could have an effect on the vowel duration, in this experiment, we regarded the rhotic vowel as a ‘phoneme’ and it would be revealed as ‘phone’.
All the recordings were made at the recording studio in one of the universities
located in Seoul. As apparatus, TASCAM (DA빲45HR) DAT recorder and an AKG microphone were used. All recordings were analysed by using Praat (6.0.36).
In order to see whether the vowel duration differs in terms of the voicing feature of following consonant and length of exposure to English, we conducted repeated measures of two빲way Anova in R (version 3.4.3) with subject group (English native speakers, English bilingual speakers, and Korean learners of English) as between subject variable and voicing features (voiced and voiceless) as within subject variable. Vowel duration was considered as a dependent variable.
3. Results
1) Analysis of vowel duration
The mean values of vowel duration for each target word of English native speakers, English bilingual speakers, and Korean learners of English are listed in Table 2. With respect to the vowel duration, we found a main effect of voicing feature [F(1,16) = 16.98, p < .01], indicating that vowel duration is significantly changed according to the voicing feature of following consonant.
But we also found a significant effect of subject [F(2,23) = 5.983, p < .01], indicating that there was a group difference in producing vowel duration. We also found a significant interaction between group and voicing feature [F(2,23)
= 3.260, p < .01], indicating that vowel difference as a function of voicing feature were different according to the speaker group. Figure 1 displays the interaction of vowel length changes among three different groups.
<Table 2> Vowel duration of English native speakers, English bilingual speakers, and Korean learners of English
(unit: m/sec) English
Native speakers
English Bilingual speakers
Korean learners of English
duff 133 143 156
duv 212 226 196
leaf 137 155 142
leave 214 236 197
teeth 122 137 117
teethe 217 227 151
bus 130 153 164
buzz 201 223 202
kiss 109 128 101
kids 155 175 136
cash 170 195 121
cas(ual) 160 180 132
mesher 105 107 141
measure 154 135 169
rich 105 110 111
ridge 201 186 146
search 173 155 177
surge 268 271 217
<Figure 1> Vowel duration changes in terms of voicing feature : English native speaker(ena), Korean learners of English(kle) and English bilingual speakers(ebi)
In terms of results of the group difference, English speakers, bilingual speakers, and Korean English learners all realized that vowels proceeding the voiced consonant were produced with longer duration than those proceeding voiceless environment. However, such a difference was not found in a word pair of ‘cash’ and ‘ca(sual)’. In the case of English native speakers and English bilingual speakers, the length of the vowel proceeding the voiced consonant was shorter than that of voiceless one when they produced ‘cash’ and ‘ca(sual)’.
However, Korean learners of English showed an opposite pattern, producing vowels proceeding voiced consonant with longer duration than vowels proceeding voiceless consonants, only for this pair. In general, this indicates that vowel coarticulation is realized by Korean learners of English to some extent.
However, the reason why Korean learners did not show the same pattern for
‘cash’ and ‘ca(sual)’ might be due to the fact that the number of syllables between this word pair is different. In other words, the vowel preceding voiceless consonant (e.g., cash) is shortened instead of getting lengthened because ‘cash’
has only one syllable, while the counterpart target word, ‘cas(ual)’, consists of two syllables. Regardless, Korean learners of English showed a similar pattern as the other groups, although their production was not like native빲like빲level.
The results also showed that the vowel duration of English bilingual speakers who have lived abroad for more than ten years have become clearer than that of Korean learners of English. In particular, for bilingual learners, the variation in the length of vowels due to the voicing feature of the following consonants was found to be quite similar to that of English native speakers.
Taken together, the results of the overall production experiment showed that changes in the length of vowels with respect to the voicing feature of following consonant is the unique characteristics of English and also can be acquired by learning. In particular, it can be seen that the longer the environment exposed to
English빲speaking environment, the more clearly the phenomena is realized.
The following figure represents vowel duration change of three speaker groups: English native speakers (Figure 2), English bilingual speakers (Figure 3), and Korean learners of English (Figure 4), respectively.
<Figure 2> vowel duration changes of English native speakers.
<Figure 3> vowel duration changes of English bilingual speakers
<Figure 4> vowel duration changes of English bilingual speakers
3. Discussion and Conclusion
The study showed that the vowel duration is significantly lengthened due to the voicing feature of the following sibilant consonants. That is, the length of vowels preceding voiced fricatives and affricates was longer than the vowel proceeding voiceless ones. The difference between voiceless and voiced condition was relatively larger for English native speakers and English bilingual speakers than Korean learners. This difference might be due to the cross빲 linguistic difference between English and Korean, such that Korean does not have voiced consonants in post빲vocalic coda position. Although the difference was not native빲like, Korean learners of English still showed a significant difference in the length of vowels depending on the circumstances of voiced and voiceless consonants. Thus, the variation in the vowel length depending on the voicing feature of following consonant can be acquired by Korean learners during the learning of a second language. However, whether the native빲likeness of the production of the following consonants affect this coarticulation effect need to be investigated in the further study.
We also found that the more the L2 learners were exposed to the target language, the higher their L2 proficiency is, as we found non빲significant difference between bilingual speaker group and native English speaker group in production of vowel length. This indicates that the length of exposure to the target language contribute to the native빲like pronunciation.
However, the current study revealed some limitations. First of all, the number of subjects was not enough. The study got the results from four English native speakers, four English bilingual speakers and three Korean learners of English.
Therefore, the future study should analyze more subjects to obtain more concrete data. Furthermore, the future study should provide more detailed
criteria when categorizing the bilingual and Korean learners of English.
Taken together, this study was able to reveal that the regressive coarticulation effect on the vowel length is also found by L2 learners’ speech, and the degree of exposure to the target language also influences on native빲like pronunciation.
References
Crystal, T.& House. A. S. “Segmental duration in connected speech signals: current results”, Journal of the Acoustical Society of America 83, pp.1553빲1573. 1988
Denes, P. “Effect of duration on the perception of voicing”, Journal of the Acoustical Society cf America. 27, pp.761빲764. 1955
House, A. S “On vowel duration in English”, Journal of the Acoustical Society of America, 33, pp.1174빲1178. 1961
House, A, S & Fairbanks, G. “The influence of consonantal environment upon the secondary acoustical characteristics of vowels”, Journal of the Acoustical Society of America, 25,pp.105빲113. 1953
Klatt,D. H. “Linguistic uses of segmental duration in English”, Journal of the Acoustical Society of America 59, pp.1208빲1221. 1976
Lehiste, I.,“The Timing of Utterance and Linguistic Boundaries”, Journal of the Acoustical Society of America, 51,pp.2018빲2024, 1972a
Peterson, G. E & Lehiste,I. “Duration of syllable nuclei in English' Journal cf the Acoustical Society cf America, 32,pp.693빲703. 1960.
Van Santen, J.P.H, “Contextual Effects on Vowel Duration”, Speech Communication 11, pp. 513 빲546. 1992
<국문초록>
한국인 영어학습자에게서 나타난 후행자음 유무성에 따른 선행모음의 길이의 변화
신동진(전주대학교)
본 연구는 모음의 길이가 뒤에 오는 자음의 음성 특징에 의해 영향을 받는지, 그리고 이 현상이 세 명의 화자 그룹 (영어 원어민, 영어빲한국어 이중 언어화자, 및 한국인 제2외국어 학습자)에 의해 어떻게 다르게 나타나는지 조사하였다. 각 피험자 그룹은 문장에 들어간 실험단어를 3회 반복하여 발화하였다. 연구 결과, 유의미한 그룹간의 차이를 나타냈으며. 후행하는 자음의 유무성에 따라서 선생 모음의 길이도 달라졌다. 영어 원어민과 영어이중언어 화자는 후행자음의 유,무 성의 특징에 따라 앞 모음의 길이에 유의미한 차이가 있음을 보여주었다. 한국 인 영어학습자의 경우에도 후행하는 자음에 따라 선행모음의 유무성에 큰 차이 를 보였으나, 다른 두 그룹만큼 차이가 크지 않았다. 본 연구를 통해 한국인 영 어학습자들은 영어를 발화 할 때, 후행모음의 유무성에 따라서 선행하는 모음의 길이를 변화 시킬 수 있었으며, 이러한 현상은 영어를 습득하는 과정에서 나타 났다고 할 수 있다.
주제어 : 한국인 영어학습자, 모음길이, 유무성,
투 고 일 : 2019년 1월 14일 심사완료일 : 2019년 2월 14일 게재확정일 : 2019년 2월 14일