A possible solution is that DQC cannot distinguishing monophthong and - PDF document

A possible solution is that DQC cannot distinguishing monophthong and diphthong. A sound with small DQC can still be a ng the same nuclear vowel and having phonological contrast with it, as the case of [ith it, as the case of [ø]. A sound with a relative high DQC can also treated as monophthong if no psound system. On the other hand, as shown in Table 2, the standard deviations of most vowel sounds are very large, which indicates that the real speech will be more complex than the abstract categories. In real speech, sounds belonging to monophthong can be diphthongized while diphthongs may sometimes monophthongized. Therefore, we may say that the terms of monophthong and phonetic terms. There may not be a universal objective boundary between monophthong and diphthong. The difference between the two types of volated in this study. It was found that [should be classified into monophthong, and [There are speaker variations for []. For some speakers, it is a monophthong; and for some speakers it is a diphthong. According to this study, there istypes in one language is determined by its phonological system. A sound treated as a monophthong in language A may be treated as diphthong in language B. Ladefoged, P. 2003. Phonetic Data Analysis: An Introduction to Fieldwork and Instrumental TechniquesMalden, MA: Blackwell. Li, Xiaofan. 1998. A study on the grammar of Suzhou Chinese. Beijing: Peking University Press. Liljencrants, J. and B. Lindblom. 1972. “Numerical simulation of vowel quality systems: the role of perceptual contrast.” Language 48, 839–862. Miller, J. 1989. “Auditory-perceptual interpretation of the vowel.” Journal of the Acoustical Society of 85 (5), 2114–2134. Wang, Ping. 1987. “Reanalysis of the phonological system of Suzhou Chinese.” Language Study 25 (1), 41–48. Ye, Xianglin. 1988. Suzhou Chinese. Nanjing: Jiangsu Education Press. Yuan, Jiahua. 1960. Hanyu Fangyan Gaiyao. (1989 2nd Ed.) Beijing: WenziGaige Press. HONETIC BOUNDARY BETWEEN MONOPHTHONGS AND IPHTHONGS IN UZHOU Table 5: Vowel DQC Vowel DQC Vowel DQC 1107.6 ou 1137.5 796.6 u 1759.5 ø 4725.5 ø 854.1 1807.6 iø 5034.4 o 868.9 uø 1904.6 u 7392.7 926.3 1915.5 iæ 7497.8 1002.3 i 8962.3 Just as we expected, the remained [] should be classified as monophthong shown Table 5. Therefore, it is not proper to just simply classify [] into monophthong or diphthong. Besides those samples having merged into [ed into [֦] still contains two variations. For some speakers, it is a The mean value for [] is still very low after those samples being monophthongs have been removed. The result indicates that the change of sound quality for this diphthong is very small even for those speakers who can distinguish the two sounds. Since the mean value is less than that of apical vowel []. The samples of [ Spectrograms of [] pronounced by two speakers. DQC of [] on left is 339.4, and DQC of [] on right is 6658.7 on right is 6658.7 џ], which were pronounced by two female speakers. However, the DQC of the two sounds shows differently. The syllable DQC of [ that on right side is 6658.7. As we ograms, the formant trajectories of the two sounds are different. [on right side is more like a diphthong. Therefore, the typical monophthong [] has A paradox appears now. If [. If [џ] or even [u] and [ʮ] should be also treated diphthong, which willphthong, which willuø] is treated as monophthong, it will be very difficult to describe the contrast between [ uø]. If those [uø] having been monophthongs were removed, the mean value of DQC of uø] will increase. The mean DQC of [] is close to the DQC of [ [֦] is quite big. [֦] changed in the new accent too and merged into the diphthong ong iø]. It seems that the case of [] is similar to the case of []. Therefore, we also checked DQC of [] for each individual speaker again. Table 4: able 4: ֦] for each individual speaker Subject DQC Subject DQC M1 1306.5 F1 738.9 M2 2343.2 F2 625.7 M3 3054.8 F3 5705.5 M4 886.5 F4 4098.0 M5 1125.9 F5 793.3 M6 877.7 F6 1440.0 M7 2419.2 F7 2564.0 M8 513.8 F8 2210.9 M9 1291.4 F9 3041.6 M10 377.5 F10 1138.6 Speaker variations for [] can also be found in Tamarked with italic fonts, their [] is not a monophthong but a why mean DQC of [ean DQC of [֦] can be found, and Figure 3 gives us three typical examples. Spectrograms of [] produced by three different speakers. The DQC of the three samples are 829.4 2403.5 4925.1 respectively. The three samples may represent three stages for [] merging into [sample on the left is still a monophthong, which should be onophthong, which should be ֦]. The sample in the middle has beenddle has beeniø] yet. It should be a middle stage from [] to []. The sample on the right has no difference to []. It is the last stage. ֦] should also be removed from the mean calculation of []. Now we can calculate the means of DQC again. HONETIC BOUNDARY BETWEEN MONOPHTHONGS AND IPHTHONGS IN UZHOU As shown in Figure 2, the difference between monophthong and diphthong is large except the diphthong [ge except the diphthong [uø] were put aside first, the classification of [] is very clear. [] are monophthongs and [Although the main question in this study has been answered, the new problem of uø] raises. If only Figure 2 is referred, [uø] may be considered not like a diphthong; instead it is more close to monophthong. onophthong. uø] merged with the monophthong [] in the new accent of Suzhou Chinese. And our result is that lt is that uø] is more similar to a monophthong. Although thpeople, it is reasonable that the accents of some elder people are also affected by the new accent. Therefore the DQC of [C of [ø] for each individual speaker are checked again. Table 3: DQC of [uø] and [ø] for m1 1496.2 443.3 f1 1524.3 997.8 m2 1946.8 844.7 f2 2079.0 1718.2 m3 646.7 994.1 f3 2046.8 1231.9 m4 1266.5 638.8 f4 3787.4 879.6 m5 2343.9 898.6 f5 1842.5 676.7 m6 1944.6 1663.3 f6 1362.2 477.1 m7 2128.2 787.1 f7 1404.6 555.2 m8 937.6 878.5 f8 2048.7 448.5 m9 975.9 880.8 f9 1660.5 805.0 m10 2223.9 653.0 f10 1487.0 610.4 As shown in Table 3, there arble 3, there aruø] and []. For DQC of those speakers marked with italic font, there is no difference with significance between ificance between uø] and [ø]. The speakers have confused these same as that in the new accent -- that [] merged into []. Actually some elder speakers did merge [] into [], however, there are also some other elder speakers merged [] into LING FENG 52 DQC of two vowel sounds. Cursor A repressound, and cursor B is the ending point. DQC from cursor A to cursor B are shown in the tables in top left of the spectrograms. ograms. ֦] pronounced by a male speaker. Point A and B are selected for this sound is 810.0. The iæ] pronounced by the same speaker. The same procedure having been 3. Results and discussion Table 2: Means of DQC and standard deviations for all monophthongs and diphthongs in Vowel DQC S.D. Vowel DQC S.D. æ 538.6368.9 1807.61159.7 588.5470.0 1827.61450.7 796.6629.3 1915.51164.2 ø 854.1505.0ou 4050.21256.7 o 868.9559.8u 4370.31536.3 1002.3616.4ø 4725.51738.7 i 1107.6763.6iø 5034.41479.2 y 1137.5545.8u 7392.71624.9 u 1759.51337.2iæ 7497.81376.2 .4 HONETIC BOUNDARY BETWEEN MONOPHTHONGS AND IPHTHONGS IN UZHOU Table 1: Test wordsVowel Test wordsMeaning Vowel Test words Meaning ‘this’ ‘book’ ‘in turn’ ‘clothes’ y y ‘winding’ i ‘coco’ u fu‘grievance’ ‘smoke’ ‘excellent’:⁴⁴ uø huø ‘sadness’ ou ou‘black’ o o ‘crotch’ ø ø ‘Europe’ All the words are associated to a high level tone [44] in order to minimize the effect of the fundamental frequency change on the vowel quality. To eliminate the effect Ś uø]. The two diphthongs must be preceded by consonant. To minimize the effect of initial consonant, the syllables preceded by glottal fricative [h] are selected for them. The word list is shown in Table 1. Randomized list was made. All the test words were placed in the following carrier ng carrier ŋəu²³¹ sə׵⁵ __________ pə׵⁵ nɛ²³¹ t] “I (am) reading ___ for you (to) listen.” Because the exact first point or last point of a vowel sound will be easily affected by many other factors (Ladefoged 2003), the start of 10 ms to the starsound are selected. The following figure is a sketch map for this procedure. However, this formula is still not eaccount. For the languages with rich inventory of lip rounded vowels, it will be better to add the F3. The auditory-perceptual space (Miller 1989) transform will be used in this study. The following equations define the auditory-perceptual space: x = log(SF3/SF2) Eq. 1 y = log(SFl/SR) Eq.2 z = log(SF2/SFl) Eq.3 SF1, SF2, and SF3 represent the frequency loprominences of the short term spectral envelope of the vowel waveform. SR is a reference frequency, which is shifted slightly by the average spectrum of the current speaker. This reference is calculated as follows: SR = 168(GMF0/168) Eq.4 where GMF0 is the geometric mean of the speaker’s F0 for the utterance. The choice of these variables and the motivation for the sensSimilar to the perceptual distance, the formula for calculating the distance change’. The equation is as follows. +(ySince all the origin formant values have been transformed into logarithm, DQC is a very small number. For the ease of use, it will be multiplied by 10000. 10 male and 10 female subjects were invite of Suzhou aged from 50 to 60 and with no history speech or hearing disorder. a quiet room using a Shure SM-58 microphone and MD Recorder. The subjects were asked to read the word list at a normal rate of speech. Four repetitions were recorded for each subject. The recording were digitized at 10 kHz with a 5 kHz low-pass filter setting with 16 bit precision in the CSL (Computerized Speech Lab) 4400 speech analysis software (Kay Elemetrics). All acoustic analyses were done in profPhonetics by Computer), which was developed ersma and Professor David Weenink of the Institute of Phonetics Sciences of the University of Amsterdam. Toronto Working Papers in Linguistics P Copyright © 2008 Ling Feng City University of Hong Kong The dialectologists have argued whether some vowel sounds in Suzhou Chinese were diphthongs or monophthongs for many years. However, there were no objective methods for judging whether a vowel sound was a monophthong or diphthong. This study is addressing my development on an objective method to calculate the degree of vowel quality change (DQC), through which a phonetic boundary between monophthongs and diphthongs can be established. We therefore can easily judge a vowel sound belonging to monophthong or diphthong by its DQC value. e. iʏ, i֦, ֦] (Yuan, 1960; Ye, 1988) in Suzhou Chinese were transcribed as [] by Wang (1987). The difference between the two transcriptions is whether the three sounds are monophthongs or diphthongs. Since the transcriptions were both based on subjective impression, neither of them has strong evidence. Before an objective criterion for judge diphthong and monophthong was established, it is difficult to say which transcription is more reasonable. In fact, both monophthong and diphthong can be treated as a combination of two parts. For monophthong, the vowel qualities for the two parts in it are quite similar; for e two parts are different. change (DQC) between the two parts can be used as a quantitative parameter for judging diphthong and monophthong. The DQC values of those typical monophthongs, which were accepted by all dialectologists, can be usedThe remaining problem is how to calculate the DQC value. Liljencrants and Lindblom (1972) provided a formula to calculvowels in fact are coexisting in a two-dimension plane. The perceptual distance is the first two formants are treated as coordinate values.