1 Syllable Final Palatal Stops and Nasals in Vietnamese 1 Malone Dunlavy Linguistics 120A 6/10/2013 I. Introduction I will be examining a change in the contrast system of Vietnamese. In Vietnamese there are 2 main dialects, northern and southern. In all dialects, the syllable structure is (C1)(G)V(V)(O) where G is an optional glide and O is an optional coda consonant (either nasal or voiceless stop). As an onset consonant, the palatal stop and palatal nasal are clearly distinct from alveolar and velar equivalents. In coda position, however, there is some debate about the phonemic status of these two. In an earlier stage of Vietnamese, there presumably was a contrast in onset and coda positions. In Thompson s Vietnamese Grammar (1965), he comments on the palatal nasal and palatal stop saying that although they have been analyzed as velar after /i/ and /e/, they should be considered palatal, for both pedagogical and orthographic reasons. He (Thompson) also states that the phonetic reality of these sounds need to be closely investigated. Thompson, in his History of Vietnamese Final Palatals (1967), notes that there are distinct differences in the pronunciation of these elements between northern and southern dialects. Thompson (1967) argues that in the north these are still phonemic and in the south they have been merged with the alveolar stop and nasal after /i/ and /e/ and merged with velar /k/ and /ŋ/ after the other vowels 1 I would like to take this opportunity to thank my informants, Dr. Nguyen Quy Thanh of Vietnam National University, Hanoi and Khoi Vo. I would also like to thank my reviewers and peer-reviewers, Professor Robert Daland, my TA Dustin Bowers, and one anonymous peer-reviewer. Any remaining mistakes in this paper are completely mine.
2 that those elements appear after. This distribution is represented is Figure 1 exactly as Thompson shows it(1967). Figure 1: Distribution of voiceless stops and nasals according to Thompson This paper aims to examine the current state of modern Hanoi Vietnamese to see if Thompson s observations still hold. Through contrastive analysis of a northern speaker and a southern speaker, I will confirm the findings of Thompson with regard to the pronunciation of these sounds between dialects, showing that in northern Vietnamese [c] and [ɲ] maintain phonemic contrast with the other voiceless stops and nasals, respectively, in coda position where in southern Vietnamese the contrast has been neutralized. II. Method To investigate this, I created a contrastive word list of Vietnamese words with alveolar, palatal, and velar stops and nasals in coda position. This wordlist incorporates only front vowels since these seem to be the only ones that provide a contrastive environment. This wordlist was read by two native Vietnamese speakers 2, one northern and one southern. I recorded these speakers repeating the Vietnamese word twice after I stated the English translation using Audacity. After recording, I analyzed and transcribed the data using Praat. I examined the 2 I also had a central speaker who participated in recording, but after informal analysis, I concluded that the data from the central speaker was uninformative with regard to the pronunciation of palatal sounds and, therefore, have not included that data in this paper.
3 transcriptions of the audio, I compared spectrograms, and I analyzed the change in F1 and F2 for all tokens using Microsoft Excel. I measured the difference in F1 and F2 for every token, the mean difference in F1 and F2 for every coda consonant, and the standard error from the mean (because I had <15 tokens per consonant and the difference per token was about the same, every environment was included in the mean difference). By comparing the 95% confidence interval for the palatal sounds with their counterparts, I can conclude that the palatal sounds are either different or the same as their counterparts depending on whether the ranges overlap. My first speaker is named Nguyen Quy Thanh. He is the Director of Education Quality for the Institute for Educational Quality Assurance in Hanoi, Vietnam and also teaches sociology at Hanoi National University. Dr. Nguyen was born in Hanoi and although he has travel abroad for education and other endeavors, he currently lives in Hanoi and has maintained Vietnamese (particularly the northern dialect) as his primary language. My second speaker is Khoi Vo. Khoi is an enrolled Junior at UCLA and has lived in the US for 4 years. Khoi s hometown is Ho Chi Minh City, Vietnam. He grew up here speaking southern Vietnamese although he has also had influence from parts of his family who speak the northern dialect. Khoi has consistently spoken Vietnamese nearly every day since he moved to the states and it remains his primary language. III. Analysis As stated earlier, Thompson claims that the palatal stop and nasal of Vietnamese maintains contrast with alveolar and velar counterparts in coda position in northern Vietnamese. He gives the transcribed examples contrasting Northern and Southern dialects shown below next to the tokens I collected from my speakers.
4 Figure 2:Thompsons observations of the pronunciation of palatal, alveolar, and velar sounds Figure 3: My observations of some of the same words as Thompson Although I found that the alveolar and velar sounds were pronounced differently after [æ], the pronunciation of palatals seemed to be in the same distribution that Thompson reported. With further analysis of spectrograms and the statistical analysis of formant change, I found his predictions to be true. I will discuss first the pronunciation of my northern speaker, then briefly discuss the pronunciation of the southern speaker, concluding with a contrast of the statistics for them both. III.a. Northern Speaker After recording my northern speaker and analyzing the data, I found that my northern speaker pronounced my wordlist precisely as Thompson predicts. That is, the palatal sounds are distinctly different from the alveolar and velar sounds. The transcription table provides a plethora of minimal pairs. These pairs and triplets, in Figure 4, are evidence pointing towards an analysis similar to Thompson s, but further evidence is needed. In addition to analyzing the transcribed data, I also looked for visible or statistical evidence that something distinctly different was happening with the tokens predicted to be palatal in contrastt to the alveolar Figure 2: Minimal pairs and triplets from the Northern speaker
5 and velar tokens. In the spectrograms, I found that the second formant for my northern speaker nearly always rose for palatal tokens while it remained constant for the alveolar and velar tokens. This is shown in Figure 5 which is the stacked spectrograms of /baɲ/, /ban/, and /baŋ/ in order from top to bottom. On top we see the palatal token experiencing a notable drop in F1 and a rise in F2. In the middle, the alveolar token has no real change in either Figure 5: Spectrograms of Banh, Ban, and Bang F1 or F2. Finally, on the bottom, the velar token shows almost no change in F1 or F2 but a noticeable fluctuation in F3 as it drops to form a velar pinch. After analyzing the data, I found that the predicted palatals are statistically distinct from the alveolars and velars. In Figure 6, the mean intervals for the change in F1 and F2 are shown for each coda consonant. Because these intervals (palatals vs. the other two) don t overlap, we can conclude that these sounds are statistically different. Although this change in formant does not prove that these sounds are palatal, the phonetic facts of these sounds could prove useful. The act of moving the body of the tongue up towards the palate causes the jaw to raise producing the drop in F1. This movement also causes the dramatic
6 rise in F2 as the bulk of the tongue moves forward. The movement of the tongue up and slightly forward has more time to affect F2 than the quick alveolar or velar movements. Given these facts, it is safe to conclude that the northern speaker was making a palatal-like sound. III.b. Southern Speaker The facts for my southern speaker were very different. To recall, Thompson stated that the predicted palatal sounds are not pronounced as palatals in southern Vietnamese. In fact, he states, broadly, that the orthographically palatal sounds are pronounced as alveolar in all coda environments. My southern speaker produced the following forms(fig.7): Exactly as Thompson reported 40+ years ago, my southern speaker produced alveolarized palatal sounds in all environments and even produced some velarized alveolar sounds after /a/. Here, in Figure 8, we see the spectrograms of the same three words /ʂǎɲ/,/ʂǎn/, and Figure 7: Minimal pairs and triplets from the Southern speaker /ʂǎŋ/. Figure 8: Spectrograms of Sánh, Sán, Sáng The last word has a very clear velar pinch, but the other two are basically the same. This fact is even further communicated in the statistics. In Figure 9, the mean change interval for F1 and F2 for each coda consonant for my southern speaker Figure 9: Mean change interval for the southern speaker is shown. These statistics are different from the northern speaker in that the palatals overlap with
7 both of the other sounds. With so much overlap between these intervals, statistically, the sounds are not different with regard to formant change. IV. Conclusion As I have shown, my northern Vietnamese speaker is producing sounds that seem to be palatal where the orthography cues palatal sounds. At minimum, he is producing sounds that are different from both alveolar and velar sounds. This is not the case with my southern speaker. This is confirmed with simple T-tests shown in Figure 10. Northern Speaker Southern Speaker F1 F2 F1 F2 ɲ vs n 0.000001 0.000000 ɲ vs n 0.154650 0.529247 ɲ vs ŋ 0.000091 0.000001 ɲ vs ŋ 0.399791 0.357175 c vs t 0.000009 0.000000 c vs t 0.526870 0.513082 c vs k 0.000025 0.000021 c vs k 0.785360 0.121752 Figure 10: T-test results of the change in formant for palatal versus counterparts The results of the t-test confirm strongly that the palatal sounds for the northern speaker are distinct from their counterparts, whereas the palatals for the southern speaker are clearly not as strongly separated. These statistics are the confirmation of Thompson's previous observations. Figure 11 shows that the palatal sounds for the northern speaker are distinct from those of the southern speaker. This undeniably confirms that something very different is happening for the northern speaker than is happening for the southern speaker. Figure 11: T-test results for northern vs. southern palatals There are a few notable flaws in this study however that must be taken into account. First, without a palatogram, it is difficult to see where the tongue makes contact with the top of the mouth. This is notable since the conclusion of this paper is about that very fact. Second, this
8 study had a very limited number of tokens. This poses several threats but most notably it begs the question of whether my sample is an adequate representation of the population. I contend that it is not a great representation, but my t-test and 95% intervals for my data are fairly strong. More tokens should be collected from many speakers and analyzed. Third, because the sample size was small, I decided to average the change in formant for every vowel, rather than each vowel individually. This is a problem because each vowel will have different effects on the formants of that vowel, thereby varying the change. In some cases the tone could affect that as well. With my data set in particular, the statistics for each vowel were extremely similar and therefore could be analyzed together. As it stands, It seems that Thompson s observations still hold. Northern Vietnamese has a palatal stop and nasal in coda position while Southern Vietnamese neutralizes the contrast there. References: Han, Mieko Shimizu. Complex syllable nuclei in Vietnamese. Vol. 6. University of Southern California, 1968. Nguyêñ, Đình Hoà. Vietnamese-English student dictionary. Southern Illinois University Press, 1971. Thompson, Laurence C. "The History of Vietnamese Final Palatals." Language(1967): 362-371. Thompson, Laurence C. "A Vietnamese Grammar. Seattle, 1965." OCEANIC LINGUISTICS SPECIAL PUBLICATIONS. Thompson, Laurence C. "The problem of the word in Vietnamese." Word 19.1 (1963): 39-52.