Sui “voiced-high” tone value distinctiveness: An areal feature

in Jiaoli JL and Northern Kam speakers are in close contact with Qiandong Miao speakers. In Qiandong Miao, all stop finals have been lost Zhang Yongxiang 1990. In fact our Jiaoli district data point, Gaorong village, was the only village in the Jiaoli region where people still speak Sui. In Sui villages surrounding Jiaoli, people have switched to speaking Miao as their first language L1. 7

4.4.2 Merger of Tones 7 and 8 in Shuiyao SY, Southern

A merger of Tones 7 and 8 on rimes with short vowels has taken place in Shuiyao. Examples are given in table 4.15. Rimes with long vowels, however, retain phonemic contrast between these two tones, for example ‘rope’ laːk⁷ and ‘child’ laːk⁸. This merger has not taken place in other Southern Sui lects. Table 4.15. Merger of Tones 7 and 8 in Shuiyao Gloss PKS JQ SY ‘dark, black’ - hop⁷ xəp⁷ ‘to blow’ dzup D hop⁸ xəp⁷ ‘to hurt’ - ȶiːt⁷ ȶit⁷ ‘to bite’ - ȶiːt⁸ ȶit⁷ ‘to exchange’ hlik D liːk⁷ lik⁷ ‘strength’ - ljək⁸ lik⁷ ‘flea’ k-hmat D m̥ɐt⁷ m̥ɐt⁷ ‘savoury, with just the right amount of salt’ - mɐt⁸ mɐt⁷

4.5 Sui “voiced-high” tone value distinctiveness: An areal feature

In terms of its tonal phonetic pitch values, Sui is noteworthy among Tai-Kadai languages in that it displays “voiced-high” tendencies as opposed to the more common “voiced-low” paradigm. This is true for all Sui dialects, including Yang’an. However, in this section we will argue that this Sui “distinctiveness” is an areal feature which has diffused across the region relatively recently, affecting not only Sui but also several non-related languages which are spoken in the same geographical region. Miao, Yao, Bouyei and Chinese dialects spoken in and around the Sui area all tend to be “voiced-high”. As we already discussed at the beginning of this chapter, almost all Sinitic, Tai-Kadai and Hmong- Mien languages underwent a tonogenetic process known as the Great Tone Split during which syllables with a voiced onset took on a tone of lower pitch or register and syllables with a voiceless or preglottalised onset took on a higher pitch. The original voicelessvoiced distinction on the onsets was then often lost. We can refer to such languages as “voiced-low” languages. Abramson 2004 provides a useful summary of a swathe of acoustic research done by different scholars confirming the underlying principle that voiced phonation on an initial stop can lead to lower F0 frequencies on the following vowel, potentially resulting in the emergence of a phonemic two-way tonal contrast. Cantonese is a classic example of a voiced-low language table 4.16. 7 These Sui are known by local people as “ 水家苗 Shuǐjiā Miáo”, lit. “Sui-household Miao”. Table 4.16. Modern tonal pitch values in Cantonese Yip 2002 A level 平 B rising 上 C departing 去 D entering 入 1 H 阴 yīn 53† 35 44 5 2 L 阳 yáng 21 24 33 3 † Also 55, an allophonic variation which has all but replaced 53 among younger speakers. However, there are some “voiced-high” languages, such as Shan, in which syllables with an original voiced onset appear to have taken on higher pitch values than those with an original voiceless onset Yip 2002:36. Within Thai, Brown 1975 shows that Northern and Central Thai dialects are almost always “voiced-high”, whereas Southern Thai dialects are uniformly “voiced-low”. Given the predominance of voiced-low languages and dialects within not only Tai-Kadai, but also Sinitic and Hmong-Mien, many scholars have argued that voiced-high tonal patternings are due to tonal “flip-flops” or “reversals” which took place subsequent to an initial voiced-low tone split Matisoff 1973, Yue-Hashimoto 1986. There are some languages that even show this “tonal inversion” at a stage when the consonant onsets still preserve their original laryngeal contrasts Kingston and Solnit 1989. Brown 1975 argued that the original voiced onset actually induced a higher pitch register in some languages and dialects at the time of the Great Tone Split. He claimed that voiced phonation on the onset can lead to a higher F0 frequency at the end of the ensuing vowel resulting in a higher overall perceived pitch. This theory is based on flimsy acoustic evidence and has not gained much credence among scholars of tonogenesis Rischel 2004. Sui most Central, Eastern, Southern and Pandong lects is clearly voiced-high in all three open syllable proto-tone categories A, B and C. Typical pitch values are shown in table 4.17. Western Sui and Yang’an dialects are voiced high in tone categories A and C but voiced-low in B see section 4.6.1 below for further discussion. Table 4.17. Typical tonal pitch values in Sui Central A level 平 C rising 上 B departing 去 D entering 入 1 H 阴 yīn 13 12 33 35 55 short 24 long 2 L 阳 yáng 31† 53 55 56 44 short 33 long † Tone A2’s starting pitch and its overall mean pitch are both significantly higher than Tone A1 see tonal pitch plots in appendix G. Voiced-high languages are certainly in the minority in southwest China. Out of 24 languages in Guangxi Zhuang Autonomous Region described in GZARMLC 2008, including seventeen in the Tai- Kadai branch, 21 uniformly conform to the voiced-low principle. The exceptions, Wusehua, Miao Rongshui and Gelao Buliu dialect, spoken in Longlin county, only exhibit voiced-high in one proto- tone category out of four: Tone B in Wusehua and Tone A in Miao Rongshui and Gelao Buliu. Out of seventeen Kam dialects surveyed by Shi and Strange 2004, fourteen consistently conform to the voiced-low principle. The three dialects which do not are all Northern Kam dialects and only one, Baojing, exhibits voiced-high in all three open syllable tone categories. 8 Out of 24 Bouyei dialects in 8 With regard to the aspirated-induced secondary tone split on y īn which is often cited as characterising the Kam language Thurgood, 1988, it is worth noting that of these seventeen Kam dialects, seven including two in Rongjiang where the standard Chejiang variety is also spoken have not undergone a secondary tone split in any tone category. Only three Kam dialects Chejiang Zhanglu 榕江县车江乡章鲁 , Tianzhu Shidong 天柱县石洞镇水洞村 and Jianhe county Mindong Xiaoguang 剑河县敏洞乡小广村 have undergone secondary tone splits in all the y īn tone categories and thus have the full set of nine contour tones. Guizhou surveyed by Wu Wenyi et al., 2007, sixteen are predominantly voiced-low i.e., at least three tone categories out of four clearly show a voiced-low based tone split. Appendix A shows typical tone pitch values for a wide range of languages and dialects in the region. As can be seen, Sui is not the only language which displays voiced-high tendencies. Mak, Ai-Cham and Then, all of them Kam-Sui languages, are strikingly similar to Sui in this regard. Furthermore, although Bouyei is usually voiced-low, it has several dialects which tend towards voiced-high. Most of the voiced- high Bouyei dialects are spoken in the same geographical region as Sui, Mak, Ai-Cham and Then. Map 4.1 shows the locations of all the speech varieties for which we could find tonal pitch data plotted onto a map of southern Guizhou and northern Guangxi. Downward-pointing black triangles show voiced-low lects and upward-pointing white triangles show voiced-high lects. We regarded a lect as “voiced-high” if at least two of its four proto tone categories unambiguously have a higher overall pitch for yáng tones than for yīn tones. A “voiced-high region” centred around the Sui area is clearly visible on the map. This region embraces several Kam-Sui languages as well as Bouyei, Miao, Yao Iu-Mien and even local Chinese dialects. Unfortunately we lack data from many of the Miao and Yao dialects in the region. Bouyei spoken in Dushan, Duyun and Libo counties generally exhibits “voiced-high” in Proto-Tai tone categories B and C. Yao spoken in Yaolu township, Libo county directly between Shuiyao SY and Jiarong JR and Miao spoken in Pu’an township just north of Sandu county town are voiced-high in Proto-Miao-Yao PMY tone categories A and B. Map 4.1. Distribution of voiced-high and voiced-low languages and dialects in southern Guizhou and northern Guangxi. Key to abbreviations is given in table 4.18 Table 4.18. Key to abbreviations in map 4.1 Letter Language Data sources A Maonan Liang and Zhang 1996 B Bouyei Wu Wenyi and Snyder 2008 D Kam Shi and Strange 2004 G Gelao Liang and Zhang 1996 H Chinese dialects Jiang 1990, Xu 1991, Zhang Fahe 1996, Li Xia 2004, Zeng 2010 J Ai-Cham Liang and Zhang 1996 K Mak Liang and Zhang 1996 L Lakkia GZARMLC 2008 M Miao Zhang Yongxiang 1990, Li Yunbing1997, Dafang County 2003, GZARMLC 2008, Wu Zhengbiao 2009 N Bunu GZARMLC 2008 U Mulam GZARMLC 2008, Liang and Zhang 1996 S Sui our own data and GZARMLC 2008 T Then Liang and Zhang 1996 W Wusehua Wei Maofan and Wei Shuguan 2011 Y Yao Iu-Mien branch Huang Hai 1997, GZARMLC 2008 Z Zhuang GZARMLC 2008, Hansen and Castro 2010 Most Chinese dialects in this region are varieties of Southwest Mandarin “ 西南官话 Xīnán Guānhuà” and only distinguish between yīn and yáng tones in tone category A “píng level” tone in Chinese phonology. Out of 32 varieties of Southwest Mandarin surveyed by Li Xia 2004:20–21, only four showed a higher starting pitch on Tone A2 yīn-píng than on A1 yáng-píng, and only one of these Danzhai, just north of Sandu had an overall pitch which was unambiguously higher on Tone A2 than on Tone A1 i.e., unambig-uously voiced-high. Typical tone values of Southwest Mandarin dialects as described by various scholars are given in table 4.19. The geographical locations of these dialects is shown on map 4.2. Table 4.19. Phonetic tone values in Southwest Mandarin dialects Jiang 1990:164–165, Xu 1991:160, Li Xia 2004:20–21. Beijing Mandarin is shown for comparison. High Tone A2 values voiced-high are delineated by double lines. Tone values are only shown for Tone D where it is preserved as a phonemically distinct tone Dialect A level 平 B rising 上 C departing 去 D entering 入 1 H voiceless 阴 yīn 2 L voiced 阳 yáng Mandarin Beijing 55 35 214 51 - Bijie 55 21 42 213 - Zunyi 55 21 53 24 - Guiyang 55 21 42 13 - Duyun 33 53 35 13 42 Pingtang 33 53 35 13 31 Dushan 33 53 24 13 42 Sandu 44 54 24 13 42 Danzhai 33 53 44 13 31 Libo 33 42† 24 13 - Huangping 55 21 42 13 31 Rongjiang, Tianzhu, Jinping 33 21 42 35 - Kaili, Leishan 33 31 55 13 - Liping 33 13 31 53 24 Guilin 33 21 54 35 - Liuzhou 44 31 53 24 - Hechi Jinchengjiang 55 21 51 35 - †Zeng 2010:44 transcribes this tone as 52, suggesting a higher overall pitch than Tone A1. The author’s own observations of Libo speech confirm this. Map 4.2. Locations of Chinese dialects in table 4.19. All are Southwest Mandarin apart from Baise † . Black triangles pointing down show voiced-low Tone A, white triangles pointing up show voiced-high in Tone A † Baise is a type of “Guangxi Baihua”. Guangxi Cantonese; Li Ping 2005 The Chinese dialects spoken in six towns in and around the Sui area are very distinctive in that they all have a high Tone A2 instead of the low Tone A2 common to almost all other varieties of Southwest Mandarin remarked upon by Xu 1991:60. In terms of other dialect features, this “island” of voiced-high Southwest Mandarin lects group together with other Chinese dialects of southern Guizhou forming a subgroup within the wider “Guiliu 9 speech” area Li Lan 2009:81. Li Lan 2009:73–74 notes that there are thirteen Chinese dialects in southern Guizhou Qiannan prefecture which are significantly different from and have low mutual intelligibility with other Southwest Mandarin lects. In particular, they preserve aspiration on píng A tones and non-aspiration on non-píng B, C and D tones when they occur with fully voiced onsets. She does not specify the precise locations of these dialects although they certainly lie within the same geographical region as the voiced-high lects in map 4.2. Wu Weijun 2010:110 devotes a large section of her paper on “dialect islands” to this particular region. She says that the Chinese dialects of Duyun, Dushan, Pingtang, Sandu, Majiang and Danzhai are significantly different from other Chinese dialects in the area, not only in terms of their onset and final reflexes, but also in terms of their tones: they all preserve a distinct Tone D although the original stop- finals have been lost and they all have a high yáng-píng A2 tone. 10 She says that a dialect island such as this, covering a relatively large contiguous geographic area and so distinctively different from surrounding dialects, is a rare phenomenon. The migratory or language contact factors which must have given rise to it are, in her words, “as yet undiscovered.” 9 The term “Guiliu 桂柳 ” is formed from the first characters of Guilin and Liuzhou, the two largest urban centres in Guangxi where this dialect is spoken. 10 Majiang county 麻江县 is about 20 km north of Duyun and 40 km west of Kaili. Even so, it could well be that the voiced-high features of both Chinese and minority languages such as Sui, Bouyei, Miao and Yao in the same area are somehow linked and to some extent contact-induced, although the process by which it happened is unclear. It is probably no coincidence that Tones A1 and A2 in Sandu and nearby Chinese dialects are phonetically identical to Tones C1 and C2 in most Sui dialects, viz. 33 and 53 respectively. It is clear that the “voiced-high” tonal tendency is an areal feature which has diffused across languages that are not necessarily closely related genetically. This was possibly initiated by a tonal “flip- flop” or “reversal” Matisoff 1973, Yue-Hashimoto 1986 occurring some time after the Great Tone Split which then spread across the area. This idea is in harmony with Rischel’s 2004:622 suggestion that tonal reversal in northern Thai Southwestern Tai, as observed by Brown 1975, “is a local innovation which spread to neighbouring dialects and languages.” Thus the similarity of Yang’an Sui to the other Sui dialects in terms of its phonetic tone realisation is not the result of any close genetic affiliation. In fact, shared innovations indicate that Yang’an Sui is more closely related to Kam, Mulam and Then. Rather, “voiced-high” appears to be a regional tonal characteristic which Yang’an Sui acquired along with many other neighbouring languages and dialects. It is also possible that, instead of resulting from a post-Great Tone Split tonal flip-flop that diffused across the area, the areal “voiced-high” tonal tendency is due to an as-yet unexplained tonogenetic process which caused all of the languages in this particular region to split along “voiced-high” lines instead of according to the more common “voiced-low” principle. Oral histories indicate that the Sui have lived in this geographical region for at least one thousand years Lei 1985, Lei and Yang 1988. The Great Tone Split probably did not happen until a later point in time. Further investigation into the migratory history of the other minority groups in the area would shed further light on this possibility. Incidentally, there is one “voiced-low” variety of Sui marked on map 4.1, lying southwest of the main Sui area and just south of the Guizhou-Guangxi border. This variety is spoken in Longma village, Liuzhai township, Nandan county 南丹县六寨镇龙马村 . According to Yu 2008, the Sui inhabitants first settled here around one hundred years ago after migrating from Libo and Sandu counties. Most of the places that Yu mentions as their ancestral home villages now lie within the Southern Sui dialect area Shuipo, Shuichuan, Shuili SL, Jiuqian JQ although she does say that some originally came from the Yang’an area. Table 4.20 shows the tone values for this particular variety. Table 4.20. Modern tonal pitch values in Sui Nandan county, GZARMLC 2008:782 A level 平 C rising 上 B departing 去 D entering 入 1 H 阴 yīn 24 33 35 55 short 35 long 2 L 阳 yáng 31 42 13 21 short 21 long Further investigation is needed to confirm the precise frequencies of the tones in this location. If it can be proven that this variety of Sui is consistently voiced-low, it leads one to speculate as to whether these phonetic tone values are closer to the “original” Sui tone values directly after the Great Tone Split. If so, the subsequent tone reversal now evident in the Sui homeland must have occurred very recently indeed, within the last one hundred years or so. Given the proven stability of Sui dialect features over the last fifty years Stanford 2011, this seems highly improbable. Perhaps more likely is that after moving to a predominantly voiced-low area, these recent Sui migrants adopted the tonal characteristics of closely related languages spoken around them even within Longma village there are over twice as many Zhuang-speaking people as Sui, Yu 2008:6.

4.6 Variation in Sui phonetic tone values