Tai branch
Mixed (Tai and Kra)


1. Internal phylogeny and spread of Kra-Dai
It now appears that Kra-Dai (also known as Tai-Kadai) consists of perhaps 7 or 8 branches, depending on whether Biao and Lakkia form a subgroup together. It is still unclear how these branches fit together. They may form a rake-like structure as with Paul Sidwell's classification of the Austroasiatic languages, or they may form a more nested structure as with Gerard Diffloth's Austroasiatic tree. Additional research on Kra-Dai internal phylogeny is currently being carried out by Peter Norquest and Weera Ostapirat.
  1. Kra
  2. Biao
  3. Lakkia
  4. Jiamao
  5. Hlai
  6. Kam-Sui
  7. Be-Jizhao
  8. Tai
The validity of the Kam-Tai grouping (Kam-Sui + Tai, sometimes containing other branches as well) is still debated. Kra may the earliest branch to split off from Proto-Kra-Dai, or it may have split off at about the same time as the other Kra-Dai branches.

Based on geographic and linguistic evidence, my hypothesis is that Proto-Kra-Dai was spoken in the Pearl River Delta region and the coast of Guangdong and Guangxi, perhaps around 3,000 years B.P. (Map 1). Kra-Dai most likely spread upstream via the Pearl River's tributaries and via the coast, with its expansion motivated by rice agriculture (Map 2).

Map 1: Dispersal of Kra-Dai branches. Green: Kra; red: Tai; orange: Kam-Sui; light brown: Hlai.

Map 2: Dispersal of Kra-Dai and Hmong-Mien branches.

Map 3: Double-crop rice regions in China and the distributions of the Kra-Dai (KD) and Hmong-Mien (HM) phyla, as well as the putative Austro-Tai and conjectured Yangtzean superfamilies.

I personally consider the following two possible dispersal scenarios for Kra-Dai to be the most likely ones.
  1. Kra and Hlai represented the first wave (or pulse) of westward expansion from the Pearl River Delta region, followed by a second Kam-Tai westward wave represented by the Tai, Kam-Sui, Biao-Lakkia, and Be-Jizhao branches. The last major expansion is that of Tai, which occurred approximately 1,500-2,000 years B.P. (i.e., after the Qin conquest of Guangdong), absorbing other language branches in Guangxi and Guizhou such as Biao-Lakkia, Be-Jizhao, Pakanic, and Kra.
  2. All Kra-Dai branches spread out through riverine routes in the Pearl River drainage basin about simultaneously, resulting in the lack of a deep nested structure in the Kra-Dai family tree. This parallels the Mekong riverine spread of Austroasiatic as proposed by Blench & Sidwell (2010).

As Kra-Dai speakers spread westward, they would have encountered speakers of other language phyla and assimilated them. This can be seen in non-Kra-Dai loanwords that are found only in certain branches. Kra has Tibeto-Burman ('flower', 'hair', and 'mouth') and Northern Austroasiatic ('water', 'meat') loanwords, while Biao has Hmong-Mien loanwords ('liver', etc.). Be and Jizhao have Eastern Austroasiatic loanwords (such as 'two', 'moon', and 'grass'), and Hlai has Malayo-Polynesian loanwords such as 'liver' and 'snake'.

These branches and languages were also in contact with each other, and formed linkages. Jizhao is one such example. My analysis of Jizhao shows that it is genetically most closely related to Be, but has many lexical items of Biao-Lakkia origin that are not found in Be. However, both Be and Jizhao show various similarities with Hlai, although Be and Jizhao are genetically closer to Tai and are not closely related to Hlai.

I consider Biao and Lakkia to form a single branch, but both split off from Proto-Biao-Lakkia quite early, similar to how Pramic and Khmu both split from Proto-Khmuic (or Mang and Pakanic from Proto-Mangic) relatively early, resulting in lower cognate percentages. Some shared lexical innovations between Biao and Lakkia not found in other Kra-Dai languages include 'stone', 'house', 'heavy', 'I', 'neck', 'weep', and so on. Although there have been earlier proposals considering Biao and Lakkia to be closely related to but not part of Kam-Sui, I have found no convincing evidence of Biao-Lakkia having a special relationship with Kam-Sui. In fact, I have found that Biao and Lakkia are just as similar to Tai as they are to Kam-Sui. Biao-Lakkia would have formerly been spoken across much of western Guangdong, and was later pushed into their residual areas due to the expansion of Yue Chinese. Biao then borrowed heavily from Yue Chinese, while Lakkia experienced moderate borrowing from Tai.

My tentative internal phylogeny of Kra-Dai is provided below, with subgroups organized geographically from north to south.
  1. Northern Kra-Dai linkage
    1. Kra
      1. Northern
      2. Southern
    2. Greater Kam-Sui linkage
      1. Kam-Sui
        1. Mulam
        2. Nuclear Kam-Sui
      2. Biao-Lakkia
  2. Southern Kra-Dai linkage
    1. Be-Tai
      1. Be-Jizhao
      2. Tai
        1. Northern
        2. Central-Southwestern
    2. Hlai
    3. Jiamao

However, Peter Norquest (p.c.) has suggested that Biao and Lakkia may each form separate branches within Kam-Tai.

Below is a computational phylogenetic analysis of Kra-Dai branches making use of 58 lexical items that I performed in December 2018. The subgrouping of Hlai and Jiamao is due to areal influence, and so is Be with Hlai.

2. Kra internal classification
Below is my tentative classification of the Kra languages as of August 2014. The classification of Southern Kra is based on Edmondson's notes on Kra classification (Edmondson, Jerold. 2011. Notes on the subdivisions in Kra. Published as Geyang yuyan fenlei buyi 仡央语言分类补议 in Journal of Guangxi University for Nationalities. 广西民族大学学报. 33.2.8-14.)

Northern Kra
    1. Hezhang Buyi
    2. Lachi
    3. Gelao [see below]
Southern Kra
    1. Guangxi Buyang (Yalhong)
    2. Laha, Paha
    3. Qabiao
    4. Yunnan Buyang: Ecun, Langjia, Nung Ven

Northern Kra languages have phonologies resembling those of neighboring Tibeto-Burman languages likely due to contact with Nasu, Tujia, Caijia, and Longjia, with tendencies for word compounding, loss of final stops, and lack of vowel length contrast.

Southern Kra languages have phonologies resembling those of Tai languages and Yue Chinese, final stops, and vowel length contrast. Laha of Noong Lay, like Saek and Baisha Hlai, has final -l, possibly due to contact with earlier forms of Lolo-Burmese languages and Austroasiatic languages (most notably Vietic) that tend to preserve final liquids (Ostapirat 1995).

The 2013 ASJP tree also provided very similar results for Kra.

2.1. Gelao internal classification
As of 2014 (with Dongkou Gelao from Li & Yang (2016) added in 2018), my tentative classification of the Gelao languages is as follows. Lachi is a sister clade of Gelao. See also the Gelao-Lachi spreadsheet posted at the bottom of this page.

1. Red Gelao
    a. Vandu
    b. A'ou
2. Core
    a. Dongkou Gelao
    b. White Gelao (Telue)
    c. Central Gelao
        1. Hagei
        2. Qau

Below are the computational phylogenetic results (as of August 2014) using SplitsTree 4.0, based on this basic vocabulary spreadsheet of Gelao and Lachi lects that I had compiled and coded for cognate sets.
Neighbor-joining (NJ) network:

The 2013 ASJP tree also provided very similar results for Gelao.

2.2. Clause-final negation in Kra
Many Kra languages, as well as Yang Zhuang of southwestern Guangxi, are also notable for having clause-final negation, an unusual grammatical feature that is not found in any other language of the China-Vietnam border region (Edmondson 2008; Jackson 2015). In Southeast Asia, this is quite atypical for Sinitic, Hmong-Mien, and Kra-Dai languages, but is very typical of some Tibeto-Burman languages, including Tujia and many Kuki-Chin-Naga languages. Based on evidence from toponyms and Zhuang folk traditions, Jinfang Li (1999) has also speculated about the possibility of Yang Zhuang having a Buyang (Kra) substratum.

Pre-verbal negators in Kra languages (ultimately of Tibeto-Burman origin)
Paha    pi55
Pudi    ma31
Judu    ma55
Wanzi    ma55

Post-verbal negators in Kra languages (ultimately of Tibeto-Burman origin)
Yalang    la31
Ecun    naai53
Langjia    laai31
Pudi    ʔo33
Qinglong    ʔa31
Judu    ʔo55
Wanzi    tʂo21

2.3. Kra and Pakanic
Kra and Pakanic languages have mutually influenced each other. The Pakanic languages consist of Bolyu and Bugan, and form the Mangic branch of the Austroasiatic language family along with the Mang language of Jinping County, Yunnan, China.

Excerpt from Hsiu, Andrew. 2016. A preliminary reconstruction of Proto-Pakanic.

The Bolyu people are not native to western Guangxi. They have migrated with the White Gelao people from western Guizhou Province to Guangxi Province during the 1800’s, around the time of the Taiping Rebellion and Opium War during the Qing Dynasty. At this time, many ethnic minorities from Guizhou and Guangxi provinces had migrated southward to the Sino-Vietnamese border region, forming what Jerold Edmondson calls a “language corridor” (Edmondson 1998). Bolyu presence in southwestern Guizhou is attested by many toponyms there that bear the name “Lai” (来). Bolyu oral tradition claims that the Bolyu had migrated from western Guizhou Province, including the area between Anshun City and Guanling County, and Xingren County (Li Xulian 1999). Another source of the Guizhou origin of Bolyu is that Bolyu has various loanwords from White Gelao, a Kra language which is spoken today in Judu village, Liuzhi Special District, west-central Guizhou Province, but used to have a much wider distribution throughout western Guizhou. Gelao loanwords include tshe¹ ‘person’, from Judu Gelao tshu²¹⁴ (which is in turn borrowed from Loloish languages), and the post-verbal negator ʔo², from Judu Gelao ʔo⁵⁵ (Ni 2010). Post-verbal negation is typologically rare in southern China, and is a defining feature of Kra languages. On the other hand, sporadic Pakanic loanwords have also been found in White Gelao of Fengyan, Malipo County, Yunnan, such as bu³⁵mai³¹ ‘nose’, cognate with Manlong Bugan pɯ⁵⁵maŋ³¹ ‘nose’ (cf. Bolyu lɔŋ2mi6).

Bugan also displays many Kra loanwords.

Edmondson, Jerold. 2008. "Kra or Kadai languages." In Diller, A., J. Edmondson, & Yongxian Luo, eds. The Tai–Kadai languages. London: Routledge. ISBN 978–0–7007–1457–5
Li Jinfang. 1999. Buyangyu yanjiu. Beijing: Central University for Nationalities Press.
Li, Jinfang 李锦芳; Yang, Liuyan 阳柳艳. 2016. Guizhou Shuicheng Gelaoyu chutan 贵州水城仡佬语初探. In Minzu Yuwen 民族语文 2016(3):70-81.
Li Xulian [李旭练]. 1999. A Study of Lai (Bolyu) [倈语硏究]. Beijing: Minzu University Press [中央民族大学出版社].
Jackson, Eric. 2015. Two-part negation in Yang Zhuang. Paper presented at SEALS 25, Payap University, Chiang Mai, Thailand.
Ostapirat, Weera. 1995. "Notes On Laha Final -l." In Linguistics of the Tibeto-Burman Area, 18 , no. 1: 173-181.