On the Phylogeny of the Hmong-Mien languages

Conference in Evolutionary Linguistics 2012 On the Phylogeny of the Hmong-Mien languages Yoshihisa Taguchi 田口善久（Chiba University, 千葉大学）

The Hmong-Mien (Miao-Yao)languages

1.Introduction • This is a preliminary report on the phylogeny of the Hmong-Mien languages. • This study explores the phylogeny of Hmong-Mien primarily based on lexical evidence but also seeks further support from phonological evidence. • We will focus on the phylogeny of the Hmongic languages in this presentation.

2. Previous studiesPurnell (1970) • The first serious study on Hmong-Mien phylogeny Figure 1. Hmongic phylogeny by Purnell (1970: 40)

Wang Fushi (1983) “On the dialect division of Miao language” “苗语方言划分问题” • He classified the languages spoken by the ethnic Miao into three dialects based on their phonological characteristics. The term “three major dialects of Miao” has been often used for designating major subgroups of the Miao language. Wang ：Purnell 黔东方言：EasternMiao(E) 湘西方言：NorthernMiao(N) 川黔滇方言：WesternMiaoand Central (W)

Strecker (1987) Figure 2. Strecker’s classification of Hmong-Mien (Adapted from 1987)

Wang and Mao (1995) Figure 3. Classification of Wang and Mao (Adapted from 1995: 2-3)

Major issues for discussion • The three Miao languages (E, N, W) are considered to be sisters of equal level. • The positions of Pa Hng(巴哼), KiongNai(炯奈), and Ho Ne(畲)are controversial. • The overall classification is mainly constructed according to the authors’ impression, without supporting evidence, or the supporting evidence fails to distinguish innovations and retentions.

Deng and Wang (2003) Figure 4. Phylogeny of Hmong-Mien in Deng and Wang (2003)

3. Method of analysis A method of computer-based lexicostatistics that utilizes Bayesian inference is used. The software used in this study was Mrbayes(http://mrbayes.sourceforge.net/index.php). • It is a character-based method. • It identifies the best trees with credibility scores. • Its validity has been widely acknowledged in linguistics and biology (Gray and Atkinson 2003, Greenhill and Gray 2009)

Table 1. Data points and sources

Target lects (languages/dialects) (1) The lect is mentioned in the language list of Wang and Mao (1995). (2) Sufficient lexical data of the lect are available. (3) Pana 18 lects of the Hmong-Mien languages (10 lects of the Hmongic languages)

Data analysis • Meaning list used: Culturally Appropriate Lexicostatistical Model for South East Asia (CALMSEA) wordlist (Matisoff 1978). 210 meaning items. • Cognacy decision: mostly based on Ratliff (2010) Hmong-Mien language history. • Loanword discrimination: based on Ratliff (2010) Hmong-Mien language history. • 493 characters for 18 lects.

Data file (part)

The conditions for calculation • The prior probability of each tree is the same. • The rate of change is the same for all the characters. • The number of generations to be calculated is 2 million. • Sampling rate is 100 generation. • The number of chains is four.

4. Calculation result • The standard deviation of splits was 0.003. • The value of the convergence diagnostic (potential scale reduction factor) was 1.000. • From the credible set of trees, 1218 trees were sampled. The 90% credible set had 160 trees; the 95% credible set had 316 trees, and the 99% credible set had 918 trees.

Figure 5. Consensus tree of Hmong-Mien

Figure 5. Consensus tree of Hmong-Mien (part)

5. Phylogeny of the Hmongic languages (1) The consensus tree constructed by the algorithm supports the findings of previous scholars: • The closest relations between Hmong (川黔滇) and Hmyo(落泊河) on the one hand (1.00), and Pu Nu (布努) and NauKlau(瑙格劳) on the other hand (1.00) <Strecker1987, Wang and Mao 1995>. • The relatively close relationship among these four languages (0.99) <Strecker 1987> • The close relation between KiongNai(炯奈)and Ho Ne (畬)(0.97) <Mao and Li 2002>.

5. Phylogeny of the Hmongic languages (2) The consensus tree also reveals some new findings. • The positions of Qo Xiong (湘西, Northern) and Pa Hng(巴哼). The consensus tree suggests that these two languages are positioned higher in the tree than other languages. • Concerning the position of Pa Hng, Benedict (1987), Strecker (1987) and Niederer (2004) have argued that Pa Hng may have been the first language to split off from the Hmongic branch.

5. Phylogeny of the Hmongic languages (2) The consensus tree also reveals some new findings. • The positions of Qo Xiong (湘西, Northen) and Pa Hng(巴哼). The consensus tree suggests that these two languages are positioned higher in the tree than other languages. • Concerning the position of Pa Hng, Benedict (1987), Strecker (1987) and Niederer (2004) have argued that Pa Hng may have been the first language to split off from the Hmongic branch.

Niederer 2004: 137-138

5. Phylogeny of the Hmongic languages (3) • Concerning the position of Qo Xiong (湘西, Northern), Ratliff (2010) made an important finding that Qo Xiong preserves some phonological distinctions that have been lost in other Hmongic languages.

Conclusions • Pa Hng (巴哼) and Qo Xiong (湘西, N) are the first languages to separate from the branch. • The other languages are divided into three main groups: the Western Hmongic group including Hmong, Hmyo, and two Pu Nu languages; the Ho Ne group including Ho Ne, KiongNai, and Pana; and Hmu (Qiandong). Thus, it is clear that the previous classification “three major dialects of Miao” needs a revision. • Ho Ne (畬)should not be treated as comprising an independent branch outside Hmongic. Rather, we recognize a Ho Ne group (畬)inside Hmongic.

A Cladogram of Hmongic

Thank you 谢谢

Figure 6. Geographical distribution of Hmongic subgroups

Table 2. A classification of the Hmongic languages

Previous Classifications

On the Phylogeny of the Hmong-Mien languages