Ta-Li languages

by Andreas Hölzl, last updated 01.08.2026

Overview

The Ta-Li languages, also known as Cai-Long 蔡龙 languages in Chinese, are presumably an archaic branch of Sinitic. The Ta-Li languages are located in Guizhou province, China, and include:

• the endangered Caijia 蔡家 language (Glottocode caij1234),
• the extinct but documented Longjia 龙家 language (Glottocode long1417),
• the extinct and only fragmentarily recorded Lujia 芦家 language (also known as Lu or Luren) (Glottocode lure1234),
• and perhaps the extinct and undocument language of the Songjia 宋家.

The name Ta-Li (Glottocode tali1265) was chosen by myself based on two words that (at least in this form) are specific to these languages and possibly exhibit archaic features inherited from Old Chinese like the lateral:

Note that the term Ta-Li thus includes all three attested languages while the term Cai-Long only refers to Caijia and Longjia and, crucially, is based on exonyms.

That the three languages are genetically related has been suspected since the 1950s based on superficial and especially lexical similarities. Recently, Hölzl & Hölzl (2026) and Hölzl (2026: 26-33) presented clear lexical, grammatical, and structural (typological) evidence for such a relationship. Note that the phonological details (some of which are also discussed in recent papers by Lee Man Hei) still need to be worked out in more detail.

Genealogically unrelated but nevertheless connected to the Ta-Li langages are the following two languages (see Hölzl 2026 for more details):

• a Trans-Himalayan language called “Nanjing” (no Glottocode), seemingly closely related to Dafang Yi (Glottocode dafa1241) and in contact with Longjia,
• and a possibly isolated language also known as Lu or Luren language (no Glottocode).

The “Nanjing language” (also referred to as Awupo language) should not be confused with the Sinitic variety spoken in or around the city of Nanjing. The name Nanjing derives from the problematic notion that the Longjia people at least in parts came from Nanjing. Note that only one description of the Nanjing language is currently available:

• Duan, Ling & Xinglian Zhang. 2008. Yunnan, guizhou, hunan baizu qige yuyan diaocha ziliao. In Xu Lin (ed.), Dali congshu: baiyu pian, vol. 4, 2902–3076. Kunming: Yunnan minzu chubanshe. (pages 3043-3056 are on the Nanjing language)

In order to differentiate between the two languages called “Luren”, I refer to the one as Luren A (or Lujia) and to the other as Luren B. Luren B possibly consitutes a stock of its own (i.e., it could be a so-called language isolate). Hölzl (2026) proposes the term Jinshaic for that stock since the only known representative was recorded in Jinsha county.

Classification

The Caijia are not recognized as a separate minority but are classified as a variety of different minorities today. This probably accelerates the dissolution of the former Caijia communities. The Caijia language was sometimes misclassified as belonging to the Kra-Dai family.

The Longjia language is often misclassified as a Baic variety because the former Longjia today officially belong to the Bai minority. Therefore, some publications may be overlooked since titles make reference to the Bai language.

Since the former Luren are classified as Manchu minority today, both Luren languages were at some point wrongly referred to as Manchu language (see Hölzl & Hölzl 2026 for details).

The Ta-Li languages were often grouped either with Baic or with Sinitic languages. However, while the three groups could be related, Caijia, Longjia, and Lujia are clearly more closely related to each other than any of them is to Baic or Sinitic. Hölzl & Hölzl (2026: 18-30) present preliminary evidence for shared features and elements from all three languages. Note that data from Caijia and Longjia can help understand the fragmentary materials of the Lujia language (e.g., numeral classifiers), which is only possible due to their close relationship.

Given the problematic classifications of the Caijia, Longjia, and Lujia, research has been delayed and hindered by the mistaken reference to cultural and linguistic elements from the Bai in Yunnan or the Manchus in the Northeast of China.

The Ta-Li languages seem to share the preservation of the Old Chinese laterals with Waxiang Chinese in Hunan, which could represent a common retention (although Huang Hongjiang recently argued against this). However, to the best of my knowledge, there seem to be few further indications for a closer relationship between these languages (apparent lack of common innovations).

Future research is necessarily based on the comparative method and will have to demonstrate regular sound and tone correspondences between at least Caijia and Longjia, for which sufficient information is available. In a second step, the relationship with Baic and Sinitic needs to be clarified.

Typology

The three languages exhibit many features characteristic for the Mainland Southeast Asian area, including many monosyllabic words, little morphology, tones, or numeral classifiers (see Lü 2022 on Caijia). In addition, they exhibit many features characteristic for Sinitic languages, such as unmarked VO but marked OV order (positional differential argument flagging involving the same verb), prenominal relative clauses (Hölzl & Hölzl 2026). Hölzl (2026: 32-33) demonstrates a close typological similarity between Caijia and Longjia.

Status

Caijia, albeit endangered, is the only Ta-Li language still spoken today. Longjia has become extinct during the latter half of the 20th century and Lujia during the first (Hölzl 2026: 28). Songjia, if related to the Ta-Li languages, must have already disappeared earlier and remained entirely undescribed.

Available Data

Caijia

Caijia is the only language with a relatively complete description, including a grammar and publications on various aspects of the grammatical structure in Chinese and English. Despite frequently being referred to as a recently discovered language, the language was actually first described in the 1950s. (Not to mention that the Caijia people have been known hundreds of years before that time.) These data from the 1950s have never been published but some are in my personal possession and are currently being prepared for publication (see also Hölzl & Hölzl 2026 and Hölzl 2026: 27-28 for some examples and a description of the data). The first publicly available descriptions, although difficult to access, are from the 1980s. A short sketch of the language in Chinese has been published in 2004 by Bo Wenze. For many years, this remained the only widely accessible publication of the language. The main reference work today is the excellent grammar written in English by Lü Shanshan that was published in 2022.

Here is a more or less exhaustive list of publications on the Caijia language (as of 2026):

• Bo Wenze 薄文泽. 2004. Caijiayu gaikuang 蔡家话概况 [A sketch of Caijia]. Minzu Yuwen 民族语文 2. 68–81. (the first widely available sketch of the language)
• Che Guangxian 车光贤 et al. (eds.). 2003. Liupanshuishi zhi: minzu zhi 六盘水市志：民族志 [Gazetteer of Liupanshui: Ethnic section]. Guiyang: Guizhou renmin chubanshe. (contains a short Caijia word list in Chinese characters on p. 184, notably published before Bo 2004)
• GMSWSB 1982 = Caijia de yuyan 蔡家的语言 [The language of the Caijia]. In Guizhousheng minzu shiwu weiyuanhui shibie bangongshi 贵州省民族事务委员会识别办公室. Guizhou minzu shibie ziliaoji, vol. 8, longjia, caijia 贵州民族识别资料集, 110–125. Guiyang. (Unpublished manuscript.)
• Hsiu, Andrew. 2018. The Niujiaojing dialect of Caijia in Weining County, Guizhou, China. (Unpublished manuscript.)
• Hu Hongyan 胡鸿雁. 2013. Caijiahua daici xitong tanxi 蔡家话代词系统探析. Minzu yuwen 民族语文 6. 34–40.
• Lee, Man Hei. 2023. Phonological features of Caijia that are notable from a diachronic perspective. Journal of Historical Linguistics 13(1). 82–114. (an excellent approach to the historical phonology of Caijia)
• Lee Man Hei. 2024. Bai and Old Western Chinese. Bulletin of the School of Oriental and African Studies 87(3). 533–555.
• Lü Shanshan 吕珊珊. 2017. Caijiahua qizhong youbiju de leixingxue kaocha 蔡家话七种优比句的类型学考察. Fangyan 方言 3. 341–351.
• Lü Shanshan. 2018. Two locative constructions in Caijia from the typological perspective of Asian languages. Studies in Language 42(3). 600–640.
• Lü Shanshan 吕珊珊. 2019a. Caijiahua dengbiju de leixingxue kaocha 蔡家话等比句的类型学考察. Yuyan kexue 语言科学 99. 160–175.
• Lü Shanshan. 2019b. Equative and similative constructions in Caijia (Sino-Tibetan). Faits de Langues 50(1). 203–225.
• Lü, Shanshan. 2022. A reference grammar of Caijia: An unclassified language of Guizhou, China (Sinitic Languages of China 8.) Berlin: Mouton. (the first and only grammar on Caijia)
• Lü, Shanshan & Hilary Chappell. 2026. Non-verbal predication in Caijia and Waxiang: A Southeast Asian perspective. In Pier Marco Bertinetto , Luca Ciucci & Denis Creissels (eds.), Non-verbal predication in the world’s languages: A typological survey, volume 1: Eurasia, North America, South America, 95-128. Berlin: de Gruyter.
• Majewicz, Alfred F. 2023. Review of Lü (2022). Rocznik Orientalistyczny 76. 176–184.
• Nohara, Masaki 野原將揮. 2026. Hezhang xian songlinpo xiang cajiahua yinxi gaikuang 赫章县松林坡乡蔡家话音系概况 [An overview of the phonological system of Caijia in Songlinpo, Hezhang County]. In Takekoshi Takashi 竹越孝, Nohara Masaki 野原将挥 & Suzuki Hiroyuki 铃木博之 (eds.), Tianshang dafeng Mitsuaki Endo jiaoshou tuixiu jinian lunwen ji 天上大风 远藤光晓教授退休纪念论文集 [Sky above, great wind: Festschrift in honour of Professor Mitsuaki Endo on his retirement], 200-214. Tōkyō: Nihon chiri gengogakkai 日本地理言語学会. (an excellent description of a previously undescribed variety of Caijia, contains a word list with almost 500 entries)
• Zhang Yongbin 张永斌. 2018a. Binwei yuyan caijiahua liangci xitong tanxi 濒危语言蔡家话量词系统探析. Anshun xueyuan xuebao 安顺学院学报 20(2). 40–47.
• Zhang Yongbin 张永斌. 2018b. Caijiahua fuju jiqi neibu guanxi kaocha 蔡家话复句及其内部关系考察. Honghe xueyuan xuebao 红河学院学报 16(6). 89–95.
• Zhang Yongbin 张永斌. 2018c. Xin faxian binwei caijiahua xingrongci xitong tanxi 新发现濒危蔡家话形容词系统探析. Guizhou gongcheng yingyong jishu xueyuan xuebao 贵州工程应用技术学院学报 195. 102–110.
• Zhang Yongbin 张永斌. 2018d. Caijiahua cihui jiqi laiyuan xitong tanxi 蔡家话词汇及其来源系统探析. Liupanshui shifan xueyuan xuebao 六盘水师范学院学报 30(5). 1–7. (highly problematic)
• Zhengzhang Shangfang 郑张尚芳. 2010. Caijiahua baiyu guanxi ji cigen bijiao 蔡家话白语关系及词根比较 [Relation of the Caijia and Bai language and comparison of lexical roots]. In Pan Wuyun 潘悟云 & Shen Zhongwei 沈钟伟 (eds.), Yanjiu zhi le 研究之乐 [The joy of research], 389–400. Shanghai: Shanghai jiaoyu chubanshe. (largely outdated comparisons of Caijia with Baic and Sinitic)

Caijia numerals can be found in the Numeral Systems of the World’s Languages database. An online word list was put together by Lee Man Hei. The endangered languages Archive (ELAR) contains audio recordings and videos of Caijia. Unpublished Caijia data collected by Andrew Hsiu can be accessed online as well.

Longjia

Longjia was first described in the 1920s but remains almost unknown until today. While only very few data of the Longjia language have actually been published, most of which are written in Chinese and are either inaccessible or unreliable, the language has been recorded in several fieldnotes from the 1950s and the 1980s that are in my personal possession. In addition, I have non-published sketches of the language from the 1980s that are much more reliable than the published versions. An overview of most available data can be found in Hölzl (2021: 25-30, 2026: 30-31). Further unpublished data from southern Longjia (probably mostly from Anshun) have recently been discovered and are currently being analyzed. Next to lexical items, these contain many additional sentences that make the grammar more accessible. Together with Yadi Hölzl, I am working on a grammar and a dictionary based on all available data.

Lujia

Lujia (or Lu, Luren) is the least well attested of the Ta-Li languages. It was described from the 1950s until the 1980s, at a time when already no fluent speakers of the language were left. As a consequence, the language has only been fragmentarily recorded in a couple of word lists that were recently published and analyzed in Hölzl & Hölzl (2026). We are currently working on a paper addressing the cultural, genetic, and historical background of the former Luren (now classified as Manchus).

Songjia

There are currently no linguistic data on the Songjia. However, historical evidence suggests a close connection with the Ta-Li speakers.

Publications

Books

Hölzl, Andreas. 2026. Studies in East Asian Comparative Linguistics. Potsdam: University of Potsdam. (Cumulative habilitation thesis.) URL DOI

Papers

• Hölzl, Andreas & Yadi Hölzl. 2026. The lost voices of Guizhou: Rediscovering the languages of the Luren. Cahiers de Linguistique Asie Orientale 55. 119–157. DOI
• Hölzl, Andreas. 2021. Longjia (China): Language Contexts. Language Documentation and Description 20. 13–34. URL

Reviews

• Hölzl, Andreas. 2023. Review of Lü (2022). Bulletin of the School of Oriental and African Studies 86(2). 398–399. DOI

Talks

• Hölzl, Andreas. 2023. Describing a recently extinct language: The case of Longjia (Sino-Tibetan). Vielfaltslinguistik 5, 19–20.05.2023, Hamburg.
• Hölzl, Andreas. 2022. Longjia: An archaic Sinitic language. Centre de recherches linguistiques sur l’Asie orientale, CRLAO, 25.10.2022, Paris.
• Hölzl, Andreas. 2022. A typological profile of Longjia, an archaic Sinitic language. 14th Conference of the Association for Linguistic Typology (ALT), 15–17.12.2022, Texas at Austin.
• Hölzl, Andreas. 2022. The diachrony of mixed directionality in Sino-Tibetan: Rel-N & V-O in Caijia and Longjia. The 25th International Conference on Historical Linguistics (ICHL), 01–05.08.2022, Oxford.
• Hölzl, Andreas. 2022. Unusual word order features in Longjia and Caijia. Syntax Semantik Kolloquium, 26.07.2022, Potsdam.
• Hölzl, Andreas. 2021. Prenasalization in Longjia. The 34th Paris Meeting on East Asian Linguistics (JLAO), 07–09.07.2021, Paris.
• Hölzl, Andreas. 2021. The Lu(ren) languages of Guizhou, China. 9th International Constrastive Linguistics Conference (ICLC), 20–21.05.2021, Santa Margherita-Portofino.
• Hölzl, Andreas & Yadi Hölzl. 2019. The endangered languages of the Manchus. Endangered languages in Northern Asia (Internationaler Workshop), Institut für Turkologie, Freie Universität Berlin, 29–30.11.2019, Berlin.

---