Home‎ > ‎Blog‎ > ‎

Linking the Sino-Tibetan fallen leaves

Andrew Hsiu
Updated July 21, 2018
Please cite as: Hsiu, Andrew. 2018. Linking the Sino-Tibetan fallen leaves. <https://sites.google.com/site/msealangs/home/blog/fallen-leaves>.
Please note that this is a working draft that will be periodically updated.

You may also wish to see my other site:

Sino-Tibetan lexical database
A spreadsheet of Sino-Tibetan branches (ST branches database_Hsiu2018.xlsx) can be downloaded at the bottom of this page. It includes recent data of many languages not found in STEDT, as well as my own preliminary reconstructions of rGyalrongic, Gong, Tujia, Kho-Bwa, Angami-Pochuri, Zeme, Mijuish, Digaro, Dhimalish, and Raji-Raute. It can be cited as: Hsiu, Andrew. 2018. Sino-Tibetan lexical database. m.s.

1. List of "fallen leaves"

Below are some Sino-Tibetan branches that I propose are best left unclassified as "fallen leaves" (i.e., non-ambiguous coherent branches), the concept of which is further discussed in van Driem (2014).
  1. Cai-Long languages: Longjia, Caijia, Luren
  2. Kathu
  3. Gong
  4. Taman
George van Driem (2014) lists 42 "fallen leaves".
  1. Bodish
  2. Tshangla
  3. West Himalayish
  4. Tamangic
  5. Newaric
  6. Kiranti
  7. Lepcha
  8. Magaric
  9. Chepangic
  10. Raji–Raute
  11. Dura
  12. 'Ole
  13. Gongduk
  14. Lhokpu
  15. Siangic
  16. Kho-Bwa
  17. Hrusish
  18. Digarish
  19. Midžuish
  20. Tani
  21. Dhimalish
  22. Brahmaputran
  23. Pyu
  24. Ao
  25. Angami–Pochuri
  26. Tangkhul
  27. Zeme
  28. Meithei
  29. Kukish
  30. Karbi
  31. Mru
  32. Sinitic
  33. Bai
  34. Tujia
  35. Lolo-Burmese
  36. Qiangic
  37. Ersuish
  38. Naic
  39. rGyalrongic
  40. Kachinic
  41. Nungish
  42. Karenic
I have revised van Driem's (2014) fallen leaves list as follows, with some leaves consolidated, some added, and some renamed.
  1. Tibetic
  2. East Bodish
  3. Tshangla
  4. West Himalayish
  5. Tamangic
  6. Greater Magaric: Magaric + Chepangic + Raji-Raute + Dura (4 fallen leaves)
  7. Newaric: Newar + Baram-Thangmi (2 fallen leaves)
  8. Kiranti
  9. Lepcha
  10. 'Ole
  11. Gongduk
  12. Lhokpu
  13. Siangic
  14. Kho-Bwa
  15. Hrusish
  16. Digarish
  17. Mijuish
  18. Tani
  19. Dhimalish
  20. Sal: Bodo-Garo + Konyak + Kachin-Luic (3 fallen leaves)
  21. Pyu
  22. Taman
  23. Ao
  24. Angami–Pochuri
  25. Tangkhulic
  26. Zeme
  27. Meithei
  28. Kuki-Chin
  29. Karbi
  30. Mruic
  31. Sinitic
  32. Tujia
  33. Nungish
  34. Karenic
  35. Gong
  36. Burmo-Qiangic linkage: 19 fallen leaves total

Combining the 19 Burmo-Qiangic fallen leaves with the 35 non-Burmo-Qiangic leaves would give us a total of 54 fallen leaves. If Sal, Newaric, and Greater Magaric were to be split up, then that would bring the total up to 60 fallen leaves. Thus, a thoroughly comprehensive phylogenetic analysis of the Sino-Tibetan language family would need to include at least 60 taxa.

Once coherent branches are identified, it can be difficult to put them together, as we have seen with Indo-European branches, which continue to defy classification attempts despite having more than a century of rigorous research done on them.

STEDT is a wonderful resource that contains data for most but not all of these fallen branches. Unfortunately, it is not being updated another since its last release was in 2015.
  1. Proto-Luish [Matisoff 2013]
  2. Proto-Ersuic [Yu 2012]
  3. Proto-Bai [Wang 2006]
  4. Cai-Long
  5. Kathu
  6. Dura
  7. Gongduk
  8. 'Ole
  9. Lhokpu
  10. Pyu
  11. Taman
  12. Baram-Thangmi
  13. Hrusish
  14. Siangic (Koro-Milang)
  15. Kho-Bwa
Here are my suggestions for proto-languages that have yet to be reconstructed for fallen leaves that contain more than 2-3 individual languages:
  1. Proto-West Himalayish [only several proto-forms by Widmer 2014, 2017]
  2. Proto-Greater Magaric [proposed by Schorer 2016, but no reconstruction]
  3. Proto-Angami-Pochuri [most Angami-Pochuri quality data still from Marrison 1967]
  4. Proto-Zeme [most Zeme quality data is still from Marrison 1967]
  5. Proto-Nungish [currently being reconstructed by Nathan Straub]
  6. Proto-rGyalrongic [which I am currently attempting]

2. Some notes on selected Tibeto-Burman "fallen leaves"

Below are my personal speculations on the classifications of some Tibeto-Burman "fallen leaves."

Pyu appears to be an independent branch that shares some similarities with some neighboring Tibeto-Burman branches such as Meithei, Kuki-Chin, and Sal. Bradley (1997) also suggests that Pyu may be related to the Sal languages. Like Meithei, Pyu has PTB *s- > h-. It looks like Luish languages were in contact with Pyu, but are not directed related to it. I have compiled <Pyu.xlsx>, which has Pyu data from Luce (1985) and comparisons from the STEDT database.

Mruic (Mru-Hkongso), a branch consisting of Mru, Anu, and Hkongso, is definitely not Lolo-Burmese, contrary to Löffler (1966). Its autonym is reminiscent of certain Austroasiatic autonyms such as "Bru." Austroasiatic words that look similar to "Mru" include Proto-Khasic *brəəw 'person', Proto-Khmuic *-brɔʔ 'man, person', and Proto-Khmuic *kmraʔ 'person' (MKED database). Mru does not appear to have many Austroasiatic loanwords, but its phonology is highly Mon-Khmer-like with sesquisyllables, final liquids, and initial consonant clusters with liquids. A similar parallel is Kerinci Malay, which has a Mon-Khmer-like phonology with sesquisyllables and Mon-Khmer-like diphthongs, but no recognizable Austroasiatic loanwords. However, there has not yet been any solid argument for Mru having an Austroasiatic substratum. Peterson (2009) notes similarities between Mruic and Bodo-Garo. My personal view is that is in DeLancey's (2015) Central Tibeto-Burman group, but does not subgroup with Sal or Kuki-Chin-Naga languages. I have found that Mruic has some similarities to Sal languages in general, not just Bodo-Garo, which would mean that Proto-Mruic would have been in contact with Sal languages in the Chindwin River valley.

Gong is certainly not Lolo-Burmese, and may be a remnant of what was a more widespread Tibeto-Burman branch in the Myanmar-Thai borderlands located to the south of the Karenic-speaking area. It shows an unusual development from PTB *s- > ʔ- or zero, which had developed from ʔl- according to Bradley. Like Karenic, the ancestors of the Gong would have migrated down the Salween watershed (or perhaps even down the Mekong River) from the Tibeto-Burman heartland further up to the north. However, unlike Karenic, Gong word order is SOV rather than SVO (Mayuree 2006). Gong shows lexical links with Eastern Sino-Tibetan branches (Burmo-Qiangic, Karenic, Tujia, etc.) in general, but does not subgroup with any particular branch. However, the words for 'cattle' and 'horse' are from Sino-Tibetan branches in northern Myanmar, pointing to the former presence of Gong in the middle reaches of the Salween River valley (i.e., modern-day Shan State) before migrating down to its current location in the Kanchanaburi area of Thailand, located just to the southeast of the mouth of the Salween River.

Kathu shows similarities to Tibeto-Burman languages in the Myanmar-China-India border triangle, such as Nungic, Tani, Karenic, Taraon, Nusu, Naic, Qiangic, and others. My hypothesis is that Kathu (or "pre-Kathu") had split off independently from Proto-Tibeto-Burman and was in contact with other eastern branches of Tibeto-Burman in its former range, which would have been located much further to the northwest of its current distribution.

Cai-Long languages: Caijia, Longjia, and Luren of western Guizhou have no immediately recognizable close relatives, but Bai has been suggested as a close relative (Zhengzhang 2010). Waxiang Chinese is related according to Sagart (2011). However, while the superstrata of Bai and the Cai-Long languages belong to similar forms of western Old Chinese dialects, the substrata languages of Bai and Cai-Long constitute separate, independent southern branches of Burmo-Qiangic. The superstratum of Bai and Cai-Long was the first to split off from Old Chinese, even before Min (following Sagart 2011 on Caijia-Waxiang as the first branch to split from Old Chinese). These languages all have Burmo-Qiangic substrata, but they would have belonged to different branches of Burmo-Qiangic. Hence, the superficial similarities between Bai and the Cai-Long-Waxiang are due to (1) shared features due to the same Old Southwestern Chinese superstratum that is common to Bai, Cai-Long, and Waxiang, and (2) shared retentions from Proto-Burmo-Qiangic in the substratum layer. Thus, pre-Bai and pre-Cai-Long-Waxiang were each separate branches of Burmo-Qiangic. Pre-Cai-Long-Waxiang would have had a much wider geographical distribution in the past, and would have been spread across much of northern Guizhou.

3. Linked leaves

I believe that historically, many Sino-Tibetan branches formed linkages with each other, with transmission probably taking place both vertically and horizontally. They may or may not have had the same common ancestors, but they definitely did mutually influence each other through mutual contact. The following Sino-Tibetan classification is partially based on my unpublished computational phylogenetic analysis of Sino-Tibetan branches, as well as previous proposals of Sino-Tibetan groupings by George van Driem, David Bradley, Guillaume Jacques, Scott DeLancey, Nicolas Schorer, and others. I am currently compiling comparative lexical evidence for this classification. (Individual "fallen leaves" within the linkages are listed in geographical order, either from west to east or from north to south.)

In my proposed classification, the Himalayas serves as a geographical barrier separating Sino-Tibetan into a Greater Western Sino-Tibetan linkage (originating in Assam and northern Myanmar), as well as a Greater Eastern Sino-Tibetan linkage (originating in the Upper Yangzi watershed region - Sichuan, Yunnan, and perhaps also the SE Tibet area).

  1. Greater Western Sino-Tibetan linkage
    1. Far Western Sino-Tibetan ("Nepal-Bhutan") linkage
      1. Raji-Raute
      2. Dura
      3. Kham-Magar-Chepang
        1. Kham
        2. Magaric
        3. Chepangic
      4. Newaric [Turin 2004]
        1. Newar
        2. Baram-Thangmi
    2. Greater Kiranti
      1. Kiranti
      2. Tshangla
      3. Dhimalish [Grollmann 2017]
        1. Lhokpu (+ substratum of Dzongkha [van Driem 2001])
        2. Dhimal
        3. Toto
    3. Ole
    4. Lepcha [Bodman 1988]
    5. Gongduk (+ substratum of East Bodish) [van Driem]
    6. Tani (which has a Greater Siangic substratum) [Blench 2014]
    7. Miju-Meyor [DeLancey 2015]
    8. Central Sino-Tibetan branch (and/or linkage) [DeLancey 2015]
      1. Karbi
      2. Kuki-Chin-Naga branch
        1. Northern (Ao-Angami) [Coupe 2012]
          1. Central Naga (Ao)
          2. Angami-Pochuri
        2. Southern
          1. Zeme (Zeliangrong)
          2. Tangkhulic
          3. ? Koki (Kokak)
          4. Kuki-Chin
      3. Sal linkage [Matisoff 2013]
        1. Bodo-Garo
        2. Konyak (Northern Naga)
        3. Jingpho-Luish
      4. Meithei
      5. Taman [Huziwara 2016]
      6. Pyu
      7. Mruic [Peterson 2009]
  2. Greater Eastern Sino-Tibetan linkage
    1. Hrusish (Hruso-Miji)
    2. Kho-Bwa
    3. Koro-Milang (Siangic)
    4. Nuclear Eastern Sino-Tibetan linkage [Bradley; van Driem]
      1. Idu–Taraon (Digarish)
      2. Bodish (Tibetan Plateau linkage)
        1. West Himalayish [Widmer 2014]
        2. Tamangic (TGTM)
        3. Kaike [Honda 2008]
        4. Tibetic
        5. East Bodish
        6. Basum
      3. Nungish
      4. Karenic
      5. Gong
      6. Kathu
      7. rGyalrong
      8. Horpa-Lavrung
      9. "Southeast Chamdo" [Suzuki & Nyima 2018]
        1. Lamo
        2. Larong
        3. Drag-yab
      10. Burmo-Qiangic branch (and/or linkage) [Jacques & Michaud 2011]
      11. Tujia
      12. Sinitic
      13. "Donor Hmong-Mien" [see Ratliff 2010; Benedict 1987]
      14. "Donor Kra" [see Ostapirat 1995, etc.]
      15. "Donor Jiamao" [see Hsiu 2018]

Some of my personal observations (new proposals):
  • Tibeto-Burman branches that may have received Austroasiatic influence include those in the Kuki-Chin-Naga linkage, Tani, Mru, and perhaps some Lolo-Burmese languages.
  • From my own lexical comparisons, I have determined that Kuki-Chin-Naga is a branch with a common ancestral language, consisting of a Northern and Southern branch. Meithei and Karbi are not within Kuki-Chin-Naga proper, but are part of the wider Central ST linkage.
  • Sal is likely a branch as well, but Dhimalish is not part of it.
  • Idu-Taraon, Miji, Miju, Puroik, and Kho-Bwa are all early splits from Proto-Tibeto-Burman rather than isolates. All branches clearly show Tibeto-Burman forms in their basic vocabulary. Only Hrusish and Siangic look to me as if they are likely non-Sino-Tibetan languages. Hruso, with its long strings of consonant clusters reminiscent of Salish and Berber, has a very non-Sino-Tibetan look to it.
  • Digaro looks like it has Siangic + Central ST + Eastern ST elements.
  • Tamangish is similar to Bodish and may form a branch with it.
  • Gongduk has some Central ST forms.
  • Burmo-Qiangic is a linkage that has the internal diversity of Sal or perhaps even Central Tibeto-Burman. "Qiangic" is a linkage consisting of rGyalrongic, Rma, Pumi, Muya (Minyak), Zhaba / Queyu, Guiqiong, etc. that has the internal diversity of Kuki-Chin-Naga. It does not include Naic and Ersuic, which are divergent and are separate branches within the Burmo-Qiangic linkage.

Many branches in fact have at least two lexical layers, as summarized below (Note: NAA = Northern Austroasiatic):

Branch = Stratum 1 + Stratum 2
Old Chinese = EST + pre-AN?
Bai = Burmo-Qiangic + OC
Cai-Long = Burmo-Qiangic + OC
Baima = Burmo-Qiangic + Tibetic
rGyalrongic = EST + Burmo-Qiangic
Lolo-Burmese = Burmo-Qiangic + AA?
Kathu = EST + Mondzish
Tujia = EST + Burmo-Qiangic
Angami-Pochuri = EST + Kuki-Chin-Naga (CST)
Mru-Khongso = CST + NAA
Sal = CST + NAA
Kuki-Chin-Naga = CST + NAA
Dhimal-Toto = CST + WST
Kusunda = Kusunda + WST
Idu-Taraon = EST + Greater Siangic
Koro = Greater Siangic + Hruso
Milang = Greater Siangic + Padam (Tani)
Tani = CST + Greater Siangic
Miju-Meyor = CST + Idu-Taraon (EST + Greater Siangic)
Lepcha = CST + NAA
Karenic = EST + NAA
Nungish = EST + CST
Gongduk = CST + East Bodish
Tshangla = WST + Tibetic
East Bodish = Gongduk (CST + WST) + Tibetic
Dzongkha = Lhokpu (WST) + Tibetic
Ole = WST + East Bodish
Hruso-Miji = Hruso-Miji + Greater Siangic
Kho-Bwa = Kho-Bwa + Greater Siangic

In my view, based on lexical evidence, languages in the Eastern Sino-Tibetan linkage roughly diversified as follows, from earliest to latest.
Eastern Sino-Tibetan linkage
- Hruso-Miji; Kho-Bwa; Koro-Milang
-- Sinitic
--- Bodish
---- Tujia
----- Karenic
------ Gong
------- Nungish
-------- Kathu
--------- rGyalrong; Horpa-Lavrung
---------- pre-Jiamao
["Burmo-Qiangic" branches below]
------------ Bai, Cai-Long
------------- Qiangic
-------------- Ersuic
--------------- Naic
---------------- Burmish
----------------- Loloish

This is where the wave model and tree model both come into play. Combinations of vertical and horizontal transmission have complicated phylogenetic trees since the earliest life forms arose, with the best-known examples being those of multiple cross-breeding events among early human species, and constant gene transfers among Bacteria, Archaea, and Eukarya.

Other than the four major linkages listed above, the other languages would be best left as primary branches of Tibeto-Burman. Certain Tibeto-Burman languages of Arunachal Pradesh may have non-Tibeto-Burman substrata, resulting in divergent lexical items.

The NeighborNet results of my 2014 trial run of Sino-Tibetan fallen leaves using SplitsTree 4.0 is shown in the screenshot below. Cognate sets of 18 vocabulary items mainly from STEDT were included.

Phylogram using the Neighbor-Joining (NJ) algorithm:

3.1. River systems, linkages, and dispersals

How did these four Sino-Tibetan linkages (or "blocs") form? My hypothesis is that these four major linkages largely followed river systems, and in the case of Western Sino-Tibetan, the Himalayan mountain range.

- Western Sino-Tibetan: Himalayan foothills
- Central Sino-Tibetan: Chindwin River watershed
- Eastern Sino-Tibetan: Yangtze River watershed and Salween River watershed
- Bodish: Brahmaputra River watershed

Mountain ranges separated the Irrawaddy River system (home to Central Sino-Tibetan) from Western Sino-Tibetan and especially Eastern Sino-Tibetan. Passes and routes through the Naga Hills India from Myanmar have allowed for multiple prehistoric migrations from the Chindwin River region into Assam, where the Brahmaputra River system would have facilitated migrations further westward and also into the Himalayas. Some branches that originated in the Central Sino-Tibetan area but had migrated northwest into the Northeast India area are Bodo-Garo, Lepcha, Karbi, Dhimalish, and perhaps also Tani, Gongduk, and Miju.

Here is my tentative timeline of how Sino-Tibetan diversified.
- Early Neolithic (9,000 BP - 7,000 BP): Sino-Tibetan as a language family starts taking shape as hunter-gatherers spread out throughout the Brahmaputra, Irrawaddy, and Upper Yangtze river systems gradually become sedentary and adopt mixed subsistence strategies, including both foraging and early forms of agriculture. The Holocene Climate Optimum (i.e., improving climate after the Last Glacial Maximum) would have allowed for demographic expansions and increasing trade and marriage networks. The Western, Central, and Eastern linkages begin to form in their respective river valley systems.
- Middle Neolithic (7,000 BP - 5,000 BP): Various early Western Sino-Tibetan branches expand from the Brahmaputra Valley into Bhutan and Nepal, following the Himalayan foothills westward. Agriculture arrives from the east via the Upper Yangtze river system into the Greater Himalayan region. Early Eastern Sino-Tibetan branches also expand down the Yangtze River. rGyalrong, Horpa-Lavrung, and Tujia, expand upstream from the Sichuan Basin, while Sinitic expands across the Central China Plains and comes into contact with early Altaic and pre-Austronesian speakers, and perhaps also people who spoke various langauges of unknown affiliation.
- Late Neolithic (5,000 BP - 4,000 BP): A major climate catastrophe (the 4.2 kiloyear or Bond-3 event) occurs about 4,200 B.P., which causes massive flooding across central and southern China. This results in massive population dislocations across the Yangtze River valley, causing agricultural populations to disperse towards the south and west. Austroasiatic speakers, originally located in Guangxi, migrate into Indochina as a result of climate change, social changes, and introduction of agriculture from Middle Yangtze area.
- Bronze Age (4,000 BP - 2,500 BP): Sal and Kuki-Chin-Naga start expanding rapidly across northern and central Myanmar as groups of resident Sino-Tibetan speakers adopt technology and agriculture from incoming Austroasiatic speakers, also mixing with them. Austroasiatic speakers arrive from the east approximately 3,500 BP, which is when the Bronze Age of north-central Myanmar  begins. Sinitic also expands rapidly due to the expansion of the Shang Dynasty, absorbing linguistic diversity in the Middle Yangtze and helping to stimulate the formations of new Bronze Age creoles in southern China, namely Hmong-Mien and Kra-Dai. Burmo-Qiangic, associated with the Sanxingdui Culture, also expands rapidly from the Sichuan Basin using various Upper Yangtze tributaries such as the Min, Yalong, Dadu, and Jinsha river systems. Expansions from the Sichuan Basin into the Yunnan Plateau occur due to a combination of demographic pressures and searching for natural resources. Burmo-Qiangic absorbs earlier Sino-Tibetan diversity and influences branches in upstream areas such as rGyalrong, Horpa-Lavrung, and Tujia. Bodish expands across the Tibetan Plateau, aided by Sinitic technology and new cold-resistant crops introduced from the north and east via Altaic speakers.
- Iron Age (2,500 BP - 1,700 BP): Most Sino-Tibetan branches are established and are more or less in their present locations. They continue to expand and assimilate residual Sino-Tibetan branches. Sinitic, Tai, Lolo-Burmese, and Tibetic spread rapidly and assimilate branches that had diverified during the Bronze Age and earlier.

Some maps are shown below.

Map set 1: Proposed dispersal of Tibeto-Burman branches. Branch names (otherwise known as George van Driem's "Trans-Himalayan fallen leaves") are labeled in different colors according to the linkage or geographic area that they are in. Some of my proposed new "fallen leaves" in the Eastern Tibeto-Burman area are labeled in orange.

Map 2: Single-crop rice regions and Sino-Tibetan branches in China.

3.2. Burmo-Qiangic fallen leaves

Chikova (2012) believes that Qiangic may be a linguistic area rather than actual unified branch. In other words, Qiangic is likely to be paraphyletic. The comparative lexical data in ZMYYC (1991) and STEDT show that "Qiangic" has high internal lexical diversity. Based on evidence from comparative lexical data and geographical distribution, I believe that Burmo-Qiangic is best divided into multiple primary branches that had radiated out from the Sichuan Basin via the Upper Yangtze drainage basin before the Qin and Han conquests of Sichuan occurred. The Three Gorges served as a natural geographical barrier between Burmo-Qiangic and the non-Burmo-Qiangic languages to the east such as Tujia, Hmong-Mien, and the missing "Donor Miao-Yao" (or rather "Donor Hmong-Mien") branch of Tibeto-Burman proposed by Benedict (1988).

The Tibeto-Burman form *syam 'iron', reconstructed by Matisoff in STEDT, has an almost exclusively Burmo-Qiangic distribution, with sporadic loanwords in non-Burmo-Qiangic languages to the west such as Nungic languages, Apatani (Tani branch), and Deori (Bodo-Garo branch). Hence, *syam 'iron' is actually a Proto-Burmo-Qiangic lexical innovation. This suggests that Proto-Burmo-Qiangic speakers were an economically influential group that had already developed metalworking. Hence, the Sanxingdui culture of Sichuan that had existed over 3,000 years ago was very likely to have been associated with speakers of early forms of Burmo-Qiangic languages.

Sun Hongkai (2013) also notes that the geographical distributions of Qiangic subgroups correspond with specific watersheds.

George van Driem's "fallen leaves" model can be applied to Burmo-Qiangic as well. Further comparative work will be needed to figure out the relationships of these branches to each other, and whether Nungic, Karenic, and Gong (or perhaps even Sinitic) are sister branches of Burmo-Qiangic as part of a larger "Eastern Tibeto-Burman" group. This would give us 19 fallen leaves if we include extinct languages and languages with primarily Sinitic and Tibetic superstrata. If we exclude those, we would have only 14 fallen leaves.
  1. Tangut
  2. Baima [with Tibetic superstratum]
  3. Rma (Qiang)
  4. ? rGyalrong
  5. Lavrung
  6. Ergong (Horpa)
  7. Choyo
  8. nDrapa (Zhaba)
  9. Guiqiong
  10. Minyak (Muya)
  11. Ersuic
  12. Namuyi
  13. Shixing
  14. Naish (Naxi)
  15. Prinmi (Pumi)
  16. Lolo-Burmese
  17. Bai [with Old Chinese superstratum]
  18. Cai-Long [with Old Chinese superstratum]
  19. Waxiang [with Old Chinese superstratum]

I now consider rGyalrong and Horpa-Lavrung to each be independent Sino-Tibetan branches that have undergone heavy contact with Burmo-Qiangic, since they do not have many lexical roots shared by Burmo-Qiangic branches. Many rGyalrong dialects also do not display Qiangic brightening (PTB *-a > -i).  rGyalrong's location in the upper reaches of the Dadu River watershed and Horpa-Lavrung's location in the upper reaches of the Yalong River watershed have allowed them to retain conservative phonological features, unlike languages further downstream that display heavy phonological erosion.

My tentative classification of the Burmo-Qiangic languages is as follows.

  1. Tangut
  2. Baima
  3. Rma
  4. Choyo-nDrapa
  5. Guiqiong
  6. Minyak
  7. Ersuic
  8. Prinmi
  9. Naic linkage
    1. Namuyi
    2. Shixing
    3. Naish (Naxi)
  10. Lolo-Burmese
    1. Burmish
    2. Loloish
  11. Bai
  12. Cai-Long-Waxiang
    1. Waxiang
    2. Cai-Long

Tujia also has been influenced by Burmo-Qiangic, but is nevertheless a Sino-Tibetan branch.

The "fallen leaves" principle can even apply to Lolo-Burmese. Here is my list of Lolo-Burmese "fallen leaves" (10 in total; in contrast, Bradley recognizes only 5 Lolo-Burmese branches):

  1. Burmish
  2. Mondzish (Mangish)
  3. Southern Loloish
  4. Lahu-Kucong
  5. Northern Loloish
  6. Southeastern Loloish
  7. Kazhuoish
  8. Lisoish
  9. Nusu-Rouruo
  10. Lawu-Awu

The following map shows the Burmo-Qiangic branches and how their probable dispersal routes. My Burmo-Qiangic dispersal hypothesis is synthesized from earlier work by Jacques & Michaud (2011), Sun (2013), and Chamberlain (2015). Bai has been included based on Lee & Sagart (2008), Cai-Long and Waxiang based on Zhengzhang (2010) and Sagart (2011), and Baima based on Chirkova (2008).

Map 3

Map legend:
Red = Burmo-Qiangic branches
Blue = Hmong-Mien branches
Green = Kra-Dai (Austro-Tai) branches
Pink = Austroasiatic branches
Purple = non-Burmo-Qiangic Tibeto-Burman branches
Brown = Old Chinese

(Note: She of Jiangxi, Fujian, and Zhejiang is closely related Hakka, and is completely distinct from the She of Guangdong, which is a Hmongic language. The She of eastern China may have been "Para-Hmong-Mien" speakers who had shifted to Hakka as Hakka speakers from the north moved into the hills of eastern China.)

Two additional maps are shown below.

Map 4: Diversity of Burmo-Qiangic branches in the Upper Yangtze watershed

Map 5: Dispersal of Lolo-Burmese branches from the Lolo-Burmese homeland in Dian Lake

Many Qiangic subgroups are located within the Min River watershed of Sichuan, while the remaining Burmo-Qiangic diversity is concentrated in the Jinsha River (Upper Yangtze) watershed. thus, it is not unreasonable to assume that early forms of Qiangic (or Burmo-Qiangic) were spoken in the Sichuan Basin during the late Neolithic.

Likewise, Kra-Dai dispersed via the Pearl River drainage basin, and Hmong-Mien had dispersed via the Yuan and Xiang drainage basins in Hunan. Austroasiatic dispersed upstream via Mekong tributaries (Blench & Sidwell 2010), and also by coastal routes.

The internal diversity of each branch or phylum roughly correlates with the geographic size of the respective drainage basin that the phylum or branch had primarily dispersed in. Chamberlain (2015) notes that languages in Bhutan also tend to disperse via drainage basins (watersheds), and that the tributaries in a river system are analogous to subway lines, and that watersheds correspond closely to linguistic groupings. I believe that following river valleys upstream would have been the primary means of agriculturally-motivated population dispersal during the East Asian Neolithic when overland travel via roads was much more difficult than riverine transportation in sparsely populated mountainous frontier regions. Coasts and flat plains are other means for population dispersals. However, starting from the Han Dynasty and especially during the Ming and Qing Dynasties, population movements often follow roads rather than rivers, and usually occurred as a result of military operations and forced population displacements as noted by Holm (2010).

Additionally, I have noticed that Tibeto-Burman loanwords abound in Kra and Jiamao, but it is unclear which branch of Tibeto-Burman these words are from. Today, only Southern Loloish, Northern Loloish, Southeastern Loloish, and Mondzish languages are found in the region, and the Tibeto-Burman loanwords in Kra and Jiamao are ostensibly not from these branches. These Lolo-Burmese branches are all recent arrivals in northern Vietnam, Wenshan, and Guangxi within the past 1,000 years. Thus, a "missing" Burmo-Qiangic branch may have been in southern Guangxi and northern Vietnam, which was later absorbed by Tai and Vietic languages. This "missing" Burmo-Qiangic branch would had dispersed downstream via the Red River valley.

4 Other language families

4.1. A parallel to the south: Austroasiatic linked leaves

Similarly, the Austroasiatic language family has northern (Khmuic + Palaungic + Khasic), eastern (Mangic + Vietic + Katuic + Bahnaric), and southern (Monic + Aslian + Nicobaric) linkages. Blench & Sidwell (2010) argues that this is due to mutual contact, while Gérard Diffloth argues that these are coherent branches. There are 13 existent modern-day Austroasiatic branches, but there may have been as many as 22 or more branches in the past. Again, please note that these are preliminary impressions, and that I have not yet provided any well thought-out evidence for this classification. (Note: I do not consider Mang and Pakanic to form a branch; Mang displays affinities with Palaungic, while Pakanic displays affinities with Vietic.)
  1. Eastern linkage
    1. Pakanic
    2. pre-Jiamao [proposed substratum of Jiamao (Thurgood)]
    3. pre-Be [source(s) of Austroasiatic words in Be and Jizhao]
    4. Vietic
    5. Katuic
    6. Bahnaric
    7. pre-Chamic (Sidwell)
    8. Khmer-Pearic linkage: Khmeric and Pearic
  2. Northern linkage
    1. Mang
    2. Khmuic
      1. Khmu
      2. Pramic
    3. Palaungic
    4. Khasic
    5. Rongic [proposed substratum of Lepcha (Blench)]
    6. pre-Mru / pre-Kuki-Chin-Naga (?) [+ source(s) of Austroasiatic words in Mruic, Kuki-Chin-Naga, and perhaps also Tani languages]
    7. pre-Hmong-Mien (?) [source(s) of Austroasiatic words in Hmong-Mien languages (Ratliff 2010)]
    8. "Donor Kra" [source(s) of early Austroasiatic loanwords in Proto-Kra; see Ostapirat (1995)]
    9. Munda (?)
  3. Southern linkage
    1. Monic
    2. Aslian
    3. Nicobaric
    4. pre-Acehnese (?) (Sidwell)
    5. pre-Kerinci [proposed substratum of Kerinci Malay (van Reijn 1974)]
    6. Bornean (?) [proposed substratum of Greater Bornean (Austronesian) languages (Blench)]
This gives us a total of 23 Austroasiatic "fallen leaves" that would have likely existed.

Excluding languages and branches with proposed Austroasiatic substrata, we would have the following list of 13 Austroasiatic "fallen leaves," listed in "clockwise" geographical order.
  1. Munda
  2. Khasic
  3. Palaungic
  4. Khmuic
  5. Mangic
  6. Vietic
  7. Katuic
  8. Bahnaric
  9. Khmeric
  10. Pearic
  11. Monic
  12. Aslian
  13. Nicobaric

Note that Paul Sidwell (2017) has since classified Shompen within Southern Nicobaric rather than as a separate branch of Austroasiatic.

SplitsTree 4.0 NeighborNet test of Austroasiatic , based on branch reconstructions for over 50 words.

All below were drawn by Andrew Hsiu (2015).

Map 6
: Dispersal of Austroasiatic primary branches from the Mekong River drainage basin, including possible extinct branches.

Map 7: Mutual influence among Austroasiatic branches in Indochina (contact shown by black lines)

4.2. Kra-Dai and Hmong-Mien fallen leaves

My list of Kra-Dai fallen leaves (coherent branches) are:
  1. Kra
  2. Kam-Sui
  3. Lakkia
  4. Biao
  5. Be-Jizhao
  6. Tai
  7. Hlai
  8. Jiamao

Hmong-Mien fallen leaves are:
  1. Hmongic
  2. Mienic

5. Conclusion: Totaling the fallen leaves

This gives us a total of over 80 fallen leaves in mainland Southeast Asia. (In contrast, Indo-European has only 10 "fallen leaves," not including the various divergent, poorly attested extinct Indo-European languages out there like Illyrian, Messapian, Thracian, and others.)
  1. Sino-Tibetan (61+)
    1. Burmo-Qiangic (19)
  2. Austroasiatic (13)
  3. Kra-Dai (8)
  4. Hmong-Mien (2)
The relationships of Austroasiatic fallen leaves have been analyzed in detail by Paul Sidwell; Hmong-Mien by Martha Ratliff, Yoshihisa Taguchi, and myself; and Kra-Dai by Weera Ostapirat and Peter Norquest.

Sino-Tibetan is the last big nut to crack. It will be a very long and challenging, but nevertheless immensely rewarding task in front of us.

Chamberlain, Brad. 2015. Watersheds and language mapping. Presented at SEALS 25, Payap University, Chiang Mai, Thailand.

Chirkova, Katia. 2012. "The Qiangic Subgroup from an Areal Perspective: A Case Study of Languages of Muli" (Archived 2015-06-08 at WebCite). In Languages and Linguistics 13(1):133-170. Taipei: Academia Sinica.

DeLancey, Scott. 2015. "Morphological Evidence for a Central Branch of Trans-Himalayan (Sino-Tibetan)." Cahiers de linguistique - Asie oriental 44(2):122-149. December 2015. doi:10.1163/19606028-00442p02

Holm, David. 2010. "Linguistic Diversity along the China-Vietnam Border." In Linguistics of the Tibeto-Burman Area, Volume 33.2, October 2010.
Luce, George. 1985. Phases of Pre-Pagan Burma, volume 2. Oxford University Press.

Sun Hongkai, et al. 1991. Zangmianyu yuyin he cihui (ZMYYC) 藏缅语音和词汇 [Tibeto-Burman phonology and lexicon]. Beijing: Chinese Social Sciences Press.

Sun Hongkai. 2013. Tibeto-Burman languages of eight watersheds [八江流域的藏缅语]. Beijing: China Social Sciences Academy Press.

van Driem, George. 2014. "Trans-Himalayan", in Owen-Smith, Thomas; Hill, Nathan W., Trans-Himalayan Linguistics: Historical and Descriptive Linguistics of the Himalayan Area, Berlin: de Gruyter, pp. 11–40, ISBN 978-3-11-031083-2.

Zhèngzhāng Shàngfāng [郑张尚芳]. 2010. Càijiāhuà Báiyǔ guānxì jí cígēn bǐjiào [蔡家话白语关系及词根比较]. In Pān Wǔyún and Shěn Zhōngwěi [潘悟云、沈钟伟] (eds.). Yánjūzhī Lè, The Joy of Research [研究之乐-庆祝王士元先生七十五寿辰学术论文集], II, 389–400. Shanghai: Shanghai Educational Publishing House.
Andy H,
May 25, 2018, 6:20 AM
Andy H,
Jun 22, 2018, 1:06 PM
Andy H,
Nov 25, 2018, 4:14 PM