SplitsTree analysis

Andrew Hsiu

January 2019

67 lexical items

31 taxa (Tai as a control)

Bootstrap: 1,000 times

Algorithm: Neighbor-joining (NJ)

Notes:

    1. Burmo-Qiangic is not strongly supported, suggesting that this is a linkage with multiple separate branches converging,rather than a true branch. Branches are: Lolo-Burmese, Naic, Ersuic, Qiangic (Rma-Prinmi-Choyo-Guiqiong), Horpa-Lavrung, and rGyalrong. These all likely started out as separate branches that underwent intense convergence during the Bronze and Iron Ages.

    2. Similarly, "Central Sino-Tibetan" is also a convergence zone, with Sal, Mru-Hkongso, and Kuki-Chin-Naga starting out as separate branches that migrated from different areas into the Naga Hills. Mutual influence subsequently occurred.

    3. Tai and Old Chinese (as well as Tujia) group with each other due to intense influence from Old Chinese.

    4. Hrusish and Kho-Bwa are grouped together due to contact rather than inheritance.

    5. Strong support for the unity of Sal. Jingpho-Luish is definitely a valid branch.

    6. Strong support for the unity of Kuki-Chin-Naga.

    7. Various Western Sino-Tibetan branches, including Kham, West Himalayish, and Hkongso, show links to Kuki-Chin-Naga. This points to a likely westward migration of Sino-Tibetan groups from the Northeast India area.

    8. Gong is a separate branch that bears some links to Burmo-Qiangic languages, but is not actually part of Burmo-Qiangic. This points to possible historic contact with the Burmo-Qiangic with Gong being on the periphery of the linkage.

    9. Most divergent branches are Koro, Hrusish, Kho-Bwa, Idu-Taraon, Old Chinese, and Tujia; all appear to have non-Sino-Tibetan substrata, but their Sino-Tibetan superstrata are all very early splits, as indicated by the SplitsTree analysis. Tai, instead, has a more recent Old Chinese substrata, and is thus grouped by SplitsTree as part of a clade with Old Chinese.

    10. rGyalrong, which is sandwiched between the aberrant Arunachal branches and Old Chinese / Tujia, is one of the most conservative branches. An Upper Yangtze origin would best explain this distribution of divergent branches surrounding Sichuan and Yunnan.

    11. The westernmost branches do not have as much internal diversity.

    12. Koro, Hrusish, Kho-Bwa, and Idu-Taraon are all very early Sino-Tibetan splits with non-Sino-Tibetan substrata. Sinitic is indeed a divergent Sino-Tibetan branch, but so are Tujia and the four Arunachal branches. Thus, a Sinitic vs. "Tibeto-Burman" split is no more valid than a Koro / Hrusish / Tujia vs. "Tibeto-Burman" split.

    13. Idu-Taraon and Koro are placed next to each other due to the Siangic (Koro-Milang) substratum in Idu-Taraon.

    14. Karenic shows remote links to both Eastern ST (Sinitic, Tujia, etc.) and Central ST (Kuki-Chin-Naga, etc.) branches, but does not actually subgroup with any other branches.

    15. Bodish was not included since Tibetic loanwords are widespread in languages of the Tibetosphere.

    16. Except for single languages, reconstructions were used for all branches. Data for the following branches are based on my own reconstructions: Tujia, Idu-Taraon, Kho-Bwa, Gong, rGyalrong, Horpa, and Lavrung.

Confidence values around 45%-65% usually indicate linkage patterns and heavy mutual influence rather than true branch unity.

Confidence values over 80% indicate either a valid branch or very heavy mutual influence.

The longer the branch length, the less "Sino-Tibetan-looking" the branch is.