Constructing Linguistic Phylogenetic Tree 语言谱系树的构建

1. Constructing Linguistic Phylogenetic Tree语言谱系树的构建 Chenxi Shao Zihe Li 邵晨曦 李子鹤

2. Tree representation of evolution • Darwin 1859 On the Origin of Species by Means of Natural Selection or The preservation of Favored Races in the Struggle for Life. • Tree diagram • shared, derived characteristics

3. Phylogenetic tree • Zuckerkandl, E. and L. Pauling 1965 Molecules as documents of evolutionary history. J. theor. Biol 8: 357-366. • Comparison of DNA sequences

4. R. Lewin 1996

5. Linguistic family tree

6. From Science Feb. 27, 2004

7. Quantitative representation • Meyers, L. F. and William S-Y. Wang 1963 Tree representations in Linguistics. Project on Linguistics Analysis Report 3, Ohio State University.

8. How to construct a linguistic phylogentic tree? • Wang, William S.-Y. and Zhongwei Shen 1992: • Four steps: • Selection of characters • Quantization of characters • Calculation of correlation coefficients • Clustering analysis —— Selection and encoding of linguistic information Mathematical algorithm

9. Mathematical algorithm • 1. Maximun parsimony 最大俭省算法 • Example: subgrouping of Bai dialects （Wang 2006） • 2. Neighbour-joining 邻接法 • Example: subgrouping of Yi dialects （Wang 汪锋 2010）

10. 3. Average linkage 平均联结法（UPGMA法） • Example: Affinity among Chinese Dialects（Cheng, C.C.郑锦全1988） • 4. Minimum spanning 最短系连法（弗罗茨瓦夫分类法） • Example: Affinity among Chinese Dialects（Ma 马希文1989）

11. Selection and encoding of linguistic information • 1. Unique, Shared innovation characters • Classical version：the position of Armenian（Hübschmann 1875）

12. Modern version: Subgrouping Bai dialects (Wang 2006) • 19 innovation characters generalized from reconstruction. • Convert innovation to 1’s and preservation to 0’s

13. Penny in PHYLIP • Rooted tree

14. Problems: • In many cases, the defining of “innovation” is backstepping of assigning value to historical phonemes. • For many sound changes, we have no consensus on their universality.

15. 2. Retention rate of Swadesh-100 words (Wang, William S-Y.1993) • Strict correspondence among languages • Calculate the proportion of Swadesh-100 words that satisfies strict correspondence rules between each two languages • Construct a matrix of affinity/distance between each two languages

16. Example: subgrouping of Austro-Yue Languages (Chen and He 2002)

17. Neighbor in PHYLIP • Unrooted tree

18. Problems: • Much debate on Swadesh-100 words • universality • choice of words • Uniformity in the rate of change

19. Inspiration from biological study • Phylogenetic classifications of different organisms employ different segments of gene： • The Prokaryotes：16SrDNA（王洪媛、江晓路等2004） • Some mammals：18SrDNA（刘诚刚、杜志恒等2012） • Some fishes：CRY61（孙婷、刘伟等2012）

20. Difference in homonymic relationship • Scope of examination: morphemes that can be undoubtedly reconstructed to the proto- language. • Compare the differences in homonymic relationships among those morphemes between each two languages.

21. L-language, M-morpheme, innov-innovation, Rel-relationship D- difference in homonymic relationships among morphemes between each two languages

22. difference in homonymic relationships among morphemes between each two languages shows distance between the two languages • Distance matrix • Principle: difference in homonymic relationships measures unshared innovation.

23. Merits： • 1. No need of a universal list of core words. • Morphemes that can be reconstructed are basic morphemes (words) in languages in question. • 2. Generalizing structural changes into differences in homonymic relationships among basic morphemes. • Specific values of sounds are out of consideration. • 3. Weight of different sound changes are taken into consideration. • Changes involving more morphemes are more important.

24. Demerit • Borrowings through correspondence among languages in question cannot be eliminated

25. Phylogenetic trees of Naxi • Characters/Parsimony

26. Swadesh-100 words /NJ（Proto-Naxi as the outgroup）

27. D-value /NJ （Proto-Naxi as the outgroup）

28. How to make choice ? • Character Approach is the optimal choice when you have confidence in the specific value of each historical phoneme (Indo-European languages) • Swadesh-100 words Approach is optimal choice when internal contact can be identified in the group of languages in question. (Chinese dialects) • D-value Approach is optimal choice when there is no evidence of internal contact. (Naxi and many other minority languages)

29. 主要参考文献 • Hsieh, H-I. 1973 A new method of dialectal subgrouping. Journal of Chinese Linguistics 1. 64-92 • Hübschmann 1875 “On the position of Armenian in the Sphere of the Indo-European Languages”. In Lehmann ed. A reader in Nineteenth Century Historical Indo-European Linguistics. Bloomington: Indiana University Press, 1976. • Krishnamurti et al 1983 Unchanged cognates as a criterion in linguistic subgrouping. Language 59. 544-688 • Meyers, L. F. and William S-Y. Wang 1963 Tree representations in Linguistics. Project on Linguistics Analysis Report 3, Ohio State University. • Saitou, N. and M. Nei 1987 The neighbor-joining method: a new method of reconstructing phylogenetic trees. Miol. Boil. Evol. 4. 406-425

30. Wang, Feng. 2006. Comparison of Languages in Contact: the Distillation method and the case of Bai. Taipei: Academic Sinica. • Wang, William S-Y.1993 Glottochronology, lexicostatistics, and other numerical methods.收入《王士元语言学论文集》 • 陆致极 1986 《闽方言内部差异程度及分区的计算机聚类分析》，《语言研究》第2期 • 马希文 1989 《比较方言学中的计量方法》，《中国语文》第5期 • 汪锋 2010 《白彝关系语素研究》 国家社会科学基金结项报告 • 王士元、沈钟伟 1992《方言关系的计量表述》 《中国语文》第2期 • 郑锦全 1988 《汉语方言亲疏关系的计量研究》，《中国语文》第2期