![Carles li reis, nostre emper[er]e magnes Set anz tuz pleins ad estet en Espaigne : Tresquen la mer cunquist la tere altaigne. Ni ad castel ki devant lui remaigne ; Mur ne citet ni est remes a fraindre, Fors Sarraguce, ki est en une muntaigne. Chanson de Roland , 1r (11 th century)](https://image.slidesharecdn.com/bigmc-100128132625-phpapp01/85/Phylogenetic-models-and-MCMC-methods-for-the-reconstruction-of-language-history-2-320.jpg)
Phylogenetic models and MCMC methods for the reconstruction of language history
- 1. Phylogenetic models and MCMC methods for the reconstruction of language history Robin J. Ryder CEREMADE Paris Dauphine / CREST INSEE Joint work with Geoff K. Nicholls at the Department of Statistics, University of Oxford www.slideshare.net/robinryder
- 2. Carles li reis, nostre emper[er]e magnes Set anz tuz pleins ad estet en Espaigne : Tresquen la mer cunquist la tere altaigne. Ni ad castel ki devant lui remaigne ; Mur ne citet ni est remes a fraindre, Fors Sarraguce, ki est en une muntaigne. Chanson de Roland , 1r (11 th century)
- 3. La plus commune fa巽on d'amollir les coeurs de ceux qu'on a offensez, lors qu'ayant la vengeance en main, ils nous tiennent leur mercy, c'est de les esmouvoir par submission commiseration et piti辿. Montaigne, Essais , I, 1 (1580)
- 4. Tes yeux sont si profonds qu'en me penchant pour boire J'ai vu tous les soleils y venir se mirer S'y jeter mourir tous les d辿sesp辿r辿s Tes yeux sont si profonds que j'y perds la m辿moire Aragon, Les Yeux d'Elsa (1942)
- 5. Et la piaule swingue au son du ghetto, on tape la porte Chill c'est trop fort ! baisse le son merde ! j'connais A chaque fois c'est pareil tant pis il faut qu'巽a p竪te Et profite en tra樽tre des nouveaux albums qu'Rod m'ach竪te Akh辿naton, Juste une pression (2005)
- 6. What to expect Description of the data
- 7. Model of language diversification
- 8. MCMC for phylogenetic trees
- 10. Analysis of two data sets
- 13. Language diversification Languages change in a way comparable to biological species Similarities between languages indicate that they may be cousins. Most common model : phylogenetic tree
- 14. 油
- 16. Internal ages
- 17. Age of the root: 6000-6500 BP or 8000-9500 BP?
- 19. Core vocabulary 100 or 200 meanings, present in almost all languages : bird, hand, to eat, red...
- 20. Borrowing is possible (non-tree-like change), but:
- 21. Easy to detect
- 22. Uncommon
- 23. Does not introduce systematic bias
- 24. Data coding Old English: stierf綻 Old High German: stirbit , touwit Avestan: miriiete Old Church Slavonic: um牒ret鏑 Latin: moritur Oscan: ? Cognacy classes: 1. {stierf綻, stirbit} 2. {touwit} 3. {miriiete, um牒ret鏑, moritur}
- 25. Constraints Constraints on parts of the topology
- 26. Constraints on some internal ages
- 27. We use these constraints to infer rates and other ages
- 28. 油
- 29. Description of the model (1) Traits are born at rate 了
- 30. Trait instances die at rate 亮
- 31. 了 and 亮 are constants
- 32. Description of the model (2) Catastrophes occur at rate
- 33. At a catastrophe, each trait dies with probability 虜 and Poiss(僚) traits are born.
- 34. 了/亮=僚/虜: the number of traits is constant on average.
- 35. Description of the model (3) Observation model: each data point (0s and 1s) is missing with probability 両
- 36. Some traits are not observed and are therefore deleted from the data
- 43. Prior distribution on trees Our main focus is on the root age
- 44. We would like the marginal prior on the root age to be (approximately) uniform over (say) 5000-15000BP
- 45. MCMC moves Random walk on the parameters
- 46. Various moves on the tree (Drummond et al., 2002)
- 47. 油
- 48. 油
- 49. 油
- 50. 油
- 51. 油
- 52. 油
- 53. 油
- 54. 油
- 55. 油
- 56. 油
- 57. 油
- 58. 油
- 59. 油
- 60. 油
- 61. 油
- 62. 油
- 63. 油
- 64. 油
- 65. 油
- 66. Checking mixing and convergence Auto-correlations
- 67. Need statistics on the tree
- 68. Length of the tree
- 69. Root age
- 70. Presence/Absence of a few subtrees
- 71. Synthetic data True tree, ~40 words/language Consensus tree
- 72. Synthetic data (2) Death rate (亮)
- 73. Influence of borrowing True tree, ~40 words/language Borrowing: 10% Consensus tree
- 74. Influence of borrowing (2) Consensus tree True tree, ~40 words/language Borrowing: 50%
- 75. Influence of borrowing (3) Topology is reconstructed correctly
- 76. Dates are underestimated for high levels of borrowing Root age Death rate ( 亮) Borrowing: 50%
- 77. Detecting borrowing Confirmed: hardly any borrowing!
- 78. Data used Indo-European languages
- 79. Core vocabulary (Swadesh 100 or 200)
- 80. Two independent data sets
- 81. Dyen et al. (1997): 87 languages, mostly modern
- 82. Ringe et al. (2002): 24 languages, mostly ancient
- 83. Constraints
- 84. Cross-validation
- 85. 油
- 86. 油
- 87. 油
- 88. 油
- 89. 油
- 90. 油
- 91. 油
- 92. 油
- 93. 油
- 94. 油
- 95. 油
- 96. 油
- 97. Root age
- 98. Conclusions Strong support for the Anatolian hypothesis: root age around 8000BP. No support for the Kurgan hypothesis.
- 99. Applicable to a variety of linguistic and cultural data sets
- 100. TraitLab: it's free!
- 101. Questions ot叩zky sp淡rgsm奪ler vragen questions Fragen domande pytania quest探es 樽ntrebri 于仂仗仂 vpra邸anja preguntes preguntas fr奪gor vrae spurningar quaestiones 竜流竜旅 于仗仂亳 kesses sp淡rsm奪ler kl叩usimai 亰舒仗亳舒仆仆 愕悋 爐爛爐萎ざ爛爐 cwestiwnau