Monday, September 24, 2012
There’s something very North Eurasian about Indo-European
Recently we learned that Europeans, especially North Europeans, are actually very North Eurasian genetically (see here). Then, last week, scientists studying languages in a similar way to genes discovered deep affinities between Indo-European, Uralic and Altaic languages. Below are a couple of figures from that study, Dediu et al., and I have to say they look fascinating, because it seems they match my experiments with the ADMIXTURE software in trying to elucidate the peopling of Europe and the origins of the Indo-Europeans.
The MDS plot on the left focuses on the two most significant dimensions of variance, and it places the three mainly West Eurasian (aka. Caucasoid) language groups - Indo-European, Afro-Asiatic and North Caucasian - very close together. Honestly, I don’t know much about linguistics, and the theories covered in this study are way over my head. But the impression I get is that the MDS might be showing the results of relatively recent contacts between these language groups.
However, the annotated Network on the right, which is based on all the variance, reveals that at a deeper level Indo-European is actually very distant from Afro-Asiatic, and much closer to Uralic. Moreover, it appears to be part of a language family also consisting of Altaic (or Mongolic + Turkic) and Dravidian.
If that’s the case, then these results show a striking correlation with the relative distributions of two West Eurasian clusters derived from an ADMIXTURE experiment I ran as part of my project, which are shown on the map below.
Perhaps the so called Southwest Eurasians (yellow cluster) spoke Afro-Asiatic and other similar languages, while the Northwest Eurasians (magenta and cyan clusters) were a related group of people, but became very different genetically and linguistically when they moved into North/East Europe and Central Asia, and mixed with the indigenous inhabitants of those regions?
Note that all of the significantly Northwest Eurasian samples on that map speak Indo-European, North Caucasian, Uralic, Turkic or Dravidian languages. This might be some sort of a coincidence, like simply a function of geography. But then, why are such geographically distant groups like the Dravidians of India and Indo-Europeans of Ireland part of the Northwest Eurasian cluster?
Update 07/05/2013: There's a new study at PNAS by Pagel et al. which shows very similar results but with different methods.
Southwest Eurasians + Northwest Eurasians + Mesolithic survivors = modern Europeans
Dediu D, Levinson SC (2012), Profiles of Structural Stability Point to Universal Tendencies, Family-Specific Factors, and Ancient Connections between Languages. PLoS ONE 7(9): e45198. doi:10.1371/journal.pone.0045198
Pagel et al., Ultraconserved words point to deep language ancestry across Eurasia, Published online before print May 6, 2013, doi: 10.1073/pnas.1218726110, PNAS May 6, 2013