![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
rfmcdonaldI first learned of the latest claim of the detection of ancient linguistic relationships via Dienekes' blog. The affair was summarized at The Economist's Johnson blog.
This has been widely criticized, and to the best of my layman's knowledge, accurately. See the comments in one Language Hat post and linking to a general criticism of the project at Language Log, one which makes the point that reconstructions building on reconstructions are remarkable. Geocurrents' Asya Pereltsvaig and Martin Lewis criticized the project on linguistic and geographic grounds. So, alas, this effort is almost certainly misguided.
(Why alas? I suspect that much of the appeal of these projects lies in their claim to have recovered a bit of humanity's deep history, the preliterate past far beyond plausible reconstruction. Reclaiming a bit of the past through sheer ingenuity is an appealing project.)
The Washington Post reports today that linguists have discovered a handful of "ultraconserved" words, some 15,000 years old. These are said to include "hand", "give", "bark" and "ash". The paper is "Ultraconserved words point to deep language ancestry across Eurasia," by Mark Pagela, Quentin D. Atkinson, Andreea S. Calude, and Andrew Meade in the Proceedings of the National Academy of Sciences.
The Post buried the real news, though: what the new paper does is claim this as evidence that 7 modern language families, not yet conclusively shown to be related, are part of an Ur-family called proto-Eurasiatic. By their theory, the Indo-European, Uralic, Altaic, Inuit-Yupik, Dravidian, Chukchi-Kamchatkan and Kartvelian languages all share a common ancestor. The descendants of these proto-languages are spoken in a vast territory covering most of Eurasia including the Indian subcontinent today.
What the Post doesn't even brush on is how controversial this is likely to be. Historical linguists have not just established the existence of proto-families. They have elaborately reconstructed them. By contrast, the authors of the latest PNAS paper have, apparently, found just 23 words they think are shared among at least four of the seven families in the putative Eurasiatic. Clever statistical analysis can make a stab at answering how likely this is to be due to chance. But such analysis after 150 centuries of language change can hardly give certainty.
This has been widely criticized, and to the best of my layman's knowledge, accurately. See the comments in one Language Hat post and linking to a general criticism of the project at Language Log, one which makes the point that reconstructions building on reconstructions are remarkable. Geocurrents' Asya Pereltsvaig and Martin Lewis criticized the project on linguistic and geographic grounds. So, alas, this effort is almost certainly misguided.
(Why alas? I suspect that much of the appeal of these projects lies in their claim to have recovered a bit of humanity's deep history, the preliterate past far beyond plausible reconstruction. Reclaiming a bit of the past through sheer ingenuity is an appealing project.)
