The future of the Reich Lab’s studies and interpretations of Late Indo-European migrations

Short report on advances in Genomics, and on the Reich Lab:

Some interesting details:

  • The Lab is impressive. I would never dream of having something like this at our university. I am really jealous of that working environment.
  • They are currently working on population transformations in Italy; I hope we can have at last Italic and Etruscan samples.
  • It is always worth it to repeat that we are all the source of multiple admixture events, many of them quite recent; and I liked the Star Wars simile.
  • Also, some names hinting at potential new samples?? Zajo-I, Chanchan, Gurulde?, Володарка (Ukraine – medieval?), Autodrom, Облевка, Кресты, Кудуксай (Ural region, palaeo-metal?), Золкут, etc.
Ancient DNA sample bag?

On the bad aspect, they keep repeating the same “steppe ancestry” meme (in the featured image above, or the one below). I know this is the news report (i.e. science communication), not exactly the Reich Lab, but these maps didn’t appear out of the blue.

Steppe ancestry distribution in Europe, according to PBS.

Interesting for future interpretations is the whiteboard behind David Reich’s back (apparently they like to keep relevant information on whiteboards…):

Whiteboard behind David Reich’s back (at his office?).

It seems that while the Copenhagen group will still be bound (see here) by the Gimbutas/Kristiansen starting point, the Reich Lab will remain bound by Anthony’s selection of Ringe’s (2002) glottochronological model, and they will try to make genomic data fit in with it.

In fact, the whiteboard doesn’t even include Ringe’s link of Germanic with Italo-Celtic, which could maybe hint at Anthony’s recent change of heart? (i.e. Yamna Hungary -> Corded Ware). That would mean still less Linguistics (if glottochronology can be called that), and more Archaeology…

Image from Anthony & Ringe (2015). “The Proto-Indo-European homeland, with migrations outward at about 4200 BCE (1), 3300 BCE (2), and 3000 BCE (3a and 3b). A tree diagram (inset) shows the pre-Germanic split as unresolved. Modified from Anthony (2013).”

I don’t know why university labs need to do this: To select the linguistic model preferred by a single archaeologist, which happens to be the lead archaeologist of the group, and then try to make genetic data agree again and again with that model. I guess it is a strategic question, and has to do with granting continued contacts with archaeological sites, and access to samples from them?

I understand none of them will try to learn ancient languages, too much work probably. But, wouldn’t it have been more scientifish, at least, to depart from, say, three or four reasonable potential linguistic models (that is, from Indo-Europeanists), and from there discuss the best potential fits for the current genomic data in each paper?

This is, for example, how the Heyd (archaeologist) + German/Spanish Indo-Europeanist schools would look like:

Yamnaya expansion coupled with Meid’s (1975) description of three stages of Proto-Indo-European development (as interpreted by Adrados 1998) and depiction of Heyd’s proposal of Yamna expansion.

Wouldn’t you say it could have fitted the statistical and Y-DNA data seamlessly, in contrast to Gimbutas/Trager (i.e. Kristiansen today), or to Anthony/Ringe?

NOTE. I would say the mainstream German school follows Meid’s (1975) three-stage theory coupled with Dunkel’s (e.g. 1997) nomenclature. The Spanish school follows Adrados, who has repeated ad nauseam that he was the first to mention the three-stage theory in conferences and papers previous to and coincident with Meid’s proposal (see his latest JIES article, a paper available in Scribd). In any case, Spanish and German scholars have been working hand in hand in accepting and developing a general linguistic model similar to the one above.

Archaeological theories like those of Heyd or Mallory for Yamna and Bell Beaker (in contrast to Kristiansen or Anthony), and Prescott and Walderhaug for Bell Beaker and Germanic (contrasting with Kristiansen and Iversen) are compatible with this German/Spanish model.

The French school is non-existent on the homeland matter, Italian scholars seem to be behind even in the description of Anatolian as archaic (probably related to the general wish to have Latin as derived from Vergil’s Troy), Russian scholars are still working with Nostratic and Mesolithic expansions, and Leiden, as the leading IE publisher worldwide today, is full of very different ‘divos’, each with his own pet theory (some obviously agreeing with the German/Spanish model; and especially interesting is that some of them are strong supporters of an Indo-Uralic proto-language).

The English-speaking world, on the other hand, has seen the most varied models being either proposed or translated into its language, with the most popular ones being those publicized by archaeologists (Winfred P. Lehmann being one of the noteworthy exceptions), which may explain why for some people (archaeologists or geneticists) linguistics seems more like a game. It is to be assumed that these same people haven’t taken a look at the dozens of genetic papers published to date – and hundreds of archaeological papers using a bit of linguistics to support their models – , and how wrong they have all been in their interpretations, or else they would realize that genomics does (sadly) not really look like a serious discipline at all right now among most linguists, and among many archaeologists either…

Thus, instead of comparing the main theories on Proto-Indo-European (i.e. linguistics->archaeology->genetics), which would have offered the most stable framework to assess potential prehistoric ethnolinguistic identifications, they keep using a single, simplistic language tree liked by an archaeologist, and trying to fit genetic data to it, while also adapting archaeology to genetics, i.e. genetics->archaeology->linguistics; which, as you can imagine, is not going to convince any linguist.

Especially disappointing is that the world’s leading genetic lab still relies on a marginal proposal based on glottochronology, the homeopathy of linguistics… At least in that regard everyone should know better by now.

Also, they keep interacting with the wrong audience: instead of trying to engage linguists into the real homeland and dialectal quest, to keep Genomics a serious discipline among academics, they tend to discuss with politically- or racially-motivated people, which is probably also in line with strategic decisions.

In the example below, we see the main author of their recent paper on Indo-Iranian migrations seeking once again interaction, this time through “news” promoted by Hindu nationalist bigots, so that – even if that makes them look more neutral in the eyes of those who may allow access to Indian samples – , in the end, we see in genomics a fictitious revival of the “AIT vs. OIT debate” dead long ago in linguistics and archaeology (anywhere but in India).

Pretty disappointing to see these trends; so much effort and time invested in futile discussions and infinitely reworked doomed glottochronological or 19th-century models, when it is the fine-scale population structure of expanding Yamna peoples what we should be discussing now, and thus Late PIE dialectalisation with offshoots Afanasevo, East Bell Beaker, Balkan Bronze Age, and Sintashta/Potapovka; as well as Corded Ware evolution in Uralic-speaking territory.

EDIT (7 JUN 2018): Some parts of the text have been corrected or slightly modified.


7 thoughts on “The future of the Reich Lab’s studies and interpretations of Late Indo-European migrations

  1. “In the example below, we see the main author of their recent paper on Indo-Iranian migrations

    seeking once again interaction, this time through “news” promoted by
    Hindu nationalist bigots, so that – even if that makes them look more neutral
    in the eyes of those who may allow access to Indian samples – , in the
    end, we see in genomics a fictitious revival of the “AIT vs. OIT debate”
    dead long ago in linguistics and archaeology (anywhere but in India).

    There has been a lot of news about the discovery
    of a chariot and the burial of an ‘elite warrior’ from Sanauli, India
    recently. Based on the archeological context, and the wheel/chariot
    vocabulary used as a critical piece of evidence for the spread of Indo
    European, 1/n

    — Vagheesh Narasimhan (@vagheesh) June 6, 2018”

    Good thing you point this out. There is a large gap between the impartial scholarship from Europe, and the biased Hindu nationalist nonsense that comes out of India.

    I feel Michael Witzel of Harvard put it best:
    “By their very nature, that is conversation, one may and will change one’s opinion due to such conversation. The nature of conversation is after all: exchange. Of course, this is not really part of the Indian tradition or ethos: change of opinion often is regarded as “defeat.” We want to learn from such conversations. There is a serious cultural difference here, usually not noted. But very important in our context. As one Indian colleague told me, some 20 years ago, proudly: “I never change my opinion”. Well, good for him!”

    Indian people by their tradition or ethos are incapable of changing their opinions.

    Also, as James Mallory puts it, Proto-Indo European had words for bathing so it couldn’t have arose in South Asia.

  2. You are calling Indians ‘bigots’ for sharing one of the most important archaeological find in South Asia, which has direct relevance to this debate! Horses and Chariots have been an unending theme in all Indo-Aryan “migration”(as you prefer, though you expect the results of an invasion) reasoning.

    Even the linguistic supplement of Damgaard et al had this “The Indo-Aryan and Iranian languages share a common set of etymologically related terms related to equestrianism and chariotry (Malandra 1991). Since it can be shown that this terminology was inherited from their Proto-Indo-Iranian ancestor, rather than independently borrowed from a third language, the split of this ancestor into Indo-Aryan and Iranian languages must postdate these technological innovations. The earliest available archaeological evidence of two-wheeled chariots is dated to approximately 2000 BCE (Anthony 1995; Anthony and Ringe 2015; Kuznetsov 2006: 638–645; Teufer 2012: 282). This offers the earliest possible date so far for the end of Proto-Indo-Iranian as a linguistic unity.” This also had the implicit claim that the split happened in Steppe as the chariots were there, a flimsy claim, but a claim nonetheless. It lies in tatters now.

    But, it appears that people are not interested in evidence and truth but name calling when their bias is challenged.

  3. As you likely know I’ve been working with David Reich for nearly 20 years. Why do you say
    we rely on glottochronology? Reference? One of my favorite papers is titled
    “Why linguists don’t do dates” and the paper on IE that I’ve read over and over is
    Ringe, Warnow, Taylor (2002) that has no dates.

    I don’t think at all we are stuck on Anthony’s ideas, but he did write a book
    suggesting that the Yamnaya are at the base of the IE expansion. His book
    had no genetic data but it turned out that all modern IE speakers have genetics from the
    Yamnaya. Maybe he got lucky, but I don’t think so.


    1. Thanks for your comment.

      1) I have simplistically used more than once “glottochronology” in this blog to refer to any kind of computational phylogenetic method applied to linguistics. I think that all such published experiments to date are essentially pseudoscience supposedly included in studies to ‘prove’ a hypothesis, but in reality they represent unnecessary data merely shown off by the authors to offer an appearance of scientific validity to a linguistic classification or chronology. In words of Allan Bomhard on a (very recent) similar paper:

      Disregarding the technical jargon and the sophisticated reasoning, I think we can fairly summarize this paper as follows: “The results of all previous attempts at computational phylogeny have been flawed. The reason for this is that they did not produce the expected results. However, knowing beforehand what those results should be, we carefully manipulated the input data and tweaked the methodology to produce the expected results. In so doing, we succeeded.” That is it — it is no more complicated than that.

      We could discuss this for days, but if you trust this paper to read it so many times, instead of using this time to read some of the other infinite (older and more recent) papers on Indo-European languages which use methods proper of linguistics (or the very tools linguistics relies on) then we will never agree on this. I guess you have read it, but in case you haven’t I’d recommend you The Indo-European controversy, by Asya Pereltsvaig and Martin Lewis, for a comprehensive review of this subject*. I couldn’t be less interested in dedicating time to discussing this, to tell you the truth, because it seems to me like a loss of time at this point.

      *[For a general, easy to read book, I’d recommend Campbell’s introduction to historical linguistics]

      2) And linguists do dates. Maybe not in absolute terms, or at least not all of the time, but they certainly do. For that they only need a grasp of how language changes (or might change) over time. The more data they have (the more languages they know, and the better they know how languages change and the many reasons why they do) the better guesstimates they can get. Just take a look at any of the recent books or papers on Proto-Indo-European evolution and dialects, where most authors try to establish a chronological order, often in absolute terms. This is not plain guessing, and is certainly better than any of those schemes supposedly based on Swadesh lists and similar word lists.

      3) IMHO Anthony’s original contribution was, in fact, not about Yamnaya’s relevance for the Indo-European migrations. That was well established before he wrote anything about the subject. I think his main contributions had to do with the (absolute) chronology of events and the specific cultures that might have led to the formation of the Late Proto-Indo-European community, including Khvalynsk as Indo-Anatolian and its offshoot into the Balkans as Proto-Anatolian, and then Repin as the immediate precursor of Common Indo-European, which evolved into Afanasievo and Yamnaya.

      As far as I can remember, he believed that Corded Ware and Usatovo were different cultures that became Indo-European-speaking by way of acculturation and admixture (elite domination?), respectively, so the only continuous groups with Yamnaya ancestry should be western groups into the Balkans and the Carpathians, and eastern ones that evolved into Sintashta-Potapovka (representing an admixture of Poltavka with Abashevo). I’d agree that this picture is strikingly similar to what has been found about Yamnaya ancestry, but also that no patron-client relationship will be needed to explain the expansion of Indo-European languages, and that this image of a sudden split of all dialects that needed Corded Ware and Usatovo was in great part ‘forced’ by the phylogenetic tree by Ringe, Warnow, and Taylor (2002).

      I think population genomics has shown that there is as little use for cultural diffusion theories during these Eneolithic/EBA expansions, and I also think that the many different contradictory models published to date show that there is little use for computational phylogenetic methods to obtain a proper dialectal classification.

Leave a Reply

Your email address will not be published.