How a Eurasian “band of brothers” shaped the world


When I was eight years old I saw a map which genuinely confused me. I had opened up deluxe dictionary at my elementary school and saw a map of the world’s language families, and noticed that there were a group of dialects which spanned the Bay of Bengal to the North Sea. In fact, according to this map the language I had first learned to speak, Bengali, was in the same language family as English.

This was hard to wrap my mind around, but there it was in front of me. Further research at the public library confirmed this fact. And, upon further reflection it was obvious to me there were similarities…I had been learning French at school, and English, Bengali, and French, all exhibited similarities in the first ten numbers. English and French I understood in terms of a natural relationship, but Bengali?

My personal and professional interests have never been in domains where I would explore the topic first hand, but the origins of Indo-European languages have always been a hobby. I read books such as The Horse, the Wheel, and Language and In Search of the Indo-Europeans when I could. When taking in excellent works such as Empires of the Silk Road the Indo-European thread was always something I kept in mind.

But the above works take a more old-fashioned Eurasian heartland “marauders from the steppe” viewpoint. Starting about 15 years ago I began to look into a different framework: Indo-Europeans as farmers. For me begins with the 2002 paper, Mapping the Origins and Expansion of the Indo-European Language Family, which finds that “the inferred timing and root location of the Indo-European language trees fit with an agricultural expansion from Anatolia beginning 8000 to 9500 years ago” (this is the last paper I can remember reading in paper format). The model is elaborated by Peter Bellwood in works such as First Farmers, though he applies it to most language families.

But its origins go back decades, with the archaeologist Colin Renfrew. Rather than dramatic explosions from the steppe, Renfrew and colleagues suggest that the demographic expansion enabled by agriculture as a mode of production allowed for groups like Indo-Europeans to rapidly swamp their neighbors and enter into a process known as a wave of advance. There wasn’t a organized movement. Rather, farming enables the growth of population to such an extent that it was almost an undirected thermodynamic law that the original farmers would radiate outward, away from zones at the Malthusian carrying capacity and out toward virgin land.

It was a parsimonious theory, and phylogenetic techniques seem to have supported it. But then came ancient DNA to overturn the apple-cart. I won’t reshash what you probably already know, but will point to the two most relevant papers, Massive migration from the steppe was a source for Indo-European languages in Europe and Population genomics of Bronze Age Eurasia. Basically there was massive population turnover during the early Bronze Age. The genetic data aligned well with predictions you’d make from the old “marauders from the steppe” model, not the demic diffusion of farmers who were subject to high endogenous population growth over time.

Of course the Anatolian model proponents have an answer. There is a thesis whereby the steppe pastoralists derive from Anatolians, and so the European population turnover was of one Indo-European group by another. This is possible, but to my knowledge this model was never foregrounded by Anatolianists before. Rather, it strikes me as a way to “save” their framework.

So far much of the battle has been between archaeologists, who tend to favor gradualism, and often even  cultural diffusion as opposed to migration, and historical linguists and arriviste geneticists, who tend toward a more classical migration-from-the-steppe perspective.

A new paper in Antiquity takes the sledgehammer to the Anatolian hypothesis with an archaeology first tack. Re-theorising mobility and the formation of culture and language among the Corded Ware Culture in Europe. They don’t pull punches:

…the Anatolian hypothesis must be considered largely falsified. Those Indo-European languages that later came to dominate in western Eurasia were those originating in the migrations from the Russian steppe during the third millennium BC.

Why would they say this? There is a major paper coming out:

These local processes of social integration between intruding Yamnaya/Corded Ware populations and remnant Neolithic populations can be applied to language dispersal. We should expect that the transformation from Proto-Indo-European to Pre-Proto Germanic would reveal the same kind of hybridisation between an earlier Neolithic language of the Funnel Beaker Culture, and the incoming Proto-Indo-European language. This is precisely what recent linguistic research has been able to demonstrate (Kroonen & Iversen in press). In their study on the formation of Proto-Germanic in Northern Europe, Kroonen and Iversen document a bundle of linguistic terms of non-Indo-European origin linked to agriculture that were adopted by Indo-European-speaking groups who were not fully fledged farmers.

They also contend that the Neolithic language was roughly the same throughout the zone of Indo-European expansion. From what those who would know about these sorts of things have told me this is plausible, because the Neolithic farmers spread so rapidly from a small founder culture, and exhibited broad Europe-wide similarities for a thousand years. Curiously, the chart shows that Germanic languages may have been influenced by a hunter-gatherer language, which the others were not. I suspect this may have to do with the relatively late persistence of hunter-gatherers in some maritime environments facing the Baltic and North Sea.

The paper, which is open access, needs to be read in full. Here are some important points:

  • Burial type seems to be a more robust form of indicator of dominant cultural identity
  • Corded Ware males practiced exogamy
  • Corded Ware males traveled long distances
  • Corded Ware culture was initially exclusively pastoralist
  • There is a great deal of circumstantial, and some genetic, evidence that Corded Ware communities were characterized by having women who were clearly from the Neolithic farming population
  • There was intergroup violence as a function of culture
  • The Corded Ware and Neolithic populations persisted near each other geographically, though the Neolithic groups seem to have retreated to uplands
  • The Corded War engaged in a wholesale pattern of landscape sculpting, burning down forests to produce pasture

Neolithic Y lineages, such as G2, are far rarer in Northern Europea today that R1a and R1b (in contrast, the hunter-gatherer I seems to have gone through an expansion just like R1a and R1b). We already have a model for what went on here, the Iberian settlement of the New World. Among mestizo populations there are huge skews of mtDNA and Y, with the former almost all Amerindian (with some African) and the latter almost all European (with some African).

The Corded War are the ancestors of the German peoples who we see emerge into the light of history during antiquity. What these data are telling is that the Germans are the product of a massive period of biological and cultural amalgamation and synthesis between indigenous groups and intrusive populations from the steppe. The archaeological data indicate that the intrusion was male mediated. The “battle axe” culture probably lived up to its name. And they weren’t likely exceptional….

22 thoughts on “How a Eurasian “band of brothers” shaped the world

  1. @Razib, “The Corded War are the ancestors of the German peoples who we see emerge into the light of history during antiquity.”

    People of mixed Steppe/Local Neolithic Europeans are the ancestors of Germans but Corded Ware certainly isn’t the sole ancestor. If anything Corded Ware is the ancestor of Slavs and Balts.

    Scandinavians have a lot of R1a-Z284 (the same type of R1a found in Swedish Corded Ware) and so certainly have a big amount of Corded Ware ancestry. However the small amounts of R1a in other Germans could be of Slavic origin, indicating most Germans don’t have a lot of Corded Ware ancestry.

    We don’t even know how West German speakers(Germans, Dutch) are related to North German speakers(Swedes, etc.). Who are the Germans? We don’t know. So if German origins are a mystery, how can we say they descend from Corded Ware.

    Slavs, Balts, and Norse have a lot of Corded Ware ancestry but other Europeans probably don’t.

    1. We don’t even know how West German speakers(Germans, Dutch) are related to North German speakers(Swedes, etc.). Who are the Germans? We don’t know. So if German origins are a mystery, how can we say they descend from Corded Ware.

      i seen plenty of data. they’re very genetically close. i can double check again…. (the Y data is interesting, but remember that it has fast drift compared to autosome; like 4x).

      1. @Razib,
        “i seen plenty of data. they’re very genetically close”

        I agree they are but we don’t know exactly how they’re related and if any of their common ancestry is from Corded Ware. R1a-Z284 is rare in West Germanic speakers but popular North Germans like how R1b P312 is rare in North Germanic speakers but popular in West Germans.

        R1b P312 looks like a non-Germanic(Celtic?) substrate in West Germans, so maybe R1a Z284 is a non-Germanic(Corded Ware?) substrate in North Germans.

    2. Corded ware are pre slavs and pre balts – what we call the german language came from scandinavia and it came late – probably it started seeping in with the first marauder bands coming over denmark in the late Roman republic time

      Ther is zero evidence of scandinavians in central europe before that time.

  2. Great papers, particularly the Kristiansen one. The current question seems to be the extent to which the migrating Yamnaya people brought their own women with them. These papers argue for a heavily male-dominated migration, while the Reich teams looks like it’s holding out for a genuine mass migration. What do you think, Razib?

    1. To clarify: nobody disputes that the migration was tilted toward males or that local Neolithic women were subsumed into the group.

    2. so i had dinner with iosef and iain last fall and they expressed skepticism. but the question i have is can their inferences reject the hypothesis of sex bias to such an extent, or are they suggesting that issues with the X chr make it hard to conclude from it that there was a bias? i need to look at iosef and amy’s latest exchanges.

      i lead toward the idea that they didn’t bring many women. but weak confidence.

      1. I dunno, seems likes there’s enough ancient mtDNA to be more confident. I think it’s as you say — they brought some, but not many women. Either that or some farmer mtDNA has been under crazy selection. 🙂

  3. I have a hunch regarding the earliest split on the IE language tree, Anatolian, and why this has thrown people off. The clue is in a single sample from Lazaridis 2016, Anatolia_Chalcolithic. It’s dated to ~4,000 BC so it should predate the major explosions out of the steppe — *yet is already has a significant amount of CHG and EHG ancestry.*

    Since we have earlier genomes from the same exact site, and they look like all other farmer genomes, then the only way to explain this is an early movement from the Steppe. Makes me kind of surprised this sample hasn’t received more attention.

    It seems plausible to me that the Anatolia_Chalcolithic population spoke some sort of Anatolian language. This is consistent with its early branching as well as early IE presence in the area, but does not preclude that it was the later movements that mattered demographically and linguistically — at least to most Euros. Anatolian-derived people might still be responsible for spreading IE genetics in the middle east / north africa / southern Europe.

    1. The limited historical evidence available suggests that Indo-European Anatolian languages like Hittite were confined to a handful of quite small enclaves in Anatolia as of ca. 2000 BCE, after which Hittite rapidly expanded with Hittites ruling essentially all of modern Turkey, northern Iraq, Syria and much of Lebanon by 1200 BCE when their empire experienced the Bronze Age collapse. One important factor in this empire’s success was its ability to embargo iron metallurgy knowledge that was unavailable to other cultures until after the Bronze Age.

      In my view, the seemingly highly basal position of the Anatolian language in the Indo-European language family is an illusion due mostly to strong influence of a healthy substrate language that was very different from that of the other Indo-European languages. Elsewhere the substrates were probably less different and the substrates were associated with distressed civilizations that were culturally weaker in Europe and South Asia than the Hattic culture of Anatolia. Historical linguists have a long history of underestimating how much language change is due to contact with languages that are significantly different and of overestimating how much language change is due to random drift. Outliers like Iceland suggest that language change due to random drift is a much smaller factor than commonly assumed.

      Instead, the pre-Hittite people of Anatolia were probably from an earlier non-Indo-European wave of first metalworkers who expanded ca. 4000 BCE from the Caucasus Mountains and vicinity (an obvious place to pick up CHG and EHG genes), replacing Anatolian first farmers. These first metalworkers probably gave rise to the Hattic culture and possibly also to some of the other contemporaneous cultures of the Aegean and Zargos Mountains.

  4. The Anatolian hypothesis is certainly dead.

    But, the outrageous hubris of the claim of the Kristiansen paper that the CWC spoke proto-Germanic, or even pre-proto-Germanic, discredited the paper in my eyes before I even finished the abstract, and it was also a very short paper that was very thin on original, competent analysis. Proto-Germanic a.k.a. Old Norse is a historically attested well understood language, with a point of origin more or less in the vicinity of Denmark. No language family has been subject to more systematic linguistic study. This is the language family in which historical linguistics was pretty much invented. It is at least 1000 years younger, and realistically, more like 2000 years younger than the ca. 2500 BCE date the authors claim, on linguistic grounds. CWC is far more likely to have spoken the much older proto-Balto-Slavic language as this is a much older language family in the right geographic location, contrary to the graphic associated with the paper.

    The Kroonen and Iversen Neolithic substrate theory is plausible. But, that substrate was probably laundered through intermediate Indo-European languages in many cases. And, the claim that substrate influence was the source of farming terms in modern Indo-European languages also neglects the extent to which it is possible to reconstruct proto-Indo-European farming terms that extend to areas that were not touched by the Anatolian EEF wave people of Europe. It is at least equally plausible that, contra Kroonen and Iversen, that non-Indo-European farming terms had a substrate influence on proto-Indo-European which incorporated those terms and then spread them into its daughter languages.

  5. The Neolithic timeframe is dead but it doesn’t mean the location is. The steppe brought a subset of IE to Europe while Greek and Anatolian came directly from the Caucasus.That’s kind of what the theory is and we need Greek and Anatolian DNA to be sure among other places.I agree the steppe looks more probable at the moment.

  6. The statement that these Aegean languages “came directly from the Caucasus” implies they didn’t take the steppe route, but instead skirted the southern Pontic shore. This may hold for Anatolian, with its bevy of Hurrian and Hattic loans. But Greek has its own substrate, with no Hurrian or Hattic in it.

    It is more likely that proto-Greek grew up with proto-Armenian and ancient Aryan (“Indo-Iranian” if you like), in the Ukraine steppe, before rushing south from the Balkans.

  7. One of the hardest elements of the steppe invader hypothesis for me to wrap my brain around is how Indo-Europeans kept on winning battles with the (presumed) remnants of “Old Europe” for millennia. I mean, within the recorded history of antiquity, for example, there were plenty of non-Indo European languages. You had Etruscan, Raetic, Anqitanian, all of the non-Indo European Paleo-Hispanic languages, whatever the Sardinians spoke, Lemnian, and the various presumably non-Indo European languages spoken in Crete and Cyprus. And that is just those languages which we have a recorded history of. You could argue that much of this later loss was directly due to the expansion of the Greek and Roman cultures of course, but we really have no way of knowing how many similar “Old European” groups were in northern Europe in the same time period. Presumably less, judging by the DNA evidence, but absence of evidence is not evidence of absence.

    To explain why it seems so weird, I consider the example of the Plains Indians within the North American context. Through introduction of horses to the Pueblo Indians, horses spread northward by the 17th century, and completely transformed the way of life of the tribes, with both hunter-gatherers and farmers abandoning their old ways, and a very well-developed pastoral cultural toolkit evolving within only a few centuries.

    Given the expansion of Indo-European was a far slower unwinding than that of European colonialism – that even after 3,500 years of expansion, non Indo-European groups still existed on its fringes which were being absorbed – one would think that if the key component was a cultural toolkit, it would have been copied by then. This is particularly the case because by the later periods of the expansion Indo-Europeans assuredly no longer saw themselves as one distinct people, and probably didn’t distinguish between other tribes of IE-speakers and the remaining “old European” groups at all. It makes me wonder if it was something else other than the cultural toolkit which explained the continued success of the language family in expansion past its early push into Europe (Corded Ware) and the eastern steppes.

    1. Snowball effect? Once a large area is covered by one language family, statistically more likely that new expansions will happen to begin in that area, increasing the language family area more at the edges; rinse and repeat. Individual branches of Indo-European have been thoroughly replaced – Celtic once spanned from Portgual to Turkey after all. Really it is Romance, Slavic, and Germanic in modern Europe, the others are more at the level of Finnic or Basque.

      Until the recent expansion of Russian I daresay IE lost ground to Uralic – certainly in Hungary, but probably also in Finland and Estonia. In Asia Turkic is far in the lead of course.

    2. “one would think that if the key component was a cultural toolkit, it would have been copied by then”

      i think the key thing about horses is not having horses but time actually in the saddle – which peasant farmers don’t get. the eventual solution to this problem was the feudal type system where knights, sipahis etc were given a village of peasants to support them while they hunted all day.

      also

      “One of the hardest elements of the steppe invader hypothesis for me to wrap my brain around is how Indo-Europeans kept on winning battles…”

      maybe they weren’t battles. if they had a mobility advantage and practised hit and run raiding then they could have cleared land in front of them – only advancing as the farmers retreated.

      no battles, just pastoralist raiding and gradual farmer retreat to better defensive ground.

    3. I think it’s just a matter of more mobile cultures being better able to adapt to changing climate conditions than sedentary ones. I don’t think it’s any different from the various waves of expansion of Semitic languages from ~3,000 BCE to 600 CE, or Germans penetrating the Roman Empire’s borders and settling down in the 4th and 5th centuries, or Turkic and Mongolian expansions. The main story of recorded history up to say 1,500 CE seems to be that of mobile cultures expanding at the expense of more sedentary ones, regardless of linguistic affiliation.

  8. We shouldn’t omit that the main reason why archaeologists like Kristiansen now turn back to the migrationist explanation of the Corded Ware is the brute force of the ancient DNA evidence which leaves no room for other interpretations. Archaeologists now have to come to terms with it, whether they like it or not, and that paper is an early expression of this process. The upcoming linguistic paper for sure was just a minor factor for the publication of this paper.

    1. Prehistoric archaeology and linguistics are just as alien to each other as archaeology and paleogenetics. Archaeologists who have studied both subjects, like David Anthony, are clearly in the minority.

    2. Well, in the end you have to remember that linguistics, archeology and genetics are independent of each other. You can learn a language that was not your biological parents’ language, and you can adopt a cultural practice that your parents did not observe. Obviously, the three tend to correlate, since cultural communities typically share the same language and also reproduce more with each other than with those outside the group. But ultimately they do not depend on each other. So genetic evidence does not prove or disprove linguistic or archeological claims.

  9. “Neolithic Y lineages, such as G2, are far rarer in Northern Europea today that R1a and R1b (in contrast, the hunter-gatherer I seems to have gone through an expansion just like R1a and R1b).”

    How sure are you that R1b isn’t a hunter-gatherer lineage as well? Lipson et al has those Blatterhole MN farmers with R1b and ~40% hunter gatherer ancestry, and I think it’s possible that the surge of R1b was at least partly connected to the resurgence of WHG in neolithic Europe. There’s Els Trocs in the early neolithic and Villabruna in the Upper Paleolithic too, so I think R1b as a marker shared between EHG and WHG is pretty likely. Don’t you?

Comments are closed.