Lactase and language: the spread of the Yamnaya

« previous post | next post »

[This is a guest post by Doug Hitch]

I have had a theory for a number of years about the success of the IE (now Yamnaya) people in populating the world. Here I would like to survey some of the basic reasons for their demographic spread.

Populations in all species prosper when there is adequate food. If there is a surplus of food, the population will grow to meet it. When there is a shortage, populations shrink. There are well known population cycles for lemmings and rabbits. A peak in hare population is followed by a peak in fox population. Then, with more predation, the hares diminish, followed by a decline in fox numbers. In northwestern North America the Athabaskan populations were always small, often facing starvation. One group went south and adopted corn and beans, and later sheep. There are now more Navajo than all the other Athabaskans combined.

The Yamnaya were mutants. They developed lactase persistence and could live on milk as adults. They could turn grass into food through animal husbandry. This is what gave them an advantage over other populations. In times of famine they could survive without killing their animals. A few years ago I dug up stats on global dairy production. The areas of highest production are also those with the highest concentration of Yamnaya people. (An exception was the rapidly increasing production in China.)

The cow is still sacred in India, even in the south where there is less ability to digest unfermented milk. I sometimes think the Avesta is just about cows. They really go on and on. It is astounding how many distinct dairy products are available to modern Yamnaya. I recently tried skyr, the Icelandic stuff that is half yogurt and half soft cheese. I suspect there are many Yamnaya regional products that I have never heard of.

There has lately been more and more evidence from human genomics to support the idea that Yamnaya are milk-eating mutants. I suspect some geneticist will eventually connect lactase and language.

The Asiatic herders that now dominate the steppe also consume much dairy but it is usually fermented which removes lactose. This might partly explain the later expansion of Turks throughout the steppe.

There appear to be other cases of language expansion because of better food. The Bantu were better farmers and spread from West Africa to the east and south. Food is likely a big part in the expansion of the Austroasiatic and Austronesian peoples (both at least partly through rice). Polynesians brought a rich food complex to Hawaii, including taro, yam, sweet potato, bananas, sugar cane, dogs, pigs and chicken. Food availability may partly explain why there are more Inuit on Greenland than on Baffin Island. I suspect the Iroquoian languages pushed north from the Ohio valley to occupy the Great Lakes area because the Iroquois had farming.

The next time you enjoy an ice cream cone, you are holding in your hand the stuff that changed the world. I suggest pistachio.


Selected readings


  1. Victor Mair said,

    July 16, 2020 @ 7:31 am

    I think this post constitutes the kernel of a future best selling book.

  2. Thomas Hutcheson said,

    July 16, 2020 @ 7:41 am

    This is the thesis of "History of English" podcast, although he also suggests that horse domestication multiplied the size of herds.

  3. Brian Spooner said,

    July 16, 2020 @ 7:43 am

    Interesting theory. Similarly, Iran became historically established because of Iranian (or from the geographical name of their language, Persian) technological success at irrigation, which gave them a dependable food supply in an arid area.

  4. Robert T McQuaid said,

    July 16, 2020 @ 8:04 am

    The book The 10,000 Year Explosion by the politically-incorrect Gregory Cochran and Henry Harpending attributes the spread of the Indo-European languages to lactose tolerance.

    The book Who We Are and How We Got Here by David Reich charts the spread of the Yamanaya, but attributes it to microbes taking the lead, similar to the invasion of the Americas by Europeans.

  5. Keith said,

    July 16, 2020 @ 8:26 am

    @Thomas Hutcheson

    If you have ever tried to outrun a cow or a sheep (yes, I've done both, but only won against the cows because they stopped chasing me in order to chase somebody else), you will understand how either being on horseback or having a dog at your command makes controlling a herd or a flock possible.

    It seems to me that controlling livestock necessarily went hand-in-hand with domestication of the horse in some cultures and the dog in others.


  6. Guillaume Jacques said,

    July 16, 2020 @ 8:30 am

    On this topic, see the article "Milk and the Indo-Europeans":

  7. Laurent Sagart said,

    July 16, 2020 @ 8:33 am

    This is exactly the subject of a recent paper:

  8. John from Cincinnati said,

    July 16, 2020 @ 9:18 am

    The C class ("substantial") Wikipedia article on Yamnaya Culture would benefit, I believe, from these scholarly insights. One Wikipedia section in particular becomes disputed. In Yamnaya_culture#Physical_characteristics, the final sentence reads

    Despite their pastoral lifestyle, there was little evidence of lactase persistence.

    Here at Language Log, however, the post asserts

    The Yamnaya were mutants. They developed lactase persistence and could live on milk as adults.

    I'm not comfortable making Wikipedia edits in this area, not being armed with citations (other than the current Language Log post) to satisfy the Archaeology Project monitors. Perhaps other readers would like to step in.

  9. David Marjanović said,

    July 16, 2020 @ 10:22 am

    The C class ("substantial") Wikipedia article on Yamnaya Culture would benefit, I believe, from these scholarly insights.

    Or, rather, lactase persistence came a bit later, as the archeogenetics papers have been saying.

    Of course milk plays a large role in a pastoral lifestyle – but fermented like in the West-Eurasian agricultural lifestyle, not fresh. Evidence of adults drinking milk is not known before the Indo-Iranian soma/haoma rituals; it must be older than that, but need not be older than the Corded Ware culture.

  10. jih said,

    July 16, 2020 @ 11:11 am

    What research on ancient DNA (by the Reich group) seems to be showing is that the arrival of people with Steppe ancestry in Western Europe was (sometimes) accompanied by a strongly biased type of population replacement. According to this research, in Iberia, 90% of Y-chromosomes were replaced, even though a much smaller part of the total genome was affected:

    Maybe there is some peaceful way this could had happened, but, unless somebody can come up with a plausible alternative explanation, this suggests that the expansion of IE-speaking tribes in Western Europe had more to do with war technology and violence than with their ability to drink milk.

  11. Philip Taylor said,

    July 16, 2020 @ 11:49 am

    So Herr Hitler might have been right all along, then, with his belief that Aryans were indeed the Herrenvolk ("master race"). What a shame that he could not have found some way to discuss the idea scientifically and dispassionately, as is being done here, rather than use it as an excuse to attempt to rid the world of all the Knechtvolk

  12. Rodger C said,

    July 16, 2020 @ 12:03 pm

    In northwestern North America the Athabaskan populations were always small, often facing starvation. One group went south and adopted corn and beans, and later sheep.

    Another group went south and did no such thing, viz., Apaches.

  13. cliff arroyo said,

    July 16, 2020 @ 12:47 pm

    " Apaches"

    The Navajo used to be considered one of the many Apache sup-groups. So it might be most accurate to say that a number of Athabaskan groups went south, one adopted corn, beans and later sheep and quickly outnumbered all the other Athabaskan groups….

  14. Christian Weisgerber said,

    July 16, 2020 @ 1:03 pm

    Related, from PLOS Biology's Unsolved Mysteries:

    "Why and when was lactase persistence selected for? Insights from Central Asian herders and ancient DNA"


    The genetic adaptation of humans to the consumption of milk from dairying animals is one of the most emblematic cases of recent human evolution. While the phenotypic change under selection, lactase persistence (LP), is known, the evolutionary advantage conferred to persistent individuals remains obscure. One informative but underappreciated observation is that not all populations whose ancestors had access to milk genetically adapted to become lactase persistent. Indeed, Central Asian herders are mostly lactase nonpersistent, despite their significant dietary reliance on dairy products. Investigating the temporal dynamic of the −13.910:C>T Eurasian mutation associated with LP, we found that, after its emergence in Ukraine 5,960 before present (BP), the T allele spread between 4,000 BP and 3,500 BP throughout Eurasia, from Spain to Kazakhstan. The timing and geographical progression of the mutation coincides well with the migration of steppe populations across and outside of Europe. After 3,000 BP, the mutation strongly increased in frequency in Europe, but not in Asia. We propose that Central Asian herders have adapted to milk consumption culturally, by fermentation, and/or by colonic adaptation, rather than genetically. Given the possibility of a nongenetic adaptation to avoid intestinal symptoms when consuming dairy products, the puzzle then becomes this: why has LP been selected for at all?

  15. J.W. Brewer said,

    July 16, 2020 @ 2:15 pm

    Turning milk into clarified butter alias ghee apparently removes most of the lactose, which makes it interesting to consider the important role of ghee in the cuisine of India where (in one common narrative) Yamnaya invaders with a pro-dairy culture conquered a pre-existing non-Yamnaya (and lactose-intolerant) society.

  16. E. N. Anderson said,

    July 16, 2020 @ 2:46 pm

    As an anthropologist I have to remind all that "race, language and culture" are three different things. Yamnaya was a culture, not a genetic unit, and the fact that I probably have Yamnaya ancestry (and lactase persistence) does not make me a Yamnaya. And my Finnish relatives handle milk just fine in spite of not being IE speakers. That said, surely being able to digest milk did help the IE expansion. But steppe peoples and most Near Eastern and South Asian people do just fine without lactase, thanks to Lactobacillus, which usefully turns milk into yogurt, qurt, kumiss, and countless other wonderful products.

  17. Victor Mair said,

    July 16, 2020 @ 4:04 pm

    Robert Drews, Militarism and the Indo-Europeanizing of Europe (Routledge, 2017)

    Amazon / Google Books description — from the publisher:

    This book argues that the Indo-Europeanizing of Europe essentially began shortly before 1600 BC, when lands rich in natural resources were taken over by military forces from the Eurasian steppe and from southern Caucasia. First were the copper and silver mines (along with good harbors) in Greece, and the copper and gold mines of the Carpathian basin. By ca. 1500 BC other military men had taken over the amber coasts of Scandinavia and the metalworking district of the southern Alps. These military takeovers offer the most likely explanations for the origins of the Greek, Keltic, Germanic and Italic subgroups of the Indo-European language family.

    Battlefield warfare and militarism, Robert Drews contends, were novelties ca. 1600 BC and were a consequence of the military employment of chariots. Current opinion is that militarism and battlefield warfare are as old as formal states, going back before 3000 BC.

    Another current opinion is that the Indo-Europeanizing of Europe happened long before 1600 BC. The "Kurgan theory" of Marija Gimbutas and David Anthony dates it from late in the fifth to early in the third millennium BC and explains it as the result of horse-riding conquerors or raiders coming to Europe from the steppe. Colin Renfrew’s Archaeology and Language dates the Indo-Europeanizing of Europe to the seventh and sixth millennia BC, and explains it as a consequence of the spread of agriculture in a "wave of advance" from Anatolia through Europe. Pairing linguistic with archaeological evidence Drews concludes that in Greece and Italy, at least, no Indo-European language could have arrived before the second millennium BC.

  18. Victor Mair said,

    July 16, 2020 @ 4:20 pm

    From Sunny Jhutti:

    Jats are known as Doodh-putt (Milk Sons) in Punjab. Dairy and wheat are a large part of our diet.

  19. Chester Draws said,

    July 16, 2020 @ 8:15 pm

    the expansion of IE-speaking tribes in Western Europe had more to do with war technology and violence than with their ability to drink milk.

    These need not be unrelated.

    “Amateurs talk about tactics, but professionals study logistics.” Or perhaps you prefer "An army marches on its stomach". Either way, better food allows for easier conquest.

  20. Chris Button said,

    July 16, 2020 @ 10:03 pm

    I still think its fascinating how Old Chinese 酪 *rák "fermented milk" might be somehow connected with the very restricted Proto-Indo-European root *g(a)lag‑ ~ *g(a)lakt‑ "milk" (as discussed elsewhere on LLog and cited in the OP). Its phonetic 各 *kák does hint toward some kind of disyllabic form with a velar onset followed by a liquid onset. That the same velar presyllable was dropped in PIE in forms like "lactic" versus "galactic" would (presumably) be considered coincidental?

    Then again, what about Mongolian ayraq ? There was a uvular *-aq rhyme in Old Chinese, distinct from *-ak, which would have been the more obvious choice. And the ay- in the front doesn't have any velar component. Nonetheless, if we assume metathesis of the -y-, it does nicely parallel Inscriptional Burmese rjɐk in rjɐk tɐk "buttermilk", which later merged with the (presumably) unrelated rɐk in ərɐk "liquor" from Arabic 'araq. A metathesized -j- in 酪 *rák wouldn't have worked since that would give *rác (from **rákʲ in turn from **rják) in accordance with the evidence that any pre-OC palatal (and separately labial) elements were thrown over onto any velar codas shifting -k, -ŋ, -ɣ to -c, -ɲ, -j respectively.

  21. Andreas Johansson said,

    July 17, 2020 @ 3:08 am

    The annoying thing about the Drews book is that he pays no attention to the genetic evidence, which should be required in a 2017 book.

  22. David Marjanović said,

    July 17, 2020 @ 8:43 am

    The other annoying thing about the book would seem to be this:

    Pairing linguistic with archaeological evidence Drews concludes that in Greece and Italy, at least, no Indo-European language could have arrived before the second millennium BC.

    There is good evidence that Greek and Italic have an Indo-European substrate: there seem to be whole branches of IE that have submerged. Maybe Drews is right that certain IE branches expanded around 1600 BCE – that just wasn't the first such expansion.


    the very restricted Proto-Indo-European root *g(a)lag‑ ~ *g(a)lakt‑ "milk" (as discussed elsewhere on LLog and cited in the OP)

    This really doesn't look like a PIE root; I still recommend the quote from Glenn Gordon's defunct blog Paleoglot in this post.



    Drinking cow's milk like a boss?

  23. gds555 said,

    July 17, 2020 @ 10:02 am

    Given that, over the last few thousand years prior to the modern era, the Indo-European languages generally strengthened their various presences in the Western (Pontic-Caspian) division of the Great Eurasian Steppe, but lost much of what presences they had in the Central (Kazakh) and Eastern (Siberian-Mongolian-Manchurian) divisions, one could perhaps say that it was a case of one steppe forward, two steppes back.

  24. Sean said,

    July 17, 2020 @ 10:16 am

    Andreas: Robert Drews of Vanderbilt University was born in 1936. The archaeogenetic data will be fundamental in 10 or 20 years, but right now we are building up communities which know the bioscience and the linguistics or the bioscience and the archaeology and can really sift through it. Right now, there are often not many independent people who can give an informed opinion, just the team and one or two rival teams (and there is 150 years of pseudoscience and tendentious arguments in this area, where people who did not understand the details but trusted the experts got egg on their faces). So its not fair to criticize an 81-year old with an old-school humanities background for putting this evidence to the side.

    If someone were 31 and setting out for a career researching the Bronze Age and Indo-European expansion, I would agree, they should learn enough statistics and enough archaeogenetics to have an opinion on these studies (or find a co-author with those skills). But one human being can only do so much, and right now, there are not a lot of people with both skill sets who can give an independent opinion on how these fit within our fields' ways of understanding the world.

  25. Andreas Johansson said,

    July 17, 2020 @ 10:44 am

    @David Marjanović:

    Drews uses "Indo-European" in a narrow sense to exclude Anatolian, calling the larger family "Indo-Hittite, so there's not necessarily a contradiction if Garnier and Sagot uses the term in the more usual wider sense. Indeed, Drews does think that Indo-Hittite languages were spoken in Greece before the arrival of Greek, having got there from an Anatolian Urheimat.

  26. Chris Button said,

    July 17, 2020 @ 12:08 pm

    @ David Marjanović

    This really doesn't look like a PIE root

    Exactly. Kind of reminiscent of horse isn't it:

    The mystery continues …

  27. Slumbery said,

    July 17, 2020 @ 1:21 pm

    I do not have the time to dig up the genetic papers, but from memory LP frequency was not very high among Yamnaya people. Also it was not overwhelmingly high in the following Bronze Age groups either. LP reached its current frequencies only during the Middle Age. I'd wager to say that even in the early Iron Age the majority of Europeans was still non-LP.

    So whatever was the secret of the Steppe folks, it was probably not lactase persistence.

    I think their secret was mobility and aggressivity. In a time when the farming societies were in crisis due to climatic shift, soil degradation, and pandemics, they became vulnerable and an invasion of mobile pastoralists toppled them in certain regions, then domino effect.

    @E.N. Anderson
    "Yamnaya was a culture, not a genetic unit, and the fact that I probably have Yamnaya ancestry (and lactase persistence) does not make me a Yamnaya."

    Of course nobody is Yamnaya today (or anything else from the past as we are not our ancestors, partial or not), but I think this is a bit of a non sequitur. Based on the available genetic samples the population of the Yamnaya culture was as homogeneous as you get, considering the extent of their territory. There are well defined ethnic groups in modern Europe that are much more heterogeneous genetically than them. This of course does not mean that they were clones with no diversity, but the current data the very least compatible with an assumption that they recognized some level of kinship all over the territory covered by their material culture.

  28. Jerry Friedman said,

    July 17, 2020 @ 3:12 pm

    gds555: I'd thank you for pointing that out, but I'm afraid it would be a faux pas.

  29. gds555 said,

    July 17, 2020 @ 4:33 pm

    Jerry Friedman: I suppose it’s also worth pointing out that, having most likely had their homeland in the Western (Pontic-Caspian) division of the Great Eurasian Steppe, the Indo-European languages have established a significant presence in the great majority of all the steppes on Earth, supporting Lǎozi’s famous dictum that a great journey begins with a

  30. gds555 said,

    July 17, 2020 @ 4:41 pm

    [cont.] single steppe.

    (Note also that the steppes that are substantially inhabited by the Indo-European language family include the vast North American one commonly known as the prairie, which extends from central Manitoba. Saskatchewan, and Alberta down to central Texas. There’s also a prairie exclave in southeastern Texas, giving rise to a situation that I suppose could be referred to as the “Texas Two-Steppe”.)

  31. Doug Hitch said,

    July 17, 2020 @ 5:07 pm

    The key genetics papers may be two from 2015, summarized here At the bottom there is a short but detailed audio clip from an interview with Morten Allentoft. The studies suggest lactase persistence was uncommon at the time of Yamnaya expansion. The authors of "Milk and the Indo-Europeans" (see above) were aware of these suggestions but still argued in favor of lactase persistence being instrumental in the expansion.

  32. Sean said,

    July 18, 2020 @ 4:29 am

    Slumbery: think about it this way. If someone tossed you in a time machine and dumped you in Ukraine in 3000 BCE, would the first members of the Yamanaya archaeological culture you met see you as members of their community or strangers? If they picked up someone from that village and dumped them where your parents grew up, and after learning the barbarous local language they started to feel amorous, would they have to consider a lot of kinship rules and family alliances going back generations or just whether they had an itch for one of the locals? One set of my ancestors were all living in Ireland in 1840, but have not lived there or had social and marriages ties with anyone living there since 1850- does it make sense to call me Irish?

    Humans can't see genetic relations, and they can only track a few hundred people's kinship networks, so they have to rely on proxies. And they are very clever at finding reasons why someone with the right parents should actually be out, and reasons why someone with the wrong parents should really be in.

  33. maidhc said,

    July 18, 2020 @ 5:15 am

    A sort of parallel development can be seen in the cultivation of maize in the New World. Maize cultivation came from MesoAmerica (modern Mexico, Guatemala and nearby places). Those cultures also developed the process of Nixtamalization.

    Wikipedia: Unprocessed maize is deficient in free niacin. A population that depends on untreated maize as a staple food risks malnourishment and is more likely to develop deficiency diseases such as pellagra, niacin deficiency, or kwashiorkor, the absence of certain amino acids that maize is deficient in. Maize cooked with lime or other alkali provided niacin to Mesoamericans. Beans provided the otherwise missing amino acids required to balance maize for complete protein.

    Maize culture spread up the Mississippi River as far as Ontario and New England. However nixtamalization did not always accompany it, so that modern archaeologists can identify the advent of maize culture by evidence of malnutrition,

    In a famous set of prison experiments in the US in 1930s, it was demonstrated that a diet of corn bread and fatback bacon (typical of the US southern rural population) led to niacin deficiency. Whereas in Mexico, where nixtamalization was common, such problems did not occur.

    Not a genetic issue, but a cultural one though.

  34. Victor Mair said,

    July 18, 2020 @ 5:53 am

    From Alan Kennedy:

    This makes me wonder if there are cultural variations in the consumption of human milk over time. Which cultures had wet nursing traditions? Which cultures switched from human to animal milk over the course of the life of a child ?

  35. David Marjanović said,

    July 18, 2020 @ 7:12 am

    one steppe forward, two steppes back


    Indeed, Drews does think that Indo-Hittite languages were spoken in Greece before the arrival of Greek, having got there from an Anatolian Urheimat.

    Oh! Interesting.

    This really doesn't look like a PIE root

    Exactly. Kind of reminiscent of horse isn't it:

    It is indeed!

  36. Rose Eneri said,

    July 19, 2020 @ 10:14 am

    In "Milk and the Indo-Europeans," Romain Garnier, Laurent Sagart and Benoît Sagot conclude that IE groups prevailed militarily over sedentary farming communities and established themselves as a durable ruling elite over non-IE language speaking peoples.

    I find it interesting that in Mr Robert T McQuaid's comment above, he refers to this phenomenon as the "spread" of the Yamanaya, yet he refers to the the spread of Europeans into the Americas as an "invasion."

  37. Philip Taylor said,

    July 21, 2020 @ 2:47 am

    I do not personally feel that there is any conflict between describing something that took place 5000 years ago as a "spread" whilst describing something that took place less than 500 years ago as an "invasion". The latter term clearly carries value judgements within it, and we are all sufficiently aware of, and informed about, the effect(s) that those more recent events had to feel entitled to make value judgements about them. Events that took place 5000 years ago, however, have left almost no folk-memories, and so one may be justified in feeling entitled to think of them (and to refer to them) in more dispassionate terms.

RSS feed for comments on this post