In the previous post, I did my best to start pinning Indo-European down in time and space, and got as far as talking about the when (siding with those scholars who hold that Common Indo-European can’t have broken up much before 4000 BC at the earliest). But what about the where?
This is just as difficult a question, maybe more so, and remains fundamentally unanswered. One view that’s currently very popular points to an area known as the Pontic-Caspian Steppes. The Eurasian Steppes are a long chain of relatively flat or rolling grasslands (with some forest), which broadly stretch all the way from Hungary in eastern Europe, to Manchuria at the eastern edge of Eurasia. The Pontic-Caspian Steppes are one zone within this chain, basically encompassing the regions north of the Black Sea (the ‘pontic-’ part means ‘sea’, and refers to this sea specifically), and extending eastwards towards the Caspian Sea (whence the other, more transparent part of ‘Pontic-Caspian’). In modern terms, this area is covered by the eastern bits of Romania, most of the southern Ukraine, swathes of southwestern Russia, and a bit of western Kazakhstan.
The Eurasian Steppe is the blue shaded bits.
This is sometimes called the 'superhighway of Eurasia', since it's a kind of east-west corridor across a lot of the continent. There are still a number of distinct zones and gaps.
There are a few reasons to imagine Indo-European on the steppes in the 4th millennium BC, but basically it boils down to two big things. One is that there are archaeological cultures of this region that seem to be a good fit for what we might think the Proto-Indo-European speakers were like culturally, and the other is that this region seems to have been the epicenter for some major cultural shifts into neighbouring areas, which could possibly be linked to the spread of the Indo-European languages.
In the rest of this post, I’m going to concentrate just on the first of these two points – but before getting to that, I really want to hammer home the point that it’s very hard to connect languages and archaeological evidence (unless we have actual writing). Humans can speak more than one language, and more often than not they actually do – our current assumptions of monolingualism may well be anachronistic for much of prehistory. And languages may move in various ways: through people actually up and going someplace else, or by people learning a new language. People physically moving around can leave traces in the archaeological record, but language diffusion often does not. Archaeologists often look at things like pottery designs, burial customs, or the ways tools are made – this is the sort of evidence available to them, and it can tell us quite a lot about certain aspects of the past. But it can be extremely hard indeed to be sure whether, say, the spread of a new type of burial (in mounds, for instance – to pick an entirely random example with no wider relevance to Indo-European) should be connected in any way to the spread of a language. Basically, all conclusions are inherently tentative here, and all we can do is try and put together the pieces as best we can.
So bearing in mind these caveats, what are the reasons so many people look to the Pontic-Caspian Steppes when searching for the speakers of Proto-Indo-European? Last time we looked a bit at the vocabulary for wheeled vehicles, to give us a hint about one aspect of the social (in this case, technological) world of early Indo-European speakers. This same principle can be applied to other areas of life, giving us a partial picture of the kind of life that speakers of Indo-European might have generally led – a picture that agrees fairly well with the archaeological evidence from the steppes.
A really basic part of this is food: we all need to eat, and food production has long been one of the central focuses of human life and labour. Anthropologists and archaeologists have a number of ways of talking about the different ways societies get their food. One common pattern in Eurasia is agrarianism, which basically has to do with tending fields of grain or other plants, which are harvested for food. Another common pattern is a animal husbandry approach, which is the practice of keeping domesticated animals around, at first mainly to eat as meat, and later on also for ‘secondary products’ such as milk, eggs, and wool.
Goats produce milk, a useful ‘secondary product’.
Both of these kinds of farming were used by the settled societies of the Ancient Near East, after the start of the Neolithic Revolution (very roughly around 9000 BC), and spread in some form or other to fairly large swathes of Eurasia. So it’s no surprise that Proto-Indo-European has at least a basic set of terms dealing with both grains and livestock.
The grain side of things isn’t too well developed. We find one main word for ‘grain’, which we can reconstruct as *jew- (or maybe *jewh₁-) – this shows up as Hittite ewan, Sanskrit yáva-, a Greek base zei-, and Lithuanian jãvas. That’s four branches, including Anatolian as well as from across the rest of ‘Common Indo-European’, so it seems safe to reconstruct this for Proto-Indo-European in the strictest sense. Exactly what kind of grain this word referred to, or if it was a generic word for ‘grain’, is hard to tell, since we find a wide range of meanings attested (barley, wheat, millet, general cereal), but it does seem reasonable to suppose that the people speaking Proto-Indo-European proper were at least vaguely familiar with cultivated grain crops of some sort. There’s also a reconstructible word for grinding, *melh₂- (think mill), which probably had specifically agricultural overtones of ‘grind grain, mill’ already in Proto-Indo-European. That said, the agrarian vocabulary for Proto-Indo-European is not great. It’s hard to reconstruct too many terms with any confidence even for Common Indo-European, much less Proto-Indo-European.
This is a bit of a contrast with terms for livestock and herding, which, especially for Common Indo-European (leaving Anatolian aside), seem to have been pretty well developed. We have a word that seems to have meant both ‘livestock’ and ‘money’ (*péḱu – English fee comes from this, reflecting the monetary sense, while the German cognate Vieh means ‘cow’), plus more specific terms for horse (*h₁éḱwos, as in equ-ine), pig (*suHs; think ‘sow’) and probably also piglets (*pórḱos, whence French-derived pork), and sheep (*h₂ówis; English ewe) along with their wool (*h₂wl̥h₁néh₂ – this notational monster is actually the source of English wool by normal sound changes).
And of course there’s cow. The English word goes back to something like *gʷow- (note the effects of Grimm’s Law), which is one of the best-attested words in the Indo-European family. Pretty clear cognates are found in every branch except Albanian. It looks like this word is also found even in Anatolian – it’s a little hard to be sure about the Hittite word, since scribes usually used a Sumerian-derived logogram here (kind of like we can write 8 for ‘eight’, which conveys the meaning just fine, but doesn’t tell the reader anything about how ‘8’ is pronounced, if they don’t already know), but in Hittite’s cousins Luwian and Lycian we find a word wawa- for ‘cow’, which is more or less what we’d expect *gʷow- to have turned into in these languagse. Alwin Kloekhorst suggests that the Hittite word was *kuwāu-.
Cows were probably central to the way of life of most Indo-European speakers. We have a couple of other bovine terms, like the yearling calf, *wételos, or the ox, *uksén-, as well as a verb for ‘to milk’, *h₂melǵ-, which seems to have had a specifically verbal sense ‘to milk an animal’ (rather than referring primarily to human milk, which is much less culturally specific).
Taken together with the wheeled vehicle words from the last post, this suggests a society not just with cows, but with the full range of ‘secondary’ products: exploiting the wool of sheep, using oxen or horses to pull vehicles, and drinking the milk of cows (and possibly goats). Cows also seem to feature exceptionally prominently in the mythologies of many later Indo-European speaking societies, from India to Europe – I’m generally hesitant to use this kind of evidence (cultural traditions often develop in ways only very loosely related to linguistic history), but in the bovine sphere it’s really too blatant to ignore.
Auðhumla, a primordial cow from Norse mythology.
(Described by Snorri Sturluson in the Prose Edda, Gylfaginning ch. 6 c. 1220.)
In any case, we seem to be looking for a society where cattle play a central role, where livestock herding in general is very important, where animals are used not just for meat, but for a range of other products, and where grains and cultivation are, while not unknown, of secondary importance. And while it’s not a guarantee from the linguistic evidence, people often look to the model of nomadic pastoralism, where speakers of Indo-European were relatively mobile, moving regularly from pasture to pasture, often seasonally.
Nineteenth-century anthropological ‘theory’ (which was often hardly better than very literate navel-gazing) put nomadic pastoralism within a sequence of societal ‘deveopment’: people started as hunter-gatherers, ‘improved’ to being pastoralists, and finally settled down to be proper ‘civilized’ agrarians. Aside from questioning the idea that such transitions are necessarily improvements, we also now understand that this kind of ordering doesn’t match the actual attested trajectories of what people actually do. In particular, many nomadic pastoralist societies seem to grew out of older agrarian ones: sometimes settled people, growing crops and tending some livestock, find it more advantageous to switch to relying on the livestock primarily, and relegate crop-tending either to a secondary task, or leave it to neighbouring peoples from whom they can get cereals by trade, threat, or force.
Talk of horsey nomads from the Steppe is liable to conjure images like this, based on much later models such as the Mongols.
This is pretty anachronistic.
The picture of hordes of mounted warriors sweeping past and whirling around to shoot arrows behind them depends on much later technologies, like smaller bows and eventually stirrups. At the early dates we're talking about, horses might help people raid farther out, but the actual fighting probably wouldn't have been from horseback. The numbers of people involved in a raid would like also not have been very large.
For thinking about the speakers of early Indo-European, this kind of thinking aligns things pretty well with the nomadic pastoralism common on the Eurasian Steppes. For the broad time period we’re talking about, based on the previous post, there’s two archaeological ‘cultures’ that need mentioning: one called the Sredni Stog culture, from about 4400 to 3400 BC, and another called the Yamnaya, from around 3300 to 2500 BC. These cover more or less the same region (the Pontic-Caspian Steppes), and have a good deal in common, including a strongly pastoral economy, and have long featured as major candidates for matching at least partly with speakers of early Indo-European.
Map of the Yamnaya (zone 1) and surrounding areas.
(From Anthony 2007, p. 302)
Of course, this is all fairly general, and there are lots of question marks. For one thing, any correlation of a reconstructed language with specific archaeological ‘cultures’ probably is very much partial. Even if the ‘Steppe hypothesis’ is broadly right, there’s no reason to think that everyone associated with the these ‘cultures’ archaeologically actually spoke early Indo-European, or that everyone who spoke Indo-European would be labelled ‘Sredni Stog’ or ‘Yamnaya’ by archaeologists today. The suggestion is just that there might have been a pretty good fit between Indo-European and these early Steppe ‘cultures’.
This fit is good enough that the Steppes have features in nearly every hypothesis about the spread of Indo-European in the last half-century, one way or another. But – there’s always a ‘but’ with this sort of stuff – this hasn’t stopped scholars from having serious disagreements, and for good reason. The big issue is this: even if we accept that these Steppe cultures (Sredni Stog and/or Yamnaya, depending on the date) included people who spoke a form of early Indo-European, and who were important in spreading some early dialects around, does this mean they necessarily spoke Proto-Indo-European as such?
The answer obviously has to be: no, not for sure. For instance, there are those apparent differences between Anatolian and the rest of ‘Common Indo-European’. Even though Anatolian shares in the general Indo-European words for cows and sheep (animals which were common even in very settled agrarian societies in western Eurasia), its vocabulary for both pastoralism and wheeled vehicles seems to be rather different than other branches (the old verb for ‘to milk’ isn’t found, for instance). There are a few ways these differences could have come to be. Maybe Proto-Indo-European was very pastoral, but Anatolian lost a good deal of this vocabulary at some point. Or, perhaps, ‘Common Indo-European’ was spoken by a society that became more pastoralist after it had lost contact with the Anatolian branch.
Some scholars have argued that Proto-Indo-European proper may in fact not have been spoken in the Steppes, but instead in some neighbouring area – perhaps a bit to the south, in the Caucasus region. At some point (so this hypothesis goes) there was a split, and a form of early Indo-European made its way north, onto the Steppes, where it was widely adopted by people practicing either the late Sredni Stog or early Yamnaya economies. This was ‘Common Indo-European’, while the Anatolian branch would (according to this view) derive from the dialects that did not move north, and which instead later percolated west into Anatolia. (It’s of course also possible there were other early Indo-European or cousin dialects kicking around the region too, and which were later lost without a trace – it’s good to remind ourselves periodically that our evidence is very patchy indeed!)
Is this approach right? I don’t know. This sort of research is very hard to do, for a bunch of reasons: it involves trying to coordinate evidence from very different disciplines (archaeology, historical linguistics, and now also genetics), as well as incorporating new research, which is ongoing in all of these areas, into a kaleidoscope of competing models. And more generally, we have to balance credulity with scepticism. If we’re too sceptical, then there’s no point in even trying to solve this puzzle, since it can’t be done without some speculation – and if there are real discernible connections to be found, we’ll miss them if we don’t look. But if we’re too credulous, we’ll end up with overly simplistic equations of different kinds of evidence, and produce really nice charts in books and articles that have little to no useful bearing on reality. Currently the latter pitfall seems to be a bit more dangerous, to me, and I’ve seen absurd talk of things like ‘Indo-European DNA’ and the like – which is not only deeply inaccurate, but disturbingly reminiscent of ideas (popular not all that long ago) about an ‘Aryan race’.
My own preference in all this is to accept that the Yamnaya cultures, at least, probably included a lot of early Indo-European speakers of some sort – this seems to be a pretty robust conclusion (granting that this is an inherently speculative area of study), supported by a lot of evidence, and incorporated into many different models (even ones that are otherwise thoroughly at odds with each other) – but I would prefer to wait and see about whether we should understand this to be Proto-Indo-European, or just some early dialects. At any rate, looking to Yamnaya seems to be a pretty useful starting point for thinking about the Indo-European languages in Europe, including Germanic, and that’s where we’ll pick up for the next post in the History of English.
It’s a bit hard to write a good ‘further reading’ for this post, since this is a vast and rapidly evolving topic. I’ll have to be extremely selective, and apologize for leaving out many excellent and important works. Also, some of the more recent work will come up in the next post (I'm also saving basically all discussion of genetics till then too).
There’s a long history to trying to place the speakers of Proto-Indo-European – if you want a long survey of older ideas, and the way the debates have evolved over time, Jim Mallory has a useful (though sometimes rather catalogue-like) article, A Short History of the Indo-European Problem.
For current research, the key archaeological reference for the ‘Steppe hypothesis’ in its modern form is Anthony’s The Horse, the Wheel, and Language (also mentioned in the last post). For more details on trying to reconstruct the society of Proto-Indo-European speakers from their vocabulary, a reasonably accessible starting point is Mallory & Adams, The Oxford Introduction to Proto-Indo-European and the Proto-Indo-European World – a lot of its specifics should be taken with a grain of salt, as the authors tend to err slightly generously on the side of including material as ‘Proto-Indo-European’ even when it may not be quite sufficiently supported for the proto-language proper, but it’s useful nonetheless, and fairly widely available. These two have also produced the voluminous Encyclopedia of Indo-European Culture. A classic earlier work is Emile Benveniste’s Le vocabulaire des institutions indo-européennes, translated into English as Indo-European Language and Society.
On concepts like pastoralism, there’s an enormous anthropological literature. I’ve turned to Alan K. Outram’s chapter ‘Pastoralism’ in A World with Agriculture, 12,000 BCE–500 CE. Also really helpful for me is Nicola Di Cosmo’s book Ancient China and its Enemies, particularly the chapter ‘The Steppe Highway: The Rise of Pastoral Nomadism as a Eurasian Phenomenon’. No doubt someone more expert in this field than I am could provide a fuller or more nuanced set of recommendations, but these two chapters should be reasonable starting points for thinking about pastoralism and its possible relations to the question of trying to pin down the speakers of early Indo-European.
The suggestion that the Steppes were a ‘secondary homeland’ was outlined by Thomas Gamkrelidze and Vjačeslav Ivanov in their 1984 book Индоевропейский язык и индоевропейцы: Реконструкция и историко-типологический анализ праязыка и протокультуры, translated into English by Johanna Nichols as Indo-European and the Indo-Europeans: A Reconstruction and Historical Analysis of a Proto-Language and a Proto-Culture. Gamkrelidze & Ivanov argue that Proto-Indo-European was spoken in ‘the fifth to fourth millennia B.C. within eastern Anatolia, the southern Caucasus, and northern Mesopotamia’ (p. 791), and are the ones to suggest a link with the Kura-Araxes culture. They see a group of later dialects moving up north onto the Steppe, and dispersing further from there. (This book also contains an extensive, though often controversial, investigation of Indo-European ‘linguistic palaeontology’, as discussed in the previous post, including an attempt to reconstruct an Indo-Euoporean word for ‘lion’.)
A similar case is made in a recent paper by Damgaard et al. (2018), who posit a location for Proto-Indo-European proper south of the Caucasus, and a Steppe dispersal for the rest of Indo-European. This paper is accompanied by an important ‘Linguistic Supplement’ by Guus Kroonen, Gojko Barjamovic, and Michaël Peyrot. This ‘Supplement’ is a useful synthesis of a number of aspects of the problem, and is my source for the mention (in the previous post) of place-name evidence for Anatolian going back to c. 2400 BC. I think the conclusions the authors draw from this are a bit too strong – because this overlaps with the very end of the Yamnaya period, they argue that ‘scenario in which the Anatolian Indo-European language was linguistically derived from Indo-European speakers originating in this culture [i.e. Yamnaya] can be rejected’. This doesn’t quite work, in my view, since Yamnaya was very long lasting, and a bit of overlap on the late end doesn’t rule out a derivation of Anatolian from the early Yamnaya, some eight centuries earlier. And in any case, the classic archaeological models already seek the origin of Anatolian in pre-Yamnaya Steppe cultures (e.g. Anthony 2007, ch. 11), so this new material would seem to confirm the standard models more than to disturb them.
That aside, both these newer studies (Damgaard et al. and Kroonen et al.) and the older work of Gamkrelidze & Ivanov are very interesting (though they don’t agree in all points, even some important ones), and are a good reminder that however important the Steppes and the Yamnaya (and their predecessors) may have been in spreading some Indo-European languages, this does not necessarily mean that this is where Proto-Indo-European proper was spoken.