The Paleolithic Indo-Europeans


Back in the days when J.R.R. Tolkien was studying what was then called philology, the history of Indo-European was seen as the key to a remote and romantic era, a time of of great migrations and epic conquests.  That sweeping vision of past glories was what first attracted me to historical linguistics as well.

It was taken for granted in the early 20th century that the prehistoric past could best be understood in terms of warfare and colonization, just like the present. Wherever archaeological evidence suggested a change in culture, the assumption was that one people had replaced another -- or, at the very least, had subjugated another and become their rulers.  And the wide distribution of certain language families was taken to mean that their original speakers had been particularly powerful and ruthless warlords.

In particular, the presence of Indo-European languages everywhere from England to India was assumed to have been a product of the invention of horse-chariot technology shortly after 2000 BC.  The original Indo-Europeans were imagined as a horde of aristocratic Bronze Age warriors who came hurtling out of the steppes, overwhelming the simple peasant cultures of Europe and even toppling the supposedly decadent high civilization of the Indus Valley.

Despite its troubling racist overtones, that point of view was still dominant when I went to college in the 1960's. However, by the 1970's it had started to lose ground.  I remember being particularly startled when I read a book called Bronze Age migrations in the Aegean; archaeological and linguistic problems in Greek prehistory (1973) and discovered that there hadn't actually been very much Bronze Age migration in the Aegean.  Even the Mycenaeans -- who had previously been considered a prime example of invading Indo-European chariot-warriors -- were now reassessed as a purely local development.

That reassessment created real problems.  If the ancestors of the Myceneans were already living in Greece by 2300 BC -- before the invention of the horse-chariot -- they could not have arrived as horse-chariot warriors.  And if the chariot-warrior explanation of Indo-European expansion no longer held true for the Greeks, then perhaps it no longer held true anywhere.

So what was the secret of the Indo-Europeans?  If they were not the masters of an irresistible new form of military technology, then just what was the special advantage that had enabled them to expand so dramatically?

By the 1980's, it was also becoming clear that the conventional date for the Indo-European migration had to be off by not merely a few centuries, but thousands of years.  The earliest known Indo-European languages -- Mycenaean Greek, Hittite, and Sanskrit -- were already far more divergent in the second millennium BC than the offshoots of Latin, such as French and Italian, are today.  This suggested that their common ancestor must have been spoken not around 3000 BC, as formerly assumed, but well back in the Neolithic.

Such a radical redating suggested an equally radical solution to the problem of Indo-European dispersion.  The central premise of this new hypothesis, as presented by Colin Renfrew in Archaeology and Language: The Puzzle of Indo-European Origins (1988), was that the secret of the Indo-Europeans was agriculture.  They were, he argued, the people who originally brought the Neolithic to Europe from Anatolia.  It was not force of arms but rather the ability of farming to support a greater population that had enabled them to outbreed and eventually absorb the small Mesolithic hunting bands.

This new scenario seemed both plausible and exciting.  Not only did it greatly expand the historical depth of Indo-European linguistics, but its image of peaceful agriculturalists generously accepting more primitive hunters into their society was very much in tune with the political biases of the time.  Although Renfrew's suggestion of Anatolia as the Indo-European homeland was never universally accepted, it did seem as though a Neolithic hypothesis of some sort would ultimately provide the best solution to the puzzle.

However, in recent years, the agriculturalist theory has been undermined in turn by the hard facts of genetic analysis.  It seems that the Neolithic farmers who entered Europe from the Near East and North Africa were the source of no more than 20% of present-day European DNA, with the other 80% going back to the Paleolithic.  Apparently the farming folk, rather than multiplying rapidly and assimilating small bands of primitive hunters, were themselves the ones who were assimilated.  And, as Renfrew himself had pointed out, except in the special case of imperial conquest -- which was unknown before the rise of civilization -- it is unheard of for the language of a limited number of intruders to supplant that of the natives.

The DNA evidence also creates problems for the alternative theory that Indo-European was originally the language of certain inhabitants of the Balkans, who acquired agriculture from the east at an early date and spread it throughout the rest of Europe.  It seems that Europeans just haven't moved around very much since they reoccupied the northern part of the continent at the end of the Ice Age.  For example, when a nine thousand year old skeleton from Cheddar, England was subjected to DNA testing in 1997, it turned out that a local schoolteacher was an almost direct descendent.

In light of the DNA evidence, it is now being acknowledged that all the earliest agricultural societies in Europe show considerable similarity to the non-farming cultures that preceded them.  It seems like an obvious conclusion that if there was both genetic continuity and cultural continuity during this major transition, there must have been linguistic continuity as well.

But if the spread of Indo-European can no longer be attributed to either Bronze Age conquest or Neolithic population replacement, what does account for it?   The more precise our knowledge of DNA patterns grows, the harder it is to fit an Indo-European migration in anywhere.  Indo-European has been reduced to a kind of ghostly presence, with no firm ties to either history, archaeology, or genetics.  Instead of being the essential key to the thought and actions of past times, it has become an irrelevance -- almost an embarrassment.

And yet this image of a stealth invasion is ultimately not tenable.  The appearance of the Indo-European languages cannot have merely left faint and ambiguous traces -- it must have been an absolutely epochal historical event.  The unavoidable question, then, is how far back in time do we have go to local the arrival in Europe of the ancestors of today's languages -- and by what means did they become dominant?


I eavesdrop on the online discussions of the professional Indo-Europeanists from time to time, and although they keep arguing about the "where" of an Indo-European homeland, they don't seem to be coming up with any new approaches to the problems of "when" or "how."  In the meantime, radically new ideas have been percolating up from an unexpected direction -- not among the Indo-Europeanists themselves, but among the Uralists.

The Uralic family of languages includes Finnish and Hungarian, but most of its members are found to the east of Europe, in the Ural Mountains and northwestern Siberia.  Unlike the Indo-Europeanists, the Uralists never seem to have gotten their heads filled with romantic stories about great migrations and heroic conquests.  Instead, their central theme is, "We have always been here" -- and now the DNA evidence appears to be backing that up.  It looks as though speakers of Uralic have lived in pretty much the same place since the Paleolithic, only moving a ways south and then back north again in response to the advance and retreat of the ice.

Language Families of EurasiaBut if Uralic speakers were already in roughly their current location by the Paleolithic, it becomes difficult to see how Indo-European speakers could have slipped past them and gotten into Europe.  The obvious conclusion would be that the Indo-European languages have also been in Europe since the Ice Age, a surprising conclusion but one that matches well with both the DNA evidence and the archaeology.  That is the premise of the new and still controversial Paleolithic Continuity Theory of Indo-European.

At present, the adherents of this theory seem to be primarily concerned with marshaling arguments in favor of the long-term stability of the languages and cultures of Europe.  They're not very interested in historical change and have almost nothing to say about the problem of origins.  In order to address that larger question, it is necessary to step outside the Indo-European framework entirely and look at the broader picture of which Indo-European and Uralic are only a small part.  And that is where things start to get really interesting.

In recent years, certain linguists with the courage to look for really deep relationships have suggested that Indo-European, Uralic, and the Altaic family of languages (which includes Turkic and Mongolian) are all members of a single, ancient language group.  This so-called Eurasiatic group, if it is authentic, is so old that the commonalties among its members have been reduced to a small core of basic vocabulary plus certain grammatical elements.  And yet there are good archaeological and genetic reasons for believing in its reality.

However, even this ancient Eurasiatic family would be recent compared to a still more tenuous grouping whose members are scattered in a small number of isolated regions, most of them either in mountainous refuges or on the borders between larger families.  These so-called Dene-Caucasian languages give every impression of being the remains of a formerly widespread language group that was overspread and largely replaced by Eurasiatic.

The Dene-Caucasian languages of Europe and Asia are marked in dark red on the map above.  The westernmost member of the group is Basque, with its center in the Pyrenees Mountains between Spain and France.  Further east are the languages of the northern Caucasus Mountains, located east of the Black Sea.  One isolated language in the mountains of Pakistan and another in Siberia seem to be part of the collection.  And finally there is the great Sino-Tibetan family, which includes Chinese, Tibetan, and Burmese.

It is generally accepted that any common ancestor of the Dene-Caucasian group would have to date back to the Paleolithic -- but it has never been clear just how far back.  If we accept the possibility that Eurasiatic goes back at least to the late Paleolithic, then Dene-Caucasian must be older yet.  So let us start by hypothesizing that there were two separate expansions of language across northern Eurasia during the Paleolithic, one significantly earlier than the other, and see whether or not that theory fits the facts.

As it happens, there were two major cultural traditions which reached Europe from the east in this time frame.  The first was the Aurignacian, which arrived around 45,000 years ago.  The Aurignacians were the earliest modern humans in Europe, which had previously been inhabited only by Neanderthals.  The second was the Gravettian, which displaced the Aurignacian starting about 29,000 years ago.

The DNA evidence -- at least in the female line -- also seems to reflect both an earlier migration, which accounts for about 10% of contemporary European mtDNA, and a later one which accounts for another 65%.  It is widely accepted that these two components can be plausibly associated with the Aurignacian and the Gravettian cultures.

So let us go one step further and assume that all three lines of evidence -- archaeological, genetic, and linguistic -- are telling the same story, and see if we can use them to form a coherent narrative.


Let us start by imagining the world as it was 50,000 years ago.  It has already been 30,000 years or more since modern humans emerged from Africa, initially crossing the narrow mouth of the Red Sea from Ethiopia to the southern tip of Arabia.  These early humans were primarily denizens of the seacoasts, who moved quickly along the shores of the Indian Ocean and also spread inland along the great rivers.  By 70,000 years ago, they had gotten as far as New Guinea and the coast and rivers of China.

At that point they halted, or even retreated somewhat, as the onset of the most recent ice age had rendered much of the world unsuitable for habitation by modern humans.  Especially between 65,000 and 55,000 BP, all but the most favorable areas were locked in bitter cold and drought.  In time, however, the climate began to moderate, setting off a second great wave of migration.  One of the most important centers of expansion for this second wave lay in western Iran, between the Persian Gulf and the Caspian Sea, where the people of the Zagros Mountains had developed the set of radically new stoneworking techniques that characterize what is known as the Upper Paleolithic.

Upper Paleolithic migrationsAccording to Stephen Oppenheimer in The Real Eve, both archaeology and genetics suggest that about 50,000 years ago, people from that area began moving out in all directions.  Some went east towards the Indus Valley, where they mingled with the earlier inhabitants  Others travelled in a northwesterly direction, up the Tigris and Euphrates Rivers into what is now Kurdistan.

From Kurdistan, these Upper Paleolithic migrants spread still further.  Some went west to Anatolia (modern Turkey), and it was their Aurignacian culture that reached Europe before 45,000 BP.  Others went north into the Caucasus Mountains, where they replaced the local Neanderthals after 35,000 BP.  Still others went south along the Mediterranean coast about 30,000 BP and then spread across North Africa, where they gave rise to the Berbers.

Although Oppenheimer does not say so, these ancient Iranians seem like excellent candidates to be the original speakers of Dene-Caucasian.  Both the period of their migrations and many of the locations in which they settled show a precise match to what is known or hypothesized about the proposed members of that ancient language group.

At the same time, other Upper Paleolithic migrants were spreading northward into Central Asia.  Some probably made their way up the Indus Valley while others may have gone overland through Afghanistan.  As the climate grew warmer and moister after 45,000 BP, they were able to settle in the vast open steppes of Siberia, which had become well-watered grassland, full of big game.

There they expanded rapidly to both east and west. By 40,000 BP, one group of Upper Paleolithic people was living in the region of Lake Baikal in east-central Siberia.  Another had reached as far as the River Don in southern Russia, where their finely-crafted tools have been found at Kostenki (a site located in the suburbs of my father-in-law's birthplace of Voronezh.)

The ancient migrations of these Paleolithic wanderers can still be seen reflected in the Y chromosomes of their male descendants.  Just a handful of related Y haplotypes, all of them derived from a very early Indian lineage, are dominant among Europeans, northern and eastern Asians, and even Native Americans.  The two haplotypes which are found in a majority of the men in Europe and in the Americas respectively are particularly closely related to one another and seem to go back to a common ancestor in Central Asia about 40,000 BP.

However, no equivalent pattern is apparent in mitochondrial DNA, which is handed down from mother to daughter.  The mtDNA of European women derives entirely from the Middle East, while that of women in eastern Siberia and the Americas derives mainly from eastern Asia.  This suggests that the Upper Paleolithic was spread primarily by small bands of hunters who traveled fast and light, taking wives as opportunity afforded whenever their wanderings brought them far enough south to make contact with existing settlements.

As they spread out, these migrants probably gave rise to the several language families that show some distant affinity to Dene-Caucasian.  One such group is Eurasiatic, which is likely to have been spoken by those early dwellers on the Don.  Another has been proposed to consist of Korean and Japanese.  A third includes certain ancient languages of the northeastern tip of Siberia, along with Innuit, the language of the Eskimos.

The most distant member of this clan may be the vast language family which accounts for the majority of Native American languages.  This family, which is said to show similarities to both Eurasiatic and Dene-Caucasian, is thought to have had a common ancestor about 40,000-30,000 years ago, and its most likely point of origin is near Lake Baikal, where certain modern peoples still display the same distinctive mix of DNA types as Native Americans.


Please send any comments to Cory Panshin

Return to Worlds of Wonder

Return to Trogholm



Background courtesy of Eos Development