Tuesday, February 14, 2017

Indo-European spread

Found this Creative Commons gif by Joshua Jonathan which I've unpacked into individual images so I can discuss each in kind.  This seems to be the latest and best understanding of the Kurgan Hypothesis for Indo-European, which has now pretty much eliminated any credible vestiges of dissent.  Not that the Anatolian Hypothesis doesn't still have its adherents, and other (less serious) theories don't still surface from time to time, but they aren't really credible anymore.  Mallory in 1989, Anthony in 2007 and subsequent archaeogenetic research have pretty much wrapped the theory up nicely.  So here, I present, the prehistoric spread of the Indo-European languages, as best as we understand them.  I deleted a few of the images because they were either unnecessary or redundant.


The first image in the gif is actually a bit misnamed.  It says Yamna (or Yamnaya) culture, but this predates the Yamna culture, and would refer to the Sredni Stog and other neighboring cultures instead like Dnieper-Donets II, Khvalynsk, Samara, etc.  and not extend as far to the east as it does.  The time depth here is roughly 4,500-4,000 BC.  This is the earliest period of common Proto-Indo-European.


At the end of the fifth millennium, there was a minor climate change; becoming a bit cooler and drier.  The farming communities of the early Balkans, which bordered on the steppes and which are not considered to have been Indo-European, were significantly depleted in the Danube valley, and some archaic Proto-Indo-Europeans moved into the area, spreading a more mobile economy with the domesticated horse, and (presumably) early Indo-European culture, including language.  This would have happened roughly 4,000 BC, but probably took a few hundred years to come to full fruition.  Anthony and Gimbutas have interpreted this as both the presence of a fair number of steppe intruders to the region, but also their establishment as an economic (and probably political) elite among the remainder of the "Old European" inhabitants in the region, who became "Kurganized"—i.e., they adopted much if not most of the steppe culture, including the language—but it was a far cry from complete population (i.e. genetic) replacement.  Anthony calls this new culture the Suvorovo-Novodanilovka complex, and it persists in the Danube for some time.

Although the timing is not clear, it's believed that this group (or at least some of them) eventually made their way southward into Anatolia and are the source of the archaic Anatolian language family; the most "primitive" or "archaic" of the Indo-European families, and thus the one that had to have left the common Proto-Indo-European sphere the earliest.  The Anatolian-speaking Hittite Empire was established in 1,600 BC, but clearly the Anatolian speakers had been in Anatolia for quite some time by then, and diversified into a variety of related languages (Luwian, Palaic, Lycian, Hittite, Mycian, etc.)  This move is believed to have happened by at least 3,000 BC.  Of course, if the proto-Anatolian branch of archaic I-E separated itself from the main body of the steppes and came into the Danube around 4,000 BC and didn't end up in Anatolia until 3,000 BC, obviously they spent some time in the Balkans and eastern Anatolia before ending up where they ended up.  Plenty of time for them to have swallowed up substrates, both genetic and linguistic, and been influenced by new linguistic and cultural neighbors in various ways.


The Yamna cultural horizon, from 3,300 to 2,500 BC, is usually considered the "classic" period of Indo-European unity (although the Anatolians had already split off by then.)  On the eve of the Yamna horizon, a wide variety of clearly closely related cultures spread throughout the region, and the farthest eastward of these, the Repin culture, is believed to have been both the source of the classic Yamna cultural tropes, and the source for the far-flung Afanasevo culture, which appears at the same dates as the Yamna horizon.  Despite the vast distance, archaeologists have long been fascinated with the almost identical material culture between Afanasevo and Yamna, and more recent research suggests that the burials include individuals who are indistinguishable genetically between the two cultures as well—they were clearly a completely transplanted eastern wing of the Yamna horizon.  This also partly explains the relative archaism of the Tocharian languages, in spite of their late attestation, although it needs to be pointed out that the Tocharian languages do not actually appear in the same place as the Afanesevo culture—although it's widely believed that they must have been derived from it, because there aren't any other candidates that can be taken seriously.

Further cooling and drying of the steppes may have been partly responsible, but the development of full blown nomadic pastoralism, including mobile homes (on wagons) and the invention of the chariot seems to have been what set the Yamna apart from the less committed type of pastoral nomadism that preceded it.  It allowed for more freedom, economic opportunity (greater ability to take advantage of ecologically marginal terrain) and as a result, probably resulted in a prestige social system as well, which swept up all of the steppe cultures into a remarkably unified horizon.  It may well have displayed the advance of a prestige dialect too—nobody ever believed that the entire unified Indo-European language was the same through this phase, and marked dialectical differences may well have existed, but the Yamna horizon might have sanded many of those differences down; up until the languages started spreading geographically shortly afterward.

Long-term settlements of mixed agriculture and hunter-gatherers that had existed on the steppes in river basins for centuries, if not millennia, finally are left behind as the culture becomes firmly "committed" to pastoral nomadism; although there is an east-west cline to many of these developments, with the east being the most nomadic and mobile, least agricultural, and most patriarchal (based on burials, at least).

Between 3,300 and 3,100 BC, the Usatovo culture developed further to the west; Anthony interprets this as local Balkan and Old European peoples developing a client-hybrid relationship with "overlords" from the steppes, who were themselves moving more and more into their territory, and therefore importing more and more of their culture.  Kurgans spread westward into Hungary and elsewhere.  To my knowledge, at least, there isn't really any good archaeogenetic research on the Balkans and the potential spread of Yamna genes into the area like there are for some other regions, but because the Balkans were relatively heavily populated (compared to areas north and east, for example) it seems unlikely that total population replacement was going on.


On the other hand, the Corded Ware horizon, which spreads like wildfire across northern Europe (2,900-2,400 BC), has been shown to have been at least 75% genetically Yamna, so it does represent a very definite migration of steppe populations into another area, where they largely replace or swamp the genetics of whomever was there before.  Exactly how the Corded Ware developed from the Yamna horizon is still somewhat unknown, although it's clear based on genetics (as well as numerous cultural traits) that it did.  Likely the new pastoral economy was so productive and efficient that it led to significant population growth in the steppes and new territory (lebensraum) was needed.  Because northern Europe was relatively lightly populated, it was easy territory for the new Indo-European dialects, including the Indo-European peoples and cultures to take over.  There is some evidence that there is a Uralic substrate in at least some of this territory, which makes sense as Estonian, Finnish, Lapplander and other Uralic languages appear historically on the northern fringe of Indo-European.  And there may well have been any number of anonymous substrates—we know of at least some non-Indo-European languages that appear to be vestiges from pre-Indo-European days (Etruscan, Basque, Tyrsenian, etc.) and there's long been suggestions that you can see evidence of substrate influence in the languages themselves; some of the dramatic sound changes that led to the development of Germanic, for example, have been proposed as what you'd get if non-native speakers attempted to speak "broken" Indo-European.

In addition to the Corded Ware, the Balkan Area seems to have thrown off its first branches; proto-Greeks are estimated to have arrived in Greece by late in the 3rd millennium BC.  By 1600 BC, they have evolved into the Mycenaean palace civilization that lasts until the Bronze Age collapse.  The Armenians also traditionally have been historically believed to have crossed Anatolia at some point in here to arrive in their historical seats fairly early.  The connections of Armenian are confused—it was initially believed to be a very eccentric form of Iranian, but was later shown to be an independent language stock that was heavily influenced by prolonged contact with Persians and Parthians and other Iranian languages and peoples.  It's now captivated linguists with possible early unity with Greek, suggesting that the two languages developed together apart from the rest of the I-E family for at least some time.  Although Armenian and Greek are both presumed to have relatively early migrated out of the Balkan area, to be honest, the paleo-Balkan linguistic situation is very poorly known, and while all kinds of Indo-European languages clearly resided there (Thracian, Dacian, Illyrian, proto-Greek, proto-Phrygian, proto-Armenian, etc.) the relationships (if any) between these is poorly understood.  Just about the only thing that is known is that the Thracians, Dacians and Illyrians were known peoples from the historical period who fought against both Greeks and Romans before being linguistically subsumed, and somehow out of that mess the Albanian language appeared.

In the east, on the other hand, and this diagram simplifies a bit, the Andronovo horizon is in some ways equivalent to the Corded Ware—a broad horizon that encompasses a number of regional variations—although it's a bit more recent than the Corded Ware, flourishing from about 2,000 BC to 900 BC, and supposed to have been a development of the Sintashta culture which preceded it (2,100-1,800 BC) which was a development of the Poltavka culture (2,700-2,100 BC) which is so similar to the Yamna horizon which preceded it that it's clearly just a temporal extension of it.

Andronovo does account for the spread of the Indo-Iranian branch, although exactly how Indo-Iranian split into Indic and Iranian is still anyone's guess.  It also suggests why Indo-Iranian is one of the "youngest" Indo-European language; it was among the last to develop from PIE, in the same environment both culturally and literally as PIE itself developed. Some of these late isoglosses suggest that a late PIE community that had shed Anatolian, Tocharian, and much of the European languages still influenced, to some degree, Indo-Iranian, Graeco-Armenian, possibly Thraco-Daco-Phrygian and even the Baltic and Slavic languages before they all finished going their own ways.

By the end of the period covered on this map, the break-up of Proto-Indo-European is complete—although we're still a pretty far cry from developing into the recognizable pattern of Indo-European families that we know.


BMAC, or the Bactria-Margiana Archeological Complex, sometimes called the Oxus civilization, is an urban settled farming community of the neolithic (2,300-1,700 BC) that has its roots to the southwest in the Near East.  It had contacts with the Harappan civilization, and the historical Elamites, and the BMAC people are presumed to have been the inhabitants of a cosmopolitan trade nexus of sorts.  Described by Soviet archaeologists and not really well-known in the west until the collapse of the Soviet Empire, BMAC has a complicated history with regards to Indo-European studies.  Although the base population is not to be derived from the neighboring Andronovo culture, and the original inhabitants are not believed to be Indo-European, contacts with the steppe are well known and appear to intensify around 2,000 BC.  In fact, the related Tazabagyab culture appear to be steppe Andronovans who settled down and adopted some BMAC cultural traits, including irrigation agriculture.  By 1,800 BC the BMAC cities shrink in size, steppe pottery appears more frequently, and kurgan burials appear in the highlands outside of the urban centers in larger numbers.  It appears as if the Andronovo presence becomes more concentrated, and they gradually kind of overwhelm the BMAC centers, although not without picking up a fair bit of BMAC culture (and presumably genetics) along the way.  Mallory states the implied mainstream position, that the BMAC was a Central Asian urban membrane through which Indo-Iranian, or at least Indic pass before appearing on the other side as some of the proto-Indic cultures such as Swat, Cemetery H and Painted Grey Ware.  Iranian is more complicated.  Iranian plateau cultures speaking Iranian languages certainly appear on the southern side of BMAC in later years such as the Medes, Parthians and Persians, but the steppe Iranians such as the Saka, Scythians, Sarmatians, Alans, etc. certainly remained on the steppes longer and couldn't possibly be part of the same wave that later emerged as Indic.

Parpola interprets these various archaeological complexes and cultures with linguistics as follows, although this is only one interpretation and does include (by necessity) some just-so stories:
  • Catacomb and Poltavka cultures (2,800-2,000 BC) (followed Yamnaya culture)—late PIE on their way to becoming Proto-Indo-Iranian.
  • Srubna and Abashevo cultures (2,000-1,800 BC) (followed Catacomb and Poltavka)—proto-Iranian—the historically attested Cimmerians are suggested to derive from this horizon.  Given that the Thracians are often associated with the Cimmerians, and Herodotus himself seems to suggest that the Thracians originally came from the northern Black Sea region, this would lend credence to some theories that Thracian was more closely related to Iranian than to other Balkan languages that originated in the Balkans as the result of earlier waves of "kurganization"—but given that Thracian is so poorly known, this remains a tantalizingly speculative discussion.  However, it's worth pointing out that many of the Balkan languages, including languages that are believed to have originated in the Balkans like Greek and Phrygian, are often considered to be late separators from PIE, and therefore have more in common with Indo-Iranian than other western European languages would.
  • Petrovka and Sintashta cultures (2,000-1,800 BC) (sub-units of Andronovo)—proto-Indic
    • This is a bit at odds with the mainstream, who does not see Andronovo as differentiated Indic yet, but most likely still both Indic and Iranian before they split from each other.  It also gives the Iranians a somewhat more vague and anonymous position from which to later emerge, but there are some precedents for accepting it.  Zoroastrianism is often interpreted as a new cult, interpreted as a wave of Iranian Avestan speaking peoples swarming over a territory previously occupied by early Indic speaking peoples with whom they then recognized a cultural distinction.
  • BMAC and Cemetery H (1,900-1,400 BC)—Dasas, i.e., non-Indo-Europeans who were swarmed and defeated militarily or culturally by the early Rigvedic civilization.  
  • Alakul-Federovo (Andronovo variants), early Swat, and late BMAC cultures (1,800-c. 1,400 or so BC)—hybrid "proto-Sauma-Aryan", i.e., practitioners of the soma cult, which was common to both Indic and Iranian peoples, but which was almost certainly picked up from the non-IE BMAC peoples, at least partly.    Interpreted as largely culturally Indo-Iranian or Indic, yet with some genetics and traditions picked up locally.
  • Early West Iranian Grey Ware (1,500-1,000 BC) the source of the Indic element among the Mitanni—a non-Indo-European Hurrian kingdom that was a neighbor and rival to the Hittites during much of the Bronze Age, but which has recognizable Indic linguistic influences.  Identified as an advanced wave of Indic speaking chariot warrior aristocracy that established themselves as a superstrate over an indigenous Near Eastern population.
  • Late Swat, Punjab and Painted Grey Ware cultures (1,400-800 BC)—Classical Rigvedic Indian culture, and the eventual "conquerors" of the subcontinent who brought Sanskrit to India, where it displaced languages of presumably Dravidian and possibly even Munda affinity which were previously spoken there.
  • Yaz, Seistan (1,400-1,100 BC)—proto-Avestan, i.e. early Eastern Iranian; so similar to Sanskrit that it might still have been mutually intelligible.
  • Gurgan Buff Ware and West Iranian Buff Ware (1,100-1,000 BC)—proto-Persian and proto-Median.
Regardless of whether this identification is correct, in some way or other, Andronovan filtered through BMAC to emerge on the other side as the languages of the Iranian and Indic families—even the Iranian languages that remained on the steppes (Scythian, Saka, Sarmatian, etc.) were influenced by it.  One notable item of interest is the possibility that the Andronovo hybridized or creolized to some degree with Uralic peoples who lived to the north of them.  There is both linguistic and maybe even some genetic and archaeological evidence brought to bear to support this story, but it remains unproven, yet interesting.

The last region covered here is the Tarim basin.  Archaeologically represented by a striking set of burials occupied by corpses that naturally mummified in the dry, salty environment, the oldest of which date to about 1,800 BC (but which stretch for the better part of 2,000 years) it's unclear exactly what all of them were, but they clearly represent peoples from a variety of extractions.  The most interesting, of course, are those who have a very pronounced European physical type, including red, brown and blond hair, tallness, blue and green eyes, woven tartans that look like they could have come straight off of proto-Celtic looms in prehistoric Austria, and presumably their languages—many centuries later two groups of Indo-European languages emerge in the area, Tocharian, an independent family with a number of very archaic features that cannot be derived from the Andronovo/Indo-Iranian group in any fashion, and eastern Iranian languages, particularly Khotanese Saka and some Indic religious texts.

Physically, the Tarim mummies (those that aren't clearly Asian in physical features, that is) come in what appear to be two clusters, a "primitive" and robust proto-Europoid physical type, especially to the farthest east and the oldest mummies, and a more gracile "Mediterranean" physical type, interpreted as the bearers of the Saka language.  The Chinese also referred to a number of barbarian tribes that lived here as well as to the area north of them that appear to have European physical features, including the Wu-sun and the Yuezhi.  Lacking any other candidates, the Afanasevo and Pazyryk cultures are presumed to have been the vector by which the Tocharian languages entered the Tarim Basin, and the late Andronovo cultures are presumed to be the bearers of Saka and other Iranian languages.  As with how the Andronovo turned into the historically attested Indic and Iranian groups, though, or what languages the barbarian tribes referred to by the Chinese spoke—we don't really know.


Of course, if by spending all of that time talking about the development of the East I've implied by omission that we understand how the Corded Ware and other Balkan cultures turned into recognizable European Indo-European groups, I have to make sure and disabuse you of that notion.  Very little is known about the dispersal of any of these cultures either, and their ties to various linguistic groups that later appear.  We don't even know (again, to the best of my knowledge) of any studies of archaeogenetics of the Balkan groups.  Somehow, though, the Balkan groups turn into Dacians, Thracians, Illyrians, and almost certainly the linguistic forebears of the Greeks, Armenians and Phrygians.  And somehow the Corded Ware culture turned into the Baltic, Slavic and Germanic groups (at least we presume so, since they covered the same physical area.)  And somehow, the Italic and Celtic speaking peoples, and maybe some other languages who's affinities are unclear, appear out of that mess too.

The Single-Grave culture, a variant of the Corded Ware found in Scandinavia, northern Germany and the Low Countries is proposed to be ancestral to the Bell Beaker culture (2,800-1,800 BC), a culture that spread to various places over much of western Europe, including the British Isles, has been proposed as both proto-Celtic as well as proto-Celtic/Italic/Germanic/Balto-Slavic.  Its connections to cultures that later are recognizably Celtic or Germanic (such as Hallstatt and Jastorf) is unclear, though.

Most likely the Bell Beaker culture represents European language groups that were on their way to differentiating.  This process is very poorly understood, as the various Bronze Age cultures of Europe are not easily derived from each other, and linguistically lots of connections between the European languages are presumed to still have existed.  The Italic languages (best known by Latin and the subsequent Romance languages, but which earlier were more diverse) are often believed to have maintained a period of unity specifically with the Celtic languages, as the Indic and Iranian languages did, perhaps.  The Baltic and Slavic languages are presumed to have done so too.  Germanic is harder to place, with some commonalities with the Balto-Slavic group, but some with Italo-Celtic too.  There's even a serious proposal that a completely anonymous Indo-European family existed which has since disappeared without a trace other than its influence on the sounds and grammar of Germanic and Celtic, called Nordwestblock in the Belgian region which separated Germanic and Celtic development zones.  And languages that are attested but very poorly known, like Venetic, Rhaetic, Ligurian, Lepontic, North Picene, Messapic, Illyrian, Thracian, and Dacian are anyone's guess as to their closest relatives.  Some may pass muster as Illyrian, Italic or Celtic languages—some may not even be Indo-European at all.

The problem in Europe is a little different than that in Asia—there are plenty of cultures and plenty of languages, but making a sensible matching of one to the other eludes us.  Big horizons like the Nordic Bronze Age, the Urnfield horizon, or the Atlantic Bronze Age defy geographical matching to later linguistic groups; where we have sharply defined lines linguistically when they first emerge in the historical record, we do not see anything like them in the archaeological horizons, which means that all kinds of linguistic groups are thrust uncomfortably into the same archaeological horizons and cultures.  On top of this, there remain a dizzying array of European language groups that are difficult to classify, and may belong to many or even none of the well-known Indo-European groups.  And contacts between Baltic, Slavic, and probably Thracian, Greek and Armenian with the developing Indo-Iranian languages appear to have happened too.

This dispersal of the Corded Ware and Indo-Europeanized Balkans into something recognizable from historical seats of known language families remains one of the bigger unresolved and possibly unsolvable problems of Indo-European research.  And most likely it can't be accurately resolved, really—although we can certainly hope to understand it better than we do now.  Today we have no choice but to try and locate the differentiated late Western PIE languages—proto-proto-Celtic or Italic or Germanic, etc. more or less where they emerge when we first have historical data to place them.  But most likely these proto-proto-languages; differentiating PIE stocks, jostled each other around and ended up covering different areas when they appear than they did when the start.  We see plenty of this in the historical record: Celtic has swept across most of Western and Central Europe and into the Balkans and even Anatolia when they first appear in the historical record, and can be pretty reliably tied to the Iron Age La Téne culture.  But the Romans conquer their provinces of Hispania and Gaul and even Brittania, replacing the Celtic languages there with vulgar Latin.  Later, Germanic invaders cover much of the same territory; Brittania (minus Wales) speaks Germanic English now, and Austria, the proposed homeland of the Celts, speaks Austrian German.

And that's just one example.  The spread of Romance, Germanic, and Slavic languages across much of Europe that formerly spoke some other Indo-European language in the historical period should give us pause and make us suspect that the same probably happened in prehistoric times as well, and because its prehistoric, it's probably unrecoverable, at least with any real confidence.  

No comments: