search this blog

Monday, October 30, 2017

Genetic and linguistic structure across space and time in Northern Europe


I feel that I need to do a double take, and demonstrate more obviously why my new PCA, the one that I introduced in the recent Tollense Valley warrior blog post (see here), should prove very useful for analyzing both genetic and ethnolinguistic links in Northern Europe between modern-day populations and ancient samples, particularly those from late prehistory to early history, which is when the main ethnolinguistic groups that today dominate Northern Europe formed. Judging by some of the reactions in the comments, not everyone was convinced, so let's try this again.

Below is a new version of the said PCA that focuses on several ancient individuals who, based on their archaeological contexts, should show strong genetic affinities to modern-day speakers of Celtic, Germanic and Slavic languages in Northern Europe. These are three Iron Age samples from what is now England, one Iron Age sample from what is now Sweden, and two Medieval samples from what is now Bohemia, Czech Republic, respectively. The relevant datasheet is available here.


And clearly these ancients do show the expected genetic affinities considering where they cluster relative to modern-day Northern Europeans in the two most significant dimensions of genetic variation. Moreover, despite the fact that the Anglo-Saxon and English Iron Age samples were all excavated from sites in eastern England, the Anglo-Saxons cluster between the English Iron Age individuals and the singleton Scandinavian Iron Age sample. This of course makes perfect sense, considering that the Anglo-Saxons were Germanic speakers with recent ancestry from very near to Scandinavia.

So everything seems in good order, and for now it's very difficult for me to consider that those Tollense Valley warriors who cluster alongside modern-day Slavic speakers on my PCA are not ethnolinguistically closer to them than to Celtic and Germanic speakers.

On the other hand, my standard PCA of West Eurasian genetic variation does a comparatively lousy job at matching ethnolinguistic origins with genetic structure, at least in Northern Europe. Note below, for instance, that the same Celtic and Germanic samples from England and Scandinavia form a tight cluster between the two Slavs from Bohemia. Hence, based on this PCA it would be very difficult, perhaps impossible, to correctly predict the ethnolinguistic ties of these ancients just by looking where they cluster relative to modern-day Germanics, Slavs and so on. Right click and open in a new tab to enlarge to the max.


But this is not surprising, because this PCA is based on a wider, more diverse range of populations, and so rather than being dominated by relatively recent, ethnolinguistic-specific genetic drift within Northern Europe, it's much more reflective of deeper, more basic genetic relationships across West Eurasia.

See also...

Tollense Valley Bronze Age warriors were very close relatives of modern-day Slavs

Ancient genomes from NE Europe suggest in tandem spread of Siberian admixture and Uralic languages into the region >3,500 ya

50 comments:

Anonymous said...

I see two Slavic clines:
PC2: Baltic - Uralic
PC1: Baltic - (descendants of) Hungary_BA
Any other sensible explanation than Slavs emerging as Balts expanding in these two directions?

Davidski said...

It seems, rather, that this structure already existed during the Bronze Age in the southern Baltic.

https://3.bp.blogspot.com/-yvHaAqTg--g/WfHMXKgF9AI/AAAAAAAAGLE/4XnXXO9pp-YAETU6nixx7KqyMADyA8WewCLcBGAs/s1600/Tollense_Valley_warrirors_PCA.png

Synome said...

It looks like a pretty good model, and I agree with the statement that the Slavic clustering individuals are closer to modern day Slavs than modern day Germanic or Celtic speakers.

If I had to make a hypothesis about this battle, my take is that it is related to the expansion of the warlike Urnfield culture into Pomerania and the formation of the Lusatian culture (1300 BCE-500BCE), which itself belongs to the Urnfield system.

Gimbutas believed that the pre-Lusatian occupants of this area were related to western Balts, a culture and linguistic group that is today lost. These people would indeed be closer to Slavs than Celts or Germanics.

In "The Balts" (1963):

'During the Early as well as the Middle Bronze Age, the territory occupied by the Baltic culture had reached its maximal size. In the west, it covered all of Pomerania almost to the mouth of the Oder, and the whole Vistula basin to Silesia in the south-west. During no subsequent age was this culture found so widely; piece by piece it began losing its border lands
...

It was not until the second vigorous
central European expansion, before and around 1200 B.C., that its culture was
greatly affected. Then the entire south-western corner of the area in question —central, eastern, and southern Poland — was apparently occupied by the central
Europeans.'


I don't know how much later Slavic immigration changed the genetic landscape of this area. Maybe someone else here could have a go at that. I'm guessing it had a measurable impact.

Simon_W said...

According to C14 dating of one of the genotyped BA individuals, the battle took place around 1007 BC, that's long after the beginning of the Lusatian culture! Hence the battle dates to the LBA, and that's the cultural situation at the time of the LBA:

https://jpst.it/16QH2

As you can see, the battle took place in the area of the Nordic Late Bronze Age culture. But given the proximity of the Lusatian culture and the Slavic-like DNA of some of the individuals, while other individuals cluster with Germanics and Celts, it's tempting to speculate about a clash of Nordic/Germanic locals with Lusatian invaders.

The Lusatian culture developped out of the pre-Lusatian culture (Vorlausitzer Kultur), which started with BZ B (1550 BC). The latter was a Polish variant of the Tumulus culture. Trzciniec influence from the east and further Tumulus influence from the south led to the change from pre-Lusatian to Lusatian, starting with BZ C (= 15th century BC). Especially this eastern Trzciniec influence makes sense in the light of the new ancient DNA findings.

Simon_W said...

As the Lusatian culture spread from the south northwards, it cannot have been Germanic. Here in the MBA map, there's still a big gap between the early Lusatian culture and the Nordic MBA:

https://jpst.it/16QJb

It's logical that they cannot have had closely related languages.

As for the stuff Gimbutas wrote about the Balts, I think that's most of all based on the hydronymy which shows a river with apparently Baltic etymology, the Persante, in Pomerania:

https://jpst.it/16QJt

But Gimbutas also speculated that the Pomeranian culture of the early Iron Age (kultura pomorska) was ethnically Baltic, because it expanded from eastern Pomerania. I find that highly speculative.

Simon_W said...

Still, there is still the problem that the Goths apparently were not the only Germanic tribe on the territory of present-day Poland. In particular there seem to have been the Rugii on the shore of the Baltic sea, presumably in the form of the Oksywie culture, and the Burgundian-Vandal-Lugian complex.

According to Ptolemy, the Burgundians inhabited all the land between the river Suevus (which can only be the Oder) and the Vistula: "...and the Suevi Semnones, whose boundaries beyond the Albis [the Elbe] extend from the area we mentioned towards the east up to the Suevus river, and the Burguntae, who inhabit from there to the Vistula."

And the Lugians were living south of the Burgundians, according to Ptolemy, up to the upper Vistula: "Back below the Semnones the Silingae have their seat, and below the Burguntae the Lugi Omani, below whom the Lugi Diduni up to Mt. Asciburgius; and below the Silingae the Calucones and the Camavi up to Mt. Melibocus, from whom to the east near the Albis river and above them, below Mt. Asciburgius, the Corconti and the Lugi Buri up to the head of the Vistula river"

Ptolemy doesn't mention the Vandals, but according to Pliny the Burgundians (and the Goths) were sub-groups of the Vandals, while the Vandals were one of the five major Germanic "races", the other four being the Ingaevones, the Istaevones, the Hermiones, and the Peucini aka Bastarnae. Tacitus mentions a similar account of the Vandals as being one of the main branches of the Germanics.

Tacitus and Ptolemy also mention several sub-groups of the Lugii, which hence must have been a large people, not a small marauding tribe.

Anonymous said...

@Simon_W

"According to C14 dating of one of the genotyped BA individuals, the battle took place around 1007 BC,"

No. This date was not calibrated. Already much has been done the combatants with radiocarbon dating, they all show the dating between 1200 calBC - 1250 calBC.
Only calibrated date is true date.

Synome said...

@Simon_W

I don't know why I thought the date was around 1250 BCE. Was that written somewhere on this blog? 1000 BCE is definitely too late for the formation of the Lusatian culture.

With that in mind, your NBA/Lusatian clash seems like the stronger hypothesis.

It seems the earlier Nordic Bronze Age didn't quite extend into this part of modern Germany--at least that's the impression I get from the maps and my reading so far. Is there an accepted date by which the Nordic Bronze Age culture began to spread more directly into this area? If it corresponds closely to the battle date, that would be notable.

I lean towards Lusatian being a mixture of Baltic like and Italo-Celtic like peoples.

Synome said...

Oh, well there is supernord with the calibrated date.

I guess the more likely scenario might hinge on the dating.

André de Vasconcelos said...

Also, according to wikipedia, "The authors of an article in 2017 say that the origin of some of the men may have been in Bohemia, and that 'the non-local combatants are not from northern Germany and must have traveled long distance to reach the battlefield at Tollense'."

If this, and the 1007BC date are correct, could some of these people be from Urnfield culture?

Teper said...

"The Balts" (1963): ...

That's ridiculous. Gimbutas cannot be taken seriously.

Teper said...

@Simon_W

As the Lusatian culture spread from the south northwards, it cannot have been Germanic.

It's logical that they cannot have had closely related languages.

If the battle was a result of an expansion from the South-East, then I think we might just run into the first trace of the "mythical" tribe of Venedians.

https://postimg.org/image/8xcw1koqm3/

As those people were fire worshippers and thus cremated their dead, such battles are the only possibility for us to examine their remains. We are lucky I guess.


As for the stuff Gimbutas wrote about the Balts, I think that's most of all based on the hydronymy which shows a river with apparently Baltic etymology, the Persante, in Pomerania:

Parsęta river has nothing to do with Balts. Its name has a clear Slavic ethymology. It comes from the Slavic (or rather Indo-Slavic, as it exists also in Sanskrit) root "prs" which has the meaning of "to sprinkle" or "to splash". It's present in such modern Slavic words as "pryskać" (to splash/sprinkle), "parskać" (to snort), "prysk"/"bryzg" (splatter) etc.
All those absurd linguistic theories and word derivations simply are produced by people who have no clue (and don't want to have a clue) about Slavic languages.


But Gimbutas also speculated that the Pomeranian culture of the early Iron Age (kultura pomorska) was ethnically Baltic, because it expanded from eastern Pomerania. I find that highly speculative.

A nonsense, out of doubt. Balts came into being in the East Baltic, where CWC assimilated some Mesolithic substrate and where later mixing with N1c Uralics happened. They never exceeded that area.
Pomeranian culture, on the other hand, originated from a subset of the Lusatian culture.

Arza said...

Re: Gimbutas

Since the Iron Age Germanics were pushing West Slavs out of their homeland. In the Middle Ages Lithuanians for a moment conquered East Slavic homeland.

Then, Germanic and Baltic historians, archaeologists and linguists created a theory which says, that the land once conquered by the Germans was ancient Germanic land and land once conquered by Lithuanians was ancient Baltic land. And Slavs were just recent invaders out of nowhere.

Why don't we trust the true scientists?
Ah, it must be _our_ ethnocentrism!

Re: Hydronymy

As Tworzyciel already explained, hydronymy in Poland, Belarus, Western Ukraine or Lithuania is preserved since the CWC times. Claiming that all the names are of Baltic origin without even checking for parallels in Slavic or Indo-Iranian (sic!) languages is just ridiculous.

Gaspar said...

@twasztar

Flat bed grave culture clearly represnets west-baltic people of more than 3000yo according to russian studies. these are found along the coast of modern poland and old prussia

Venedi lived on the baltic sea coast, east of the nogat river ( no where near the vistula river) , they where absorbed into gothic society, the remnants turned into the warmians who was an old-prussian tribe who eventually mixed in modern times with the masuria who where an old prussian tribe called galidians

Balts where in modern Poland way before any ancient germans or slavs ever appeared

Ryan said...

What do you guys make of those Finno-Ugric samples that cluster between the Germans and Slavs? Any idea where those samples are from? Are they IE speakers that got assimilated into Finnic society?

Synome said...

The ultimate source in Gimbutas "The Balts" about Baltic river names west of the Vistula is this:

L. Kilian, “Baltische Ortsnamen westlich der Weichsel,” Altpreussen, IV, 3 (1939), pp.
67-68; H. Krahe, “Baltische Ortsnamen westlich der Weichsel?,” Altpreussen, 1943: I, pp. 11-12.

I don't suppose anyone can track these down? I couldn't find anything.

Teper said...

Oy vey! German "studies" from the Nazi times! That certainly must be ze truth! :)

Arza said...

@ Simon_W

there is still the problem that the Goths apparently were not the only Germanic tribe on the territory of present-day Poland

Frankly speaking we still don't know for sure if any Goths were in Poland in the first place.
Gocie, Kociewiacy. Names of the Slavs living in the area of supposed presence of the Goths.

In particular there seem to have been the Rugii on the shore of the Baltic sea

Róg in Slavic means horn. Cape Arkona comes to mind immidiately. Or maybe Rugievit, who was worshipped there.

According to Ptolemy, the Burgundians inhabited

No, it's according to Germanic scholars. Ptolemy says here about Phrungundiones, not Burgundians.

the river Suevus (which can only be the Oder)

Suevus and Odra? It doesn't look even remotely similar. Isn't this rather attested Slavic name of river Soława - Souava (Saale)?

Suevi Semnones

Interestingly in Tacitus Germania they appear as Semones or Senones. Now compare this with Old Polish name Siemion (don't confuse this name with "Simon").

and the Burguntae, who inhabit from there to the Vistula.

Buguntae here, not Burguntae.

Lugi Diduni up to Mt. Asciburgius

https://de.wikipedia.org/wiki/Dadosanen + multiple toponyms with "děd" in Jesenik Mountains aka Ash Mountains.

BTW wasn't this word borrowed from West Slavic into Proto-Germanic?!?
https://en.wiktionary.org/wiki/Reconstruction:Proto-Germanic/daudaz
From Pre-Germanic *dʰowHtós

"Pre-Germanic", yeah, right.

https://en.wiktionary.org/wiki/Reconstruction:Proto-Slavic/d%C4%9Bd%D1%8A
https://en.wikipedia.org/wiki/Dziady
Dziady is an ancient Slavic feast commemorating the dead ancestors. The Polish word means "grandfathers".

Dziady/deads are tomorrow.

Corconti

https://en.wikipedia.org/wiki/Krkono%C5%A1e#Names
"Alternative linguistic theories mention a connection with the pre-Indo-European word Corconti, which is first listed by Ptolemy"

How all of the above looks in the light of the newest aDNA data? Doesn't it simply... fit?


Teper said...

@Gaspar

Balts where in modern Poland way before any ancient germans or slavs ever appeared

Sure thing! There were ancient Balts from the Atlantic to the Urals, but then almost all of them somehow magically vanished into thin air (I suppose that was the same kind of magic as with our favourite "Eastern Germanics" :) ), ancient Baltic land has been seized by those pesky Slavs, and that's why now there's only 5,5 mln of poor Balts in the world possessing 0,13 mln km^2 of land, while there's 360 mln of genetically more diverse Slavs possessing 18,7 mln km^2 of land :)

@Ryan

Aren't those Hungarians? They are basically language-shifted Pannonian Slavs with some German input.

Arza said...

@ Simon_W et al.

http://jassa.org/

Awesome blog about ancient sources touching the topic that we discuss here.

Matt said...

Didn't comment on these aspects before, will say I agree not so sure about the inferring linguistic aspects exactly for the Welzin samples - for analogy, Hungary BA sit in the "Slavic" region and Bell Beakers in the "Celtic" region, but it's pretty debatable to me whether they actually had an ancestral linguistic relationship. (E.g. did the Bell Beaker populations, if they had a single IE language, speak anything ancestral to Celtic, or was that from later movements as many linguistic-archaeological models suggest? It's not clear). YMMV.

(Like, I also guess these are kind of pretty complicated linguistic evidence issues, and hard ones to hash out well in this kind of forum setting, and I don't see these comment threads as tending to produce a whole lot of light. But obviously I have no authority to say what people should and shouldn't discuss.)

But! I would say that this all feels like it weakens my expectation that the major IBD linked expansion in the early Middle Ages that is roughly associated with Slavic (and Baltic) languages has that much to do with the signatures that come out in these kind of unlinked PCA between recent modern Europeans... This looks like more evidence that this signal proceeded that, and is not just a result of founder effect from that time...?

zardos said...

I want to note that Germanic is clearly between Celtic and Slavic. Not just on this PCA, but in a way also linguistically between Centum and Satem, yDNA wise between R1a and R1b. Actually Germanic is one of the few, if not the only Indo-European group in which R1b and R1a are both very numerous in its core areas.
Probably Germanic was at the centre of a development or between two centres at its birth and BBC and CWC clearly overlap in its area. We still dont know for sure when and where Proto-Germanic came into existence.

Synome said...

Yes, it's all still very murky at this point. Perhaps that's why we've been reduced to quibbling over hydronymy and such. There isn't enough yet genetically to narrow our search.

Still, the Slavic related cluster in the PCA is very interesting, and I don't know how many would have predicted its existence at this place and time beforehand.

Arza said...

@ Matt

not so sure about the inferring linguistic aspects

I think that in case of the Welzin samples the only ones who care about the language of WEZ are Western Europeans. Everything, but the Slavic. Even Open Genomes, who usually fights with racism and discrimination, suddenly tried to make Nordics out of Pannonians.

And I think that no one here tries to make Proto-Slavs out of Bronze Age Hungarians or WEZ. We're just joyfully highlighting the fact, that we have found another Central European Bronze Age population that shares recent drift with us.

kind of pretty complicated linguistic evidence issues

Not really. Everything is beautifully pro-ste.

But obviously I have no authority to say what people should and shouldn't discuss.

So can we go back to a discussion about DNA and can we leave behind the "6th century AD" mantra, "Baltic hydronyms" that are not Baltic, "Suebi" who in all manuscripts of Germania are written Suevi or "Ptolemeus wrote that there were Burgundians in Poland" despite he didn't? I hope so.

Re: Medieval expansion.

In the paper that you have linked they wrote, IIRC, only about the gradient going to the Balkans. Additionally the analysis lacked neighbouring Germanic populations for comparison.

No one says that there was no migration in the early Middle Ages, but it was rather a minor one. Not the first one and not the last one. It looks like it originated somewhere around northern part of Carpathians. Going by genetic gradients Slovenians should live in Slovakia. Also there was a movement in the north-eastern direction which is now (possibly) confirmed by Sungir6 sample (autosomal, Y-DNA, possibly mtDNA). This also may be the reason why Western Ukrainians seem to be more "western" than Poles.

Why all of this happened (including shifts in material culture and depopulation)? Downfall of Rome, bubonic plague, invasions from the East. But around this time somewhere near Slovakia also something positive happened. Heavy plow was invented.

The heavy plow and the agricultural revolution in Medieval Europe
https://doi.org/10.1016/j.jdeveco.2015.08.006
More systematic evidence on the evolution of shares is given in Henning (1987) for South Eastern Europe, which encompasses parts of the Balkans as well as Hungary and Slovakia. Henning shows that from the 3rd to the 6th century there is no systematic asymmetry in the shares found, but concludes that for the period from the 7th to the 10th century there is a strong “overweight of left-sided asymmetry” (1987, p. 55). This is consistent with White's view that Slavic tribes had the heavy plow from around AD 600.

and

That the heavy plow explains more than two fifth of the increase in productivity observed in the High Middle Age in Denmark is not unreasonable given the large amount of clay soil in this country.

That was their advantage.

In the previous thread Simon_W (thanks!) posted a map of Slavic migration inferred from the hydronymy. I'll add that this is only the youngest toponymic layer as I highly doubt that they considered Odra or Wisła to be of Slavic origin. But what is interesting in this map is that it quite nicely matches some genetic data.

Overlayed on a migration rate map:
https://s6.postimg.org/hajvo1701/migmig.jpg

On a I2-Din distribution map:
https://s6.postimg.org/cbwd9hvhd/mig_I2din.jpg

As one can see this migration barely affected West Slavs. Czechs and Slovaks look here like South Slavs that never moved South. Additionally Lusatian Sorbs seem to be also affected (and they are linguistically connected to Czechs) so maybe they are not the best example of proper West Slavs, like Obodrites or Veleti. This pattern of migration can also explain similarities between Prague and Korchak cultures. Sungir6 autosomal and Y-DNA also matches (although he's pretty late, so we need to be cautious).

Makes sense?

Slumbery said...

Ryan:
Those must be Hungarians, at least the position on the PCA suggests that.

Matt said...

@Azra: I think that in case of the Welzin samples the only ones who care about the language of WEZ are Western Europeans. Everything, but the Slavic.

Well, in the original thread, the first comment to identify some of the Welzin samples with Slavic speakers was EastPole ("October 24, 2017 at 12:52 AM"), and others then commented in response, and it all snowballs from there. I don't really see it that it's a totally agnostic discussion about language on the side of commentators from Eastern European / ethnic Eastern European backgrounds, though it is on tangents at times.

"Baltic hydronyms" that are not Baltic

I think I may be to blame for introducing this to the discussion, though generally this tends to get out of control (to be honest, I just mentioned it as the supporting evidence that Baltic languages may have had a wider sphere than present, and was not expecting a kind of "I don't think that's good evidence / true point" response and not strong tangents about it and Gimbutas's "ethnocentricism" and such).

We can leave the discussion of languages in the Bronze Age and later aside; before that though, just as a last thing I'm still unclear what your model is on distribution of the Slavic languages is before the early Middle Ages. Do you have any mental list of which regions you think they widespread were distributed in and which they weren't? Sorry if this seems like a simplistic question.

Ryan said...

@Slumbery - Aha! Good call.

Unknown said...

David, I have a question from a Swedish archaeologist friend who wants to know where in Sweden the ancient Swedish individual's remains were found. Can you advise?

Davidski said...

@John Massey

The Nordic Iron Age sample is from the Oxie 7 site just southeast of Malmo.

Unknown said...

Thank you. Much appreciated.

Anonymous said...

I think immense caution needs to be exercised here, for two reasons.

One is that, while genes and languages often go together, they don't always.

The other is that at the time periods being discussed, there were no such things as "Celts", "Germanics", "Slavs", etc. If we're talking about 1200BC, we're probably at least 400 years before any noticeable 'Celts' (and iirc the linguists wouldn't even be comfortable with that), and maybe 1400 years before the era of Proto-Germanic, and 1700 years before Proto-Slavic. And by comparison, we're only about 1000 years AFTER the breakup of Proto-Indo-European (which is only a bit more than the difference between Scots and English).

At this time, there would have been pre-Celtic, pre-Germanic, and pre-Slavic languages, yes. But they would have been much more similar than today - we're talking about differences like those between, say, Portuguese and Italian. The European families would all have been in contact with one another, there would doubtless have been intermediate forms and areas of mixing, and it's likely that a person from one language area could understand some of what was being said by someone from another language area, and could quickly learn the language if they wanted (if, say, the other tribe became dominant in that area). This would have been a more fluid situation, between much more similar languages, than we might imagine. In this environment, it's not really plausible to pin specific IE branches to any specific prehistoric populations from this time period, I don't think. It would be like guessing whether someone today spoke Spanish or Catalan based on their genes - there may be tendencies, but they won't be very reliable.

The other thing to point out in that regard is that we have no idea how many branches of Indo-European there really were. Germanic, Slavic and Celtic did not split up until the 1st millennium AD, so it's possible that in 1200BC there were only these three languages, spoken over large areas due to recent expansion. But it's also possible that each of these branches had already split up repeatedly, and it's just that only one language from each branch survived to have modern descendents.
And it's ALSO possible that there were, say, a dozen different major branches of IE in Northern Europe at the time, and that only Baltic, Slavic, Germanic and Celtic ultimately "won" and took over the area. [Slavic, Germanic and Celtic all observably had massive expansions in relatively recent history, and we have no idea what they replaced in their new territories. Slavic may have expanded over near-Slavic, non-Slavic IE, or non-IE (we know pre-IE survived into historic times in at least Iberia and Italy), and we just don't know.]


(It's important to bear in mind that when a linguist says something like Baltic and Slavic split at a certain date, what she has in mind is not that two tribes of people started speaking two different languages at that date. What she means is that a clear sound change propagated through only part of the linguistic community, dividing it in two in a way that is retrospectively reconstructable, so that all samples from that point on can be sorted into one of two groups. But that doesn't mean that anyone at the time is even aware that they've "split". For instance, by that metric, Standard Southern British English and General American English have already "split", due to clear changes in the vowel systems and elsewhere. But if three centuries from now the British and Americans have developed into totally distinct Americ and Britic 'ethnolinguistic groups', it would be rash to retrospectively identify what dialect someone in 2017 spoke by their genes. Particularly when your samples actually came from, say, Ireland!)

Anonymous said...

Matt: regarding where the early Slavs were before the early middle ages: well, the technical answer is 'nowhere'. Before the early middle ages, we'd be dealing with "pre-Slavs".


What we know from history is that the Slavs suddenly appeared, swarmed over huge areas of the Empire, and within a hundred years appear to have dominated a vast expanse from Slovenia to Denmark to the Urals. Their armies were said to be immense, and they crushed all opposition. But until their sudden appearance, everyone we can identify in central and eastern Europe seems to have been linguistically either German or Iranian (and the Germans themselves had expanded into the area only a few centuries earlier).

Linguistically, however, we know that all the modern Slavic languages, over their entire range, date from a single language spoken around AD500, with no noticeable dialectical variation at that time (the dialects developed after the expansion, in the period 500-1000, after which they'd broken into separate languages). Proto-Slavic appears to have been a highly conservative language, not greatly differentiated from the Baltic languages, with most of its distinctive features developing during the post-expansion period. The Baltic languages are themselves highly conservative relative to the rest of IE (in many ways, at least).

The combination of conservativism and lack of internal diversity strongly suggests that the first Slavs were a small community occupying a small area. Where that area might be is unknown, but presumably somewhere in central or eastern Europe - people have suggested everywhere from the Baltic to the Black Sea.

The fact that Slavic expanded so amazingly rapidly, and continued to undergo common changes for another half a thousand years, suggests that the expansion primarily reflects its adoption as a lingua franca by a large multi-ethnic political/cultural confederation, with secondary migration into neighbouring areas (like Greece, Dalmatia, etc). Most of what were 'Slavic' genes by 700AD, therefore, would have been held by speakers of Iranian, German, Baltic, Venetic, Dacian, Thracian, Unknown Branch of IE (and so on) languages in 300AD.

Matt said...

@vacuouswastrel: Appreciate the effort, though have to say I did actually already understand that what you've presented is something close to the consensus model (and that "the technical answer is 'nowhere'. Before the early middle ages, we'd be dealing with "pre-Slavs").

My understanding is that some of the posters in these threads (Arza, EastPole, Twazstar) have an alternative model where some or all of the following hold true (I'm not sure they all totally agree, but I'm pretty these are positions some or all advocate):

- The reconstructed Proto-Indo European language did not actually exist (in any of the forms reconstructed by the comparative method). Instead a language which was fairly close to Slavic and/or Baltic existed. All reconstructions of proto-Indo European have erred in that they assume too much divergence from proto-Slavic.

- Strong substrate effects in different branches of Indo-European merely give the appearance that these Indo-European languages have been evolving separately from a PIE ancestor. In actual fact, they all descend from a language much more like proto-Slavic-like language with strong substrate effects which were avoided in Slavic. (This is possibly even more the case for the very divergent Anatolian and Tocharian reconstructions).

- The language immediately ancestral to the languages classified today as Slavic language was spoken by a geographically wide ranging community of people well across Eastern European (from Bohemia to Western Russia and down to the Balkans in the south) well before any early Middle Ages migration, possibly even by the early/late Bronze Age, across much of what is currently recognised as the "territory" of today's Slavic speakers. Today's distinct Slavic languages were in place separately well before the early Middle Ages.

I think this is the argument though I may have got things wrong in the details.

Comment re: where the speakers of Slavic were in early Middle Ages was really I wanted to try and understand the spatial distribution at least in Azra's take on this model.

@Davidski, would it be possible to put out the Global 10 for the Welzin samples? I would like to run them against some of these other PCA together.

Anonymous said...

Ah, sorry - I didn't see any such theory explicitly laid out above.

The thing is, there's no fruitful discussion to be had with people who believe your first two points. The crucial flaw in the theory that IE descends from anything similar to proto-Slavic is that we know it's not true.

We know this because distinctions made in other branches are not recoverable from proto-Slavic, or even from proto-Balto-Slavic. Most obviously:
- PBS had only two stop series, because two of the original three series had merged. If other branches evolved from PBS, they'd have to all independently split the voiced stops into two series, with identical distributions in each language, and then in the case of many languages re-merge at least two of the series again (but not the same two in each language).
- PBS had only a single short non-high back vowel. That means the other branches would have to split this into *o and *a independently in most of the other branches, for no apparent reason, all in the same way.
- the ablaut system is much less productive than in PIE

These proofs are suplemented by a whole heap of other probabilistic reasoning: almost all the changes between PBS and PIE are much more plausible in one direction than the other. For instance, take the three sibilants of PBS. Either:
- *k' assibilated (a strong stop weakening to a fricative)
- *s backed adjacent to RUKI.
OR:
- a sibilant spontaneously strenghtened everywhere to a stop
- two sibilants merge, in such a way that the distribution in the parent language is perfectly predictable from the surrounding consonants in the daughter language.

Clearly, there's a 99.9% chance that it was the former.

Additionally, pre-PBS words are directly attested in early borrowings into Uralic.

[It is theoretically possible that Slavic was a single language spoken across the whole of central and eastern europe for thousands of years with no dialectical variation. "Theoretically" in the sense of "not actually", because that's not how language works.]

Jaap said...

Vacuouswastrel: Thanx for a very helpful comment! And thanx Matt, for a very helpful reply. Still out of my depths really. Difficult stuff to wrap your mind around. I would like to mention the Eastern connection. Closely related (autosomally) Steppe clans migrated east (R1a making it to the Indus valley, R1b founding Afanasievo)and west (R1a CWC, R1b BB). Bit of a riddle. R1a: 'Hey, we're off to India first thing tomorrow! Care to join us?' R1b: 'Mwah. It's Altai for us. See you guys around!' And they did, but not in India. It seems hard to get past the comic-book narrative. Clans kept apart/split already on the Steppe (and came back together again elsewhere in many places). But where and when the splits occurred in the present language-distribution is only coming into focus in the last 2 millennia. Historical information as to what went down 3 millennia before that is hard to come by. Archeology is the only provider here, and it's difficult to read. Plus the Vedanta, but here we need our Indian friends (and EastPole, of course!). Avesta? Out of my depths!
But here's my point: Sanskrit and Avestan sources are by implication (and by implication only!) 5 millennia old.
Then there's the Irish sources! Fourteenth century manuscripts, bwah! But how seriously can they be taken? I very much doubt they are the phantasy of monks, although the famous Irish narrative slant is bound to twist the matter a bit. Or a bit much, actually. They may have a much more ancient 'implied date'. Rubbish until one starts spotting patterns with the eastern branch ... For example the use of memory as a vehicle, plus the training involved. Sanskrit still taught (and remembered) today! The Celtic, Germanic and Slavic worlds ignored writing for more than a thousand years! Minting? Fine! Writing? Nah! 'Ve 'ave our vays, ya no'. Gaulish was never recorded in writing for exactly this reason.
For a language split to be visible after 5000 years is incredible! But the R1a - R1b dichotomy, plus the cultural and linguistic relatedness between Sanskrit and Slavic, plus a host of other clues, makes this conclusion possible. Late PIE is practically identical with Proto-Slavic. Or what the hell, Pre-Proto Slavic ...
Vacuous Wastrel, can you put me straight here? I'd welcome that!

Jaap said...

Ah! Just saw your reply to Matt. Must study that, but for now can only assume the matter settled!

Slumbery said...

The more I think about this, the less I think that there were Slavs in this battle. The main argument is that a considerable part of those people is within the modern Slavic variation. But the modern Slavic variation is a result of the early Medieval expansion that led to mixing and increased the genetic variation of Slavs. The pre-expansion Slavs had to have a considerably smaller variation, and would cover a considerably smaller area on this PCA. And it stands to reason that this smaller area would be the left side, adjacent to modern Baltics, who are less mixed. The unmixed early Slav from Bohemia is also there on the PCA.
I think nobody who is in the right half of the modern Slavic variation on this PCA can be a pre-expansion Slav. Pretty much all of the Welzin samples are outside of the area were pre-expansion Slavs should be suspected (the only one that is not too much to the "right" is too Uralic).

If I have to guess, these were some Corded Ware derived cousin lineage that might have spoken a related dialect, but definitely not ancestral to the Slavic "proper" group that later expanded.

EastPole said...

@Slumbery
“The more I think about this, the less I think that there were Slavs in this battle. The main argument is that a considerable part of those people is within the modern Slavic variation. But the modern Slavic variation is a result of the early Medieval expansion that led to mixing and increased the genetic variation of Slavs. The pre-expansion Slavs had to have a considerably smaller variation, and would cover a considerably smaller area on this PCA.”

I disagree with you. It is contrary to logic. Early Slavic societies which resulted from mixing of very divergent genetic groups like Corded Ware and Tripolye/TRB were definitely more divergent then modern Slavs. It took long time of intermarrying within Slavic group to converge to modern Slavic variation. I am talking about Northern Slavs because in the South it was different.
Early Slavs could fell anywhere between Corded Ware/Trzcinec and BA_Hungary.
I am still very surprised that as early as Bronze Age from Tollense to Ural we have very genetically uniform populations who are also similar to modern Slavs.
Let’s wait for Y-DNA and if those ‘Tollense Slav’s' had Y-DNA more related to Slavic than to Germanic Y-DNA then they most likely were real Slavs.

Teper said...

@Slumbery

You are entitled to your opinion, but your post is nothing more than a heap of unsubstantiated assumptions.

EastPole said...

@Matt
“My understanding is that some of the posters in these threads (Arza, EastPole, Twazstar) have an alternative model where some or all of the following hold true (I'm not sure they all totally agree, but I'm pretty these are positions some or all advocate):

- The reconstructed Proto-Indo European language did not actually exist (in any of the forms reconstructed by the comparative method). Instead a language which was fairly close to Slavic and/or Baltic existed.”

It is not an alternative model, it is a mainstream view. Reconstructed PIE is not a real language, there is absolutely no evidence that it was spoken or understood by anybody:

“Firstly, it is useful to distinguish between the hypothetical, reconstructed ‘language’ which is the result of the operation of the comparative method on the IE languages, what we shall call ‘reconstructed PIE’, and the unattested spoken language from which we presume all the IE languages derived, which we shall refer to as ‘the spoken IE parent language’. Reconstructed PIE may have some features in common with the spoken IE parent language, but it is not the same as it, and is not a real language. Reconstructed PIE is a construct which does not have an existence at a particular time and place (other than in books such as this one), and is unlike a real language in that it contains data which may belong to different stages of its linguistic history.”

“Indo-European Linguistics: An Introduction” By James Clackson

https://books.google.pl/books?id=DJDjNp6wODoC&printsec=frontcover&dq=Indo-European+Linguistics:+An+Introduction&hl=en&ei=zk11TZq5A5HpOcTXuJcK&sa=X&oi=book_result&ct=result&redir_esc=y#v=snippet&q=Reconstructed%20PIE%20may%20have%20some%20features%20in%20common%20with%20the%20spoken%20IE%20parent%20language%2C%20but%20it%20is%20not%20the%20same%20as%20it%2C%20and%20is%20not%20a%20real%20language.%20&f=false

Matt said...

@vacuouswastrel, thanks for taking the time to comment btw.

@EastPole, that is a statement that reconstruction of PIE is necessarily imperfect though; it's not a statement that the reconstruction does not at all capture the features of an actually existing ancient language, and that reconstructed PIE is no more or less viable than what I understand you to be proposing (which is very much not the mainstream view).

EastPole said...

@Matt
According to great linguist Oleg Trubachyov proto-Slavic most likely originated around 3000 BC. Trubachyov also stresses uninterrupted origin of Slavic (which excludes mixing and development outside PIE homeland). Kortland and other linguists also think that Slavs remained in PIE homeland.

It is also puzzling how close Slavic languages are to Vedic Sanskrit and how many ‘PIE’ words can be explained in Slavic languages:

http://www.anthrogenica.com/showthread.php?10241-How-Robust-is-the-P-I-E-model/page3&p=225770#post225770

Now it all seems to be confirmed by genetics. It is simply not true that there is a consensus in linguistics and that vacuouswastrel represents mainstream. But you can believe what you want. I am here to learn something new from David, not to fight old propaganda which evidently doesn’t fit genetics.

Ric Hern said...

Maybe there should be a distinction between Genetic Slavic and Linguistic Slavic just like there should be a distinction between Genetic Irish and Linguistic Irish.

Most Irish today speak English yet they are direct descendants of the original Irish of 2000 bC. as DNA shows. Does this make them less Irish ?

Ric Hern said...

When we look at Early Tocharian, which is proposed to have split just after Hittite, we see that some words are closer to Slavic, others closer to Germanic and yet other closer to Irish etc. The same we see in Iranian languages.

The question we have to ask ourselves is when did Slavic become more Slavic than Proto-Indo-European ? When did it cross the 50/50 line ? Because there is no written evidence of such early events it is basically impossible to say.

This is why I would rather only prefer to talk about Genetic Slavs. In other words Genetics of todays Slavic Speaking people.

Ric Hern said...

Which of todays Slavic Speaking people was the Most Isolated Group both Culturally and Genetically since 2000 bC. ? Which area between Poland and the Urals was the most Isolated ?

Matt said...

@Ric, well "linguistic Slav" or "linguistic Irish" is quite a simple term - they're a group that speaks a Slavic language after the divergence of proto-Slavic from Baltic (and other IE languages), or that speaks a form of Celtic that's ancestral to Gaelic (Goeidelic), after the divergence of Goidelic from Brythonic.

But genetically it's much more complex. In Ireland, there's a strong continuity of Ireland_EBA, the Rathlin individuals, and present day Irish people. However, continuity is perhaps better with British_Iron_Age individuals, and the Rathlin individuals look subtly different in the broad autosomal picture for present day Irish, as they look to have more steppe ancestry.

So this is tricky enough with the Irish, who are a small homogenous island population that has probably experienced relatively little gene flow post early Bronze Age.

It's even trickier for Slavs, who are a large and autosomally varied group, and for whom there are ancient individuals like BR1 and BR2 from Hungary who look share lots of specific drift with them, but who look even more different from most Slavic populations in the broad autosomal picture than Ireland_EBA are from Irish.

This is also true for the Tollense individuals, who in broad autosomal West Eurasian picture still look different than present day individuals, including present day Slavic individuals - See Davidski's https://4.bp.blogspot.com/-0X53ULQMZ6U/WfhwdhmT9zI/AAAAAAAAGNQ/zs34s9OzA00HFftgKc-s10cus_wWLAwGQCLcBGAs/s1600/WEZ_West_Eurasia.png. Like Hungary BR1 and BR2 and the rest of the Hungarian cline, they look to have a richer WHG+EEF balance, and less steppe ancestry than present day peoples from Northern Europe, though there may be a few who overlap perfectly.

At what point does a population / individual then become a "genetic Slav"? Cutoffs are even more difficult in genetics than language, not because we have no evidence, but because the what evidence we have will not be amenable to clear cutoff points when they are part of a continuous process of development and change (language or genetics!).

It might be easier to just say that these individuals we find are likely to be "ancestral to the Irish" or "ancestral to Slavic peoples", rather than try to find a cutoff for a "genetic Slav" or "genetic Irish".

Ric Hern said...

@ Matt

What you say makes sense however when we are talking of even Older connections regarding Proto-Indo-European then it bacically throws everything out of the Window....

If Gaelic or Slavic can not be connected to specific Genetic Signature, how then does it make any sense to try and link Proto-Indo-European which is thousands of years older to a Specific Genetic signature ?

Are everybody then basically wasting their own time ?

Matt said...

@Ric, well the reconstruction of the proto-Indo European urheimat, at least the urheimat ancestral to all or most living forms of IE, should have a lot more evidence behind it.

But I think the people, who are not wasting their time, are trying to use the ancient genetic evidence as an extra-plank to understand the expansion of the Indo-European languages and to understand prehistory, or just trying to understand the relatedness of present day and ancient people without being too committed to any specific population labels. If they're trying to use it to define "genetic Indo-Europeans" (as opposed to genetic Yamnaya cluster or genetic Steppe_EMBA cluster and so on), then yeah, that might be a waste of time.

Unknown said...

@Davidski,Hi! I'm YoonHee kim from Korea.​
I saw your post about Ancestry! I have lots of question about that and wonder about database you are used.
I have many attention about analysis ancestry. In your ancestry, your report presents many populations.
so I wonder about what kind of database you are used.
​Can I give me some information about database your used ?
​I was finding many kind of database about ancestry, but I didn't find exact data.
​If you can give me some information, please give me the answer.
​thanks
yh

Davidski said...

@김윤희

Two main ones I use are these...

Human Origins (and you can sign a waiver to get extra samples)

http://polishgenes.blogspot.com/2016/06/poles-in-new-human-origins-dataset.html

Estonian Biocentre Human Genome Diversity Panel

http://eurogenes.blogspot.com/2016/09/estonian-biocentre-human-genome.html

Plus you can find many other datasets at these two links...

http://evolbio.ut.ee

https://www.ncbi.nlm.nih.gov/gds