My proposed tree of Indo-European languages

[Leandros]

Its funny how your proposed language tree has many languages that we dont have any archeological findings for.
You think Greek was born by some fictional eastern or balkanic branch? Where are the findings?


Y Haplogroup: I2(I-S17250)
mtDNA: U5b2a

 

[Maciamo]

It doesn't quite explain why Tocharian isn't within a centum branch, unless you assume IE started out as centum, with satem emerging later.

Same as for Albanian and Armenian. Proto-Tocharian was originally an R1b offshoot from Yamna to the Altai (5300 years ago). But R1a tribes moved en masse to the Central Asian Steppe and Altai (Sintashta/Andronovo, Scythians), mixing with the descendants of Afanasievo and satemising those who managed to retain their language (apparently only in the isolated Tarim basin, where both R1b and R1a are now found among the Uyghurs).

 

[Maciamo]

I agree with your main point, but disagree on the particular example you used. The bulk of the Norman-influenced (but not necessarily triggered) changes happened during the transition from Old English to Middle English, ranging from ~1000 to ~1350 AD. By 1400 the really significant Anglo-Norman influence had virtually ceased, and it seems that Norman French ceased to be the dominant language of the courts and elite by the late 1300s and by the late 1400s wasn't even spoken in significant numbers any longer, so I think it's unlikely that the main changes that happened from Middle English to Modern English involved significant Anglo-Norman influence, especially because the most transformative changes in phonology and grammar seem to have consolidated during the 16th century, well after French ceased to be a formal court/state language. On the other hand, you're absolutely right about the very fascinating development of the Japanese language.

I think that we mean the same thing but express it differently. I read three books on the history of English and they all said that it was the fusion of Middle English with the Norman French that formed Modern English. Wikipedia says that Middle English existed between 1150 and 1500. Originally Middle English was a purely Germanic language similar to modern Dutch. French (and Latin) words permeated progressively Middle English throughout the Late Middle Ages, so that by the 16th century Modern English appears as a Romance-Germanic hybrid. It's hard to tell how quickly French words were adopted between 1150 and 1500 because it depends on the social class and whether people were more urban or rural. Rural peasants kept speaking an nearly unaltered Germanic version of Middle English quite late in comparison to the ruling class.

As for how Goidelic could've come to be without any massive migration into Ireland, I'm not completely sure, considering the very new and still unpublished Reich study about Iron Age Britain, that we will really find no fine-scale change at all between Bell Beaker and Anglo-Norman/Viking Ireland, though it may definitely have involved only an influx of a very similar (genetically) people.

They will find some migration to Ireland in the Iron Age. For example, historical sources mention that some Menapians from Belgium settled in Ireland. But that was in Late La Tène and it was only a minor migration (accounting for part of the R1b-U152 in Ireland).

But even if that did not happen, as I said in my 1st answer, I think a language shift with a significant degree of hybridization (kind of a Celtic "Portuñol") is much more feasible when you have a local people who speak a language still very closely related to other language that has more cultural and economic prestige and is spoken in the neighboring region, making that shift very easy and actually almost imperceptible because the foreign influences get into the local language slyly almost as if nobody could notice, e.g. see how Galician has become clearly less and less Portuguese-like and more and more Castillianized in the last 500 years, gradually over the generations (that neighbnoring powerful language could maybe have been an early Urnfield/Hallstatt Celtic language in Britain, maybe also initially spoken in the Atlantic shore of the continent).

Another possibility, maybe even more plausible in fact, is that Goidelic is simply the Bell Beaker Para-Celtic that became gradually and heavily "Celticized" through several generations due to long-term influence from a language so similar that a re-convergence (maybe combined with widespread bilingualism) became almost inevitable, like that "Castillianized Galician" you may often hear in bilingual urban and cosmopolitan areas of modern Galicia.

But of course here I am speculating to try to fit this supposed early divergence of Goidelic with the linguistic comparisons and glottochronological evidences that do not support a very ancient split between P-Celtic and Q-Celtic.

A combination of these two hypotheses could very well explain how Goidelic remained closer to P-Celtic. Trade with Britain and Brittany may have gradually brought a re-convergence of Goidelic toward P-Celtic.

I suppose that when you talk about Goidelic you are referring to medieval to modern Goidelic, as there was no written source in the Iron Age, let alone the Bronze Age. If that's the case, then the influence of Lat Tène tribes (like the Menapians) would already have altered Goidelic, even without taking into account the gradual re-convergence that would have happened through trade with other Celtic parts of Europe during the Late Bronze (e.g. Atlantic Bronze Age) and Iron Age.

 

[Zanatis]

The situation with Albanian is the same. The Balto-Slavic influence most certainly dates from the Gothic and Slavic migrations. After all, modern Albanians have about 21% of Slavic Y-DNA, 5% of Germanic Y-DNA and that is without counting the 27% of E-V13, which is at least partially Gothic or Slavic. That doesn't mean that Albanian language itself descends from the same R1a-Z280 branch as Balto-Slavic. It would be like saying that French evolved from a branch in between Italo-Celtic and Germanic languages because it has influences from both! No, obviously French is Italic, and the Germanic influence is merely loanwords from the Franks who came much later.

When looking at modern languages like Albanian without knowing much about their ancient ancestors (like Illyrian), all you see is the cooked dish, but it's not always easy to guess the ingredients, and even less in which order those ingredients were added.

Yeah it would seem so if R1a and I2a were evenly spread, but most of it comes from a region with heavy Vlach and minor Bulgarian presence, while other regions have 5% or less. I'm not saying there's no Slavic influence though as there is indeed a lot. What needs further investigation is many words common between Albanian and South Slavic but absent from West and East Slavic.

As for the Germanic haplogroup, it seems that they're mostly Norman so I don't believe they had much of an impact on the language.

On the other hand, we also need higher resolution results for R1a to be able to call it all Slavic. It might very well be all Slavic, but the recent results of our member Dibran here made rethink about it.

 

[Maciamo]

Yeah it would seem so if R1a and I2a were evenly spread, but most of it comes from a region with heavy Vlach and minor Bulgarian presence, while other regions have 5% or less. I'm not saying there's no Slavic influence though as there is indeed a lot. What needs further investigation is many words common between Albanian and South Slavic but absent from West and East Slavic

Albanians have mixed South Slavic ancestry, of course, as it is the South Slavs that spread over all the Balkans. What's your point?

As for the Germanic haplogroup, it seems that they're mostly Norman so I don't believe they had much of an impact on the language.

Are you saying that the Germanic haplogroups among Albanians are of Norman origin? AFAIK Albanian I1 is mostly Z63, which is typical of places settled by the Goths. There is also some I1-P109 as elsewhere in the Balkans, and although this branch is more widely Scandinavian and was found among the Vikings and Normans, it could also be Gothic or Varangian, not necessarily Norman.

On the other hand, we also need higher resolution results for R1a to be able to call it all Slavic. It might very well be all Slavic, but the recent results of our member Dibran here made rethink about it.

Don't forget that the Goths brought different clades of R1a from assimilated Slavic tribes in Poland and Western Ukraine than the ones that came a few centuries later with the South Slavs.

 

[Ownstyler]

That's possible, but how do you account for the Celtic Y-DNA in Greece then? Or is it all Roman? (possible as the two are closely linked)

Honestly I had not paid that much attention to the explanation of Celtic DNA in Greece because I just assumed its mainly from the Celtic invasions of the 3rd century BC. I just read the part about R1b-S28 more carefully and I agree on the Italo-Gallic split and the Hallstatt culture, but I still think the Celtic DNA in Greece is mostly from these later migrations.

If you compare the map of R1b-S28 to the maps of the Celtic invasions of the 3-6 centuries BC, first you can see that the Celtic tribes that migrated were concentrated in the south of France (https://en.wikipedia.org/wiki/Galatia#/media/File:Volcae_Tectosages.jpg), where this haplogroup is widespread, and second, a striking similarity especially in eastern Europe (https://en.wikipedia.org/wiki/Galatia#/media/File:Celtic_Expansion.svg). Note the lack of R1b-S28 in the central-northern Balkans and especially the perfect match of R1b-S28 in Turkey with old Galatia.

 

[Maciamo]

Honestly I had not paid that much attention to the explanation of Celtic DNA in Greece because I just assumed its mainly from the Celtic invasions of the 3rd century BC. I just read the part about R1b-S28 more carefully and I agree on the Italo-Gallic split and the Hallstatt culture, but I still think the Celtic DNA in Greece is mostly from these later migrations.

If you compare the map of R1b-S28 to the maps of the Celtic invasions of the 3-6 centuries BC, first you can see that the Celtic tribes that migrated were concentrated in the south of France (https://en.wikipedia.org/wiki/Galatia#/media/File:Volcae_Tectosages.jpg), where this haplogroup is widespread, and second, a striking similarity especially in eastern Europe (https://en.wikipedia.org/wiki/Galatia#/media/File:Celtic_Expansion.svg). Note the lack of R1b-S28 in the central-northern Balkans and especially the perfect match of R1b-S28 in Turkey with old Galatia.

There is annoying little data for deep clades of any haplogroup from Greece. But the FTDNA project shows that the R1b-P312 that could be Celtic are concentrated in central Greece between Athens and Corfu. In fact, almost all the samples from Central Greece are R1b-P312, including one L2 in Athens and another one in Chalkida. There is also a DF27 in Epirus. This corresponds more or less to the region that the Celts invaded (Thermopylae, Delphi). But since the Celts were defeated and ousted from Greece, it is hard to believe that the majority of paternal lineages in that region today descend from that ill-fated Celtic invasion that only lasted a few weeks or months.

The Dorians are said to have come from Epirus, Macedonia and northern Greece, so it's equally possible that many settled in central Greece on their journey south. There is a lot of R1b-M269 in the southern Peloponnese and a bit in Crete, including an R1b-L2 member on this forum who isn't on the FTDNA Project. Unfortunately there is very little info on the R1b clades in the Peloponnese. That could have settled the matter.

 

[Expredel]

There is annoying little data for deep clades of any haplogroup from Greece. But the FTDNA project shows that the R1b-P312 that could be Celtic are concentrated in central Greece between Athens and Corfu. In fact, almost all the samples from Central Greece are R1b-P312, including one L2 in Athens and another one in Chalkida. There is also a DF27 in Epirus. This corresponds more or less to the region that the Celts invaded (Thermopylae, Delphi). But since the Celts were defeated and ousted from Greece, it is hard to believe that the majority of paternal lineages in that region today descend from that ill-fated Celtic invasion that only lasted a few weeks or months.

https://www.eupedia.com/forum/threads/35967-R1b-subclade-table-based-on-59K-FtDNA-samples

I've got some information on greece in that topic, and I can delve deeper if you're interested in certain subclades. Greece appears to have the highest percentage of L23 in Europe, though it's unclear when it arrived.

 

[Ownstyler]

There is annoying little data for deep clades of any haplogroup from Greece. But the FTDNA project shows that the R1b-P312 that could be Celtic are concentrated in central Greece between Athens and Corfu. In fact, almost all the samples from Central Greece are R1b-P312, including one L2 in Athens and another one in Chalkida. There is also a DF27 in Epirus. This corresponds more or less to the region that the Celts invaded (Thermopylae, Delphi). But since the Celts were defeated and ousted from Greece, it is hard to believe that the majority of paternal lineages in that region today descend from that ill-fated Celtic invasion that only lasted a few weeks or months.

The Dorians are said to have come from Epirus, Macedonia and northern Greece, so it's equally possible that many settled in central Greece on their journey south. There is a lot of R1b-M269 in the southern Peloponnese and a bit in Crete, including an R1b-L2 member on this forum who isn't on the FTDNA Project. Unfortunately there is very little info on the R1b clades in the Peloponnese. That could have settled the matter.

I might be wrong but I thought there was only a fraction of Y-DNA in that part of Greece that belongs to R1b-S28. If it's actually a majority then maybe a more substantial migration was responsible. The invasions lasted for a few years though. And even if they were defeated at Delphi, the Celts won at Thermopylae so they probably ravaged the countryside.

 

[APO276]

Some R1b subclades in Greece can be from people who moved here after 1204.
(4th crusade related)

 

[Dibran]

As for the Germanic haplogroup, it seems that they're mostly Norman so I don't believe they had much of an impact on the language.

On the other hand, we also need higher resolution results for R1a to be able to call it all Slavic. It might very well be all Slavic, but the recent results of our member Dibran here made rethink about it.

Don't forget that the Goths brought different clades of R1a from assimilated Slavic tribes in Poland and Western Ukraine than the ones that came a few centuries later with the South Slavs.

Probably the case with me. My case even has a Albanian founder effect, so far TMRCA of 1200ypb(with me and my match not the clade). I am even testing the rest of the novels as we speak to refine it. My relation with all other L1029 seems to be between 300BC-0AD. My cluster is basal belonging to L1029-B-Western which has the highest occurrence in Germany and Poland, including highest diversity in these areas. Idk how well tested they are, but even Southern Sweden which has practically no L1029, has more diversity than elsewhere. They also seem to belong to B-Western. L1029 experienced a big depopulation. As I understand, it formed around 1400BC. Though the oldest TMRCA seems to be between 2000-2300ypb for all modern carriers.

 

[Zanatis]

Albanians have mixed South Slavic ancestry, of course, as it is the South Slavs that spread over all the Balkans. What's your point?

I meant a region within Albania where R1a and I2a peaks and that region is South-Eastern Albania. The other regions have below 5% of these haplogroups.

Are you saying that the Germanic haplogroups among Albanians are of Norman origin? AFAIK Albanian I1 is mostly Z63, which is typical of places settled by the Goths. There is also some I1-P109 as elsewhere in the Balkans, and although this branch is more widely Scandinavian and was found among the Vikings and Normans, it could also be Gothic or Varangian, not necessarily Norman.

You're right and you would know better, it's the recently tested Albanians that are coming up as possibly Norman according to the Albanian project that confused me. Nevermind.

Don't forget that the Goths brought different clades of R1a from assimilated Slavic tribes in Poland and Western Ukraine than the ones that came a few centuries later with the South Slavs.

I don't know about Poland having been already Slavic at that time but the fact that Goths could have brought Slavic R1a is highly probable and completely logical, and similarly the distribution of I2a could have been caused or brought by Goths too and to me this makes more sense since I'm linking them to a Latin speaking Balkan population.

Not to mention that the distribution of I2a doesn't fit the Slavic migration at all with ratios of I2a:R1a in Bosnian Croats 7:1, Serbs 2:1, Montenegro 4:1, Croatia 1,5:1, Macedonia 2:1, then we have Kosovo 1:2, Albania 1:1, Bulgaria 1:1, Greece 1:1, Aegean Greeks 1:3 and very interestingly Hungary 1:2 despite being adjacent to Serbia and Bosnia.

In my point of view R1a is spread more or less evenly and its distribution is explainable, where Northern Greece has more R1a than Bosnia, Bulgaria, Macedonia, and twice more than Montenegro, and that still makes sense, but you cannot say the same for I2a where the chances of a founder effect specifically in Western Balkans and specifically in mountainous settlements is next to impossible.

 

[Ownstyler]

I meant a region within Albania where R1a and I2a peaks and that region is South-Eastern Albania. The other regions have below 5% of these haplogroups.


You're right and you would know better, it's the recently tested Albanians that are coming up as possibly Norman according to the Albanian project that confused me. Nevermind.


I don't know about Poland having been already Slavic at that time but the fact that Goths could have brought Slavic R1a is highly probable and completely logical, and similarly the distribution of I2a could have been caused or brought by Goths too and to me this makes more sense since I'm linking them to a Latin speaking Balkan population.

Not to mention that the distribution of I2a doesn't fit the Slavic migration at all with ratios of I2a:R1a in Bosnian Croats 7:1, Serbs 2:1, Montenegro 4:1, Croatia 1,5:1, Macedonia 2:1, then we have Kosovo 1:2, Albania 1:1, Bulgaria 1:1, Greece 1:1, Aegean Greeks 1:3 and very interestingly Hungary 1:2 despite being adjacent to Serbia and Bosnia.

In my point of view R1a is spread more or less evenly and its distribution is explainable, where Northern Greece has more R1a than Bosnia, Bulgaria, Macedonia, and twice more than Montenegro, and that still makes sense, but you cannot say the same for I2a where the chances of a founder effect specifically in Western Balkans and specifically in mountainous settlements is next to impossible.

I thought you meant that I2a moved into the western Balkans before R1a. Now it seems you're saying it's been there all along. Are you? Because if so, you have to take into account it's higher diversity further north and east.

 

[Ygorcs]

It doesn't quite explain why Tocharian isn't within a centum branch, unless you assume IE started out as centum, with satem emerging later.

This is exactly what mainstream IE linguistics consider most likely: IE started out as "centum" or, more specifically, with a centum-like distinction in its stop consonants, and later some dialects began to spread the satem phonetic development. Few linguists these days consider that centum languages are one coherent group of proto-languages necessarily more related between themselves than to satem proto-languages. What's usually proposed these days is that centum vs. satem was just one of the many possible outcomes derived from the original distinction between plain vs. palatalized or possibly plain vs. ejective stop consonants in PIE. Some of them went one way, some went the other way, and we can even speculate that those changes involved areal features and mutual influences and so on, but they do not necessarily reflect a common origin, a sort of Centum PIE vs. Satem PIE. In the case of satem in particular, palatalization and eventual fricativization are actually two of the most usual, commonplace evolutions that can happen in languages. They happened in all sorts of distinct language families, not just IE, and similar developments happened even in modern languages like Old French and modern Brazilian Portuguese.

 

[Dibran]

I meant a region within Albania where R1a and I2a peaks and that region is South-Eastern Albania. The other regions have below 5% of these haplogroups.


You're right and you would know better, it's the recently tested Albanians that are coming up as possibly Norman according to the Albanian project that confused me. Nevermind.


I don't know about Poland having been already Slavic at that time but the fact that Goths could have brought Slavic R1a is highly probable and completely logical, and similarly the distribution of I2a could have been caused or brought by Goths too and to me this makes more sense since I'm linking them to a Latin speaking Balkan population.

Not to mention that the distribution of I2a doesn't fit the Slavic migration at all with ratios of I2a:R1a in Bosnian Croats 7:1, Serbs 2:1, Montenegro 4:1, Croatia 1,5:1, Macedonia 2:1, then we have Kosovo 1:2, Albania 1:1, Bulgaria 1:1, Greece 1:1, Aegean Greeks 1:3 and very interestingly Hungary 1:2 despite being adjacent to Serbia and Bosnia.

In my point of view R1a is spread more or less evenly and its distribution is explainable, where Northern Greece has more R1a than Bosnia, Bulgaria, Macedonia, and twice more than Montenegro, and that still makes sense, but you cannot say the same for I2a where the chances of a founder effect specifically in Western Balkans and specifically in mountainous settlements is next to impossible.

There only seems to be one cluster of I2a1b-Din at present that predated the majority of South Slavic I2a1b-Din. In the case of this cluster, a Goth or Bastarnae could explain it. However this cluster so far is rare, and only present in Greeks and Jews with TMRCA of 2200ypb. Most I2a1b-Din in the Balkans is very young. If I am not mistaken they belong to clades 1500ypb. Majority of Serbian and Bosnian I2a1b-Din seems to stem from PH908. Specifically further downstream clades with a TMRCA of 900-1100ypb. Very young. Based on the current data, I am willing to bet this theoretical "Greek" Cluster of I2-Din is the one that arrived before Slavs. For all we know, Albanian I2-Din could belong to a similar cluster, as I read somewhere Jews had a significant presence in the south at one time.

 

[Ygorcs]

A combination of these two hypotheses could very well explain how Goidelic remained closer to P-Celtic. Trade with Britain and Brittany may have gradually brought a re-convergence of Goidelic toward P-Celtic.

I suppose that when you talk about Goidelic you are referring to medieval to modern Goidelic, as there was no written source in the Iron Age, let alone the Bronze Age. If that's the case, then the influence of Lat Tène tribes (like the Menapians) would already have altered Goidelic, even without taking into account the gradual re-convergence that would have happened through trade with other Celtic parts of Europe during the Late Bronze (e.g. Atlantic Bronze Age) and Iron Age.

Yes, that makes sense. As for the dating of the Goidelic language I'm referring to, I'm mostly referring to the most well attested period of it, the medieval Old Irish, but also, partly, to the earlier and apparently at least partially "fossilized" Primitive Irish, which is significantly distinct from the later Old Irish despite only a few centuries separating them, which is to some linguists suggestive of Primitive Irish being something of a formal, erudite version of the Goidelic language, probably preserving more archaic forms (more or less like Classical Latin vs. Vulgar Latin) as opposed to the eventual acceptance of more vernacular, "contemporary" forms of the language in the Old Irish period. Primitive Irish looks much closer to the continental (mostly P-Celtic) Celtic languages and in some ways still extremely similar to languages like Gaulish, which is why some linguists can't grasp the possibility of a very early divergence of Goidelic from the rest of the Celtic languages. I think the truth must lay somewhere in between your hypothesis and mine, probably some kind of language re-convergence or hybridization originally involving two similar languages which still had partial mutual intelligibility, effectively creating a seemingly "new" Celtic language out of a Para-Celtic language stratum.

 

[Maciamo]

Yes, that makes sense. As for the dating of the Goidelic language I'm referring to, I'm mostly referring to the most well attested period of it, the medieval Old Irish, but also, partly, to the earlier and apparently at least partially "fossilized" Primitive Irish, which is significantly distinct from the later Old Irish despite only a few centuries separating them, which is to some linguists suggestive of Primitive Irish being something of a formal, erudite version of the Goidelic language, probably preserving more archaic forms (more or less like Classical Latin vs. Vulgar Latin) as opposed to the eventual acceptance of more vernacular, "contemporary" forms of the language in the Old Irish period. Primitive Irish looks much closer to the continental (mostly P-Celtic) Celtic languages and in some ways still extremely similar to languages like Gaulish, which is why some linguists can't grasp the possibility of a very early divergence of Goidelic from the rest of the Celtic languages. I think the truth must lay somewhere in between your hypothesis and mine, probably some kind of language re-convergence or hybridization originally involving two similar languages which still had partial mutual intelligibility, effectively creating a seemingly "new" Celtic language out of a Para-Celtic language stratum.

All Insular Celts (R1b-L21) spoke Q-Celtic languages originally, and P-Celtic probably started replacing Q-Celtic in Britain when the Hallstatt migrants arrived c. 500 BCE, followed by Belgic tribes in the first century BCE. As you explained it yourself, Q-Celtic speakers could easily have adopted the new Gaulish dialect and turned it into Brythonic if the continental Celts enjoyed a higher cultural prestige. The same thing happened when Latin replaced (quite easily) Celtic languages in northern Italy, Gaul, Iberia and Britain owing to its higher prestige. This can happen quickly when languages are relatively close to one another (same family). That is also how Arabic came to replace other Semitic languages in the southern Middle East and North Africa (e.g. Coptic Egyptian and Berber).

 

[Maciamo]

https://www.eupedia.com/forum/threads/35967-R1b-subclade-table-based-on-59K-FtDNA-samples

I've got some information on greece in that topic, and I can delve deeper if you're interested in certain subclades. Greece appears to have the highest percentage of L23 in Europe, though it's unclear when it arrived.

R1b-L23/Z2103 surely came with the Mycenaean Greeks, but could also have come from Trojans, Luwians, Lydians and Lycians in western Anatolia, especially when these became Hellenicised and the populations merged during the Roman, Byzantine and Ottoman empires.

According to your table, about 65% of the R1b in Greece (among those who tested for subclades) are L23/Z2103 and the rest if mostly Celtic P312 (25%) with a bit of Germanic U106 (5%) and Near Eastern V88 (5%). In absolute frequencies this translates as 10% of R1b-L23/Z2103 and 4% of R1b-P312 in average for the whole of Greece, but a few percents more in South Greece (Peloponnese) where R1b is the highest.

 

[Maciamo]

I might be wrong but I thought there was only a fraction of Y-DNA in that part of Greece that belongs to R1b-S28. If it's actually a majority then maybe a more substantial migration was responsible. The invasions lasted for a few years though. And even if they were defeated at Delphi, the Celts won at Thermopylae so they probably ravaged the countryside.

Central Greece has the lowest percentage of R1b (10%) of any Greek region, but based on the few FTDNA samples (probably not representative) almost all of it is Celtic.

 

[Zanatis]

I thought you meant that I2a moved into the western Balkans before R1a. Now it seems you're saying it's been there all along. Are you? Because if so, you have to take into account it's higher diversity further north and east.

Well I'm not really claiming anything but simply checking with more knowledgeable members if this theory could be considered too because almost everyone has decided to call it Slavic, brought by the Slavs, spread by the Slavs, exclusively Slavic. But that in itself it's still just a theory and IMO not a very complete one.

It would make it more difficult if we use modern regional borders (like the Balkans) but I'd say centered around the Carpathians and radiating in all the nearby regions only, since it's a young clade. So it makes no difference if it was first in modern Ukraine or Romania/Moldova since those regions weren't inhabited by Slavs in antiquity.

That said, looking at its distribution it makes more sense that it was first the earlier barbarians that pushed I2a further South and West rather than having been absorbed first by R1a migrations from further North and spread around in the same time as R1a.