Tracing the Spread of Celtic Languages using Ancient Genomics

[Riverman]

Thanks again to Kaltmeister for recommending this presentation by Detlef Jantzen. Very nice, informative presentation even there was nothing spectacular new. Jantzen still favors the theory that in the Tollense valley a trade caravan was ambushed by a group of locals (maybe members of the Mecklenburg group), explaining the quite diverse Sr-isotope, archaeological and DNA finds.
Interesting is that they really plan to analyze the DNA from all around 90 skulls together with 14C measurements to be sure that this was a single event.

What I don't get about this caravan idea is, that the post-1960 archaeologist are notorious for underestimating the size of the armies in prehistoric and historical battles. Like "large armies didn't exist" in this or that time and couldn't get provisioned etc. Then we find the first evidence for a larger battle in the Bronze Age, in the rather sparsely settled North, and suddenly, one of the biggest documented battles from that area, in that time frame, being presented as a robber attack on "a caravan"?
This doesn't make sense at all.
Armies are supposed to be smaller than the estimate for the total involved manpower based on the small portion of excavated remains, and that's most likely just a tiny fraction of the forces involved, with many remains being not excavated yet or being lost, with hundreds if not thousands more surviving or getting caught etc.

If that was no large battle, but just "a caravan", what were the large battles like? And of course we know that there were clashes of these cultural groups and Urnfield alliances tried to expand all the time. The observable diversity is simply the result of the Urnfield networks using allies, auxiliaries and highly likely even mercenaries (professional warriors with a specific ethical code being a signature feature of the Urnfield world).

But alternative theories might be more popular in some circles or sound more "sophisticated".

Reminds me of the recent presentation of the Bronze Age bottleneck of yDNA in the FTDNA-group on Facebook. A known fact for long, but confirmed with way more data than in the past. However, the explanation of "stratified societies" and "rich men" doesn't add up. Because quite obviously, the male lineages which expanded big time were moving with people, with clans, tribes, alliances based on ethnic, ideological and religious traits in common. Not by "rich men" buying women or gaining some sort of social prominence. It was cousins of various grades in warbands which expanded with their patrilinear kin and replaced those which lost against them.

Because the patrilinear clans were exogamic and assimilated foreign females in different ways, the mtDNA didn't suffer from the same kind of diversity reduction. Very fascinating is in any case also how widespread that pattern was in the Old World - going much beyond the sphere of direct Indoeuropean influence.

This topic is directly related to what we debated before, because many of the lineages which experienced a bottleneck or went effectively or nearly extinct didn't so immediately. Take e.g. given the branches of the Western Carpathian Tell cultures which got tested so far: Many of these did survive the initial "steppe shock" and the transition to the Bronze Age, they just didn't make it to the modern period.
A lot of the diversity still present in the Early Bronze Age was further reduced in subsequent events. It was not a single big event in just one generation or even a couple of generations which caused it all and everywhere. The dynamic went on for much longer, well into the formation of proto-states and states at least.

[La Tene]

This article was once discussed at length on Anthrogenica, but I'm not sure everyone is familiar with it. The book it refers to ("Ancient Celtic Placenames in Europe and Asia Minor," Blackwell ed.) is, I hope, still available.

This article discusses a problem in integrating archaeology and philology. For most of the twentieth century, archaeologists associated the spread of the Celtic languages with the supposed westward spread of the ‘eastern Hallstatt culture’ in the first millennium bc. More recently, some have discarded ‘Celtic from the East’ in favour of ‘Celtic from the West’, according to which Celtic was a much older lingua franca which evolved from a hypothetical Neolithic Proto-Indo-European language in the Atlantic zone and then spread eastwards in the third millennium bc. This article (1) criticizes the assumptions and misinterpretations of classical texts and onomastics that led to ‘Celtic from the East’ in the first place; (2) notes the unreliability of the linguistic evidence for ‘Celtic from the West’, namely (i) ‘glottochronology’ (which assumes that languages change at a steady rate), (ii) misunderstood place-name distribution maps and (iii) the undeciphered inscriptions in southwest Iberia; and (3) proposes that Celtic radiating from France during the first millennium bc would be a more economical explanation of the known facts.

https://www.cambridge.org/core/journals/...2484F8E8C3

The concept of Pre-Germanic, as discussed in this text, specifically chapter 4,  Substrate alternations in Celtic by  Paulus S. van Sluis, refers to a pre non Germanic language (as in a non Germanic substrat language that existed in places where Germanic would later be spoken), or to an earlier phase of Germanic preceding proto Germanic but post dating Paleo Germanic? 

Sub-Indo-European Europe

https://www.degruyter.com/document/doi/1...ml?lang=en

5.2 Pre-Germanic (page 119)

Kuiper’s second oldest layer, A2, is somewhat less widespread. This layer is characterized by initial clusters kn-, kl-, a/i/u-vocalism in derivational suffixes, prenasalization, geminate-simplex stop alternation and voicing alternation.51 Boutkan (1998) furthermore adduces disyllabic stems of the shape C®VCVC- and various semantically neutral extensions in labial, dental and velar stops and ✶ l as a feature of this substrate layer, and perhaps this phenomenon may be equated with dental-velar-zero alternation (Section 4.3). Some words in the Celtic substrate lexicon contain one or more of these features. Table 8 contains words with initial velar plus resonant (✶ KR-), vowel variation in non-initial syllables (✶ V~V), and words with disyllabic stems of the shape C®VCVC-. Substrate alternations otherwise found in these words are the n- and st®-suffixes for flora, liquid alternation and various types of vowel alternation, but the a-prefix is not found. All of the words given in Table 8 except PUBIC HAIR have comparanda in Germanic. Most words are otherwise confined to Balto-Slavic and/or Italic, but GARLIC, HOLLY and SILVER have wider distributions. One may conclude that this pre-Germanic layer primarily influenced Germanic, and its influence on Celtic and other languages neighbouring Germanic was less intense. The Celtic substrate lexicon confirms Kuiper’s and Boutkan’s findings that velar-resonant clusters and disyllabic stems belong to a Germanic substrate; if they had collected words sharing these shapes through mere chance resemblance, then Celtic substratum words with the same shape would not be expected to have Germanic comparanda to the same extent. Among the words given in Table 8, CUP/HEAD was borrowed into Celtic at an early date, as is shown by its PIE treatment of ✶ p. Other words appear borrowed into Celtic or other IE branches at a later date, however. GARLIC and PUBIC HAIR both require multiple reconstructions within Celtic, while SILVER requires multiple reconstructions within Balto-Slavic. The geminate-simplex stop alternation found...

'...in FREQUENT/MANY also suggests borrowing at a later date. All in all, both strata identified in Section 4.13 appear represented in the Celtic substrate lexicon of preGermanic origin, but later borrowings predominate. Pre-Germanic likely also accounts for the one Finnic comparandum in the Celtic substrate lexicon, i.e. OATS. It is not likely that pre-Germanic was itself Finnic (pace Hyllested 2010: 123–124): its characteristic ✶ kn/l- clusters are incompatible with Finnic phonotactics, where such initial clusters are not allowed. Hyllested (2016) argues that there was a Finnic substrate in Celtic not shared with Germanic. However, the one word also found in Finnic also has comparanda in Germanic, so the existence of a Finnic substratum in Celtic but not Germanic cannot be confirmed. Pre-Germanic was most likely spoken in prehistoric northern Europe. This location is suggested by the fact that comparanda are found in northern European languages, plus Italic. The presence of this otherwise northern European substrate in the Italic lexicon suggests that the IE language that would later develop into Italic had some presence north of the Alps, or alternatively that this substrate language reached south of the Alps.

Page 124, notes

The large amount of Italo-Celto-Germanic substrate words with pre-Germanic features suggests that an undifferentiated Italo-Celtic was in contact with pre-Germanic, while the pre-Germanic words shared with Germanic alone suggests that this contact between Celtic and pre-Germanic endured after this split

Page 125

These findings can feed into the debate on where the Celtic homeland was located. The two competing hypotheses are the traditional hypothesis, which places Proto-Celtic in the eastern Alps of the Late Bronze Age, and the Celtic from the West hypothesis which sees Proto-Celtic as the result of dialect levelling among Indo-European speakers facing the Atlantic over the course of the Bronze Age (Sims-Williams 2020, with references). As discussed under its lemma, borrowing of HOLLY is estimated to date to around Proto-Celtic and the absence of its referent from eastern Europe suggests that Proto-Celtic developed no further east than the Alps. Influence from the pre-Germanic substrate appears strongest before Proto-Celtic and appears located in northern Europe. These insights are difficult to reconcile with a Celtic homeland in the eastern Alps. An early Celtic homeland along the Atlantic, however, would suggest early contact with Vasconic, but what is found instead is that such mutual borrowings postdate Proto-Celtic. A third option offered by Sims-Williams (2020), with the Proto-Celtic homeland in second-millennium BCE Gaul, is the most compatible with the Celtic substrate vocabulary.

Page 177

In contrast to these linguistically unsubstantiated ideas, it is evident that Proto-Celtic, or the earliest stages of Celtic, had considerable linguistic contact with pre- or Proto-Germanic (cf. Van Sluis et al. 2023; Koch 2020). This supports traditional hypotheses about the geographic position of Proto-Celtic. Taking into account the staging area for Germanic probably in northern Central Europe and the comparatively shallow time-depth of Proto-Germanic, the contact zone for the two branches is best assumed to be first-millennium B.C. Continental Central Europe, not the Atlantic seaboard

[La Tene]

...

The author of the Proto-Italic paper seems to flirt with the idea of migrants from the lower Tisza area making there way to Italy. He references a so called "Tisza Site Group", which upon further research seems to be non other than the Belegis Group. What do you make of this? Belegis migrants heading west? Is the author no aware of the Khisinau Korlateni culture?

Integrating and Dividing in a Late Bronze Age Society: Internal Organization of Settlements of the Tisza Site Group in the Southern Carpathian Basin, 1600–1200 b.c.

https://www.tandfonline.com/doi/full/10....161#d1e509

Ive read posts by others on Eupedia that link the Nagyrev and Vatya cultures to proto Italic speakers and Terramare as well, but such a suggestion seems very odd to me. What do you think? Do the descendants of the Vatya culture represent the "Pre Gava style" in the middle Tisza/Lower Mures?

[Riverman]

...

The author of the Proto-Italic paper seems to flirt with the idea of migrants from the lower Tisza area making there way to Italy. He references a so called "Tisza Site Group", which upon further research seems to be non other than the Belegis Group. What do you make of this? Belegis migrants heading west? Is the author no aware of the Khisinau Korlateni culture?

Integrating and Dividing in a Late Bronze Age Society: Internal Organization of Settlements of the Tisza Site Group in the Southern Carpathian Basin, 1600–1200 b.c.

https://www.tandfonline.com/doi/full/10....161#d1e509

Ive read posts by others on Eupedia that link the Nagyrev and Vatya cultures to proto Italic speakers and Terramare as well, but such a suggestion seems very odd to me. What do you think? Do the descendants of the Vatya culture represent the "Pre Gava style" in the middle Tisza/Lower Mures?

Belegiš I and Belegiš II-Gava did expand in all directions at its height, including to the West, including Italy.

Especially items and customs from Belegiš II-Gava teached Italy. However, their impact was much bigger on specific sites, especially territories later associated with Ligurians and Venetic people in my opinion.

I see no connection to the Italics and even in Ligurians and Venetic people more of an influence.

The Pre-Gáva horizon is a complex matter with many influences, but central are Suciu de Sus culture and related people (Igrita etc.) which received influences from the Tumulus culture people directly and indirectly via Piliny.

I think that Suciu de Sus was E-V13 dominated ans Proto-Thracian to North Thracian.

Vatya and Nagyrev appear to be dead ends rather, even though Nagyrev had a stronger Bell Beaker related element in its mix.

Italic is more likely to have moved more directly through the Alps from a TC source IMHO.

[Hodo Scariti]

Everything is very interesting... I'm still wondering about the role of other U152 subclades other than L2 in Celtic languages expansion... It seems that Z56 and Z36 are in a very tiny percentage among the total... nearly inexistent...

[Riverman]

R-Z56 seems to have a pretty strong bias towards Italy I'd say, especially if going by the ancient DNA record. R-Z36 looks like having an overall more Northern and Eastern run, with many branches in Eastern Central Europe. That's something I would expect to find in Celts in Central Germany, Eastern Germany, Czechia and Poland.
Both look to me like being secondarily involved, compared to R-L2, but that's just something I would say at first glance and it might be proven wrong eventually.

[Hodo Scariti]

R-Z56 seems to have a pretty strong bias towards Italy I'd say, especially if going by the ancient DNA record. R-Z36 looks like having an overall more Northern and Eastern run, with many branches in Eastern Central Europe. That's something I would expect to find in Celts in Central Germany, Eastern Germany, Czechia and Poland.
Both look to me like being secondarily involved, compared to R-L2, but that's just something I would say at first glance and it might be proven wrong eventually.

Many thanks! I thought the same: Z56 more "italo-centric", and Z36 more eastern.(even if Z36>Z37 seems, today, very "italian").

[Mitchell-Atkins]

R-Z56 seems to have a pretty strong bias towards Italy I'd say, especially if going by the ancient DNA record. R-Z36 looks like having an overall more Northern and Eastern run, with many branches in Eastern Central Europe. That's something I would expect to find in Celts in Central Germany, Eastern Germany, Czechia and Poland.
Both look to me like being secondarily involved, compared to R-L2, but that's just something I would say at first glance and it might be proven wrong eventually.

Many thanks! I thought the same: Z56 more "italo-centric", and Z36 more eastern.(even if Z36>Z37 seems, today, very "italian").

I just did a different approach for analyzing U152 subclades; specifically L2, Z36, Z56, and PF6658.

I limited the analysis to these countries. 

England
Germany
France
Italy
Ireland
Switzerland
United Kingdom
Scotland
Sweden
Poland
Spain
Italy (Sardinia)
Czech Republic
Netherlands
Norway
Hungary
Wales
Russian Federation
Northern Ireland
Finland
Austria
Belgium
Ukraine
Portugal
Denmark
Belarus
Romania
Lithuania
Greece
Slovakia
Luxembourg
Bulgaria
Slovenia
Serbia
Lebanon
Turkey
Croatia
Italy (Sicily)
Belgium (Flanders)

With the exception of bottom 3 (9 each), all had at least 12 U152 samples.

There were 7589 samples from the these countries of which 
L2=4769
Z36=776
Z56=754
PF6658=386

The median number of samples was 765.  Because L2 skews the numbers so much, I weighted each subclade based on a median number of 765 samples i.e. to get L2 4769 to 765, divide by 2.480065.  I took the L2 raw total for each country and divided by 2.480065.  I repeated this process for the other 3 U152 subclades, so that each totaled 765 total samples.  With the weighted totals for each country I identified which subclade(s) were most common for each country.

The results. 



Even with the weighted reductions, L2 still is the most common subclade in Nordic countries and England.  Its also #1 in Wales with Z56 in a close 2nd (L2=9, Z56=7).
L2 also comes in first in Flanders and Slovenia.

Z36 makes a strong showing in the Balkans/northeast Mediterranean and is first in Germany, Switzerland, Czechia, Hungary, Romania, Greece, and Turkey.

Z56 is first in many Eastern European countries including Poland, Lithuania, Russia, Belarus, Ukraine, Slovakia, Austria, and Luxembourg (L2 and PF6658 were close seconds in Lux.)

PF6658 had an interesting pattern, appearing almost Phoenician/Mediterranean (don't take that too literally), with a 1st place in Lebanon, Sardinia, Croatia, and Ireland.

Portugal, Spain, Scotland, Belgium, Netherlands, Italy, Serbia, and Bulgaria had too close to call results so I used hatch marks to show the top finishers.

It appears as if Z36 is more western Alps focused, while Z56 more eastern Alps (Switzerland vs Austria).  With the weighted results, both were dominate and nearly equal in Italy (136 for Z36, 135 for Z56, 83 for PF6658, and 51 for L2).

An oversimplification of the weighted results is L2 is more Western and Northern European, Z36 is more Central European, Z56 more Eastern European, and PF6658 more Southern European.

[Pylsteen]

Don't forget that Z56 includes 3 Ashkenazi clades (as opposed to 1 L2 and 1 small Z36), boosting its Eastern European numbers; looking at the most important subclades, the image to me is roughly Z145 (mixed focus Italy vs Upper/Middle Rhine), Z44 (focus Rhone?), BY3544,BY3538 (focus Upper/Middle Rhine); Z56 clades seem to have spread mostly during the Urnfield period.

Wonder if more is known already about the clades off 2 new found Britain IA Z56's (poss. Belgae? don't remember which study)

Z36 als seems roughly focused around the Upper Rhine/ West Alp / Piemonte area, although more East of the riverd than Z56. Out on top of my head, the larger amount of newly  born clades was during the MBA/Tumulus period, slightly earlier than Z56,

In any case, both did not boom as much as L2 during the Nell Beaker period.

[Riverman]

Don't forget that Z56 includes 3 Ashkenazi clades (as opposed to 1 L2 and 1 small Z36), boosting its Eastern European numbers; looking at the most important subclades, the image to me is roughly Z145 (mixed focus Italy vs Upper/Middle Rhine), Z44 (focus Rhone?), BY3544,BY3538 (focus Upper/Middle Rhine); Z56 clades seem to have spread mostly during the Urnfield period.

Wonder if more is known already about the clades off 2 new found Britain IA Z56's (poss. Belgae? don't remember which study)

Z36 als seems roughly focused around the Upper Rhine/ West Alp / Piemonte area, although more East of the riverd than Z56. Out on top of my head, the larger amount of newly  born clades was during the MBA/Tumulus period, slightly earlier than Z56,

In any case, both did not boom as much as L2 during the Nell Beaker period.

That's why I described it differently, since I thought the Ashkenazi branches are even more in favour of an Italian-French/South Western origin, like ancient DNA suggests as well.
Bigger Ashkenazi branches can pretty much skew the numbers, especially for areas like Poland.

[Manofthehour]

Don't forget that Z56 includes 3 Ashkenazi clades (as opposed to 1 L2 and 1 small Z36), boosting its Eastern European numbers; looking at the most important subclades, the image to me is roughly Z145 (mixed focus Italy vs Upper/Middle Rhine), Z44 (focus Rhone?), BY3544,BY3538 (focus Upper/Middle Rhine); Z56 clades seem to have spread mostly during the Urnfield period.

Wonder if more is known already about the clades off 2 new found Britain IA Z56's (poss. Belgae? don't remember which study)

Z36 als seems roughly focused around the Upper Rhine/ West Alp / Piemonte area, although more East of the riverd than Z56. Out on top of my head, the larger amount of newly  born clades was during the MBA/Tumulus period, slightly earlier than Z56,

In any case, both did not boom as much as L2 during the Nell Beaker period.

Z46 (Z46>Z48 to be more specific since Z48 holds the vast majority of it) not Z44 though that's a subbranch within it. 
Z48 may be skewed to western alps including the Rhone area as well. S4634 seems to have been a major Urnfield lineage, perhaps spreading from Southern Germany or the Alps. The Late Iron Age Veronese (Vescoville US3231) sample seems to have fit the autosomal profile of what one might expect of a Rhaetic or Venetic person.

Z44 is older and more elusive at this point. A Tumulus era lineage who's spread may fit Urnfield patterns of migration as well. (Germany, Italy, Spain etc...)

[Roslav]

Yet we can assume the majority in both instances, despite the immigration of Daco-Thracians and Illyrians respectively, was speaking some sort of Centum language, if not Celtic.

From Frög we have no results, unfortunately, but from the Unterkrainische we have a couple of results and the majority turned out to be Tumulus culture derived R-L2, with one J-L283 Illyrian outlier. Yet we can see, even in that context, that the Unterkrainische group was mixed.

This.

Dolenjska/Unterkrainische Gruppe is particularly interesting. There is quite a bunch of ancient R-L2-Z49 from that region. Three samples (800-500 BC) from the Dolge Njive burial ground belong to R-BY164359 (probably a single clan, Time Tree TMRCA at 1600 BC), which is quite an interesting clade. It includes a sample from Roman Ovilava near Linz (on the Danube), two German testers from the Middle and Upper Rhine area and one Czech tester. This is a good example of the Rhine-Danube connection which I have mentioned before. And it also provides a perspective on the timescale of the dispersal. 

Another sample from the same site tested positive for FGC20809, which is part of a block of SNPs that dates back to around ca 1000 BC pointing to an Urnfield/Hallstatt migration. There is a bunch of Czech testers who also belong to this clade along with a Medieval Hungarian sample (Abasar 450). 

I only briefly read on this subject, but it looks like there is a plethora of opinions on the ethnic and linguistic composition of Bronze&Iron Age Pannonia (and all the various Eastern Hallstatt groups). The Western border of the much later province of Pannonia cuts right through the Kalenderberg and Sulmtal groups, touching the Dolenjska/Unterkrainische group. Personally, I think it is hard to imagine that Kalenderberg, and to lesser extent also Sulmtal and Frog represented non-Italo-Celtic (or related) populations by as late as 500-450 BC if even Dolenjska Group appears very similar in terms of Y-DNA (which of course does not exclude later incursions from the La Tene culture zone). Strong cultural influences form the Mediterranean sphere, whether Illyrian or Venetic, do not essentially preculde this.

Quite a few if not most of the sources that I read claim that originally inhabitants of Bronze Age Pannonia have likely been speaking other languages than Proto-Celtic (some see Proto-Italic as having originated in this region, others see it as Illyrian territory), but in the end, by 500 BC they seem to appear pretty similar genetically. The key issue of how far Proto-Celtic (and other, related languages) extended in the Hallstatt era has not really been resolved and probably it will never be known for sure. The same can be said about Venetic. Most of the studies that deal with Pannonian linguistics rely mostly on scanty and much younger anthroponymic and toponymic data.

Overall however, the consensus seems to be that initially there were non-Celtic languages present in most of Pannonia (Illyrian or “Ostalpenblock”, Venetic or “North-Adriatic complex/group”) and possibly even Noricum. A historical consensus, dating back to Pan-Illyrian hypothesis and early XXth century, saw Pannonians as Illyrian speakers, but I am not sure how relevant this is today. I don’t really have the time to quickly review the literature or gather all those articles. I can try some time in the nearest future, but here are a handful of examples. A fairly recent, critical article which touches on the subject of Pannonian language can be found here. The author relies on works of Peter Anreiter, especially "Die vorrömischen Namen Pannoniens". Anreiter is a frequently cited scholar when it comes to ancient Pannonia and its languages, like in this article which repeats a common view that Pannonian personal names resemble those from central Dalmatia (but the conclusion presented in the article seems mostly based on Roman period data, as indicated on the map on page 12).  

Here, when talking about parts of Southern Slovenia and Croatia, the author quotes a Croatian linguist, Radoslav Katicic, and mentions the “North-Adriatic” anthroponymic province (in opposition to Noric, Illyrian and Dalmato-Pannonian), which includes territories inhabited by the Liburnians, which he considers “cognate” to Histri and Veneti. Here it is suggested that the original population of Pannonia was related to Illyrians ad even Noricum was not originally Celtic-speaking. While the authors of this paper appear to dispute this view and also point out that despite potential linguistic differences, Pannonians have been considered to be “descendants of the late Bronze/early Iron Age Urnfielders”.

Another linguist dismisses the idea of a separate “North Adriatic group”, and suggests presence of both Venetic and Celtic dialects in the Emona/Ljubljana area, although at a much later time Article 1 ,  Article 2

So there is no clear answer really. It looks like there is evidence of another centum language being spoken in Pannonia, but the history of its origins, evolution or its full extent is largely speculative.

[Pylsteen]

Z46 (Z46>Z48 to be more specific since Z48 holds the vast majority of it) not Z44 though that's a subbranch within it. 
Z48 may be skewed to western alps including the Rhone area as well. S4634 seems to have been a major Urnfield lineage, perhaps spreading from Southern Germany or the Alps. The Late Iron Age Veronese (Vescoville US3231) sample seems to have fit the autosomal profile of what one might expect of a Rhaetic or Venetic person.

Z44 is older and more elusive at this point. A Tumulus era lineage who's spread may fit Urnfield patterns of migration as well. (Germany, Italy, Spain etc...)

you're right, I shouldn't write purely out on top of my head,

looking at possible Celtic-related U152 ancient samples besides L2 in the Discover tree:

Ancient samples (BC) in the FT DNA discover tree:

Z56>S47
- Vescovile 3231 (ca. 300-1 BC, Cenomani, North Italy Late Iron Age)
Z56>FT15337
- Le Peyrouv 163 (ca. 400-200 BC, La Tene, France)

Z36
- Radosevice 15048 (480-390 BC, La Tene, Czech)
- Chemin de Coupetz 20817 (300-200 BC, France, Gaul Iron Age)
- Zamárdi 25519 (300-100 BX, Hungary, La Tene)

and let us not forget Z193 (PF6658)
- Radosevice 15042 (290-250 BC, La Tene, Czech)

We might get the best insights by focussing on subclades of subclades instead of sticking to the branches that date to the Bell Beaker period.
It won't be easy, especially considering all back-and-forth migrations of L2.

[Roslav]

R-Z56 seems to have a pretty strong bias towards Italy I'd say, especially if going by the ancient DNA record. R-Z36 looks like having an overall more Northern and Eastern run, with many branches in Eastern Central Europe. That's something I would expect to find in Celts in Central Germany, Eastern Germany, Czechia and Poland.
Both look to me like being secondarily involved, compared to R-L2, but that's just something I would say at first glance and it might be proven wrong eventually.

Many thanks! I thought the same: Z56 more "italo-centric", and Z36 more eastern.(even if Z36>Z37 seems, today, very "italian").

I just did a different approach for analyzing U152 subclades; specifically L2, Z36, Z56, and PF6658.



An oversimplification of the weighted results is L2 is more Western and Northern European, Z36 is more Central European, Z56 more Eastern European, and PF6658 more Southern European.

Just as Pylsteen pointed out, one has to be careful when it comes to Poland, Belarus, Lithuania and Ukraine. PF6658 and Z56 are almost non-existent in Eastern Europe, in particular North of the Carpathians. Z56 is vastly dominated by the Ashkenazi lineages and there are also quite some German surnames to be seen in this set. Looking at various FTDNA projects R-Z36 (and ZZ45) appears more frequent in Eastern Europe, especially in the South. However, there are still some Germans disproportionately represented in this lot. 

Generally, in North-Eastern Europe there are only single, dispersed, isolated and usually very old lineages that genuinely appear ethnically Slavic (for example, for Polish Z56 I found only 3 such subclades on Time Tree, two under PF6658 and 3 plus one historical for Z36). Care has to be taken though, as such single lineages could also have arrived at a very early stage, predating Urnfield and maybe even Tumulus, or they could be late arrivals, brought by Germanic tribes or Medieval German settlers that got assimilated. 

That said, Eastern Z36 generally appears "younger" and more recently connected to Western subclades, and it is also much more frequent South of the Carpathians, including the Balkans, in stark contrast to PF6658 and Z56 (which I guess makes it a much better candidate for an Eastern Celtic marker). In fact, given its overall global frequency, I would not even consider Eastern R-PF6658 to be Celtic per se.

[nomad01]

Yet we can assume the majority in both instances, despite the immigration of Daco-Thracians and Illyrians respectively, was speaking some sort of Centum language, if not Celtic.

From Frög we have no results, unfortunately, but from the Unterkrainische we have a couple of results and the majority turned out to be Tumulus culture derived R-L2, with one J-L283 Illyrian outlier. Yet we can see, even in that context, that the Unterkrainische group was mixed.

This.

Dolenjska/Unterkrainische Gruppe is particularly interesting. There is quite a bunch of ancient R-L2-Z49 from that region. Three samples (800-500 BC) from the Dolge Njive burial ground belong to R-BY164359 (probably a single clan, Time Tree TMRCA at 1600 BC), which is quite an interesting clade. It includes a sample from Roman Ovilava near Linz (on the Danube), two German testers from the Middle and Upper Rhine area and one Czech tester. This is a good example of the Rhine-Danube connection which I have mentioned before. And it also provides a perspective on the timescale of the dispersal. 

Another sample from the same site tested positive for FGC20809, which is part of a block of SNPs that dates back to around ca 1000 BC pointing to an Urnfield/Hallstatt migration. There is a bunch of Czech testers who also belong to this clade along with a Medieval Hungarian sample (Abasar 450). 

I only briefly read on this subject, but it looks like there is a plethora of opinions on the ethnic and linguistic composition of Bronze&Iron Age Pannonia (and all the various Eastern Hallstatt groups). The Western border of the much later province of Pannonia cuts right through the Kalenderberg and Sulmtal groups, touching the Dolenjska/Unterkrainische group. Personally, I think it is hard to imagine that Kalenderberg, and to lesser extent also Sulmtal and Frog represented non-Italo-Celtic (or related) populations by as late as 500-450 BC if even Dolenjska Group appears very similar in terms of Y-DNA (which of course does not exclude later incursions from the La Tene culture zone). Strong cultural influences form the Mediterranean sphere, whether Illyrian or Venetic, do not essentially preculde this.

Quite a few if not most of the sources that I read claim that originally inhabitants of Bronze Age Pannonia have likely been speaking other languages than Proto-Celtic (some see Proto-Italic as having originated in this region, others see it as Illyrian territory), but in the end, by 500 BC they seem to appear pretty similar genetically. The key issue of how far Proto-Celtic (and other, related languages) extended in the Hallstatt era has not really been resolved and probably it will never be known for sure. The same can be said about Venetic. Most of the studies that deal with Pannonian linguistics rely mostly on scanty and much younger anthroponymic and toponymic data.

Overall however, the consensus seems to be that initially there were non-Celtic languages present in most of Pannonia (Illyrian or “Ostalpenblock”, Venetic or “North-Adriatic complex/group”) and possibly even Noricum. A historical consensus, dating back to Pan-Illyrian hypothesis and early XXth century, saw Pannonians as Illyrian speakers, but I am not sure how relevant this is today. I don’t really have the time to quickly review the literature or gather all those articles. I can try some time in the nearest future, but here are a handful of examples. A fairly recent, critical article which touches on the subject of Pannonian language can be found here. The author relies on works of Peter Anreiter, especially "Die vorrömischen Namen Pannoniens". Anreiter is a frequently cited scholar when it comes to ancient Pannonia and its languages, like in this article which repeats a common view that Pannonian personal names resemble those from central Dalmatia (but the conclusion presented in the article seems mostly based on Roman period data, as indicated on the map on page 12).  

Here, when talking about parts of Southern Slovenia and Croatia, the author quotes a Croatian linguist, Radoslav Katicic, and mentions the “North-Adriatic” anthroponymic province (in opposition to Noric, Illyrian and Dalmato-Pannonian), which includes territories inhabited by the Liburnians, which he considers “cognate” to Histri and Veneti. Here it is suggested that the original population of Pannonia was related to Illyrians ad even Noricum was not originally Celtic-speaking. While the authors of this paper appear to dispute this view and also point out that despite potential linguistic differences, Pannonians have been considered to be “descendants of the late Bronze/early Iron Age Urnfielders”.

Another linguist dismisses the idea of a separate “North Adriatic group”, and suggests presence of both Venetic and Celtic dialects in the Emona/Ljubljana area, although at a much later time Article 1 ,  Article 2

So there is no clear answer really. It looks like there is evidence of another centum language being spoken in Pannonia, but the history of its origins, evolution or its full extent is largely speculative.

I think the problem lies in different definitions of Pannonia. The tribes roughly around the Sava river are considered Illyrian-related, and around the Drava river were the proper Pannoni, Italo-Celtic people coming here with the Tumulus or Urnfield invasion.