104 thoughts on “The problematic of substrates – A case study of Iberia”

1. Lots of food for thought, thank you. You’ve made very interesting and different application of what the aDNA findings thus far can tell us and what possibilities are now more or else less likely in light of them.

   I would like to know if many linguists have started asking the same questions, of how much of the presumed IE vocabulary is originally IE (PIE). It’s an intriguing idea that more might be known of the preceding languages in parts of Europe if certain subsets of words were not taken for granted as being IE rather than IEised, in consequence of incomplete consideration being given for making the assumption in each case.

   Also, great work by Rob and his colleagues on that Megalithic paper. Patrilineal aspects to Megalithic culture controverts what some earlier popularised suppositions on pre-IE Europe led me to believe, so in that respect this finding was surprising, but it does make sense.

2. @ak2014b

   Thanks. As with any post concerning linguistics I was afraid I wouldn’t be able to make my point get through correctly, but it seems you really got it so that’s a good sign.

   I would like to know if many linguists have started asking the same questions

   Me too. But if they haven’t (not sure how many follow closely these latest aDNA studies) then I hope this post can reach a few of them and stimulate these kind of reflections. I’m eager to see what aDNA can bring to linguistics too.

   But if that’s a too ambitious on my part, I hope that at least the post can serve the other purpose of giving some insights into the linguistic problems (of substrates, but even beyond that) to the people that do follow these studies but are less linguistically inclined.

3. >I won’t extend in explaining the extremely unlikeliness of the first scenario because I’m sure that no one will agree with it. So let’s look at the second one to see how likely that can be.

   Not to be contrarian, but why do you consider this unlikely? Isn’t this exactly what one expect given the ancient DNA – linguistic change due to aggressively male-biased replacement but retention of the native toponymy. Just look at the toponymy of the Americas.

4. @Marko

   I thought no one would be arguing for something like Renfrew’s Anatolian Hypothesis. Though in this case it would be with a twist to make it a bit more complicated: steppe migrations replacing IE languages, and later IE languages re-expanding again to replace steppe languages.

   I guess it’s possible, and if this is the scenario you’d favour, maybe you’d like to elaborate on it? I’d welcome an alternative such as that one, as long as someone could make it credible.

5. The use of “white” in river names isn’t that odd, it could mean “clear” or “silver”. Several fish in Germanic languages have names derived from the same root (Dutch “elft” and “alver”). Anyway, there are a lot of American rivers called “White River” and German creeks called “Weissbach”.

   On account of the possibility of a previous IE becoming Iberian speaking: Why not? It actually fits the scenario of BBC being IE speaking but eventually picking up the Iberian languages very well.

6. @ Alberto
   I had to keep my last post about hunter-gatherers brief, but I suspect the rise of WHG in Iberia as the Neolithic progressed was mostly of exogenous origin (e.g. from France).

   Generally in north-central Europe, much of the WHG/ EEF admixture occurred in the post -LBK (post-5000 BC) period. In areas such as Brittany, I think hunter-gatherer chiefs must have been marrying into Neolithic communities (for a good overview ”Dating Women and Becoming Farmers: New Palaeodietary and AMS Dating Evidence from the Breton Mesolithic Cemeteries of Téviec and Hoëdic. Rick J. Schultin”). A similar situation can be seen in ALPc, which despite close similarity to Starcevo, the male lineages are almost all I2a2.

   In other areas, there was somewhat of an LBK collapse, e.g. middle Rhine. The interactions in the Michelberg culture seem interesting, and hopefully full genomic data will clarify who was being sacrificed in those pits. From 4000 BC, groups like TRB and GAC were expanding southward at the expense of older farming groups.
   At the same time, Mediterranean Europe had its own set of interesting but low key interactions during the Neolithic-Copper Age. What the brief overview suggests is that ”WHG/ farmer interaction” was a variable affair, depending on time & place, with resulting different sociolinguistic vectors.

   Therefore I think by 3000 BC, Europe would have been quite linguistically diverse. The unfolding evidence would suggest that the linguistic diversity must have been virtually erased during the BB transition; and given the nature & degree of change its hard to imagine any of the Megalithic langauges of Iberia surviving beyond 2000 BC.

7. This is an excellent article, real joy to read — and you have pointed out an obvious thing that I have somehow not been really aware of all these years — if you have so many reconstructed roots, and you’re not much restricted in meaning, you will likely find *something*. Any river can be big, small, white, black, red, dark, deep, slow, fast, cold, named after deers, ducks, otters…

   BTW the PIE word for water is heteroclitic (has an -r/-n alternation, which is a very archaic feature) so it’s not likely a loan from Uralic. It’s (in my humble opinion) a relic from the Indo-Uralic era.

8. @epoch

   “The use of “white” in river names isn’t that odd, it could mean “clear” or “silver”.

   Yes, I don’t find it specially odd. My point is that it would be preferable to have an explanation to justify such name. When we have a “red” river (for example in Spain we have one called Río Tinto), the reasoning behind it is usually clear. This applies to most other names or rivers or seas (think Yellow River, Yellow Sea, White Sea or Dead Sea). However, if those “white” rivers happen to carry dark waters and show nothing to justify the name “white”, it becomes more of a stretch to prefer such etymology. It’s all about trying to be more strict with assumptions.

   “On account of the possibility of a previous IE becoming Iberian speaking: Why not? It actually fits the scenario of BBC being IE speaking but eventually picking up the Iberian languages very well.”

   I elaborated about the problems of such scenario in note ***. I don’t think that “eventually picking up Iberian languages” is a good enough explanation to overcome those problems.

9. @Rob

   Yes, I think that those interactions between WHG and EEF need to be more carefully analysed than what I could do here (and what the current data allows, though we’re making progress thanks to researches like the mentioned one you participated in). Especially in the fringes of Europe (Atlantic, North and Baltic Seas).

   Though as you agree in your comment, it might ultimately be a moot point, given that those languages most likely disappeared by 2000 BC or soon after. It may still explain several widespread words that appear in different language families, though.

10. @Daniel N.

    Thanks! Very glad to see this can stimulate to rethink some assumptions that may need revision.

11. In reality, I bet Indo-European was actually spread with Y DNA I2 farmers straight from Anatolia up towards Cucuteni-Trypillia and from there to Sredny Stog (and Corded Ware), but this is all after the main advance of G2a farmers (one example is the Dimini invasion theory: Sesklo would be the G2a wave, Dimini this I2 wave). Yamnaya thus non-IE, and perhaps Mediterranean/Atlantic I2 farmers actually spoke some old IE language (idk about that).

    All evidence, though, does point towards the Bell Beakers speaking Vasconic. I have my own personal theories, but I wouldn’t be surprised if we could go back in time only to see a Yamnayan speaking… Sino-Caucasian (and for whichever culture spawned L23, perhaps Orlovka or Khvalynsk, speaking Sino-Vasconic; and Central Asian M269 having branched off early to form Burushaski perhaps even as part of the Hissar culture – all this is structured according to Starostin but ignore the dating and Na Dene being the earliest split). I don’t know why people see Afanasievo as more likely Tocharian than the Tarim basin folk given the Tocharians were not around the Yenisei for one thing (and I think Afanasievans spoke Sino-Yeniseian). The two things hardest to explain would be Sino-Tibetan and Na-Dene (in the grander context of Dene-Caucasian). Sino-Tibetan, perhaps, spread with Afanasievo’s influence on introducing metallurgy to the Far East. Na-Dene is, however, heavily associated with Y DNA C-M217 and not to that Native American R1b, so I’m not really sure. Really though, this is exactly the same thing with e.g. Chinese (not associated with R1b), yet there is a semi-plausible hypothesis for Chinese contained within the Dene-Caucasian hypothesis ultimately from Afanasievan R1b (and of course, not all R1b, but only specific subclades). The Yenisei natives today are also almost exclusively Y DNA Q and C, when we know there was a likely elite Z2103 presence in Afanasievo. Almosan however is sometimes lumped into Dene-Caucasian with Na-Dene and that is associated with R1b. Now, it would be incredible to know which subclade this branched off from, however nobody is willing to test even a single Ojibwe R1b sample in-depth. All we know from available studies is that this R1b is almost entirely M269 and different to the Western European variety.

    That point though, to me, is astounding. Absolutely nobody wants to look at those R1b subclades in-depth. What does this say about the state of aDNA researchers? Could it be the case that they are heavily influenced ideologically, rather than being brutal scientists? Listening to any talk from absolutely any of them, Reich included, certainly gives that impression. Want to see a similar example? This is one of the first results that came up for me when researching the Old Copper Complex of the Americas (first known examples of metallurgy, during the 4th millennium BC (probably not as old as 4000 BC though!) and right on top of Algic territory i.e. the area of Native American R1b): https://web.archive.org/web/20190424131553/https://www.tf.uni-kiel.de/matwis/amat/iss/kap_a/advanced/ta_1_3.html

    This is all working back from the idea that Yamnaya perhaps didn’t speak IE (while Corded Ware did), as their L51 cousins likely spoke Vasconic; and because Anatolian is extremely archaic and lacks the word for wheel in PIE as well as having lots of words relating to farming and other things that Steppe folk would be unfamiliar with.

    Thankfully, however, the truth for most of this will be available eventually. We may never find out if the Solutrean hypothesis of Y DNA I/C seafarers was legit or not given the questionable status of Anzick-1 and easy monopolisation of remains from that period, but there are way too many samples out-and-about in private collections from later periods to be censored from examination forever. It only takes one sample. And it’s only a matter of time.

12. @Alberto

    In your explanation you state that this scenario requires 1400 year of isolation and inbreeding. But that is obviously not true, as the area where Iberian and Basque were spoken were pretty large, about half of Iberia plus Aquitaine.

    There is also this. You propose that one of the reasons Iberian isn’t so similar to Basque is because Basque is a very drifted language, and another is the limited nature of the surviving Iberian inscriptions. However, we also have the Aquitanian language, with a similarly limited nature of surviving inscriptions, but thightly related to Basque. So if Basque is a very drifted language, it was already very drifted before the Romans came.

    That means two things: Basque and Iberian started to take different paths way before the Romans came and secondly, the numeral system which is so similar apparently did not drift. Hence the numeral system must be a loan. But the numeral system is the main reason to consider a genetic link. Whatever the link may be, by the time we can attest them they must already have been growing apart for a long time. Is 1400 years enough for that?

13. Epoch,
    The Bell Beaker culture began to arrive in Iberia c. 2500 BC. So, by the time the Romans arrived there, the languages which the former brought would have been diverging for over 2,000 years.
    On the related note, it seems surprising for one Bronze Age group to have the need to wholescale borrow a basic counting system from another.

    Alexander.
    I suspect that IE might have expanded through Europe somewhat later, perhaps still unfolding during the second millenium BC. If so, the process went beyond certain lineages or farmer vs steppe models.

14. @Alberto

    Also, on determining PIE roots. A system similar to what you propose was already used for reconstruction of PIE, with Sanskrit taking the position of Group B. Now, maybe that has the flaw that Anatolian wasn’t always included, which led to the distinction between early PIE and late PIE. But that doesn’t free the roots for your use of them, doubting their origin. If they weren’t PIE, they at least must have been *late* PIE for else there weren’t any cognates in Sanskrit.

15. @Rob

    “On the related note, it seems surprising for one Bronze Age group to have the need to wholescale borrow a basic counting system from another.”

    But not unique. See Chinese counting system.

16. Irrelevant but harmless side-note:

    Why is there not an established discord server for anthrogenetics? These blogs certainly serve a more formal and organised purpose but the lack of instant messaging undeniably really slows general discussion down. This is obviously self-promotion (mods pls) but there’s nothing personal about this, no power trips or anything, I just think it would be good for everyone: I’ve recently created a discord server for people to talk about this kind of stuff more informally. Link here – https://discord.gg/QRbVFGF

    No matter your opinion it’ll be essentially impossible to get banned unless you’re actually being disruptive (spamming etc.), so there won’t ever be scenarios like with certain other websites restricting discussion of “unpopular” ideas. The only other relevant caveat to this is that hypertribalistic behaviour would get labelled as disruptive and is bannable as it always ends up destroying valuable discussion: what I mean by this is the kind of things like Balkan flame wars polluting everything it touches. Otherwise, it’s a free speech anthrogenetics server and nothing besides, to complement the more formal and laid-out discussions on sites like this. Hopefully, that means maximum discussion!

17. @epoch

    “In your explanation you state that this scenario requires 1400 year of isolation and inbreeding. But that is obviously not true, as the area where Iberian and Basque were spoken were pretty large, about half of Iberia plus Aquitaine.”

    I have a hard time following you. If what you are proposing is that non-IE languages were widespread, then that’s the same thing I’m proposing.

    If you could explain more clearly (to me and the other readers) who were (according to you) the IEs and who the non-IE during the Bronze Age in Iberia (and Western Europe as a whole), when and how did those IE pick up Iberian languages and how we ended up with the linguistic landscape that we know during the Iron Age, that would make it much easier to understand what you’re saying and agree or not.

    “Also, on determining PIE roots. A system similar to what you propose was already used for reconstruction of PIE, with Sanskrit taking the position of Group B.”

    What do you mean by “was already used“? That someone used it at some point but then it was ignored by the rest of the linguists? Or do you mean that this is the golden standard followed by all linguists for determining if a root is PIE or not, until today?

18. @Alexander

    Why is there not an established discord server for anthrogenetics?

    I wouldn’t have the time to keep up with those discussions, so I can’t refer people to go there. For me it’s easier and more useful to keep this more formal format here, but if others want to create and participate in those other ones, it’s fine for me if they create one and post a link to it here.

19. @Alberto

    I propose what you consider unlikely, that BBC was originally IE speaking, picked up local non-IE languages that existed in Iberia and that the Celtic languages entered Iberia during the Iron Age. Your objection against is this:

    ” Some sub population could have adopted the local language for one reason or another. One problem though, is that after that, we need such population to keep that non-IE language for the subsequent ~1400 years in spite of them being surrounded by IE speakers (who in their surroundings would have spoken the same IE language and would have been able to communicate easily). This scenario requires a geographical (more likely) or some sort of ideological (less likely) isolation of the non-IE speaking population. ”

    I consider that untrue because Olalde shows that two groups lived together for 6 centuries before the R1b group prevailed, so there is ample time for all kind of dynamics. At the point the R1b group on Iberian east coast prevailed they were speaking Iberian languages because they somehow switched language. That is a pretty large area, large enough to have enough critical mass to sustain itself. The IE substrate fits such a scenario neatly.

20. @epoch

    Thanks for clarifying.

    That’s not the scenario that i consider unlikely. What I consider unlikely is that Bell Beakers replaced non-IE languages across Iberia (and Western Europe) and that later (some 1500 years later) Iberian languages expanded from some unknown, small population.

    What you are proposing (Bell Beakers adopting local languages in large areas of Iberia and at least SW France) cannot explain the presumed IE substrates seen in these areas.

21. @Alberto

    To the best of my knowledge PIE reconstruction required a word or root to exist in Western IE languages and Sanskrit. That means that you can’t just throw away etymological explanations, since even if the method wasn’t perfect, it at least required existence of a root in a language that never set foot in Iberia.

    You also may state that since so many roots exist etymologies may have a larger chance of being coincidentally right, but that is never enough to throw away 50% PIE etymologies.

    In other word, to consider Alba a coincidental mismatch is special pleading.

22. @Alberto

    “What you are proposing (Bell Beakers adopting local languages in large areas of Iberia and at least SW France) cannot explain the presumed IE substrates seen in these areas.”

    Off course it can. Why not?

23. I think epoch raises a good point in that any number of language sweeps might have occurred after the initial expansion, with a Vasconic group replacing earlier Indo-European layers after R1b reached near-fixation frequencies in Iberia. Of course this might be said about any place and time where differentiated populations encountered each other. I guess that’s one of the limitations of ancient DNA.

    What irks me most about the steppe hypothesis is the lack of earlier IE languages in the wider European region (not considering the Balto-Slavic area). Under the steppe hypothesis one would expect to find at least one highly differentiated survival like Albanian (which persisted in an area constantly inundated with invading tribes) in Western- and Northern Europe, but instead there are strange substrate phenomena like Insular Celtic, Lakelandic/Laplandic, Schrijver’s A2 etc. and the Iron Age IE languages. I guess one such candidate might be Elymian, but southern Italy isn’t really western in the genetic sense.

24. @epoch

    To the best of my knowledge PIE reconstruction required a word or root to exist in Western IE languages and Sanskrit.

    If that was a real standard it would be great, since that’s what I was proposing. But it’s not really what I generally see (for example, *gel-) , but maybe some linguist can confirm or deny that.

    As for the substrates, they are more resilient than spoken languages (that’s the point of using them to try to figure out previously spoken languages in the areas). So if Indo-European speaking Bell Beakers adopted local languages, why would the largest part of the substrate become IE and remain so for thousands of years, when no one actually spoke that language there (except maybe part of the population during a short transitional period)?

    What those substrates require is a longstanding presence of IE speakers in the area, with a short lived Iberian speaking presence.

25. @Alberto

    >What those substrates require is a longstanding presence of IE speakers in the area, with a short lived Iberian speaking presence.

    Aren’t the farmers the only group fulfilling that requirement? This would have to be qualified, of course, since there is a high likelihood of language shift with the strong introgression of WHG and male biased replacement. I find it unlikely that G2a Anatolians and I2a megalith builders for instance would have spoken related languages.

26. @Alberto

    “As for the substrates, they are more resilient than spoken languages (that’s the point of using them to try to figure out previously spoken languages in the areas). So if Indo-European speaking Bell Beakers adopted local languages, why would the largest part of the substrate become IE and remain so for thousands of years, when no one actually spoke that language there (except maybe part of the population during a short transitional period)?”

    Well, as you said yourself: The hydronyms and toponyms are more resillient than spoken languages. I don’t see a problem there. The picking up of languages wouldn’t have been an ad hoc process where the entire Iberian area switched all at once.

    I have no idea what the dynamics were. But pre-steppe Chalcolithic Spain already became different from previous farmer cultures and the link below show that these cultures acquired a warrior culture with status burials. That means some incoming BBC groups could have taken over existing power structures in those 6 centuries.

    https://rd.springer.com/content/pdf/10.1007%2Fs10963-018-9114-2.pdf

27. Since Villar contends that old IE toponymy is more persistent in the south, remind me was this site tested yet?

    https://bellbeakerblogger.blogspot.com/2018/05/a-prince-and-his-twenty-wives-garcia.html

    I wonder if the big guy there has G2a, I2a or a stray steppe lineage. This seems to be the site where those Chalcolithic social changes are most evident.

28. @Marko

    Aren’t the farmers the only group fulfilling that requirement?

    Yes, I think that the only reasonable explanation for those substrates would be if the Early Neolithic Farmers brought Indo-European languages from Anatolia (as in Renfrew’s hypothesis). So if the latter turns out to be correct, then probably the substrates are correct too. Conversely, if the Anatolian Hypothesis turns out to be incorrect, the substrates are probably incorrect too (and this is the assumption I’ve gone for in the post).

    Who knows, maybe as more data comes out we’ll have to revisit the Anatolian Hypothesis as the only viable option left. With the twist that the early IE languages of Europe went extinct and replaced later by others closer to the core area (which actually would be Northern Mesopotamia, rather than Anatolia itself).

    But this is quite speculative right now. There are many problems for the Anatolian Hypothesis that would need to be addressed before we can consider it a serious contender again.

    BTW, I don’t think we’ve seen DNA from that man yet. Given the date it’s not possible for him to be an R1b/Steppe guy though.

29. The dynamics in Iberia really are quite fascinating ; and definitely worth a closer look; in light of some suggestions above.
    With regard to chronology; it appears that BB first “lands” in Meseta (and not the Tagus estuary; as has long been posited) c 2500 BC. They certainly appear amidst other groups; which themselves appear to have been quite diverse (even featuring the anticipated individuals with North African affinities). However, this doesn’t seem to have resulted in a fusion or free -for-all; but instead a brief coexistence (where the different groups used separate burial locations even within the one complex ); followed by the ascendency of one particular group. By 2200 Bc; the main chalcolithic groups in the south had all “collapsed”; and the new dominating horizon (El Argar) had no roots in the former; whether in site topography or burial styles. All earlier Los Millares sites show abandonment as do the earlier wealthy tholoi tombs (by 2400 BC). Again in the south it is the same group as those in the north which rose to dominance; as im sure all aware from the uniparental data

    And whilst the odd non-L51 lineage hovers around until 2000 BC; the transition in Iberia; in terms of power structures; is eerily rapid and complete. The most economical explanation is that there would be no great incentive or opportunity to commit to language switching (but of course theoretically possible). This is not a case of Visigoths integrating, in some way or form, into local Iberian Romanitas.

30. Marko, the issues as I understand them with the Anatolian hypothesis are that many linguists have a problem with the slow rate of linguistic change implied by it. From a paleo-demographic and ‘culture-historical” perspective, too much happens between the arrival of early farmers from Anatolia until the attestation of IE groups; and it is difficult, for ex, to envision the obvious similarity between Balto-Slavic & indo-Iranian if they diverged in, say, 7000 or even 5000 BC.
    In the East Balkans and parts of Anatolia there is in fact settlement discontinuity for up to 500 years after 42/4000 BC. So any scenario will have to take this into account.

    At the same time, the scenario of multiple arrows emerging out of the steppe might require revision, because it seems that some of the movements are associated with offshoots where later attested languages are arguably non-IE (BBC) or became locally extinct (Afansievo).

31. @Marko, is the lack of earlier IE languages in the wider European region (not considering the Balto-Slavic area). what’s so odd about that?

    The wider European region (and I would guess we mean Europe north of Balkan Mountains, given this blog) is a linguistic spread zone marked by either repeated and known radiations (Slavic / Germanic / Romance, possibly Celtic), or at least frequent contact preventing formation of divergent varieties (the ‘Celtic as developing through contact’ theory of Andrew Garrett).

    If you had a topology like the Caucasus or Pyrenees, it might be odder. Mountainous topology preserves languages better than flatland (see PNG for another illustration), warm climates tend to preserve better than cooler (empirically – https://www.sciencedirect.com/science/article/pii/S0278416598903282).

    Analogous, within linguistics around Sino-Tibetan, there are arguments against Sino-Tibetan originating on the North China plain. But saying “But this massive Chinese Empire on the North China plain, today, only speaks one variety and many isolated peoples in the southwest hills speak very, very many” is not really one of them, alone! (There are known historical and geographic-demographic processes which make that a fairly bad piece of evidence).

32. @Matt

    Disagree, a landscape like Fenno-Scandia is favorable for preservation, that’s why not-so-distant paleo-language layers are fairly obvious there. Sweden is flat, but peoples and languages don’t diffuse as readily through the northern forest as they would in Iberia for instance. Hence the dynamic seen in ancient DNA.

    Germanic/Saami-Finnic/Baltic Y-DNA shows this as well, in that it is quite disjointed and marked by local founder effects, in contrast to Western European populations which were for the mostpart fathered by the L51 guy.

33. To use the examples of Germanic, Slavic & even Roman, they inform that these were demographic – migratory events (c.f. political-cultural for Roman along with native client management, re-settlement & colonization); and those languages are indeed quite differentiated within the I.E. tree.
    Whilst Celtic is just on the cusp of proto-history, that ancient authors were noting their movements around Europe, and there is an expansion of ”Celtic material culture” speaks volumes. I think recent scholars have been a little too deconstructionist in their approach to such questions, and now might have to back-peddle a bit. The expansion of classic Celtic might have had a dialect levelling effect in some areas, whilst representing a wholly novel introduction of I.E. in others. Even with fine-grained correlative archaeo-genetic analysis, it might not be entirely clear.

34. I suppose considered in isolation the language argument isn’t very strong and Matt was right raise those criticisms.

    I should have expressed myself clearer in that it is the fact that where the Iron Age languages (Celtic, para-Celtic, Germanic) one finds non-Indo-European languages (Tartessian, Aquitanian, Iberian, Rhaetic, Etruscan) and not differentiated IE survivals as one would expect if BB spread IE already in the EBA. It would require a highly selective replacement of old IE languages.

35. @Marko

    Yes, the selective replacement of Bell Beaker languages in R1b heavy areas of Western Europe, with not even one surviving into the Iron Age sounds strange enough to consider it highly unlikely that the pre-Celtic Western European linguistic landscape was large IE (of unknown branch).

    It’s much easier to imagine that it was largely Ibero-vasconic, and that Celtic expansions replaced large areas were those languages were spoken, but still in other areas were not replaced.

36. @Alberto

    “Yes, the selective replacement of Bell Beaker languages in R1b heavy areas of Western Europe, with not even one surviving into the Iron Age sounds strange enough to consider it highly unlikely that the pre-Celtic Western European linguistic landscape was large IE (of unknown branch).”

    There is at least one theory that proposes a non-Celtic, Non-Germanic Indo-European substrate in the North West of Europe: the Northwestblock Theory. It is based, in part, on hydronyms and toponyms.

    “It’s much easier to imagine that it was largely Ibero-vasconic, and that Celtic expansions replaced large areas were those languages were spoken, but still in other areas were not replaced.”

    Why much easier? The Vasconic substrate theory by Theo Vennemann has been rejected by most linguists because Venneman compared European hydronyms to *modern* Basque, which skews the results. To propose a NW Block actually makes pretty much sense.

37. Yes I think its difficult to argue that Vasconic goes back to Solutreans, or that Megalithic Europe spake Semitic. However, Koch, Schriver have also pointed to a Vasconic substrate in Celtic. It is also interesting that linguists generally view Celtic & Germanic to be distinctive, apart from some obvious recent loans, despite their geographic proximity. It suggests that they have had quite different formations, population wise. But I dont know any specifics of substrate study to comment on whether it is correct or not.

38. One of the interesting things about Celtic in light of the ancient DNA evidence is the increasing non-IEness as one approaches the North-West European fringe. A good summary can be found in Matasović, Ranko, “The substratum in Insular Celtic”. As for the continental Celtic languages and the supposed Vasconic substrate, is there any recent work on this (apart from Koch)?

    Germanic is more complicated. The agricultural substrate should be investigated (and some of it is quite puzzling, for instance why would the Germanics have borrowed words for bull, cow, goat etc.?). More esoterically, Shrijver’s A2, “the language of geminates”, which doesn’t affect Proto-Germanic but its offspring. Finally I’d also like to see a discussion of the archaic Germanic loans in Proto-Albanian that Vladimir Orel lists in his book.

39. Marko: “Under the steppe hypothesis one would expect to find at least one highly differentiated survival like Albanian (which persisted in an area constantly inundated with invading tribes) in Western- and Northern Europe”

    Albanian itself might have ended up fully Romanized in the long-run if it weren’t for the Avaro-Slavic disruption of the imperial system in the Balkans though. It already shows very pervasive Romance influence, of both the Dalmatian and Eastern Romance varieties, after all. The early Albanian-speaking population that emerges after all these events is quite small and only expands considerably towards late medieval times. That kind of linguistic survival seems like the result of a streak of pretty good luck (and partially likely helped due to mountainous landscapes in central-northern Albania and thereabouts) and to generally be more the exception than the rule in Europe. I’m not sure we should necessarily expect that to be the case a priori then. Had things turned out a bit differently, linguists might be attempting to reconstruct more hypothetical substrates (akin to nordwestblock, alpine etc. in western Europe) than studying a living language. Who knows? The general lack of early writing in most of Europe obviously doesn’t help either.

    Are varieties like the various poorly attested western European ones that linguists aren’t able to neatly categorize as either Italic or Celtic a really different story to the extinct Balkan branches anyway?

    Btw, in the most recent G25 spreadsheet, I notice an interesting newly added sample “TZA_Zanzibar_First_Millenium_I0588” (supposedly from the Skoglund et al. https://www.cell.com/cell/fulltext/S0092-8674(17)31008-5 but it doesn’t seem to be in it) that’s Iberian-like and is dated c. 800 AD in the Harvard database. Could there be Iberian individuals in Zanzibar that early (say, movement within the Islamic world) or is it more likely that it’s misdated?

40. 2713/5000
    I see a problem in these debates that I am surprised. It is being taken for granted that the falsifiable hypothesis (because it is only that) that the Iberian numeration is virtually identical to the Basque or Basque one is already a theory or a proven fact.

    Apparently no one is considering that it could simply be a mere forced construct. Analyzing the same in depth, it is easily discovered that everything is merely speculative. There are very few elements that actually lead to think that the sequences, which have been artificially biased (without in some cases respecting the divisions themselves using points in the same Iberian texts) are really words, in this case, lexemes for numerals. In fact, we reach the absurdity of assuming as if it were something natural that in texts where the Iberians use symbols for numerals (vertical bars), which is what they always do, they also write – supposedly – the same numerals in their form lexical That contravenes the principle of economy.

    But most importantly, a recent study has shown that the Iberian letters used as acrophones in the two Iberian dice found to date correspond to the acrophones (principle of acrophony) of the main sounds of the words or used for numerals in Altaic -Turkic languages, but not with Basque or Indo-European or Afrasian numerals.

    It would be too coincidental that through the hypothesis that the letters are acrophonic signs of the lexemes for numerals of those two dice only coincide with numeral lexemes of Altaic-Turkic languages, especially when it is verified that the few and very brief bilingual texts ( Ibero and Latin) found to date, can only be explained (the Iberian text as a translation of the Latin text) through the same Altaic-Turkic languages. It would also be very coincidental that everything that to date has been reconstructed as nominal suffixes in the Iberian language, as well as its most important verbal paradigms, can be explained equally through the Altaic-Turkic languages, and in most of the cases can not be explained from any other family of languages.

    Apparently, nobody (except two Spanish authors) had taken into account that the Altaic-Tukic languages ​​(which Russian linguists consider to be very old, rather than the same Indo-European) could be part of the language of those early Kurgan peoples, or even a important language of some communities of the first farmers that arrive from Anatolia to central and western Europe, and that, genetically, we see arrive to Iberia in a relatively short time with the same genomic profile.

    A cordial greeting,
    Ariel

41. One thing I found quite interesting is the name of the river Danube. It is clear an IE kind of name.

    The Danube delta was the key region from which stemmed the EEF societies that influenced the culture of the steppe people ( Cucuteni, Gulmenita, Hamangia broadly speaking ) so the steppe people must have known this river by the native name of the locals . If the EEF were not IE then it means that the steppe tribes renamed the river in IE . But how likely is that after interacting for nearly two millennia with the EEF people the R1b rich clans would have a rationale to change the name of this much famous river?
    More likely they would have retained the name of the locals……but then it means that the EEF danubian societies were IE……

    I think the name of the Danube is just an example among the hundred of thousand of IE place names all over the continent.

42. Egg; Matasovic argues Lusitanian is just Celtic; but even if it’s not; it’s probably unlikely they began to separate 2500 prior their atteststation; although I’m aware we can’t date words in a lab
    I think we’ve learned what we can from Western Europe (although I realise France is still somewhat of a blank box); and data from elsewhere will help contextualise it further

43. 1. Danube is usually considered to come from Scythians. Your EEF argument requires linguistic continuation in an area where archaeology shows discontinuity. Not only that but you’re under the impression that if it’s not EEF then Steppe when the current data is not actually in favor of these two groups.

    2. Many rivers have more than one name including the Danube ,which was called Istros.

44. Speaking of linguistics, anyone has the original Sogdian hypothesis before the academic mafia attack?

45. @ Vara
    Something like Old Europe’s ”Out of ECE” is a viable model at present, IMO.

    It seems like Prof Nicholl’s has changed her mind about out of Sogdiana. Indeed, the shift from Namazga to BMAC requires somethign more western.

46. @Rob

    As someone who was meddling with the Indo-Hittite in Balkans, I really do not see IE reaching Western or Eastern Europe early anymore.

    “word/root can be considered PIE if it is attested (with regular reconstructions from the putative PIE root) in at least one language in each of these two groups (Group A and group B), and being attested in a minimum of three of them in total.”

    ^ Using this approach with the known IE religions, folk stories and texts we can easily construct the original PIE culture and see how it evolved to the known groups. There are many IE elements that are lacking but I can’t go over it at the moment.

    Yes, I agree that the movements that brought many new technologies to “Turan” were from the west. I am just looking for her original hypothesis.

47. @Egg

    I do think that in Albanian we have something quite different, in that it really doesn’t belong with any of the other attested branches. An alternative explanation might be that the Balkans were in fact Indo-Europeanized by steppe peoples much earlier than Western Europe, contrary to what Reich and colleagues have been suggesting. I’ll have to admit that I don’t really have a good understanding of the dynamics in Chalcolithic Southeastern Europe yet.

48. @Vara

    Do you have a write-up of your ideas anywhere? I’d always been interested in those comparative models as thex pertain to religion and folkloric motifs – based on these before ancient DNA my suspicion had been that the IEs had their ancestral home somewhere in northern Mesopotamia. Alberto Green in his “The Storm God in the Ancient Near East” describes Anatolian IE religion as a sort syncretism of Caucasian, Semitic and Sumerian influences – I think that could be extended to the entirety of the IE family. Not sure if something like this can be supported with DNA though.

49. Late to the party (have been, and will still be celebrating my father’s 90th birthday), seeing that there is lots to comment about.

    First: Congrats Alberto – you aimed into the proverbial wasp net, in an extremely well thought and balanced way. [Also congrats to you, Rob, on the Megalithic paper – a caveat here that it is IMO to Britain-centered and doesn’t sufficiently reflect N. Germany, e.g. early dates of the Büdelsdorf Megaliths, ignored in the paper, or the fact that N. German dolmens rather represent family (farmstead) than community burial sites].

    Let me start on “white“.

    Alberto: You obviously have never stood on the banks of the Elbe in Hamburg (but I offer you a tour, should you ever make it there) – otherwise you wouldn’t have questionned whether an Albus = “white river” ethymology could make sense – in Hamburg, it certainly does. However, there is also Czech Labe for the Elbe, which connects it to another element of Old European Hydronomy (OEH) – Lippe, Luppe, Ptolemy’s Lupfurdum (Leipzig?, Dresden?), etc.
    Intriguing in this respect is that the Laba, the largest confluence of the Kuban in N, Caucasia, apparently shares the same root. There isn’t any historically attested presence of Indo-Europeans along the Laba prior to the 18th century Russsian expansion. Even to date, nearly half of the inhabitants of the Laba catchment area seem to speak NW Caucasian as first language.
    There are several possible explanations for the Laba-Labe-Elbe-Albus homomophony:
    a. Chance occurence : This would of course posit a significant blow to the whole concept of Old European Hydronomy;
    b. An ancient IE layer in the NW Caucasus replaced by later migrations. That ancient layer would plausibly have been represented by Maykop (the eponymous city lies some 40 km west of the lower Laba). The source of replacement would remain speculative – neither archeology nor aDNA have so far provided indication of such replacement ever having taken place in the Laba basin.
    c. The Laba-Labe-Elbe-Albus hydronymy isn’t derived from PIE, but instead goes back to a proto-language that a/o has also shaped NW Cauacasian.

    I tend towards option c. In fact, the whole connection of Old European Hydronymy to PIE looks extremely tenouus, since there is hardly any hydronomic root that can be connected to modern IE. Take, e.g., hypothetic *eis- “to move quickly” as reflected in hydronyms such as Isar, Eisach, Isère, and in that Curchin paper also related to the Catalonian town of Aeso (Plinius). I am not aware of any modern Romance, Germanic or Celtic language having any word that can plausibly be connected to *eis- “to move quickly”. The same applies to OEH terms like Seine/ Saone/ Sinn (Main), or Leine/ Lahn.

50. @ Vara
    an ultra late scenario for IE in north- Europe has been discussed here before. How would you sketch that out ?
    By which means did IE arrive to Europe from Sogdiana or environs ?

51. Frank; congrats to your father; that’s quite a feat ! As for Germany; I have much interest but I’m sure the local labs will bring forth some much needed data.

52. @Frank

    You can’t be serious there. The Elbe area has been Germanic up until the Slavs arrived in its valley, roughly 600 AD. All Roman authors call it Albis. Romans actually visited the river more than once, so they must have known its name from firsthand contact. Then, 600 years later, when the Slavs came Albe become Labe and you don’t see that it’s a loan?

53. @Marko

    “I do think that in Albanian we have something quite different, in that it really doesn’t belong with any of the other attested branches.”

    Am I getting it right that your basic point is that Albanian seems very distant to all existent IE branches, unlike the rest which seem to have more recent common ancestry with at least one more branch?

    Let me think of a few potential counterpoints or reasons it might not matter too much:

    – Attested very recently while having undergone a lot of changes and having remained relatively reclusive for a period of a few hundred years in the Middle Ages might be biasing an assumption about how old its divergence from any particular branch is.

    – Plausibly assuming they ultimately spring up from a similar procedure (the interplay of steppe and Balkan populations after the emergence/intrusion of Yamna in the EBA) Greek and Albanian might have split as far back as the early-mid 3rd millennium and not necessarily been geographic neighbors until much later again. The extinction of all other Balkan branches might be biasing our assumptions a bit too, especially since Albanian seems to be of the inbetween variety in the centum/satem split, like Armenian and unlike neatly centum Greek. Since you mention Orel, my recollection is that he thinks Albanian is most related to Balto-Slavic on one hand and Greek on the other. It might have originally developed as a more northern language in some early contact with emerging Balto-Slavic. Taking all those together, Albanian might have split from any other extant branch much earlier than the rest, then, even if we assume a similar timeline for the general early IEzation taking place in Europe.

    – Germanic also often has a varying position on trees and linguists seem to have some trouble properly situating it, no? It seems to be the closest analogue to Albanian, even if one supposes a more recent divergence from other branches.

    @Rob

    That’s true and the poorer attestation doesn’t help these endeavors. I was also thinking about some other western branches that linguists, from an outsider perspective, seem to have had some trouble relating to more prominent ones e.g. Ligurian, Venetic and North Picene even though I’m not exactly sure what the latest consensus on those, if we can broadly speak of one, is.

    @FrankN

    Happy birthday to him and here’s to 10 more.

    As a further aside, similar to the one about the Zanzibarian, I notice the inclusion of the Botocudo in the newer G25 spreadsheets. I have to say I didn’t even have recollection of the study they initially appeared in and I REALLY have to ask: are those genuine?

    A less out there sample which is a bit more pertinent to the IE discussion, which I think might have been discussed here before after its inclusion in G25 but I’m not recalling right now, is the very low quality Kum4 which already seemed interesting in the paper and appears to have some sort of steppe ancestry (under the assumption that it’s dated properly so the prominent ENA stuff in it is probably not genuine). Unless it’s truly some early one off, it does seem to strengthen the view that Anatolian might have appeared in Anatolia already quite early contra views that make its migration there quite late.

54. @epoch, I suppose literally, if you presuppose a relatively recent Celtic expansion that replaced the languages of Western Europe (and this is a big if), then you could suppose they could well have been speaking any sort of language non-IE or IE, and that includes ones related to Basque.

    However, I would think there are distributional grounds to suspect pre-Celtic IE. If you presuppose any sort of split and spread of IE wherein it’s at this time common in Central Europe (Italic/Celtic/lots of dead branches), Northern Europe (Germanic/Baltic), Eastern Europe into Central Asia (Slavic/Iranic), South Asia (Indo-Aryan), the Balkans (Greek/Albanian/Thracian/Dacian), Anatolia (Anatolian languages/Armenian/Phyrgian), East-Central Asia (Tocharian), by this time, then it’s a bit of an unexplained exception to suggest these movements stopped cold at NW Europe until Celtic, and not really so implausible that the phenomena of its general dispersal would take it into NW Europe (whatever this phenomena was). There’s certainly no grounds for an ecological barrier.

    On the other hand, let’s say we do suppose the “IE expansion phenomena stopped cold” idea, then it’s hard to give any justification about for why a non-IE language in Western Europe generally would be related to Basque or Iberian, even less brought by the Bell Beaker culture.

    Even if these two are distantly related, there’s no firm evidence for a <2500 year old coalescence of Basque and Iberian suggesting that they were introduced by a migration at this time depth rather than much earlier (and the top Basque linguists don't think so*), no evidence of relationship of either to Tartessian, and nor is there a far flung connection between Basque and any other isolate in Northern Europe that would suggest a formerly wider distribution of languages in the family. (For instance, if Pictish were attested and turned out to be very like Basque, rather than the Brittonic like language that all its onomastics suggest it was, then you could probably claim with some sense for a wider Vasconic distribution, with a Celtic "wedge" moving into this. But no such evidence exists.) Even if you presuppose that West Europe were not speaking IE, there's no more evidence that they spoke a recently Basque or Iberian related language than some totally unknown language.

    *Yes, any individual person can claim that they are related to this recency and extraordinary rates of lexical and grammatical change that are unusual but possible have for some reason occurred and hidden this, but I would personally prefer to go with the more consensus position and a less edge case assumption.

55. Regarding the “white river” thing, there’s evidence of it in Greek as well e.g. Alpheios. Noticed rapids in parts of a river could potentially provide an extra link, even if it isn’t a generally too prominent feature.

    Also I have to agree that Labe seems just like the result of metathesis in Slavic rather than anything deeper, which also means that the river name must have entered Slavic relatively early.

56. @Egg

    Yes, it is very likely that Albanian migrated south recently – Orel mentions extensive contacts with Balto-Slavic and Germanic and hypothesizes that Albanian would have been situated in the southern Carpathian area. That’s one of the reasons why I think that irrespective of the original homeland, some kind of delayed Indo-Europeanization of Western Europe is likely. Under the steppe hypothesis this would probably be either due to the loss of the IE language by BB already on the North European plain, or because only CW (pushed back by BB) was IE. If that’s true I think I’d slightly favor the latter (in the absence of ancient DNA evidence)

57. Hey Frank, do you think it’s plausible that Indo-European spread to Sredny-Stog (and thus later to Corded Ware) with Trypillia and that Yamnaya and its Volga predecessors spoke a Dene-Caucasian language? I wrote a lengthy, somewhat rambling comment above – just wondering about the major flaws there other than Na-Dene requiring movement into the Americas (which imo is actually obvious from the archaeology but also the presence of almost certainly pre-colonial Native American R1b).

    Also, which haplogroups would you associate with the invasion of the people of the black and grey pottery (the Dudesti-originated wave which ultimately led to Dimini, Vinca, Trypillia, Hamangia, Boain etc.)? Still just G2a? Or maybe I2a1b/E-V13?

    Congratulations with your father by the way 🙂

58. Sorry for late replies. It’s Orthodox Easter here and i’m out for long weekend.

    @Ariel
    I never heard that theory about Iberian being connected to Altaic. Theoretically it would be possible if Altaic had an origin in the western steppe and arrived to Iberia with Bell Beakers (and to Mongolia with Afanasievo related cultures that reached that area). However, being Turkic languages widespread and therefor better known than Basque (history, reconstrction of proto-language) , it would seem strange to not have much more evidence in between the steppe and Iberia (a turkic substrate would be much easier to identify than an Iberian or Basque related one). In any case, if you have a reference for the hypothesis let us know, to check it.

    @FrankN
    Thanks. And congrats to your father. I hope he stays healthy and that you can follow his steps to reach such age.

    I certainly haven’t seen the Elbe in Hamburg region, and in general don’t doubt that rivers can be called ‘white’. It was a comment meaning that one has to be careful about etymologies that deviate from an original meaning. Might have been more clear if instead I asked about the meaning of the root in Sanskrit (skillful, master). Can we say that it derives from the presumed original meaning of ‘white’? I guess we can if we allow enough flexibility. But once we allow such flexibility the number of pairs of words that can be said to derive from the same root is countless. So my question would be if in this specific case we have some sort of evidence about the evolution of the meaning from white to skillfull in Indo-iranian languages. If we indeed have it, then no problem. But if we don’t then I do see a problem there. Just finding a word that could theoretically derive from the same root (if indeed the Sanskrit word can be regularly derived from the PIE root), but has a makedly different meaning doesn’t seem good enough to me. Some higher requirements should be placed or else we end up with a reconstrcted proto-language that is orders of magnitude richer than any known language and with the problems I’m trying to highlight in the post.

    @Matt

    if you presuppose a relatively recent Celtic expansion that replaced the languages of Western Europe (and this is a big if)

    You mean that that is the low probability scenario while some other is significantly more likely? If not, going with the most likely doesn’t seem such a big ‘if’ (given the linguistic and archaeological evidence).

    it’s a bit of an unexplained exception to suggest these movements stopped cold at NW Europe until Celtic

    Not necessarily accurate. What the data suggests is that earlier IE languages expanded to Western Europe (as far as Portugal) before Celtic. These would be, for what we know, para-Celtic languages and would have arrived in the late Bronze Age, which seems consistent with the timing of other expansions further east. We can’t say too much about them, since only Lusitanian could count as the one that survived into the Iron Age. Maybe Pictish is another, but we don’t really know. We must suppose that their impact was moderate, since only one survived.

    On the other hand, let’s say we do suppose the “IE expansion phenomena stopped cold” idea, then it’s hard to give any justification about for why a non-IE language in Western Europe generally would be related to Basque or Iberian, even less brought by the Bell Beaker culture.

    The justification is based on the data available. It’s the most likely scenario because it’s based on what we see after the Celtic expansion. That doesn’t mean it’s necessarily correct. But any other proposal would be based on nothing at all, so why would one prefer it?

    I’ve also explained in the notes my point of view about possible language diversity pre-Bell Beaker. It’s complicated to know with much certainty, though. As for the time of divergence of Basque and Iberian, also quite difficult to say precisely unless we get more data. If the theory about the numbers is correct, a lower chronology like it would be for a Bell Beaker origin seems preferable. If it’s not correct, then no way to know.

59. I don’t think the suggestion is that there were no IE -speakers in Western Europe before Celtic; or that IE expansion halted at the Rhine for 2, 000 years; which would be absurd. However; broadly speaking, something of an MLBA expansion of IE in Western Europe; with at least some of the pre-IE langauges expanding with the BB network is certainly consistent with the evidence; and the comparative analysis Alberto has shared is very powerful to help resolve some possible language relationships.
    In any case; “did BB speak IE or not” is probably not the question we should be asking, but rather – why did some of the post-BB oucumene go on to be IE-speaking and some not; and dare I say; that answer is becoming more clear to us.

60. Alberto: You mean that that is the low probability scenario while some other is significantly more likely?

    No, I’m not qualified to say which is the higher or lower probability scenario, but there are clearly multiple viable scenarios for consideration in play (at least as I judge by presentation by Koch and Garrett, as an outsider to the field). Hence, “big if” (to imply an open linguistic problem, not that a Celtic ‘homeland’ and expansion from there is less likely per se).

    The justification is based on the data available. It’s the most likely scenario because it’s based on what we see after the Celtic expansion.

    See, I think this is based on scaling what we see after by the Iron Age in Aquitaine and East Iberia up into a much larger sphere which we have no distributional evidence ever existed or much less recently expanded (no indication of extinct varieties, no positive evidence of an expansion, recently to the time of the Beaker phenomenon, or even much earlier, as much as a remote early neolithic connection between these languages seems possible, given a possible but obscure relationship).

    If the theory about the numbers is correct, a lower chronology like it would be for a Bell Beaker origin seems preferable.

    I have to disagree with that; It seems fairly less than preferable to postulate a >2500y divergence of languages, based on only the numeral system, if they appear to share little to no basic lexicon or grammar as far as the people who are in the most expert position to say view it.

    (If we were future archaeolinguists working with Thai and Mandarin, and we had evidence of essentially no convergence in their lexicon and grammar (which is true, beyond some sprachbund effects and some borrowing from Sinitic to Thai), would we suppose that since they share the basic numerals (which they do), the lack of any, let alone recent lexical and grammatical relationship they must have had a recent relationship and it must be due to an extraordinary rate of linguistic change? It seems that we wouldn’t be very parsimonious future archaeolinguists.)

61. If we are limit ourselves to a kind of low-res, autosomal -only analysis, without referencing collateral evidence (e.g. the long-known rhythms of European prehistory, anthropology, uniparental markers), it might seem like there’s ”no positive evidence of an expansion, recently to the time of the Beaker phenomenom’. But in reality much happened afterwards. A striking example is the Elp culture in Netherlands, as summarised by Fokkens, who participated in Olalde 2018; and there are a some tantalizing hints from the few samples that suggest a shift from P312 to U106 during the BB to Middle Bronze Age transition. Such evidence is also emerging from England, not to mention future work in the mainland, where there was quite literally wave after wave of cultural-demographic movements and ping-ponging during the Tumulus, Urnfield, Halstatt & La Tene periods. For Iberia, if one (from some reeason) doesn’t like the idea of Urnfield migrations, there alwasy the MLBA dynamics surrounding the rise of Cogotas / demise of El Argar; not to mention an entire archaeological horizon in northern Iberia linking to the La Tene culture. In light of the wealth of data, it would be very peculiar to claim that nothing much happened in western Europe after the arrival of BB; which is probably why paleogeneticsists are now focussing to the Bronze Age & more recent periods.

    As for linguistic models, convergence, etc; they need to be channeled via aDNA & archaeology, because the latter reconstruct the human actors which spoke the words. Therefore, pre-aDNA models are part of an arsenal which we are now lucky to draw from and adapt to what the data is telling us. In any case, the idea that Celtic, Italic or Greek emerged due to contacts of pre-existing language dialects is consistent with ongoing mobility, even if de Vaan has suggested that Garrett’s suggestions are somewhat overstated. Whatever the case, I don’t think G.’s models imply a need for immobility nor that the said convergence ocurred over the entire swathe of western Europe.

62. @Matt

    (Sorry again for late reply)

    “No, I’m not qualified to say which is the higher or lower probability scenario, but there are clearly multiple viable scenarios for consideration in play (at least as I judge by presentation by Koch and Garrett, as an outsider to the field). ”

    That sounds like you’re underestimating your qualification for having an opinion about the subject. The expansion of Celtic is a relatively simple and clear question (compared to other much more difficult questions like the dispersal of IE languages as a whole, where there is not enough data to know with any certainty and many authors have proposed many different scenarios and yet you do seem to have an opinion about what is more likely and less likely).

    I haven’t seen the presentation you mention, but otherwise from what I know, when it comes to Andrew Garrett, I do see some value in his ideas (in the context of arguing against an unrealistic scenario where PIE in the steppe was evolving into different branches and the speakers of each of those branches would then migrate to the attested location of each branch). I find it quite reasonable that he reacts to this and proposes a more complex model of divergence and convergence.

    However, I don’t think that today it can be it can be taken at face value. It’s a purely linguistic hypothesis without any historical context and it would imply PIE expanding to all the locations where later IE languages are attested, and evolving regionally into the branches attested in each region. Going to the Celtic case specifically, we’d have PIE arriving to each region of Western Europe (Iberia, France, British Islands and North Italy – not sure what he thinks about other places like Poland, the Balkans or Anatolia-) and evolving into different Celtic dialects through divergence from more eastern regions but mutual convergence throughout the said regions. Your knowledge of genetics can easily tell you that this sort of isolation-by-distance cannot work this way. You’d have a hard time turning PIE into a Celtic dialect in Iberia and another one in Great Britain or North Italy, simultaneously (even with France acting as a bridge). But even a harder one explaining the borders with Germanic and Italic.

    His theory is not specifically about Celtic, though. It would apply to Baltic and Slavic too (and every other branch, really), but the late attestation of these languages does not allow for an outlier to show his point.

    When it comes to Koch, I don’t think his ‘Celtic from the west’ has much value. It’s a one man’s theory (I suspect with some sort of preference behind it) that has never gained any traction given it’s implausibility.

    So I’m really not sure why you would consider yourself unqualified to have an opinion about this and therefor put the same weight in these theories as in the more traditional one of Celtic expanding in the Iron Age from Central Europe.

    “See, I think this is based on scaling what we see after by the Iron Age in Aquitaine and East Iberia up into a much larger sphere which we have no distributional evidence ever existed or much less recently expanded ”

    Here again you’re underestimating your knowledge about the genetic prehistory of Western Europe. You know well the colonization of EEF from the Early Neolithic and you know well the large scale replacement during the transition from the Chalcolithic to the EBA that took place by an homogeneous population (the R1b/Bell Beakers) with an origin in the steppe. It’s not reasonable to think that Bronze Age Western Europe spoke a large variety of unrelated languages. We have to main possibilities: that the Bell Beakers replaced the large majority of the Neolithic languages with their own (and if any survived initially, it would have been gone during the next 1500-200 years) – this would be the higher probability one. And the second, lower probability one is that Bell Beakers for some reason shifted from their own language to one of Neolithic origin at some point and then proceeded to replace the rest of Neolithic languages with the one they adopted.

    Both scenarios lead to a very low linguistic diversity in Bronze Age Western Europe. Then there are other possibilities, but their probability becomes too low too fast. For example, Bell Beakers arriving to each region of Western Europe carrying their own language but then shifting in each region to the local language. Since such a language shift has low probabilities of happening given the level of population replacement (almost total when it comes to male lineages), this happening in each region (or most of them, at least) is extremely unlikely given the way statistics work (you know, like having an option with 10% likelihood of being true and needing 10 or more of those low probability options to be true at the same time – you can do the math). And this is without even taking into account that if what we would be proposing is that the steppe was PIE and the origin of the IE language spread, arguing that such thing happened in the area where they arguably had the largest impact is totally unreasonable when you’d have to argue that they did impose their language (thanks to whatever mechanism they had to make it so successful – like feasting with heroic poetry, as Anthony suggests) in areas where they had no or little genetic impact, and no cultural impact whatsoever.

    “I have to disagree with that; It seems fairly less than preferable to postulate a >2500y divergence of languages, based on only the numeral system, if they appear to share little to no basic lexicon or grammar as far as the people who are in the most expert position to say view it. ”

    Here, on the contrary, it seems you’re overestimating your ability to judge such a complicated matter as the precise relationship between Basque/Aquitanian and Iberian. You would first need to have read all the recent research about it, which is mostly from the last 10 years and mostly available in Spanish. Had you done so, you’d probably think that while there are similarities in both languages far beyond the numerals (in fact, the numerals are a recent addition to a very long tradition of Vascoiberismo), you’d probably also admit the difficulty of having a high level of certainty about it given the quantity and quality of the data. So the conclusion would probably be that this is something that no one really knows for sure (but needless to say, aDNA clearly reinforces the argument of genetic relationship due to the difficulty that other scenarios would pose).

    (If you want a couple of papers with English abstract: here and here).

    In any case, if you can reference the people who are in the most expert position according to you, I will look into it.

    As for the numeral system, there are several reasons why the experts prefer the genetic relationship over the loan. And one of them is parsimony, which dictates that when you can’t be sure about something (like in this case the exact relationship of both languages as a whole) and need to make an assumption (in this case about the numeral system), you should prefer the statistically common option (the rule) over the statistically rare one (the exception). Nevertheless, you refer to the exception while invoking the principle of parsimony.

    (Finally, I hope this is not one of those rare occasions where we just have to agree to disagree -as it happens to me when trying to debate with epoch, for example- until more data comes out. But if it is, then probably just let me know and we can leave it there rather than making even longer a debate that won’t lead us anywhere).

63. Alberto: Appreciate the reasonable tone of your counterargument, I do think our ideas may probably be too divergent and firmly held here for too much further discussion. I’ll try and give a response below, it will be long and you may well think much of it totally spurious given different ideas.

    Alberto:Here again you’re underestimating your knowledge about the genetic prehistory of Western Europe.

    On your point about neolithic-chalcolithic diversity, I’m not totally sure about this. Dispersals from Anatolia may have involved groups from Anatolia which were structured in different language groups which are beneath the resolution of our ability to detect, and there are open questions about admixture scenarios in which European HG may have passed over their languages. (We don’t know enough about the fine scale dynamics; various different European groups are mostly EEF with minority HG, but if some were formed by augmenting HGs with a constant trickle of farmers with larger population sizes, they’d probably keep their HG language, etc.)

    From world wide evidence, it does seem like we do see expansions of farming through NE Asia, SE Asia and PNG where linguistic diversity of stocks survives early farming (in NE Asia, even after significant historically attested leveling linked to expansion of polities, farming is associated with at least two surviving isolates Japanese and Korean, in mainland and insular SE Asia with various different stocks that appear to have survived the transition to farming in South China).

    In terms of how much diversity there was by the chalcolithic I still mainly go by Donald Ringe’s estimate (http://languagelog.ldc.upenn.edu/nll/?p=980) on this subject:

    Before the arrival of speakers of IE languages in the Medi­terranean, the linguistic situation must have been even more diverse; a reasonable estimate would be more than thirty languages—possibly many more—grouped into more than twenty families belonging to at least fifteen different stocks.

    Of course the rest of the continent can be expected to have been somewhat less diverse linguistically, but only somewhat less. Given the number of areas that should have promoted modest diversity—the Atlantic coast, the Alps, the Balkans—it would be no surprise if the rest of the continent together exhibited a linguistic diversity similar to that of the Mediterranean region, with little overlap of families or stocks between the Mediterranean and the rest of the continent: perhaps sixty languages in Europe altogether, representing some forty families and thirty stocks . This is not an extreme estimate. Note that the archaeologist David Anthony, who is willing to contemplate “language communities” (isolated languages or families of closely related languages) spread over territories the size of Yugoslavia or even France, estimates that there must have been between twenty and forty such communities in Europe in the late Neolithic period (Anthony 1991:196-8).

    I think he would have known all about Cavalli-Sforza’s theories of a large replacement of population in Europe during the Neolithic when he wrote this, so it’s not coming from an “anti-migrationist paleolithic continuity before IE” stance. (Though when he says different stocks, I tend to take him in the mode of the linguistic mainstream that needs strong evidence through the comparative method to establish a stock and which is skeptical of deep reconstruction through restricted lexicon and mass comparison).

    Alberto: Your knowledge of genetics can easily tell you that this sort of isolation-by-distance cannot work this way. You’d have a hard time turning PIE into a Celtic dialect in Iberia and another one in Great Britain or North Italy, simultaneously (even with France acting as a bridge). But even a harder one explaining the borders with Germanic and Italic.

    Specifically on Garrett, I’d reference http://linguistics.berkeley.edu/~garrett/BLS1999.pdf directly. As I understand it, the argument is that changes which are identified as proto-Celtic on the basis of reconstruction from living languages are not actually shared by archaic but apparently Celtic languages (languages with enough Celtic features to be clearly not unrelated general IE) when attested in Western Europe. This pushes to higher probability the idea that you have a wider para-Celtic speaking community and linguistic changes which occur in piecemeal across it, rather a well defined sequence of innovations in a single small area of Western-Central Europe then expands.

    I may be underestimating myself but honestly I really don’t have a good intuition about if this is possible – comparable situations are any convergence of features among Romance languages, post-Rome, giving the appearance of “ancestral” features that are actually lacking in Latin; I know from Garret+Chang that this happens to some degree through homoplasy in the lexicon, but I don’t know if it is reflected in parallel independent sound and grammar changes.

    Also note, a sharp Germanic and Romance border can be explained by Roman expansion (which is a unquestionable event for which we have more than an archaeological horizon) and Germanic expansion…

    Alberto: For example, Bell Beakers arriving to each region of Western Europe carrying their own language but then shifting in each region to the local language.

    Within Iberia and France, it seems to me it really depends on what the local dynamics are, and I’m not sure we know about it enough.

    I take the view that languages tend to follow the majority of the community. So if the dynamics were about the augmentation over time, in a number of waves, into a linguistic community, of migrating individuals, whether they are male or female, high status or low, the language will tend probably not to change (cf Austronesian retention in far Oceania despite wave of Papuan males). If the dynamics favour the slow integration of local individuals over time into the migrating community, they will keep the migrating group’s language instead. (Others take a view that high status migrating powerful males will always “impose” their language, regardless of the numbers they are dealing with and whether they are moving into another community or others are moving into theirs, or that specific mechanisms of “elite dominance” associated with cultural sophistication change things. I think, at the moment, more that it’s pretty much a numbers game, and dependent on the numbers moving in each direction in any given generation, which are hard to tell if we have snapshots which are subject to burial bias and only known at a low resolution in terms of generations and in my opinion possibly losing low level admixture.)

    I think it’s plausible that you have much of the former dynamic, even if there is a bias towards males and towards high status for in-migrating males. The Bell Beaker burial package has a patchy distribution within Iberia, as well, and is not so clearly necessarily the cultural ancestor of all the later cultures (even if genetically contributed to them)?

    (Bell Beaker Blogger hints here at why he believes the distribution of migrating Beaker culture was what it was and what effects this may have had – https://bellbeakerblogger.blogspot.com/2019/04/murderous-mesetans-take-booty-olalde-et.html. Read his comment to Gaska also.)

    Alberto: Here, on the contrary, it seems you’re overestimating your ability to judge such a complicated matter as the precise relationship between Basque/Aquitanian and Iberian. You would first need to have read all the recent research about it, which is mostly from the last 10 years and mostly available in Spanish. Had you done so, you’d probably think that while there are similarities in both languages far beyond the numerals (in fact, the numerals are a recent addition to a very long tradition of Vascoiberismo), you’d probably also admit the difficulty of having a high level of certainty about it given the quantity and quality of the data.

    I can’t speak to having enough knowledge, but I will continue to hold to the null hypothesis that demonstration of a family relationship (particularly a recent <2500 year old family relationship) needs quite a strong argument from the comparative method and generally to expect quite a high convergence in basic lexicon.

    I still believe it would be less probable that there is a recent relationship and it is not fairly evident already after all the study that has happened to date. I don't think you're wrong to be interested in suggestions of grammatical similarities but this is hard to weigh up overall as to whether it provides any kind of evidence of a recent family relationship (I would note your second abstract does not seem to mention Basque specifically however).

    For my references on the current consensus on Vasco-Iberian, I found these works (all within last ten years I think although some may be older, and largely by Spanish authors) and previously provided to Kristiina: https://imgur.com/a/MfAGenB (it did not persuade her either!)

    On the number system specifically, I think there's a suggestion that the sharing of lexemes is also present in words due to urbanisation and trade ("money", "silver", "town"), and it seems like linguistic borrowing of numbers is not uncommon in situations where language contact is mediated by commerce or contact with a more commercially sophisticated culture: https://imgur.com/a/Uni7ZMI

    Re; parsimony, I will push back on that; I'm stating that if there is a family relationship, it is more parsimonious that *within the languages* it should be present in all categories of words within those languages, and particularly the basic ones that relate to universal constants of the human environment, not one particular class for exceptional, unexplained reasons.

    I agree it's less parsimonious on a world scale to presume a shared numerical system is borrowed rather than a family relationship because it is less generally the case and more rarely the case, but this is outweighed in my opinion by the absence of any attested languages at all which share a numerical system but no other lexicon, and yet are still agreed upon to have a recent family relationship. There just aren't any at all that I'm aware of. Should it just happen to be the case that this *never* happens in the linguistic record, but just happens to happen once in the case where we have limited evidence to test it?

    The main point is here that it has to be led by a consensus linguistic reconstruction of a recent family relationship between Aquitanian (and Basue) and Iberian. We can't be led by the idea that we believe there *ought* to have been a shared language that spread in the EBA in Iberia based on our interpretation of the genetics and archaeology, and then put together the idea of a recent Vasco-Iberian (even in Southern France and Iberia) without reference to whether linguistics generally reconstructs this! (Although in any case this wouldn't explain Tartessian unless that is also sharing the same low time depth family relationship). The linguistic reconstruction of a 2500 BCE Vacso-Iberian split must lead, not follow.

64. @Matt

    Thanks, I also appreciate the extensive reply and argumentation. I will say first of all that I can agree with your general attitude of being cautious and asking for detailed analysis of each case, for example when it comes to possible language shift with the arrival of Central European Bell Beakers in Iberia, where you say we’d need more resolution. Indeed, till we have such resolution, a probabilistic approach is the best we can do, but if you don’t find that good enough (even if overall statistics should end up being honoured with enough cases under study), that criticism is ok for me.

    My main issue may be that I just tend to see this sort of high standards when something may pose some risk (even collateral, as is the case here) for a specific hypothesis (about the IE homeland, which this post is not about) while complete flexibility is allowed for anything that may support it. Which is something that becomes very tiring after many years and many people doing it (and obviously this is not something I’m saying about you specifically, but about the issue in general). For example I really didn’t expect anyone to complain about my much more hand-wavy dismissal of something that could support the Anatolian Hypothesis (but hey, Marko did, which I kind of appreciate and had to say ‘Fair enough. Even with its problems, we can’t definitely dismiss that hypothesis and may need to revisit it at some point’). And fortunately no one ever though I’m anti-Anatolian (which of course I’m not, even if I find problems with that hypothesis due to its chronology, and overall think it’s probably wrong).

    With all this out of the way, and back to your comment, while as I said I can agree with the general critical attitude I have to disagree with some of the specifics. I’ll try to answer to those more relevant to the post in the next comment, and try to keep it informative rather than as any sort of counter argument.

65. @Matt

    (…continued)

    Regarding the current views on Basque-Iberism, unfortunately this is an ongoing research and therefor things are still a bit preliminary (even if the theory itself is very old), and it seems that the literature available in English is still a few years behind. For example, as said above Francisco Villar only changed his mind and accepted the genetic relationship between these languages as the most parsimonious explanation in 2014 (in a book I quoted above in the first note marked with a red asterisk). This book is not available in English, and it may take time since it permeates into English literature (especially books) as an important opinion from someone prestigious and without interests in the subject.

    In the same way, the book in English that you quote (and which is published in 2019) has this chapter about “The Vasco-Iberian Theory” written by Eduardo Orduña, but I don’t know when that was originally written. E. Orduña is responsible for first studying the numerals and finding the connection, and indeed his thoughts back then were that this was a loan. However, as years passed and he continued to deepen the study of both languages, he had to change his position and prefer a closer relationship (genetic) between them. Which is important, since this does not come from someone who started trying to prove a relationship, but it comes from someone who after dealing for years with the subject had to change his mind and accept what he initially rejected. I’ll quote (and then translate) some lines from the beginning of this paper by him from 2013:

    “ 1. ¿PRÉSTAMO O PARENTESCO GENÉTICO?

    En un trabajo reciente (Orduña 2011) me he manifestado, en contra de lo que afirmé en mi propuesta inicial (Orduña 2005), en favor de que la semejanza de los numerales léxicos con el vasco se explica por parentesco y no por préstamo. Tiene razón de Hoz 2011, 198, al afirmar que esta posibilidad, la preferida por Ferrer i Jané desde el primer momento, (Ferrer i Jané 2009) es más coherente, y es cierto que es en principio lo más probable, aunque el préstamo de todo el sistema no puede descartarse por completo, ya que sería lo normal si una de las dos lenguas tuviera un sistema numeral limitado a unas pocas unidades, o incluso inexistente.

    Pero no es sólo la gran similitud entre ambos sistemas, con coincidencias casi exactas en todos sus átomos, la que obliga a pensar en el parentesco genético: tenemos además una serie de coincidencias en la morfología nominal demasiado precisas para atribuirlas a fenómenos de área, de algunas de las cuales hablaremos aquí, además de otras coincidencias en la morfología verbal, que habrán de ser objeto de un trabajo posterior.

    Las coincidencias léxicas, aparte de los numerales, son extremadamente escasas si nos limitamos a las que podemos considerar probables, pero es notable la cantidad de posibles coincidencias que de momento no podemos utilizar por la falta de un contexto en ibérico que nos dé alguna pista sobre su significado, una cantidad sin duda muy superior a la esperable por mero azar, si observamos que las coincidencias con lenguas que poseen un sistema fonológico similar, como el castellano, son muchísimo más escasas.”

    My translation:

    “1. LOAN OR GENETIC RELATIONSHIP

    In a recent work (Orduña 2011) I have manifested myself, contrary to what I said in my initial proposal (Orduña 2006), in favour of the similarity of the lexical numerals with Basque being explained by relationship and not by loan. De Hoz (de Hoz 2011) is right when he says that this possibility, the one preferred by Ferrer i Jané from the first moment (Ferrer i Jané 2009) is more coherent, and it’s true that in principle the most probable, though the loan of the whole system cannot be completely disregarded, since it would be the most natural thing if one of the two languages had a numeral system limited to a few units, or even non-existent.

    But it’s not just the similarity between both systems, with almost exact coincidences in all of their atoms, the one that forces us to think about the genetic relationship: we moreover have a series of coincidences in the nominal morphology too precise to attribute them to areal features, some of which we will refer to here, apart from other coincidences in the verbal morphology, which will have to be left for an upcoming work.

    The lexical coincidences, apart from the numerals, are extremely scarce if we limit ourselves to the ones we can consider probable, but it’s notable the amount of possible coincidences that we can’t use for now because of the lack of context in Iberian that gives us some clue about their meaning, a number without a doubt very superior to what could be expected by mere chance, if we note that coincidences with languages that have a similar phonological system, like Spanish, are much more scarce.”

    So it seems that the text published in that book from 2019 was written before 2011, which is quite unfortunate.

    About Garrett’s article, that’s the one I intended to link to (but now realised the link was broken – now fixed), and my comments referred to it. I will elaborate on it a bit more since there are interesting linguistic aspects to consider, but it will have to be tomorrow since it got late now for me.

66. Matt;
    I think its important we understand the fundamental aspects of population history before moving into more difficult terrain (such as language dynamics). So I’ll stick to population dynamics & biocultural anthropology, which is the subject of my PhD, rather than straying into *comparative linguistics, loans, and numerals. I think the dynamics in late Chalcolithic Iberia are particularly interesting and a synthesis of all the data is yet to emerge, and this might be difficult because only a handful of individuals around the world might be capable to grapple with the multiple lines of evidence.

    If one looks at the samples, there are also incoming elite females from central Europe. For example there is one of the earliest steppe-admixed/ central European BB -Iberia samples, I0461, who is a female. There is also I6626 a female with rich BB set, not making the cut for analysis in Olalde, but can be shown to have ~30% steppe ancestry. So we have groups arriving-males, females, children, family units – not just ”male elites”, and their social position was communicated via males & females, adults & children. They positioned themselves in Meseta, then from there expanded further south. We can hypothesize freely as to this might have effected language shift, but we should at least get the basics right to help guide hypotheses. So, as nice as it might be to think about it, a ”male stud farm” theory misses the mark. I would also not hold much weight to the view of cultural continuity across littoral Iberia. Indeed, BB Blogger (whom you quote) is quite aware of the radical changes occurring in Iberia c. 2500 BC, because he once blogged about it.

    Regarding the distribution of the BB, ”patchy” isn’t the correct descriptor (although such a ”patchy” pattern is seen with other cultural groupings – rarely are they as confluent as some maps might indicate). In any case, the areas of BB settlement are best viewed as key nexi and loci for further expansion. Subsequent Bronze Age cultures in Iberia certainly do ”descend” from BBC- this has been understood for a long time: proto-Cogotas in central-northern Iberia derives culturally from local variety of BB styles; as does the Bronze Age in Portugal (Atalaia horizon), Galicia (Montelavar -Atios horizon). The cultural situation in El Argar might be less clear, as Frank has expressed his concerns earlier, but in reality the hallmarks of that culture – a move to single/ elite male representation, hallberds, daggers are symptomatic of BB descent; even if the BB ceramics themselves were soon dropped from the repertoire (I don;t think that really changes anything). However, I agree some more data from key El Argar sites ould help clarify our understanding.
    IN the meantime, you might find this is a quite a good book, the chapter on Iberia is easily digestable, and how the Bronze Age groups derive from preceding Beaker horizon, but then set along a path of their own trajectories, which certainly dovetails with some of the linguistic proposals outlined in the post.

    As a final point regarding Ringe’s suggestions, I certainly agree that pre-BB western Europe would have been linguistically diverse. However, pre-aDNA era guesses are just that; and the geography-based hypotheses have been shown to be incorrect (e.g. the ”Caucasus as a barrier” to linguistic expansions). We can now more clearly understand the cultural-demographic blocks in Chalcolithic Europe, and whilst not perfect or clear-cut, these can then form the basis of hypothetical linguistic communities.

67. Alberto: My main issue may be that I just tend to see this sort of high standards when something may pose some risk (even collateral, as is the case here) for a specific hypothesis (about the IE homeland, which this post is not about) while complete flexibility is allowed for anything that may support it.

    I have to agree that does often seem to be the case with many folk, particularly when treating the branching of the Anatolian and Tocharian groups. In case of those languages, there is often little evidence given to when and where and how these languages could have branched and whether any of this is particularly plausible in light of adna. As well as fair number of blithe assertions about “consensus” of time of separation of Anatolian (which doesn’t exist when you look at the primary sources given as reference), often lack of acknowledgement of Anatolian branches besides Hittite (put down to being “imposed by an elite”), arguments for Afanasievo-Tocharian that boil down to “nothing else is compatible with a steppe hypothesis, so it must be them” (more or less what Koch recently said during a recent presentation), extraordinary substrate effects to explain high lexical or grammatical divergences, and so on – despite the fact that these are precisely the early branches that must be the most informative about the IE homeland!

    even if I find problems with that hypothesis due to its chronology, and overall think it’s probably wrong

    Yes, I think the chronology and distribution of IE is the main problem (more weightily than the “linguistic paleontological” reconstruction of IE homeland from linguistic evidence primarily from the subset of IE languages that probably did expand late within a European biome and which have potential problems with homoplasy).

    As well, the Renfrevian idea of a single “farmer language” and farming linguistic expansions is doubtful to me along the lines of the same reasons as Ringe’s comment. Renfrew relies a lot on Bantu and Austronesian evidence as arguments for farming language expansions, but these seem more like examples of people who had relatively long had farming going on a tear of rapid expansion (probably connected to other new technology), in the case of Austronesian to largely essentially unoccupied territory.

    Re: Orduna, the quote I’ve underlined was more his summation of the field as a whole – note that he is reasonably sympathetic to “moderate Vasco-Iberism” (the viewpoint that there is some genetic relationship between the languages, but Iberian is not ancestral to Basque) even in 2011, rather more sympathetic that some other sources. So I don’t think it is necessarily him having a volte-face on his personal view between 2011 to 2013.

    However, I would note again that even if there is some genetic relationship between the languages, to be anything with Beaker, it has to be relatively recent to the time of Iberian and Aquitanian to be much associated with the Bronze Age turnover in Iberia, and even that doesn’t get us to a wider distribution in Western Europe associated to the BBC outside Southwest France and Eastern Iberia, without suppositions about the intrusive group never switching their language (because of male bias or somesuch) which I don’t see as ever due to get widespread linguistic acceptance (and in any case, within Iberia would be falsified by Tartessian, should it not, as seems current status, be related to Iberian or Aquitanian).

    @Rob, certainly I think females were present in some proportion in the migrating waves and it’s too simple to call it a male only migration exactly (if there ever any significant examples of such at all through all history, strictly speaking).

    Regarding words, yes, “patchy” may not be correct (implies much discontinuity rather than what I intended, to describe a complex distribution, not in a “bloc”, that is not hegemonic across the whole territory) and by mentioning “cultural descent” I have probably added heat rather than light, as it is probably a fairly meaningless couple of words, certainly within the ability of archaeology to determine from pottery continuity etc.

    We can now more clearly understand the cultural-demographic blocks in Chalcolithic Europe, and whilst not perfect or clear-cut, these can then form the basis of hypothetical linguistic communities.

    This is pretty optimistic, I’m pretty doubtful we’ll ever get right what groups should have been speaking the same or a set of different languages on the basis of archaeological and adna evidence – certainly we’ll never be able to test any of the ideas so potential to easily go wrong and manufacture huge scale phantom linguistic areas and just-so-stories that play into particular narratives is hard to error correct (unlike adna they’re not making new evidence, and we’ll never realistically be able to recover any of the languages spoken from present day language).

68. @ Matt
    Well, I guess one can be quietly optimistic if theyve studied the data for 20 years and has been lucky enough to corroborate with experts in the various disciplines 😉
    Linguistics, archaeology, anthropology, aDNA are corroborative in the study of human prehistory and it’s easy enough to decode what’s feasible and whats unlikely.
    As more data accumulates and we move into more recent periods, it would be even more complex.

    ______
    Speaking of which, this paper looks interesting https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3346985

69. Thanks to both @Matt and @Rob for the well thought comments.

    @Matt

    I still feel the need to further clarify the part about E. Orduña. His initial proposal when first writing about the similarities he found in the numerals was that they were probably a loan. This was in 2006. Then in a later work from 2011 he admitted to have changed his position about it, favouring a genetic relationship, which he further confirms in the paper I quoted and translated above, from 2013.

    In the excerpt of the book that you quoted, he does summarize the general view by saying that “Iberian and Basque were different languages, with no close genetic relationship“. He then mentions that some scholars still accept some sort of relationship (a rather vague one with some sort of “affinity”) between both languages, calling this “moderate Vasco-Iberism“. While ‘classic Vasco-Iberism’ would be mostly absent from scholarly literature and relegated to ‘pseudo-academic works‘.

    This seems in sharp contrast with what I quoted above, where he very explicitly expresses the necessity of a genetic relationship due to the very precise coincidences in nominal morphology and others in verbal morphology. So indeed there was a very clear change in his view of the relationship between Basque and Iberian from 2006 to 2011, further consolidated in 2013. And this is not just him, but a general trend, that being quite recent it will still take a few years to permeate into more generalised literature and appear in English too.

    (I may finally add, that seeking for feedback regarding these issues I sent him this post, which he was kind enough to read and comment -on personal communication- that while he can’t say anything about genetics, the linguistic part of the post looked “very correct” to him, which was quite a relief for me who am no expert in the matter, and for which I thank him again here).

    …to be continued, about A. Garret’s ideas.

70. Regarding Andrew Garrett’s ‘A New Model of Indo-European Subgrouping and Dispersal‘ (Link), there are interesting things to comment about it. The article is short enough (and accessible) for anyone to read it, so I won’t summarise it here in detail. I’ll try to go directly to the part I want to comment about.

    While he argues about the non-existence of proto-languages for the IE subfamilies, he only finds specific evidence for Celtic, Italic and Greek. But he argues that the reason why he can’t find evidence for the other branches is just the lack of early written records that would have allowed to find the corresponding evidence:

    “If we apply what we learn from cases where there is evidence to those where there is none, it follows that the Indo-European family tree with a dozen independent, highly distinctive branches is nothing more than a historical mirage.”

    So for example, Slavic languages would not be the result of the expansion of a putative Proto-Slavic language, but instead it would have formed from several dialects (initially not any different from those that would become Celtic, Greek or Germanic) in contact with each other, and these contacts is what would have made them evolve into a specific IE subfamily. The reasoning applies to all there branches, not just Slavic.

    However, there’s no mention to Romance languages (which have a similar chronology to Slavic ones) in the article, even if we could have an outlier in Romanian to show the same sort of evidence that Lusitanian provides for Celtic, or Venetic provides for Italic. The reason is very obvious: was him to doubt the existence and expansion of Proto-Romance (also known as Latin), his whole article would become ridiculous. However, and ironically, it’s a pity he didn’t go all the way into Romance languages, since there’s where he would have found the real value of his stance.

    I already commented in a previous article about the problematic that has puzzled linguists for a very long time about Romance languages: the fact that they share features not present in Latin. Features that being related to morphology are usually considered diagnostic for genetic relationship. Which even brought them to hypothesise that Latin as such never expanded, but it was already an archaic language only used for writing at the time of expansion, and Proto-Romance would have been a different dialect talked by the people. A proposal that never worked for well documented reasons. And there’s still no consensus about this problem as far as I know.

    And it’s about this that Andrew Garrett has the right answer in his ideas about areal features and language convergence through contacts. That is’ he is right about contesting the linguistic paradigm that areal features are just trivial ones and arguing that they can (potentially) permeate languages to the core features. And this is really the take away from his article, rather than trying to argue a generalised non-existence of proto-languages for every IE branch, something that might be arguable for Greek or even Italic, but seems totally unrealistic for expansive branches such as Celtic, Slavic or Romance, where the geographic extension and very evident shared features make arguing for a non-genetic relationship a linguistic suicide (in the case of Romance languages, I mean even if we didn’t know about Latin and the Roman empire was just an archaeological hypothesis). Though on a more personal note I should say that I can’t speak so much about Celtic, which I have no knowledge about, so there I just follow what I believe to be the expert’s opinions.

71. Sorry for the late reply.

    @Marko

    Thanks to my laziness I don’t have a write up yet, unfortunately, but maybe after I touch up on a few things outside of the Greco-Aryan scope.

    Though, you can get the gist of some of my ideas from comments around the blogs. I mostly critiqued the unlikely Khvalynsk scenario and the Andronovo I-I hypothesis and few things were just to get reactions out of angry people.

    Pure comparative models do not work because archaeology is needed. Every PIE model is somewhat based on weak comparative arguments that are always nitpicked, eg. see Kuzmina for some mental gymnastics. Speaking of Hittite, yeah sure there are some influences from non-IE religions but at it’s core it’s still mostly IE.

    @Rob

    By “late” I mean what is considered late by the mainstream, 1500 BCE.

    In order to find how IE spread to Europe I was thinking of something along the lines of what Alberto wrote above. I think the traditional tree model doesn’t work in respect to Italic and Celtic, and even Balto-Slavic and Indo-Iranian. For example, culturally Celtic shares much with Scythian and even Vainakh while Italic shares some features with Anatolian.

    For any Out of Europe Hypothesis, the problem isn’t Anatolian but Indo-Aryan. An explanation is needed as to how IA warlords penetrated the Zagros around 2200 BCE, perhaps had a city built for them in Mesopotamia in 1860 BCE, became a military power by 1760 BCE, and ruled several empires by 1500 BCE. There is not a single artifact that can be related to Andronovo (I am guessing your scenario follows the regular Corded Ware one). The argument that some Steppe traders managed to change the language of the advanced BMAC in 2200 BCE is ridiculous. There is no sign of conquest in the northern part of BMAC and in fact most of the conflicts in Central Asia are west of the Helmand.

    I’ll read the paper Alberto linked and maybe I’ll change my mind.

72. @all

    Off topic, but an important paper is out with samples from around Estonia here.

    The post about it won’t be immediate quite probably. My first and very quick look at it raised some flags (not that it’s the first time it happens, anyway), so I’ll take my time to digest it and comment with Kristiina to see about the post. Stay tuned for more info about it.

73. @ Vara

    Yes I see. I agree with you about an MLBA expansion of IE in Europe, that is my current feelingy. And also, based on the preliminary data in Narasinham, it looked as if the required chronology was looking just a little toight (although one can never be 100% sure about timing of languages splits or composition of Vedas).
    But somehow, the fact that Sanskrit is very similar to Balto-Slavic needs to be explained, as does the sharing of R1a-Z645
    As for BMAC, new excavations have shown how things might have ocurred. It was not a big face off between ”BMAC vs steppe”, but a decentralization of BMAC, then a series of individualised interactions between pastoralists and post-BMAC agricultural communities , varyign form interaction to avoidance. The steppe camps seem to have spread around & in between the post-BMAC villages. So youre right – no sign of conquest because there wasnt any
    BTW which is that city you mentioned ?

74. Alberto and Atrior have talked about this before regarding BS and IA, if both came from the PC steppe then you would expect BS to be closer to IE languages in NW Europe than it is to IA.

75. @Al Bundy

    Yes, that’s right. I’ll take the chance to summarize my position regarding the genetic vs. contacts relationship of languages.

    Leaving aside the cases where a genetic relationship is very clear, there are many other cases where things are much less clear. In my opinion, languages should be studied in their geographical and historical context, and not just in abstract. This is fundamental to get realistic analyses.

    I’ve talked before about the relationships between IE, Uralic and Altaic. There are obvious similarities that has led linguists to propose a genetic relationship. However, ancient DNA makes it clear that a genetic relationship is extremely complicated. We’d need to find a population (Proto-Eurasiatic) that expanded/migrated in different directions to become respectively the PIE, Proto-Uralic and Proto-Altaic communities, which would in turn expand respectively creating the modern distribution of these languages. Chances for finding a candidate for such Proto-Eurasiatic population are basically zero.

    On the other hand, we have ample evidence of the extensive (in space and time) and intensive (bi-lingual or multi-lingual confederations with a lingua franca that would change depending on the dominant groups, language shifts, etc…) contacts between Uralic, Altaic (mostly Turkic) and IE (at least Indo-Iranian) that can easily explain these similarities (even if some linguists think that some of them should be diagnostic for genetic relationships rather than areal features – something that should be phased out due to the evidence against it).

    Regarding the relationships of Balto-Slavic and Indo-Iranian it’s a similar situation. A putative genetic relationship (beyond the general IE one), would require it to be related to the R1a-Z645 ChrY connection, with a split in Eastern Europe somewhere in the CW horizon at a time when pre-Germanic, pre-Celtic and pre-Italic would also be starting to split from that same area, or even earlier. Indo-Iranian travelling all the way to Kazakhstan and then south to Iran and India would result in it being a clearly divergent branch from the “European” ones, that would form a tight cluster to the exclusion of I-I.

    But this is not what we see. Instead we see many close similarities between Balto-Slavic and Indo-Iranian, with quite lesser ones in Germanic and absent in Italic. So do we have an alternative explanation based on the historical and geographical context? Yes, we do. We now know with certainty the place and time of origin (Central Asia c. 1500 BCE) of the Scythians, their genetic and cultural influences, and their migration west to modern day Ukraine, bringing with them an early form of Indo-Iranian (maybe Iranian, but closer to Sanskrit than to any known Iranian language), and being in direct contact with the putative Proto-Balto-Slavic community (that can’t be placed before 1000 BCE, so maybe the Chernoles Culture would qualify by the time and place as early Proto-Balto-Slavic, but it might have been some other) for over 1000 years. Which clearly explains the similarities that a genetic relationship can’t (and I may add that the area also had early Greek colonies, for those interested in the cultural similarities between early Slavs and ancient Greeks).

    Nothing of the above preclude in any way that IE spread to SC Asia from the steppe. It just explains better the relationship between B-S and I-I.

76. @Alberto @Rob
    There is also the relation between Iranian and Thracian. Thracian culture has a lot of influences from Skythian. Maybe that period is when Satemization began spreading from Iranian-Sanskrit to Balto-Slavic and Balkan languages. It makes sense if you think that Iranian and Sanskrit appear as fully satem, while Balto-Slavic and Thracian are less so and Albanian, Phrygian and Armenian even lesser (they actually lie in the middle between centum and Satem). There you have a radius from the center of Satem phenomena (Skythian-Sanskrit) ) to the periphery area.

77. there are two clear dates for sanskrit

    2900 BC for the statement that krittika never swerves from the east in the shatapata brahmana

    https://nileshoak.wordpress.com/2014/02/13/nakshatra-krittka-rising-due-east-part-1/

    4500 BC as the last possible date for the AV observation in the mahabharata

    https://nileshoak.wordpress.com/2014/02/03/arundhati-vasistha-av-observation-of-mahabharata/

    please note this observation is independent of whether you believe mb war actually happened or not

    specifically only the vedic sanskritic civilization of india has a record of this astronomical event and preserved it from before 4500 BC to now showing continuity of the civilization

    there are other pretty solid datings of edits of sanskrit texts but which do not necessarily imply dating of composition that you can see in the table in this article

    http://indiafacts.org/can-veda-shakha-pravachana-rescue-bharata-itihasa/

78. In a recent comment by Claude Brixhe I came across, he also mentions this about Thracian (which I also took to have undergone “satemization”) and its relationship to satemization in Handbook of Comparative and Historical Indo-European Linguistics:

    “Βενζι for Βενδι teaches us (a) that the spirantization attested in the anthroponyms in -ζενις (Greek -γένης) given by Greek sources is old, but that it has not been generalized or that the memory of the original form is still present (cf. [Βεν]δει) and (b) that this phenomenon is highly unlikely to illustrate the satem character of this language. Rather, it may represent simply a palatalization (at least of *d/g) before a front vowel.”

    With regard to Balkan branches, Greek is clearly centum, Phrygian (which shows very strong ties to Greek), Thracian and Illyrian quite possibly centum though various opinions have been offered, Armenian and Albanian either incomplete or indeterminate/uncertain but they’re also branches attested late and having undergone many changes.

    In general, satemization doesn’t seem to have affected the Balkan branches much and seems strongest in the IE proto-groups plausibly bordering the forest steppe area.

    Even if one doesn’t go so far as to adopt Carlos’s scenario about a Uralic CWC, which I’m very skeptical about too, there are some questions about what R1a and the forest steppe groups represent at least in Indo-Iranian (vs Z2103 which also appears a fair amount, especially in Iranic speakers) which, aside from sharing some late features with Balto-Slavic like ruki and satem, is usually taken to be most related to Balkan branches.

79. @ Aigesti
    Hello, yes I think that’s possible. The influence of Iranic -speakers in Balkan are well known with Scythians, Srmatians, etc and might have been present since the MBA (e,g, Merichleri), trading some cultural items, even if they did not make a big genetic impact

    @ Egg
    Z2103 and Z93 appear to have different patterns of spread – the former via (Catacomb-Hurrian horizon), now peaking in Caucasians & modern Turks (with ~5% in Iran, & < 1% in India); the latter via central Asia (via Sintashta-Andronovo -Swat horizon)

80. Rob, I believe some East Iranic groups like Yaghnobis have good amounts of Z2103, more so than West Iranic speakers, same with some neighboring Central Asian groups. This would make a spread (solely) over the Caucasus less likely unless we think of it as some sort of founder effect in Central Asia. Either way it brings up some potential questions about the interplay of R1b steppe and R1a forest steppe groups in the possible formation of Indo-Iranian, unless we simply consider it an older expanding lineage, like all the way to Afanasievo, that survived in later groups.

    But in that case, why does Indo-Iranian seem more related to Balkan IE which plausibly seems characterized as initially expanding with typical Yamna lineages, so Z2103 and I2a2, rather than Balto-Slavic other than some late features according to linguists? Obviously the situation isn’t 1:1 with language and genetics and linguistic arguments can be very uncertain at times, but it seems weird that the post-CWC R1a forest steppe groups that ended up dominating the steppe as well should have been the ones to spread Indo-Iranian without some sort of typically R1b-Z2103 steppe related input or influence.

81. @ Egg

    Yaghnobis are a small, drifted ethnic minority. If we rely on them to prove on the connection between Z2103 and I-A, then we’re obviously reaching for rabbits out of hats 🙂 In any case, as per the Wells 2001 study, their high R1b is of the M73 type ?
    So when I suggested that Z2013 in the Near East is coming from Catacomb, over the mountains, I was referring to what accounts for the majority of cases; and also simply following the evidence we have so far (rather than, no offence, but what is a fringe theory that Z2013 = Indo-Iranian)
    But you’re right, some of the Z2013 from Afansievo surely lingered on further east after the culture dissipated as Okunevo expanded into its territory.

    ”But in that case, why does Indo-Iranian seem more related to Balkan IE which plausibly seems characterized as initially expanding with typical Yamna lineages, so Z2103 and I2a2, rather than Balto-Slavic other than some late features according to linguists?”

    The samples we might solidly link to putative IE speakers we have so far from the Balkans (albeit its a handful) are neither I2a2 nor R1b, and neither are the Anatolian Bronze Age-era individuals (although they are potentially non -IE).
    Wr.t. the position of Balkan IE tree, I’m not aware of any such definitive positions, the problem being that, apart from Greek & Phrygian, Balkan IE langauges are so scarcely attested that any definitive theory is preclusive. In any case, Ringe & Tornow have recently positioned I-A clsoer to Ba-Sl than to Greek . It really depends what theyre looking at (lexis, morpho-syntax) & what method they emply, right ?
    You might wish to refer to the commentary b/w Aigesti & myself as to why there might be I-A influences in the Balkans.

    ”Obviously the situation isn’t 1:1 with language and genetics and linguistic arguments can be very uncertain at times”
    I agree, which is why the discourse of whether R1a or R1b (or even both) were ”the original PIE” are problematic.
    But for more defined & linguistically circumscribed expansions, it is not too proposterous to make such connections (e.g. Z280 is Balto-Slavic; or some R1b-L51 expanded with Celtic; so forth)

82. @Egg. Thracian is considered a Satem language. That is a consolidated opinion and there are even some linguists lately who consider it as a part of Balto Slavic group. Illyrian and Phrygian lie in between, with both samples of Satem and Centum, just like Albanian and Armenian. That should give us an idea of the Satem phenomena. Strong in the center, less so in periphery.

83. @Rob

    The chronology is a big problem atm if we follow the mainstream scenario for IA. If we follow aDNA, we get that the Rigveda had to be composed by time travelling priests who were living in the bronze age but somehow born some time between 1200-800BCE, and that of courses contradicts the linguists and archaeologists. Basically, linguistics, archaeology and aDNA aren’t matching because the theory itself is full of fallacies. Ah, if only these fields weren’t being led on by obfuscations and lies.

    Yes, the pastoralists started pouring in during the decline of BMAC due to climate change some time between 1700-1500 BCE forming the hybrid cultures. It is plausible that language change can happen during such times. However, the hybrid influence did not reach the Zagros (nor did Andronovo) and not only that but it’s a few centuries too late according to the same mainstream theories. There is a reason Iran, where the Mitanni are supposed to come from, is being ignored in all these papers and only Swat and BMAC are focused on, no? So, me being me, I brought it up to the “consensus” club on eurogenes, the same people who think that Celtic is older than Hittite because BB started 2800 BCE while Hittite is attested a thousand years later, and after realizing that none of them actually knows the consensus they preach about Davidski proposed that because there is an outlier in BMAC that means that a few somehow changed the language of Northern Iran. A very weak theory, basically.

    It’s funny how East Europeans on the internet want Balto-Slavic to be grouped with Indo-Iranian so bad while the Russian linguists go after Greco-Aryan. Alberto pretty much covered the most likely scenario above. While I doubt the proposed close relationship between Indo-Aryan and Balto-Slavic and opt for a Middle Iranian one due to the similarities in folklore, It is highly likely that a conservative Indo-Iranian group or even an Indo-Aryan one was a part of some Scythian confederation. In the NW Caucasus we have an Old Indo-Aryan hydronym, the Kuban River (RV; Kubha), and in the Assyrian texts some Cimmerian names are can be read as a mixture of Indo-Aryan and Iranian names.

    The city is Bagdatu, which doesn’t have a clear etymology in any of the languages spoken in the region in the 1800 BCE.

84. @ Vara
    I guess I’ll have a clearer impression when the final version of Narasinham’s paper comes out
    I recall that the linguistic supplement in Damgaard argued an LBA I-A entry, although naturally some disagreed
    I recall asking Frank the same question; but he didn’t get back to me. From when exactly is that his Baghdad attested, from which source (? Assyrian manuscripts), and is it consensually accepted that it is Indo-Aryan etymology ? (Links if possible)

85. @Rob

    So the timeline is as follows (independent of the RV):

    Witzel and a few others propose that some I-I groups reached the Zagros sometime around 2200 BCE, based on a few Gutian and Lullubi names and terms. I think the arguments were the name of the neighbors of the Gutians, Tukrish being like IA Tugra, Lullubi word for god; Kiurum or Kurum from I-I *Kura, Sumerian word for horse and seal…etc. Some older arguments that claimed IA were already ruling Hurrians around 2300BCE are far from convincing though. These guys probably did not make a big impact.

    Bagdadu is attested in a text during Hammurabi’s reign and the biblical scholars follow the Aramaic etymology which wasn’t spoken there during that time. However, there is no consensus, yet. The proposed reading Hudadu is no longer accepted because it was later mentioned during the Kassite Era as Bagdadi and later Assyrian rule. Check Altbabylonische Rechtsurkunden for the original text.

    Maryannu warriors appear 1760 BCE in the Tell Leilan texts.

    After that you pretty much get Indo-Aryan names all over the Near East.

86. @Vara

    We hear about people like Herodotus’ Syginnae, which almost certainly are Indo-Iranians. Hence beside Scythians and potentially Cimmerians, we see early IA-historical influence in Europe, as far as the Danube region.
    The influence of these people on local earlier IE or non-IE people in the region should not be neglected. From linguistics, to Y-DNA down to folklore.

    In terms of Y-DNA special care is required on these Steppe related people. I see a significant, mainly technology driven elite dominance effect and extreme patriarchal founder effects. This goes so far as Herodotus describing the Royal Scythians as quite different to other Scythians and them regarding the others as slaves. They also had many slaves, blinding the males.

    Hence caution is required, knowing about these quite extreme conditions.

87. Rob, Vara: Aside from Bagdatu, we have Akkadian targumannu “translator” (-> Engl. dragoman, Fr. trucheman etc.), first attested around 2,100 BC , and generally believed to have been borrowed from IE. Vermeer 2007 opts for a Hittite origin, but Starke 1993 had already questionned that connection for lack of attested Hittite parallels, and instead rather proposed Luvian as source (similarly also Jahr 2011), However, he also concedes that the presumed PIE root *terg “to turn” is so widespread (Lat. torquere “to twist”, Germ. drechseln “to turn wood”, Ven. , Alb/Slav/NGerm torg, all “market”, etc.) that any IE language might have served as source [On the semantic connection between “turn” and “translate” c.f. Lat. verso, Engl. to convert].

    The most intriguing case IMO however is Old Sumerian (26th-23rd cBC) GAR(a)-PA-TE-si “city-state governor/ruler” – a very transparent borrowing from PIE * ĝhrdhó – poti- “settlement lord/master” (c.f. PIE *wiḱpótis->Sanskr. viśpáti “chief of a settlement or tribe”, Lit. viēšpats “lord”; PIE déms pótis “house master”->Engl. despot).
    https://www.academia.edu/3592967/Euphratic_-_A_phonological_sketch

    Both examples do not only attest early presence of IE speakers in or near Mesopotamia. They are clearly superstrate terms, and demonstrate that those IE speakers must have been of considerable political and/or commercial relevance.

    On Bagdatu , let me add that bogh “god, master” is a specific IA term, unattested for in Anatolian. It is otherwise only found in Slavic, which is believed to have borrowed it from Skythian.

88. Alberto: As concerns Garrets paper (thx for the link), let me add the following:

    1. The question of Proto-Germanic, and the respective homeland, is still unsolved. Proto-Germanic is defined by the Germanic sound shift (Grimm’s law), and was traditionally believed to have been spoken around 500 BC. However, recent research has demonstrated that the sound shift was still on-going during the early Roman Imperial period. The Waal, one arm of the Rhine delta, e.g., is named vacalis by J. Caesar, to appear as vahalis in Tacitus. Similarly, Roman writers still speak of the C(h)atti – they only became Hassi (Hesse) some time during the 2nd-6th century. This, however, is much too late to reasonably explain the considerable differentiation Germanic had reached by the early 7th cBC (Gothic, Old Norse, Old Saxon, Old English, onset of High German sound shift etc.).
    So, apparently, the situation is comparable to Garrett’s assessment of Proto-Greek: What has traditionally been believed to be a defining innovation – in this case Grimm’s Law – , seems to in fact have been a gradual spread through a pre-existing dialect continuum.
    If P.Shrijver is correct in explaining Verners Law – an other constituting innovation of Proto-Germanic – by Uralic speakers switching to Pre-Germanic, matters get even more complicated. In this case, we must look for the Germanic homeland somwehere in the SE Baltics instead of, as commonly assumed, either the Danish Isles or the Jastorf culture. This is archeologically problematic – the Oxhöft culture and the subsequent Wielbark culture (the latter commonly associated with early Goths) only reached the Vistula during the 2nd cBC, and never extended much east of it. Therefore, a completely new and different scenario would be required – with the side effect of reaping Baltic languages their assumed homeland since CW times …

89. @FrankN

    Very interesting. Do you think the influence of IEs inhabiting Anatolia might be sufficient to explain the superstrate? If not, which of the CA/EBA cultures in Mesopotamia could possibly have been IE?

90. Marko:
    Before answering your question, let me add a few more details on GAR-PA-TE-SI. Here, you can see that the term isn’t made up, but well attested in Sumerian -especially the second part, written PA-TE-SI, but usually transcribed ensi.
    http://psd.museum.upenn.edu/epsd/e1347.html

    Important in this respect is the existence of two variants: GAR – PA-TE-SI, and PA-TE-SI – GAR. In Sumerian, as in Semitic (Akkadian), Hurrian, and Elamite, the qualifier (adjective) follows the specifier (noun), resulting in constructions like Ur-Kesh “City of Kesh (Kassites?)”, Ur-Ikid “City of Akkad(ians)”, or, for Semitic, K(a)rt-ago “City-new”, Thus, PA-TE-SI – GAR is a legitimate Sumerian construction, and if only that form was attested, we could assume PIE *potis to have been borrowed from Sumerian [PIE * ghrdhó “enclosure, fort, settlement” is anyway an “areal term” with a/o parallels in Semitic k-r-t “city” and Kartvelian kart “cattle enclosure”.] However, the GAR – PA-TE-SI variant is morphologically alien to Sumerian/ Semitic/ Hurrian. It instead follows the PIE qualifier-specifier sequence (c.f. New York City, Novgorod, Ankara, Srinagar etc., and the -potis constructions cited above), and as such betrays foreign, most likely IE influence. [Kartvelian also has the qualifier-specifier sequence, but lacks the *potis root].

    Note furthermore that this is by far not the only Sumerian borrowing from IE. Even if Whittaker is distrusted (which seems to be the case, for reasons unclear to me, as the man is a well reknowned Sumerologist), Sahala 2009 has identified the following likely Sumerian borrowings from IE:

    “(6) Sum. géme (..) ‘fem. worker’ [also”prostitute”] ~ PIE *gwhen- ‘woman’; Hitt. *kuu; Luw. wanatti; Skt. gnā (ग्ना); TochB śana; Gk. gunē (γυνή); OCS žena; Ir. bean; ON kona. Logogram compound SAL.KUR (woman + foreign land) implies that the word was borrowed into Sumerian and denoted to foreign female slaves.
    [Whittaker adds early Sumerian erum ‘(male) slave; servant.’ here as likely borrowing from aryan, c.f. Proto-Uralic *orja “slave” commonly believed to have been borrowed from the same root,] (..)

    (25) Sum. ú-li-in / wu-li-in ‘colored twine/wool’ → Akk. ulinnu id. ~ PIE *wel ‘wool’; Hitt. ḫulana; Skt. ūrņā (ऊर्णा); Lat. vellus; Goth. wulla; → Finn. villa ‘wool’. More common word for ‘wool’ in Sumerian was mug → Akk. mukku. Word ú-li-in was most likely borrowed into Sumerian, and denoted to some particular type of colored twine imported from an Indo-European language speaking region. (..)

    (29) Sum. urud(a) ‘copper’ → Akk. erû ‘copper’ ~ PIE *h1reudh-ó- ‘red’; Skt. rudhira (रुति); Av. raoðita; TochA rtär; Gk. eruthros (ἐρυθρός); Lith. raudonas; Gaul roudos; ON rjóðr; PGerm. *hrauta(z) ‘iron’ → Finn. rauta ‘iron’ ~ PSD *er- ‘dark-brown color’; Tamil *eruvai ‘blood; copper’. Possibly pronounced /*oruda/. Despite the semantic inaccuracy with the PIE these words probably share a common etymology. The Sumerian initial vowel = /o?/ can be explained as a prosthetic vowel to overcome the phonotactic restriction disallowing syllable initial consonant clusters.”
    http://www.ling.helsinki.fi/~asahala/asahala_sumerian_and_pie.pdf

    This allows for some better understanding – IE speakers were obviously foreign to Sumerians, but lived closed enough to supply them with wool, copper, and slaves (most likely war captives). Sahala also discusses “Sum. kurmountain, netherworld; (foreign) land” as obvious isogloss with IE (Slav. *gora etc.), but ultimately an “areal term” or “paleo word” (c.f. Kartv. *gora “rock, mountain”, Basque harri “rock”, Proto-Sem *k-r-m “mountain”, Arin (Yen.) kar “mountain”, Arogbo (W. Ijo) kure “mountain”, Burked (Nubian) kur “mountain” Chipaya kuru “mountain”, etc. ). Still – the Sumerian association of “foreign” with “mountains” provides another clue.

    Altogether, my best guess currently is the Central Zagros, more specifically https://en.wikipedia.org/wiki/Godin_Tepe that around 3,000 BC fell victim to the Kura Araxes expansion. [However, note that KA may well have been multi-lingual, and its Levantine (Khirbet-Kherak) branch obviously switched to speaking Semitic.]

    Otherwise, Nippur sounds suspiciously Indic. I have in vain looked for any etymology. Even when considering a “Nipp-ur” segmentation (Sum. UR = “City”), the construction would be alien to Sumerian, which regularly puts the qualifier (UR) in front of the specifier.

91. @Rob

    “rather than, no offence, but what is a fringe theory that Z2013 = Indo-Iranian”

    Naturally, my point was exactly about why Indo-Iranian _doesn’t_ seem as linked to Z2103 as to Z93 despite the more common (at least in my understanding and reading) linguistic views about the closer relationship to Greek compared to Balto-Slavic. No offense taken but as an aside, everything but a steppe origin for IE is probably fringe (in a neutral sense) at this point as well but I think it’s good less common views get some discussion on some forums as well since there are a lot of unanswered questions about specific points.

    “The samples we might solidly link to putative IE speakers we have so far from the Balkans (albeit its a handful) are neither I2a2 nor R1b, and neither are the Anatolian Bronze Age-era individuals (although they are potentially non -IE).”

    Obviously my point was under an acceptance of the steppe theory. But if we mention “_putative_ IE speakers”, I think the Z2103 Vucedol sample must also be mentioned. My guess is that most people would consider that kind of sample the recipient of early IE influence, even if not IE-speaking itself.

    But apart from some individuals who had curious views about what all areas where IE languages spread would end up looking like, I don’t think anyone else was expecting areas like the Balkans to end up full of Yamnaya lineages anyway…

    “You might wish to refer to the commentary b/w Aigesti & myself as to why there might be I-A influences in the Balkans.”

    Well sure, but apart from those considerations – there have been views that accept Indo-Iranian influence all the way to southern Greece during the MBA too after all as hard as they can be to substantiate – I was wondering a bit about what’s going on if we consider some common traditional understandings of those relationships. I think Carlos, who argues for an eastern PC steppe scenario too with the known less common views wrt CWC, has brought up some good questions in regard to that that I don’t think are adequately answered just yet. The easy answer might be that the linguistic relationships don’t correlate with genetics very well or that they weren’t that correct in which branches they more frequently considered closer but we might still have some ways to go I think.

    @aigesti

    “Thracian is considered a Satem language. That is a consolidated opinion and there are even some linguists lately who consider it as a part of Balto Slavic group. Illyrian and Phrygian lie in between, with both samples of Satem and Centum, just like Albanian and Armenian. That should give us an idea of the Satem phenomena. Strong in the center, less so in periphery”

    As I mentioned, I took it as a given that the common view is that Thracian was satemized but it seems that there’s some disagreement, with the natural dislcaimer that there generally is about the less-attested branches, at least by someone who’s spent a good amount of time studying the Balkan branches. In general, as you said and as I also previously mentioned too, all less-attested Balkan branches have seen various views on their “satem” status (and for some well-attested like Albanian and Armenian their status is a bit uncertain since some of their development obscures the satemization too), since specific examples could argue for either, but the basic point on which we’d agree is that it didn’t affect the Balkan branches as much as Indo-Iranian and Balto-Slavic i.e. the branches that plausibly spawned closer to the forest steppe/steppe area after the rest had moved further away.

    As for the relationship to Balto-Slavic, Thracian, Dacian and even Albanian have been argued to occupy the linguistic space between it and something like Greek, which lies on the other end of the Balkan branches, sometimes but considering how different developed early historical Balto-Slavs and early historical Thracians look like so far, or even the entities that expand towards the respective areas during the Bronze Age, in order to assume it was even part of the Balto-Slavic group proper we might have to assume that entity broke up very early, basically while still at the steppe/forest steppe area. If the Y-DNA continues not to mesh as well, I think it will make that kind of identification even harder.

92. @ Egg

    I would define ‘fringe’ as what is not consistent with the data. On the other hand, being open to alternative interpretations when there are potential holes in a particular theory, even if a well popularised one certainly is not. Perhaps you are familiar with how academia works- stick with the status quo, make small additions, and if new data arises, see how that can fit into your model (even if you they have not really understood it). Anything too different is risky. And I appreciate their predicament, but frankly, I am not constrained by such bonds and will happily call out vacuous approaches. The ”consensus” model is mandated by a vocal few scholars who, frankly, arent even specialists in the regions necessary for said reconstructions.

    Which takes me to your next point

    ”Obviously my point was under an acceptance of the steppe theory. But if we mention “_putative_ IE speakers”, I think the Z2103 Vucedol sample must also be mentioned. My guess is that most people would consider that kind of sample the recipient of early IE influence, even if not IE-speaking itself.”

    That is an interesting sample, I3499 from Beli Manastir.
    He was found at the bottom of a waste pit. I would not generalise from this one sample, but in what model would you place this in for ‘the IE of SEE by steppe” . Elite conquest, mass migration, patron-client ?

    But of course, we do have more data. What would you make of the expulsion of the BB enclave in Hungary by proto-Nagyrev groups in 2400 BC, or that of Yamnaya viz Ezero a couple of hundred years earlier.
    So we must carefully dissect the data, and not merely use it to pay lip service to certain narratives.
    This is not to say that it disproves that PIE came from the steppe, but im rather outlining if one really looks at the data, the Steppe hypothesis (at present) at best works for northern Europe, or it serving as a vector for certain branches.

    ”Well sure, but apart from those considerations – there have been views that accept Indo-Iranian influence all the way to southern Greece during the MBA too after all as hard as they can be to substantiate – I was wondering a bit about what’s going on if we consider some common traditional understandings of those relationships. I think Carlos, who argues for an eastern PC steppe scenario too with the known less common views wrt CWC, has brought up some good questions in regard to that that I don’t think are adequately answered just yet. The easy answer might be that the linguistic relationships don’t correlate with genetics very well or that they weren’t that correct in which branches they more frequently considered closer but we might still have some ways to go I think.”

    Im not following your point there.

93. @ Frank & Vara

    Thanks for the info ; very helpful

    @ Frank
    I cant find the paper, but in addition to the well known movement of Jastorf from northern Germany, there were movements from Poland toward the west, associated with the Przeworsk culture. Although Im not sure (havent thought about it, tbh) about Germanic arriving from the East Baltic, then the region experiencing a language shift. It would open too many questions which might be difficult to fit

94. “Maybe, but again, your are making statements without support or qualification. Im not aware of such a concensus; indeed such a consensus cannot exist because of the nature of the evidence (linguistics)”

    A consensus in the sense of “every single person ever” probably not but that it (Indo-Iranian, Greek and Armenian from well-attested languages share a set of potentially common innovations to the exclusion of the rest) is a very commonly held-upon view overall, I’d say so from what I’ve seen. I’m not too invested in that kind of view, or even my interpretation of it, being right but I think it’s interesting that this kind of relationship doesn’t seem too prominent in the genetic data so far if we accept a steppe scenario, unless we maybe assume something like a forest-steppe Indo-Iranian dominated by Z93 neighboring a steppe Z2103-dominated Balkan/Greco-Armenian for a while before the Z93 groups end up dominating the steppe as well and nothing more than that. I think Vara (due to a similar impression from linguistics too?) argues for something similar though he considers another sort of view like the Transcaucasian one and related to J2 lineages rather than a steppe scenario. I wouldn’t agree with that at this point but the impression of the linguistic arguments is the same on me in general. Could be wrong of course but it’s still kinda interesting if a commonly held view like that was wrong in some way.

    “Concurrently, on the other side of Europe, one can argue that as BB swarmed into Iberia, replaced all local males and social structures, they found it expedient to adopt local language. But, as Matt rightly pointed out, that would make it something of a ”just so” hypothesis and a very flimsy one at that.”

    See, you brought up a good example here of my kind of thinking in the previous case since I think it’s a similar case where all currently offered theories you can come across online have some weak points that I don’t feel are (or even can be or even will ever be to one’s satisfaction) fully addressed.

    “Hence, when I say a theory is fringe, I mean at odds with actual evidence, not the feelings of certain cliques”

    But that’s the whole problem. We’re all faced with the same “actual evidence” but we can see how many different interpretations there are on similar issues all over the internet aDNA-world by people I wouldn’t necessarily consider more informed or necessarily intrinsically less biased than one another. “Actual evidence” is a tricky thing to define due to interpretation in the first place. Obviously none of us will think we’re more wrong than the others until something very obvious and hard to disagree with comes up but even when one feels that’s happened there’s disagreement by others.

    Out of curiosity, can you point towards any recent accounts that have e.g. considered together an origin of Beaker in the east along with Beaker bringing (_solely_ in your view?) non-IE languages to the west, as I understand your view to be now? Naturally archaeologists at least won’t necessarily argue for correlations with linguistic spreads but I’m still curious if this kind of general view has been explored elsewhere.

    [As a slight aside, since you linked to it, I certainly enjoyed your article about the BA as I commented back then too though I’d say you were a bit more open when writing it, e.g. “Although there is significant steppe admixture in the central Balkans, the dearth of sampling does not allow definitive answers about Yamnaya’s exact role in the genesis of Balkan & Anatolian Indo-Europeans groups” compared to now where it seems you’re _fully_ disregarding Yamnaya as a potential source of anything IE or am I misreading you?]

    “None taken”

    Hehe, I think you misread that part as “given” but no offense given either for sure.

95. @ Egg

    I’m aware of the recognised links between I-A & Greek, however whether they form a clade of not is a more difficult question, one beyond my knowledge, and one which might never be solved; nor does it really need to be. If they are not a genetic node, then the reason for the said parallels have already been adressed here

    ”But that’s the whole problem. We’re all faced with the same “actual evidence” but we can see how many different interpretations there are on similar issues all over the internet aDNA-world by people I wouldn’t necessarily consider more informed or necessarily intrinsically less biased than one another”

    I’m sure there are also sort of random theories out there, but that doesn’t concern me.
    What I understand from the current data is that (east-) Yamnaya -derived lineages dominate in areas where non-IE langauges were ubiquitous until recent pre-history (Iberia), or nothing attested until the common Era (Ireland). We also understand that these groups were exclusionary, spread by force and affected social rupture of the scale not previously seen (whatever the exact mechanism & reason). As impressive as such an expansion might be, it is limited in its ability to influence outside its own (Late Neolithic) kin group structure. We would therefore have to show that they imparted a similar effect in areas of Europe & Near East which feature early & documented presence of IE. However, that seems to be lacking (at least at present). In fact, a very different dynamic appears to have occurred just prior the attestation of I.E. languages here – marked by distinctive cultural package as well as genomic lineages & autosomal markers. Not only did the aforementioned groups make only a minimal impact here, but they were ostensibly marginalised within a short time. With such data, to hypothesise that PIE originally expanded with (east)-Yamnaya groups is not therefor not self-evident.

    ”you’re _fully_ disregarding Yamnaya as a potential source of anything IE or am I misreading you?
    Of course not, as I suggested ”This is not to say that it disproves that PIE came from the steppe”. Indeed, one understands the allure of such a link, afterall, we have a widespread expansion from there. A similar allure has come from a couple of papers focussing on Y -chromosal bursts, but written without any significant appreciation of context or a full gammut of aDNA data.
    However, a more cautious approach would break everything down into component parts. For ex, Afansievo and Andronovo were thousand years apart, genetically & culturally distinct- something Outram went to effort to point out. In these highly clannic groups, why should they all have spoken the same language ?.
    At the same time, there were certainly expansions from other regions, they are just not understood by most people; and have not been expounded so far in Indo-Europeanist discourse. All of the above dispersal events need to be understood sui generis, then integrated into a whole (new) model.

96. Rob, I think you’re right about the appreciation of local phenomena which sometimes might tend to get overlooked, especially in the early days of the steppe aDNA papers, in favor of “more steppe = better confirmation of theory” or something along those lines but as you clearly agree, there must have been an initial starting place that unites all the branches. What else do you think could effectively be argued at this point that unites all of them together except the steppe expansion that affects all IE-speaking regions in complicated ways via different paths? Even if we argue for a late sort of expansion from the Carpathian basin area, which I’m still unsure whether emanates solely/mostly cultural influence without further data, it seems hard to unite this to some other phenomenon that doesn’t involve the steppe at some point. What else would be there, the Aegean-Anatolian influence that slowly permeated the Balkans up to Hungary? Or, in a similar SE Europe kind of scenario, IE remained together at the Balkans further north until that point and very late expanding groups towards the northwest brought languages there?

    I’m needling you because I tend to see what you mention and insist on as parts of respective developments in different regions rather than something that makes sense as a uniting theory, while the steppe works much easier in that regard. Maybe I should re-read your BA post to be reminded of some things again but what’s your current stance about the potential early correlates (as complex as they might be) of IE, like e.g. the steppe theory considers R1 and the EHG/CHG component to be, if you consider that early IE arose on the interaction zone of farmer and western steppe populations? Or do you even think there are no concrete correlates at all?

97. “if you consider that early IE arose on the interaction zone of farmer and western steppe populations”

    Which I take to be your current general stance, that is.

98. Thanks, Alberto I need to catch up. I did a cursory read. You said BB need not have been IE. Is it possible that they were multilingual or that IE BB did not have written script?

99. @FrankN

    Yes, that’s one of the interesting ideas in Garrett’s paper: as long as the daughter languages are in geographical contact (what he refers to as a “dialect continuum”), there’s no need for any feature (innovation) to have been present in the proto-language. This in turn offers the opportunity of more realistic and less idealistic analyses.

    And it’s also the reason why I mention in the post the need to have evidence of shared features (or words, more specifically in the post’s context) in IE languages that are unlikely to have been in geographical contact after their split, so that we can consider a word (or feature) as PIE.

100. @postneo

     Bell Beakers (IE or not) surely didn’t have a written script. The written evidence only appears 2000 years after their arrival, when their culture proper was long gone.

     So we have no idea of which language(s) they spoke. The evidence we have suggests that Ibero-Vasconic is likely descended from them, but it could have been a language they adopted at some point along their way to Western Europe or even in WE itself at one point during their expansion. For IE or any other language we don’t really have any evidence so far.

     But this is a tangential part of the post, anyway, and not too important. The main point is about the problematic of substrates when we don’t have knowledge of the languages we’re looking for in the first place, but we do have a big knowledge of the languages spoken later throughout the area (in this case IE ones) that may have absorbed the previous languages and it’s hard to say what was part of either language unless we carefully contrast it with other IE languages spoken in unconnected areas. Thought the very main point in the post is how aDNA can be of great help here, since we suddenly have strong evidence of the extreme unlikeliness of IE having been spoken in some areas before the Roman conquest, which provides a great opportunity to separate what can be IE from what can’t.

Comments are closed.