Substack cometh, and lo it is good. (Pricing)

R1b-L21 and Goidelic Celtic

The new paper, Large-scale migration into Britain during the Middle to Late Bronze Age, did not resolve the origin of the Celts. But one thing I was curious about:

Evidence for a substantial contribution from the C/EBA population to later populations also comes from Y chromosome haplogroup R1b-P312/L21/M529 (R1b1a1a2a1a2c1), which is present at 89±5% in sampled individuals from C/EBA Britain and is nearly absent in available ancient DNA data from C/EBA Europe (Supplementary Table 9). The haplogroup remained more common in Britain than in continental Europe in every later period, and continues to be a distinctive feature of the British isles as its frequency in Britain and Ireland today (14-71% depending on region19) is far higher than anywhere else in continental Europe (Extended Data Fig. 5).

If you go online you can see the frequency of R1b-L21 varies a lot in England, with rather low frequencies in East Anglia, and higher fractions in western Britain. In Ireland, the frequencies may exceed 80% in the western counties. Lara Cassidy noticed early on that the Rathlin sample from Bronze Age Ireland, an a Bell Beaker individual, carries this mutation. On the continent, the mutation is found in Brittany, subject to migration from Britons, while in Spain it seems to be found in lower frequencies, mostly in the western provinces.

One of the insights of the new paper above is that there seems to have been an Urnfield-related migration that arrived in England around ~1200 BC. Did they bring Celtic speech? I think they were  Brythonic and P-Celtic speakers. I believe that R1b-L21 and the Bell Beakers brought Goidelic Q-Celtic languages, and there are some who argue that Celtiberian was a Q-Celtic language.

10 thoughts on “R1b-L21 and Goidelic Celtic

  1. I think you are right, Razib. People with an interest should also look into Volker Heyd’s latest article, 2019.

  2. IMO, far too great of a time depth for C/EBA to be any kind of Celtic. If Goidelic and Brythonic had been diverging since the 3rd millennium BC, we probably wouldn’t even consider them the same branch of Indo-European. L21 distribution reflects later migration history because all migrations to the British Isles follow the same path: SE England > rest of Britain > Ireland. Naturally, the demographic impact of each event is greatest in SE England, and weakest in Ireland.

    The Q-P distinction in Celtic is a single isogloss of a single sound-change. “Q” is the ancestral condition of Proto-Celtic, and the “P” shift could easily be an areal development. Quite a coincidence otherwise that “P-Celtic” occurs in adjacent Britain and France, while “Q” – the ancestral, conserved form of the sound – is limited to the periphery in Ireland and Iberia. For what it’s worth, linguists lean more towards Insular vs. Continental than P vs. Q these days.

    On the other hand, 1200BC is a plausible date for Insular Celtic vs. mainland Celtic. Still feels a bit early, considering just how conservative the earliest Irish inscriptions are (“Primitive Irish” names look more like Gaulish than even Old Irish), but I’m no specialist.

  3. The response by the Language Hatters seemed pretty similar to what Marco said (https://languagehat.com/large-scale-migration-into-britain/#comment-4360778), which seems like the most sensible response?

    I obviously don’t know a lot about linguistics but would’ve tended to agree that I don’t see how it’s likely that all the differentiating features which unite Insular / Continental Celtic could’ve developed and differentiate them from IE could’ve developed by 2500 BCE. That’s probably 500 years after the arrival of the Corded Ware horizon in Central Europe. Seems too short.

    Less confident about the idea that Iron Age Celtic must be correct because Continental and Insular Celtic, when attested are “too similar”. There are big ranges of change in language families, so I don’t think you could be too confident of dating a difference this time in this regard, i.e. distinguishing between 1000 BCE (Atlantic Bronze Age 1300-700 BCE) vs 500 BCE (Halstatt Celts 800-500 BCE). Also don’t know how tightly looking at lexicon and terms could date the difference between this Late Bronze Age and Early Iron Age phase (knowledge of the existence of iron would be around by LBA, etc.)

    I think Reich’s general comment was pretty good (“the genetic data should make you adjust your beliefs, down-weighting the scenario of early Celtic language coming in the Iron Age (and early Bronze Age) and up-weighting the Late Bronze Age.”). Weights, not definites.

    @Marco: “Naturally, the demographic impact of each event is greatest in SE England, and weakest in Ireland.”

    I think that makes sense, although it would be interesting to see any direct data from Ireland to see how this happens, if it’s possible. The leaked thesis of Lara Cassidy found a minor trend were that in the CA-Early Bronze Age there was some Southwestern divergence of samples in Ireland, and the few Southwestern Copper Age samples had enriched Neolithic ancestry and slightly less Steppe ancestry, and the one male had I2a*. It was not clear if this reflected local admixture or some sea related movement from France. A bigger independent Irish dataset in this time might help with this, although the claim here is that any admixture would be much less due to distance, and there are problems getting burials from the Late Bronze Age period that overlaps the change described in Patterson et al (cremation is univeral).

    It does seem like there is some decline in the steppe ancestry in the published samples from Ireland too. The earliest samples there are the Rathlin Island samples from the Early Bronze Age (average 1740 BCE) and they seem slightly lower in steppe ancestry than the main cluster of Bell Beakers in Britain. Based on what happens in Davidski’s Global 25 data in Vahaduo – https://imgur.com/a/cSocDV9.

    In this model, Bell Beakers in Britain (average 2150 BCE) have around 58-59% Steppe ancestry, then the EBA Irish (1740 BCE) samples around 55%, and this at some point declines to around 50% in present day Irish. Although this is not quite as dramatic as going from 59% in the published Bell Beakers in England to 48% in England and Wales, it still leaves open the potential for some change. At least going by the Global 25 there is some decline in Ireland too, so there is a question there.

    Obviously, if there was a signal of migration into Southern Ireland from a continental source at this time, and its independent from what happens in England, that would strengthens the case that Patterson’s paper makes!

    *The finding from Cassidy’s thesis seems similar to a finding in a paper about Orkney which will be upcoming soon – https://www.ebi.ac.uk/ena/browser/view/PRJEB46830?show=reads“Ancient DNA at the edge of the world”“(In Bronze Age Orkney) much of the population displayed significant genome-wide ancestry deriving ultimately from the Pontic-Caspian Steppe. However, uniquely in northern and central Europe” (well, actually uniquely for Britain) most of the male lineages were inherited from the local Neolithic.

  4. I agree with the earlier comments

    I don’t think it likely that British BB spoke (proto)-Goidelic.
    P and Q Celtic are linguistically close and Joe Eska (an eminent Celticist)makes them a clade in a phylogenetic tree of Celtic. See

    https://www.academia.edu/44813439/The_dialectology_of_Celtic

    The most up to date treatment of the origins of Celtic is
    Sims-Williams (2020) An alternative to ‘Celtic from the East’ and ‘Celtic from the West’

    Note that Sims-Williams sees little evidence that Urnfield/Hallstatt has much to do with
    Celtic.

  5. Its best not to rely upon genetics to make guesses about other disciplines yet geneticists are doing just that. Again.

    Gaeilgeoirí have always had some comprehension issues, particularly when from opposite ends of the country. While Gáidhlig and Geleg are clearly daughter languages of Gaeilge, their speakers were not always mutually intelligible to each other.

    Cymraeg is a beautiful language but is incomprehensible to speakers of Gaeilge, Gáidhlig and Geleg and that’s been the case for about two thousand years. Concerning Brythonic, its languages of Cymraeg, Bretond and Kernowek are not now inter-comprehensible. And this was certainly the case fifteen hundred years ago with their step-sister, *Pictish.

    So had Goidelic and Brythonic languages been spoken in the islands since the Bronze Age, their differences would have been even starker. Instead, Goidelic and Brythonic are clearly offshoots of Gaulish (but not Celtiberian), diverting sometime during the first centuries of the Iron Age.

    Whatever about Britain – be mindful it is a distinct island! – linguists conclude no form of Indo-European was spoken in Ireland until Goidelic-speakers arrived and that Goidelic was not spoken in Ireland until the last centuries BC. That’s not something that can be determined by geneticists. Yet here were are, again. Genetics can help make linguistic inferences, but the only way to make linguistic conclusions is by linguistics.

  6. @Jason Lusitanian probably spread to Iberia with urnfield related migrants as it does seem to be a “para” Italo-Celtic language of sorts. Hard to know what exactly BB spoke due to how old it is.
    I personally subscribe to the northwestblock theory but I wouldn’t be shocked if non-IE dialects were retained as a lingua franca of sorts as we see they do persist well into antiquity era particularly in SW Europe.

  7. @Nick Patterson

    “Note that Sims-Williams sees little evidence that Urnfield/Hallstatt has much to do with
    Celtic.”

    It is not clear why, though. He writes that Lepontic and Celtiberian inscriptions don’t fit with the proposed archaeology for Celtic from the East, on the basis that Lepontic inscriptions belong to Golasecca, not Hallstatt culture. Golasecca is, however, a descendant of Urnfield, which is followed in other places by Hallstatt C and D and then La Tene. Celtiberian inscriptions, on the other hand, are from 2-1st century BC whereas Ha A, B, C, D is 1200-500 BC, so clearly expecting them to fit makes no sense. On this basis, his confident assertion that Celtic from the East is falsified seems premature (in fact, Golasecca bolsters it). Obviously this does not say anything about Celtic from the centre, which still remains a valid theory.

Comments are closed.