I don't know how well linguists have figured out how to estimate time depth; the fact that a small tweak in the methodology (compared to Bouckaert et al.) results in a 3,000 year drop in the estimated age of the PIE split does not add to my confidence about the robustness of this field. Regardless of which hypothesis one accepts, the PIE split occurred thousands of years before the first written monuments in any Indo-European language. For the time being, the ball is on the other side of the court, which may accept this finding or come up with another tweak in the methodology that gives yet another date.
In any case, accepting provisionally that the Chang et al. date is accurate then it falsifies the Anatolian farmer/IE language hypothesis. So, steppe aficionados can declare a partial victory, because there is one less opponent to worry about. Falsification of the Anatolian first farmer hypothesis is not a complete victory, as the PIE urheimat question is not a boxing match between Kurganists and Anatolianists, but rather a mêlée with many players holding on to their swords. So, if you're willing to believe that the methodology is mature enough and they finally "got it right", you need only find a PIE split around 6,000 years ago, but you need not find it on the steppe.
ANCESTRY-CONSTRAINED PHYLOGENETIC ANALYSIS SUPPORTS THE INDO-EUROPEAN STEPPE HYPOTHESIS
Will Chang et al.
Discussion of Indo-European origins and dispersal focuses on two hypotheses. Qualitative evidence from reconstructed vocabulary and correlations with archaeological data suggest that IndoEuropean languages originated in the Pontic-Caspian steppe and spread together with cultural innovations associated with pastoralism, beginning c. 6500–5500 bp. An alternative hypothesis, according to which Indo-European languages spread with the diffusion of farming from Anatolia, beginning c. 9500–8000 bp, is supported by statistical phylogenetic and phylogeographic analyses of lexical traits. The time and place of the Indo-European ancestor language therefore remain disputed. Here we present a phylogenetic analysis in which ancestry constraints permit more accurate inference of rates of change, based on observed changes between ancient or medieval languages and their modern descendants, and we show that the result strongly supports the steppe hypothesis. Positing ancestry constraints also reveals that homoplasy is common in lexical traits, contrary to the assumptions of previous work. We show that lexical traits undergo recurrent evolution due to recurring patterns of semantic and morphological change.
Or, as I have suggested numerous times, the NW of Anatolia and LBK and many of its children were a sister of pre- PIE, all along - explaining the easy and extremely fast spread of more modern variations, at later times.
ReplyDelete"I don't know how well linguists have figured out how to estimate time depth; the fact that a small tweak in the methodology (compared to Bouckaert et al.) results in a 3,000 year drop in the estimated age of the PIE split does not add to my confidence about the robustness of this field."
ReplyDeleteIt's not a question of tweaking the methodology but of building one on conflicting data. The data on the pace of language change include examples like Icelandic, which has changed negligibly for 1,000 years, and English, which has undergone an almost psychedelic transformation in the same time. It is not a weakness in a field to refuse to crush the data out of shape on the Procrustean bed of a methodology.
In any case the pace of language change is not a central concern of the field of historical linguistics except in a relative rather than an absolute sense. It matters when splits, borrowing events or language shifts take place relative to each other; it matters very little what specific century they happen in, except as that affects the accuracy of the branching, which as I say, is defined in relative terms. I know that impacts how useful this is for integrating into archeological or genetic research, but it's not insurmountable as long as there's no conflict.
Falsifying an early out-of-Anatolia model is rather like beating a dead horse, at this point. It may make adherents of the steppe hypothesis feel better, but it fails to offer compelling evidence for their preferred narrative.
ReplyDeleteI'm reminded of David Anthony's attempt at rebutting Michael Frachetti's "new consensus" of early steppe history, a flailing, hot mess in which Anthony actually tried to claim the northern silk route was impassable to humans, even in considerably less arid times. If the steppe hypothesis cannot be made to fit our emerging picture of Eurasian prehistory, bashing away at the Anatolian model like this seems pointless and frankly comical.
That said, if this 6000 bp estimate is accurate to within a few centuries, it places the break-up of PIE in a very interesting period of history. I wish we could have more faith in their methodology.
I think that judging age by backsolving has even less meaning for languages than it does for genetics. Events and population size are what change languages, there is no magic rate of evolution. They could have been roughly indo european for 20k years for all we know.
ReplyDeleteThere's just no way to know for sure when it happened, where, or how much of an identity this ancient group represents with respect to modern europe.
@ Dienekes
ReplyDelete"PIE split occurred thousands of years before the first written monuments in any Indo-European language."
Not necessarily. As noted here, there are no "laws" governing langauge change. It can very slow or very rapid. Anatolian might be "different" just because of the more complex sociolinguistic environment in anatolia - extensive bi- and multilingualism. Sure it might also simply be older also.
Yet also recent perspectives which analyse Anatolian no longer see it to be *that* divergent ; and perhaps some of the odd features like 2-gender system, aorist, etc are not considered to be that radically different.
"Or, as I have suggested numerous times, the NW of Anatolia and LBK and many of its children were a sister of pre- PIE, all along - explaining the easy and extremely fast spread of more modern variations, at later times."
ReplyDeleteThat's completely plausible. It would mirror the spread and distribution of Latin in Europe in previously Celtic but not Germanic-speaking areas, or in Aramaic or Berber-speaking areas.
Well, Proto-Indo-European didn't fall out of the sky. It's odd to assume that there were no pre-Proto-European languages and no sisters to the reconstructed PIE language. If such languages existed, where are they? Where are their descendants? Have they all disappeared? Why didn't they leave any trace? If PIE was an isolate, a language spoken at first only by a limited group among many unrelated languages, How could they have been SO isolated, with no sisters or cousins nearby?
ReplyDeleteThis doesn't sound like the Steppes,with roaming hunter-gathers and nomadic livestock herders.
It sounds like some hidden place where one language held on. Like mountains or other refuges.
Either Chang has the rate of change wrong and there were already forms of Indo_European early on -- much earlier than he estimates -- or this language originated in a refuge, the last of its kind.
Computational phylogenetics is not the only way to produce linguistic datings, and definitely not the most reliable: http://www.elisanet.fi/alkupera/Problems_of_phylogenetics.pdf
ReplyDeleteMore certain are palaeolinguistic datings, assuming that only reliable reconstructions and meanings are applied. These also disprove the early (Neolithic) Proto-Indo-European.
"Well, Proto-Indo-European didn't fall out of the sky. It's odd to assume that there were no pre-Proto-European languages and no sisters to the reconstructed PIE language."
ReplyDeleteThis what Proto-Pontic is all about. Yes it is odd to assume no antecedents or sister languages.
"If such languages existed, where are they? Where are their descendants? Have they all disappeared?"
This kind of thing happens all the time. Latin ate all its sister languages in Italy. mandarin swamped all the sister forms of Chinese in the north. there is nothing rare about language death and no reason why a relative can't do the killing.
"Why didn't they leave any trace?"
We don't know that they didn't. Without those languages to compare, we can never know what features in the daughter groups of PIE they may have contributed.
Chitimacha in Louisiana, written off as an isolate decades ago, may be an example of this situation. The twist is that they may in fact have recovered a connection with some groups of Mexican languages. that was only possible because those Mexican languages had been documented and were available for comparison.
http://lughat.blogspot.com/2014/12/a-mexican-colony-in-louisiana-before.html
"If PIE was an isolate, a language spoken at first only by a limited group among many unrelated languages, How could they have been SO isolated, with no sisters or cousins nearby?
This doesn't sound like the Steppes, with roaming hunter-gathers and nomadic livestock herders.
It sounds like some hidden place where one language held on. Like mountains or other refuges. "
That is all solid, and your use of "refuge" is apt, since isolates are usually relics of wider groups, the other members of which are gone. It introduces a problem though - if a language is relictual, hemmed in by other languages, what mechanism is going to give it the kind of dominance that would allow it to spread the way IE has?
Glottochronology and its underlying assumption that languages change at predictable rates has always been controversial. Even in biology, the concept of "punctuated equilibrium" as a model for evolution has gained acceptance. Linguistic change probably occurs faster in contact situations. Furthermore, glottochronology is obviously much more complicated in the case of mixed languages, which I suspect Indo-European probably is.
ReplyDeleteThere are other ways to date splits in languages. In earlier posts, I noted that Indo-European and the Siouan-Iroquoian-Caddoan languages appear to share a common root for 'dog/horse' - for example Russian: сука (súka) ('female dog’, ‘bitch') and Dakota: šų́nka. A recent study of the genetics of dogs in the Americas that was published in the Journal of Human Evolution, Volume 79, February 2015, titled "DNA analysis of ancient dogs of the Americas: Identifying possible founding haplotypes and reconstructing population histories" points notes that archeological evidence for dogs in the Americas dates to only 10,500 years ago. If one assumes that the precursor paleolanguage that was the source of Indo-European was carried by a North American population that migrated back into Siberia (which could account for the Native American/"Northeastern Siberian" component in West Eurasian populations), this would tend to place an upper limit on the split between IE and North American languages.
The correspondence between forms for 'dog' between Dakota and Indo-European languages was cited by Andrew Williamson in this papers "Is the Dakota related to the Indo-European languages?" (1881) as well as "The Dakotan Languages, and Their Relationship Other Languages" (1882). Williamson had written his papers in response to a previous journal article suggesting a relationship between Dakota and the "Turanian languages" - what we would now call the hypothetical "Ural-Altaic family". Williamson pointed out that the features that Dakota shared with "Turanian" were also shared with Indo-European, and in fact "Dakotan" and Indo-European actually had more in common. Williamson also pointed to a possible relationship between Siouan and Caddoan languages which, along with Iroquoian, constitute the hypothetical "Macro-Siouan" family.
Here is another example of one of the striking likely correspondences between earlier Indo-European languages and Siouan:
Hittite: attas ‘father’ / Lakota ‘ate’
Hittite: anna ‘mother’ / Lakota ‘ina’
Indo-European and the Iroquoian languages also appear to share a common ancestral word for 'to turn' that is the basis for the words for 'wheel' as well as 'cart/car'. Forms for ‘wheel’ in non-Anatolian IE languages are derived from the *PIE *kwel- ‘turn’. Likely related forms in early Iroquoian languages (from Julian, “History of Iroquoian Languages”) include Proto Northern Iroquoian forms:
ReplyDeletePNI * { -kaɹhatho- }, * { -kaɹhathw- } 'invert, roll over, turn over'
PNI * { -kaɹhateni- }, * { -kaɹhatenj- } 'roll, turn around' ( limited to Iroquoia )
Note the resemblance to European forms such as Old Norse kartr for 'cart'.
Related Cherokee forms include:
dạgwạlela n. car, wagon
gasagwalvi adj. round
The likely Cherokee root for 'spin', ‘-gwal-‘, can be linked to *PIE *kwel- by relatively simple and straightforward sound shifts. The Cherokee form is clearly related to the *Proto Northern Iroquoian forms, with characteristic shifts k- > gw- and ɹ > l, which are paralleled by *PIE *kwel-. Note the resemblance of the Cherokee form 'dạ-gwạlela' to the Proto-Tocharian form *kwə́kwlë ‘chariot, wagon’ (Ringe 1987) and the later Tocharian B 'kokale' which is produced from *PIE *kwel- ‘turn’ by reduplication.
A question. So, where was Proto-Indo-European being spoken in say 5500 BC?
ReplyDeleteOr, if you prefer, Pre-PIE or Pre-Pre-PIE? If it wasn’t PIE yet? In 5500 BC?
If you say the Steppes, here’s why you may be wrong.
They would have been hunter-gatherers in the Steppes. And hunter-gatherers don’t stay in one place. Some farmers do. But hunter-gatherers don’t. As a rule they can’t.
One of the weirdest things about the “Early European Neolithic” as it is continually pictured is that farmers migrate and hunter-gatherers stay in one place.
Mesolithic people never travel south to Anatolia or into Armenia or Iran. They stay in Siberia or the Steppes. Which makes no sense when they can go south.
No matter where MA-1 and the ANE type was 20,000 years ago, we have good evidence that both R1a and R1b were south of the steppes in places like Armenia and Iran and eastern Anatolia well before 3000 BC. This is where the genetic variance is found. Not in Siberia. Not on the Steppes.
The reason they were there was because of the rise of food production. Not pastoralism or sedentary farming. Those are local adaptations of food production and both cattle and plants were available in the early Neolithic by 5500 BC.
The pastoralism of Steppes people depended on the new technology of animal and plant domestication. And it came from right around the Zagros Mountains. And from there it went to the Steppes.
And if it was carried BACK to the Steppes and to Iberia by people who also carried R1b or whatever, they would have been speaking the language of that huge leap in human progress.
My guess is that the language was a variant of Indo-European.
The Anatolian hypothesis is not dead. It just needs tuning. And a breather from this incredible Kurgan-Steppes public relations machine.
A little ancient DNA from Armenia instead of cold and empty Siberia might do the trick.
Unkwown wrote: "Well, Proto-Indo-European didn't fall out of the sky. It's odd to assume that there were no pre-Proto-European languages and no sisters to the reconstructed PIE language. If such languages existed, where are they? Where are their descendants? Have they all disappeared? Why didn't they leave any trace? If PIE was an isolate, a language spoken at first only by a limited group among many unrelated languages, How could they have been SO isolated, with no sisters or cousins nearby?"
ReplyDeleteInquiring about the origins of Indo-European seems to be a taboo among some linguists, sort of like asking theologians what God was doing before He created the world. :-)
Seriously, I think that originally 'Indo-European' was probably more of a cluster of closely related languages of mixed origins rather than a single unitary language, and some of them preserved more of one of the original constituent languages than others, which is why some IE languages have forms that do not have clear correspondences in other branches of IE. An example is the Greek word 'takhos' "speed, swiftness, fleetness, velocity," related to 'takhys' "swift," is said to be of unknown origin. The Proto-Iroquoian form '-takh-', ‘to run’, may in fact be related to this Ancient Greek form, which does not seem to have any corresponding forms in other Indo-European languages. Germanic has another form, 'schnell' 'quick, fast' from Old High German 'snel', from Proto-Germanic '*snellaz'. A good match can also be found in Iroquoian: Proto-Iroquoian * { -hsnuːɹiɁ } 'be fast'. In Iroquoian daughter languages with the 'ɹ' to 'l' sound shift, derived forms are Cherokee 'ùh-sìnűːlîː' adj. 'fast' and Oneida 'Oe. { -shnoleɁ } word stem for 'fast'.
Looking at the variations within Germanic languages, or between Latin/Italian and Spanish, it appears quite plausible that people who speak languages that are ~2,000 years apart (but don't have vastly different other external influences) usually will find it easy to learn the respective other one, and sometimes can even communicate from the get-go.
ReplyDeleteIf we use 7,000 BCE for the Anatolian Neolithic, then if LBK and the early SW Pontic groups (but likely not Cardium) transferred a PIE sister language into much of Europe, a "true" PIE expansion around 5,000 BCE would have been made extremely easy. This scenario also explains why Anatolian IE appears so ancient - even if it was more recently impacted by a "true" PIE.
Maykop starts just after 4,000 BCE, Leyla-Tepe a few centuries earlier. On the flip side, Yamna starts about 4 centuries later, while Sredny-Stog starts contemporaneously with Maykop.
However, the rich and highly-advanced Cucuteni-Trypillian culture predates all the above by about a millennium, thus fits the ~5,000 BCE date above, and is known - during its middle period - to have had long-standing contact with Sredny-Stog, and oulasted them by half a millennium.
Finally, its late period and decline coincide with the Globular Amphora expansion (~starting ~3,400 BCE) to the NW. Again, this is only ~2,000 years after the start of LBK spread, so we are talking fairly easy communication and spread of language innovations.
Notice that Bouckaert et al. provided support for space as well as time. I strongly suspect that the new tree topology will have a hard time distinguishing the Steppe expansion from the Anatolia expansion. I may be wrong but if true, then the burden will be on the new paper's decision to use a different combination of data as well as its use of *ancestry constraint* which I think will interfere with the clock model.
ReplyDeleteAlso, one of the reasons why people choose to do Bayesian inference is its ability to pick out the best model from many different models (given the data), and the paper hasn't actually demonstrated that their results are from the model that outperforms the previous model. The paper simply shows a different possibility, which may or may not be better. AIC, DIC, WAIC, Path sampling, Stepping stone sampling, there are several ways to do it but none of these is used in the paper, regrettably.
Sorry if this sounds intimidating..
@ Unknown
ReplyDelete- the one about pastoralism
Good points. !
@ Gary
ReplyDeleteInteresting points. BUt you'd find little support from linguists. There is no evidence that PIE was a "mixed" language, creole, etc.
It was nothing but a normal, 'natural' language which 'got lucky'
"The Anatolian hypothesis is not dead. It just needs tuning. And a breather from this incredible Kurgan-Steppes public relations machine."
ReplyDeleteUnknown... everything you said in this post is perfectly said IMHO. I'd like to know what's going on because I'm now 100% certain - based on DNA and archaeology - that my Cretan ancestors are the base stock of the Romans. So... WHY are we speaking LATIN? Did we pick it up? If so, why not Greek instead?
Did we speak proto-Latin on Crete? Italic? But Encyclopedia Britannica has just informed me there might not even have BEEN an Italic family. They could all be different languages that just BORROWED a lot from each other.
Does anybody know what they're doing in this field?
Well, we do have a very interesting example of rate of language change across at least 1500 years. That is between the Mycenaean Greek in Linear B and post Classical Greek. The distance between Mycenaean Greek and Classical Aeolic Greek on the basis of "basic cognates" and sound changes is amazingly small. That is in more than 1000 years.
ReplyDeleteBased on this, if the vestiges of Proto-Indo-European still existed in the European Bronze Age -- around say 2500 BC -- than it should have been an old form of Greek. But Celtic and Hittite are not Greek. Not by a long shot.
Indo-European languages are very old. And if one kind of Indo-European language became favored over other such languages in 3000 or 2000 BC, we would have no way detecting that because the substrate under it would they would look identical to us at this time distance.
If many of the systematic "cognates" identified by Will Chang are in fact borrowings conformed to local sound rules, then the paper will seriously underestimate the time periods involved. The borrowings will make actual variations appear to converge.
ReplyDeleteThere is some evidence of that here:
"Here, we infer the frequency of hidden borrowing among 2346 cognates (etymologically related words) of basic vocabulary distributed across 84 Indo-European languages. The dataset includes 124 (5%) known borrowings. Applying the uniformitarian principle to inventory dynamics in past and present basic vocabularies, we find that 1373 (61%) of the cognates have been affected by borrowing during their history."
Shijulal Nelson-Sathi et al
Networks uncover hidden lexical borrowing in Indo-European language evolution
http://rspb.royalsocietypublishing.org/content/early/2010/11/23/rspb.2010.1917.full
@balancemaster - here is a good summary of the other problems involved.
http://www.replicatedtypo.com/reconstructing-linguistic-phylogenies-a-tautology/4259.html
@Mike Thomas
ReplyDeleteYou wrote:
"There is no evidence that PIE was a "mixed" language, creole, etc. "
One the contrary, PIE has some interesting "dual forms" where two words exist for the same or two closely related things. There seems to be a consistent pattern of some dual forms having separate links to Siouan and Iroquoian:
'Man, hero'
*PIE *H₂ner- "man, hero” / *wiH-ro- “man”
*Proto Iroquoian *-hniɹ- 'be durable, be hard, be solid, be strong’
*Proto Siouan *wiroka ‘man (male)’
'One'
*PIE *sem / *(H)óy(H)nos
Iroquoian (Cherokee) ('-k','-kw' > -m sound correspondence) 'sagwu' (-gw>-m-) Laurentian+ 'secada' / Mohawk 'ens(-V)ka' / Oneida ós(-V)kah / Seneca s(-V)kat / Arikara as(-V)co]/ Pawnee us(-V)k’-o (Note the similarity of ancient Greek εἷς heîs derived from this root to Wichita chí?ass.)
*Proto Ohio Valley Siouan 'nǫ(ːsa)/rǫ(ːsa)', Tutelo 'no:ñs'
*PIE *egH₂ “I” / H₁me- "me (acc.)”
*Proto Iroquoian '*k(V)'-, Cherokee 'agw-/agi-'
Siouan (Lakota) 'miyé, míš, ma-, wa-, wé-, bl-'
BTW - Here is another interesting Hittite/Siouan/Iroquoian correspondence:
Swadesh list #206 ‘if' Mohawk ‘tóka’, Hittite ‘takku'; Lakota 'táku' ('what’)
@Mike Thomas
ReplyDeleteYou wrote:
"Interesting points. BUt you'd find little support from linguists."
Contemporary Indo-European scholars seem to be fixated on linking IE to Semitic or Uralic and suffer from a severe case of academic "tunnel vision". By the second half of the 19th century, many prominent linguists had jumped on the "Turanian" (aka "Ural-Altaic") bandwagon and blithely assumed that Native American languages must be related to Turkic and other languages based on superficial resemblances of Central Asian to Native American populations. On the other hand, while the findings of Reich et al of a Native American genetic component in European populations does not in and of itself constitute proof a link between Indo-European and American languages, continuing to exclude such a link based on "racial" criteria alone can no longer be supported.
Early Jesuit missionaries who were serious students of languages were struck by the resemblance of many words in the Iroquoian languages to Ancient Greek. The idea that Iroquoian languages were in some way represented from ‘a form of ancient Greek’ was taken seriously by scholars well into the 19th century, and the pioneering linguist Albert Gallatin even wrote a paper comparing the aorist in Ancient Greek and Iroquoian. Unfortunately, such speculation centered around the fanciful vision of some Bronze Age transatlantic migration by intrepid Greeks rowing past the Pillars of Hercules across the Atlantic in their trusty biremes to colonize the Americas, and not the broader issue of the possibility of a very ancient, deep relationship of Iroquoian languages to the general IE family dating back to the Neolithic.
Andrew Williamson had the misfortune of publishing his papers on the Siouan/Indo-European link in American journals with little visibility to mainstream European linguists. Likewise, Williamson and his father were not familiar with Iroquoian languages and therefor lacked a vital piece of the puzzle. Moreover, he published only a few years after the Battle of the Little Bighorn at the height of "Manifest Destiny" sentiment in the United States when there would have been little enthusiasm for the notion of the Sioux being some type of "lost Aryan tribe". Speculation regarding the possible links between Greek and Iroquoian died out around the time of the forced removal of Native Americans from the eastern United States, no doubt because such a theory did not play into the dominant white supremacist cultural narrative. (A historical footnote: the Turkish government backs up its denial that the persecution of Armenians constituted 'genocide' by comparing their actions to those of the United States - including the infamous Cherokee Trail of Tears - as a "forced relocation for national security purposes". BTW - The Armenian word for 'five' is 'hing' and the Cherokee equivalent 'hisgi', can be mapped to each other by the shift 'n' > 's' that is seen in many IE forms for 'five': ex. Greek 'pénte' and Albanian 'pesë'.)
"One of the weirdest things about the “Early European Neolithic” as it is continually pictured is that farmers migrate and hunter-gatherers stay in one place. "
ReplyDeleteUnknown,
Even weirder is the Kurgan hypothesis notion that some dry-steppe pastoralists would one day suddenly decide to branch out and become the most competitive northern cold/wet agriculturalist - entirely crazy. Especially when their horses would consume all the grains they needed for themselves, and they still had not figured out how to make hay for the 3-4 months when there was only snow and forest - no open grasslands. A merger of Globular Amphora and TRB is the best candidate to combine the proper regional seeds and animals and experience with the most effective novel practices of deforestation and forest grazing - for the northern part of PIE spread. On the steppes, of course PIE spread on the steppes... but also starting from the (N)W Pontic.
http://www.livescience.com/49603-paleo-indian-spear-thrower-evidence.html
ReplyDeleteof course their fake time is from a their very ignor-ant of history uniformitarian lego's box of toys and real magic .
http://en.wikipedia.org/wiki/Spear-thrower
"
Wooden darts were known at least since the Middle Paleolithic (Schöningen, Torralba, Clacton-on-Sea and Kalambo Falls). While the spearthrower is capable of casting a dart well over 100 meters, it is most accurately used at distances of 20 meters or less. Seven spears were found in the Schöningen 13 II-4 layer, dating from about 400,000 years ago and thought to represent activities of Homo heidelbergensis.[7] The spearthrower is believed to have been in use by Homo sapiens since the Upper Paleolithic (around 30,000 years ago).[8] Most stratified European finds come from the Magdalenian (late upper Palaeolithic). In this period, elaborate pieces, often in the form of animals, are common. The earliest secure data concerning atlatls has come from several caves in France dating to the Upper Paleolithic, about 21,000 to 17,000 years ago. The earliest known example is a 17,500 year-old Solutrean atlatl made of reindeer antler and found at Combe Saunière (Dordogne), France.[9]""
now there is some profoundly ignorant of history stuff going on there in that paragraph ! but oh well.. what the.......... . people will believe what they choose to believe .
Bouckaert et al. are not linguists anyway, they are biologists who have applied mathematical methods from biology to a linguistic problem, and who have received well founded critique from linguists for doing so.
ReplyDelete@ unknown
You invoke variance in support of your assertion that R1a and R1b in cultures like Yamnaya and Corded Ware had originated in West Asia. But what variance? STR? It's not reliable: R1a1a has the highest STR diversity in Pakistan and northern India. Yet this is all R1a-Z93, a brother clade of the European R1a-Z283. They split around 3500 BC. So South Asia is not the origin of R1a1a, even though it has the highest STR diversity. And if you instead refer to SNP diversity, it's not a conclusive sort of evidence either. A high diversity at one place may mean that many clades originated in this place - or it may mean that many clades migrated to this place... And the modern-day distribution of old paragroups isn't very conclusive either. Since they are old, they had a lot of time to migrate, more than younger clades. And in some places there may have been more turmoils, upheavals and founder effects than in others, removing old yDNA variants.
At the end of the day evidence from ancient DNA is worth many times more than all speculation based on modern-day patterns.
What is the difference between R1b1 and R1b1a-P297? Just one mutation, P297. The idea that it arose in situ on the steppe from local R1b1 is certainly not far-fetched. It may have originated in West Asia, but so far there is no evidence for this, so that's speculation. While I'm sure that there was R1b1 in West Asia too at an early date, especially R1b1b and R1b1c, it's not at all certain that R1b1a-P297 appeared in West Asia earlier than on the steppe. Future research may or may not find R1b1a-P297 in West Asia predating the one in Yamnaya.
ReplyDeleteBut isn't it curious: northern EHG in Karelia had R1a1, while more to the south, in Samara, they had R1b1. And later the more northern Corded Ware was dominated by variants of R1a1 and the more southern Yamnaya by variants of R1b1. A strange reiteration of old patterns, if both R1a1a1 and R1b1a had spread later from West Asia.
I've changed my mind, btw. Based on aDNA we really can't tell whether R1b1a existed in West Asia earlier than on the steppe, or not. The presence of R1b1 in Samara hunter-gatherers proves nothing, because that haplogroup had a very wide distribution, which may even have included West Asia. Based on modern distribution patterns (phylogeography, variance, frequency), I would rather suspect that R1b1a originated in West Asia, and then moved to the steppe together with the West Asian autosomal admixture.
ReplyDeleteAlso, to me it has become clear that the Bell Beaker people from central Germany were a mixture of three main constituents: Corded people + western steppe admixed, BR1-KO1-like people from the Carpathian Basin + people with Danish TRB ancestry. In contrast, the Iberian impact was weak. I'm drawing upon my Eurogenes K15 4 ancestors oracle analysis of the Bell Beaker individuals using all other ancient DNA kits (except the 5 Hinxtons and BR2 who are too similar to the Bell Beakers while postdating them considerably). These results are also in line with craniometric analyses of central European Bell Beaker people, who partly show a close relationship with the Danish TRB.
What this means for the origin of R1b1a2a1a2 is straightforward: An Iberian origin is unlikely given the limited Iberian impact. An origin in the Danish TRB is unlikely, as nearby Swedish farmers and hunter-gatherers didn't have it. An origin in the Carpathian Basin is likely, given the presence of Yamnaya east of the Tisza, and the presence of R1b1a2a in Samara Yamnaya.
As for this paper, I find it particularly noteworthy, that according to the analysis Greek, Armenian and Albanian split from the mainstream right after the Hittites and Tocharians. In all previous linguistic trees I've seen it was Italo-Celtic that split off third.
ReplyDeleteIf Italo-Celtic came from the Yamnaya in the Carpathian Basin (which is the most likely hypothesis imo) then an early split of Italo-Celtic from the rest of IE would mean that Greek, Armenian and Albanian cannot be from West Asia - they would have to be from the steppe too - unless there were several migrations of IEs across the Caucasus from West Asia to the steppe, which is unlikely.
However, if the Armenian-Greek-Albanian branch split off earlier than most other IE languages, then it would be at least possible that they had simply stayed in West Asia, while all the younger branches were from the steppe. And then it would be possible to explain the Bronze Age expansion of haplogroup J2 and West Asian admixture into the Balkans and southeastern Europe as associated with the arrival of some of these languages. Which makes more sense than ascribing it to the expansion of some undocumented non-IE languages.
Some late linguistic comments, mostly @Gary Moore:
ReplyDeleteI think that originally 'Indo-European' was probably more of a cluster of closely related languages of mixed origins rather than a single unitary language,
This displays a conceptual confusion. You have just a paragraph earlier wondered about an unwillingness to consider the earlier stages of IE. So, here, too, if you assume a "cluster of related languages", what will have been the origin of this situation?
The sole process known to linguistics capable of creating "clusters of related languages", or indeed even as much as a set of mutually intelligible dialects, is descent from a common ancestor — and it is this ancestor that, by definition, we call Proto-Indo-European. There are many complications that may challenge our ability to accurately reconstruct, date or locate this language, but at the very least its one-time existence is not reasonably deniable.
We also know quite well that the members of the later clusters of related languages (various stages of Proto-Celtic, Proto-Germanic, Proto-Italic etc.) have indeed interacted with one another, exchanging loanwords and other features. Their identification and analysis is regular bread-and-butter historical linguistics. It is conceivable that much further work is needed, once we arrive to slightly older "Late Indo-European" dialects like Proto-Italo-Celtic or Proto-Armeno-Greek. But there is no fundamental neglect of non-genetic commonalities going on.
PIE has some interesting "dual forms" where two words exist for the same or two closely related things (…)
Synonyms occur in all known languages (e.g. English pig, pork, guts, intestines or gas(olin), petrol) and they indicate nothing about "mixed" origin (in the creole sense). It is highly likely that the expansion of PIE proper led to the assimilation and extinction of its former sisters, but occasional substrate loanwords from such languages would not invalidate the existence of PIE itself, no more than e.g. a heavy influx of Latinate vocabulary can turn English into a non-Germanic language.
continuing to exclude such a link [between Indo-European and American languages] based on "racial" criteria alone can no longer be supported.
A link has never been excluded; it has merely never been commonly supported. It is practically impossible to disprove the possibility of a relationship though, so the burden of proof rests on anyone wishing to claim especial relationship between two languages or language families.
Your Iroquioan comparisons are perhaps interesting, but as they stand, they're a mere handful of forms from a vast ocean of vocabulary. Without more systematic evidence, we would have to leave an IE-Iroquioan relationship speculative. Moreover, probably the relationships of Iroquioan to its neighbors in North America would have to be clarified first — e.g. even a relationship with Siouan remains unconfirmed. Hypotheses of linguistic relationships "skipping over" dozens of intervening language areas are extraordinary and require extraordinary evidence.
(Albanian pesë '5', by the way, does not have a shift *n > s; it has a shift *kʷ > s before *e — a typical palatalization development similar to many other IE languages, e.g. Sanskrit pañca or Latvian pieci — followed by simple loss of *n. Hence we see that e.g. here the resemblance to Cherokee turns out to be demonstrably illusory.)