May 07, 2013

Deep common ancestry of Eurasiatic languages (Pagel et al. 2013)

From the paper:

Posterior support at internal nodes of the tree is low, as we might expect of a linguistic tree of this age, but all exceed chance expectations (SI Text) and the internal topology does not affect our estimates of the age of the superfamily. All inferred ages must be treated with caution but our estimates are consistent with proposals linking the near concomitant spread of the language families that comprise this group to the retreat of glaciers in Eurasia at the end of the last ice age ~15 kya (4, 17). The 95% CIs around the root-age are consistent with the initial separation of these families occurring before the development of agriculture beginning ~11 kya (26).

A few comments:

  • The common ancestry of Inuit-Yupik with Chukchee-Kamchatkan lends some support to the idea of Old/New World contacts postdating the initial colonization of the Americas
  • (Note that the superimposition of the tree on the map does not indicate migratory paths)
  • The deep divergence of Proto-Dravidian from the rest of the tree raises the issue of the genetic identity of the Proto-Dravidians. Today, Dravidian speakers are concentrated on the southern parts of India -with the notable Brahui exception in Pakistan- so one is tempted to associate them with the long diverged "Ancestral South Indian" genetic component whose closest living relatives live in the Indian Ocean. On the other hand, hypothesized relationships between Dravidian and extra-Indian languages, such as those postulated here might suggest that Proto-Dravidian was spoken by people more closely related to other Eurasians.
  • More generally, the hypothesis of post-glacial contacts between diverse parts of Eurasia might suggest that differentiation between Eurasian peoples did not proceed in isolation after the initial Out-of-Africa settlement. And, if there were indeed post-glacial movements, of people spreading "Proto-Eurasiatic" languages, these may be detectable by archaeogenetic means.

With the two earliest offshoots being Proto-Dravidian and Proto-Kartvelian, it would be tempting to seek some Central Asian proto-homeland for these languages; the remaining languages seem to occupy (mostly) areas that were substantially glaciated. There was of course large-scale language replacement during the Neolithic and even later time periods, so one can hypothesize that other extinct languages may also have belonged to this greater family, and it would be interesting to see if membership could be supported for any of them.

ScienceNOW has a fairly good high-level discussion. The paper is open access.

PNAS May 6, 2013, doi: 10.1073/pnas.1218726110

Ultraconserved words point to deep language ancestry across Eurasia

Mark Pagel et al.

The search for ever deeper relationships among the World’s languages is bedeviled by the fact that most words evolve too rapidly to preserve evidence of their ancestry beyond 5,000 to 9,000 y. On the other hand, quantitative modeling indicates that some “ultraconserved” words exist that might be used to find evidence for deep linguistic relationships beyond that time barrier. Here we use a statistical model, which takes into account the frequency with which words are used in common everyday speech, to predict the existence of a set of such highly conserved words among seven language families of Eurasia postulated to form a linguistic superfamily that evolved from a common ancestor around 15,000 y ago. We derive a dated phylogenetic tree of this proposed superfamily with a time-depth of ∼14,450 y, implying that some frequently used words have been retained in related forms since the end of the last ice age. Words used more than once per 1,000 in everyday speech were 7- to 10-times more likely to show deep ancestry on this tree. Our results suggest a remarkable fidelity in the transmission of some words and give theoretical justification to the search for features of language that might be preserved across wide spans of time and geography.

Link

47 comments:

Daniel Kalina said...

Maybe the language is associated with Haplogroup P y-dna

Onur said...

Once again (after the Mudrak 2009 and Turchin et al. 2010 papers), we see the finding of a close relationship between the Altaic and Eskimo-Aleut (=Inuit-Yupik) language families. The Altaic language family seems to have originated somewhere in eastern Siberia, Manchuria and/or Mongolia.

Teo. said...

It really astounds me how biologists are ready to buy stuff that 90% of linguists regard as crackpottery. I'm not even criticizing the paper yet, I'm just stating a fact. Starostin and Ruhlen's Tower of Babel reconstructions are only accepted by his few cult followers.

How can they be accepted as the starting point of anything? This is the equivalent of taking intelligent design as a starting point for a genetics study. No wonder the paper wasn't published in an established linguistics journal, but rather in the generic PNAS, which is notorious among linguists for publishing less-than-rigorous linguistics papers. It would hardly get through proper peer review as it is.

(And btw no, this doesn't have to do with resistance to evolutionary methods, which are otherwise used by mainstream linguists as well, but rather with reckless use of linguistic data.)

terryt said...

"seven language families of Eurasia postulated to form a linguistic superfamily that evolved from a common ancestor around 15,000 y ago".

I have always thought that the idea of a Eurasian superfamily made complete sense. To me the most interesting aspect here is the inclusion of Kartvelian and Dravidian in the language group. That places Dravidian as definitely having entered India from the northwest even though the evidence is usually interpreted as indicating it has indigenous origins.

"With the two earliest offshoots being Proto-Dravidian and Proto-Kartvelian, it would be tempting to seek some Central Asian proto-homeland for these languages; the remaining languages seem to occupy (mostly) areas that were substantially glaciated".

That actually makes complete sense. I was certainly under the impression that the lowland region immediately east of the Urals (the 'West Siberian Plain') was under a huge ice sheet until well after the ice had melted elsewhere. That would mean the centre of dispersal was at least south of there. Perhaps from no further north than what an atlas I have calls the 'Kirgiz steppe'?

"one can hypothesize that other extinct languages may also have belonged to this greater family, and it would be interesting to see if membership could be supported for any of them".

I can recall seeing Burushaski as being part of a Eurasiatic superfamily. And even the American languages. That last would fit:

"Maybe the language is associated with Haplogroup P y-dna"

American Y-DNA is Q, a branch within P. But the clain does fall down somewhat with Y-DNA N being apparently responsible for the spread of Uralic. But perhaps by that time U had adopted the language.

eurologist said...

Taking the study at face value, the only thing that connects these languages/ areas at the proposed time frame is steppe-tundra hunting (which at one point extended from the Atlantic coast to beyond the Altai). The problem is of course that this connection existed before LGM but the region got fragmented during LGM, with only few regions still supporting hunting on the steppe-tundra.

I see two possibilities, here. (1) One area held a significant population and was in good position to dominate post-LGM expansion. LGM vegetation maps indicate part of the Caucasus and northern Iran might have satisfied this. The northern Pontic, Dniester, and Don river regions were also populated through LGM, while southern central Siberia largely experienced a discontinuity. One should look for people who did not need to adapt their technology and lifestyle too much for a fast expansion and change to a more mobile life after fragmentation and demise of mammoth herds (and shift to reindeer and bison). The general northern Pontic/ Dniester/ Don river region fits this description.

(2) Language change might have slowed considerably in LGM refugia, as observed today in isolated populations. In that case, some similarities might actually pre-date LGM, and post-LGM contact and expansion from several refugia could have re-affirmed language similarities, even though significant genetic differences had already been established and further increased during LGM isolation drift. This might help explain a spread to the Altai and beyond (some regions of southern Siberia got resettled very early, around ~20kya).

Conversely, contact to Europe West of the Pontic broke off after the Gravettian, so I would assume only a very limited language connection survived LGM, there.

Владимир Р said...

IMHO. The maximum-likelihood hipothesis. Haplogroup R1b is the ancestor of the proto-Nostratic languages.
The expansion of R1b in the direction of C3 = PA+PCK+PIY.
The expansion of R1b in the direction of R2 = PD+PK.
The expansion of R1b in the direction of N1c = PU.
The expansion of R1b in the direction of R1a = PIE.
R1b preserved their language, which was akin to the PIE, but then finally switched to other people's languages.

andrew said...

I too am deeply skeptical. Eurasiatic is not a widely accepted concept and the notion that there are connections at a time depth of 15 kya is even more dubious.

Archaeology strongly supports the notion of Inuit as a quite recent arrival in North America from NE Asia (ca. 500 CE). Fortescue, however, argues for a link between Inuit and Uralic, rather than Inuit and Altaic. This is also supported by genetic evidence: "[T]here is a clear genetic HLA relatedness between isolated populations close to Beringia: Eskimos, Udegeys, Nivkhs (North East coast of Siberia) and Koryaks and Chukchi from extreme North East Siberia (Fig. 2), and North West American populations: Athabaskan, Alaskan Eskimos (Yupik) and Tlingit."

Talking about a proto-Dravidian language with an approximate time depth of 4500 ya having connections with other major language families which aren't much older themselves at a time depth of 15,000 ya, however, seems like utter speculation to me.

Some of the big methodology problems include:

1. If you include only seven language families and evaluate them based on highly speculative proto-languages created for them, you are necessarily going to see some connection at some time depth. There is no indication that any effort was made to evaluate error that could arise in the proto-language creation process or that this was tested against any other alternative language families.

2. There is no good evidence of any statistical merit regarding the rate or process by which "ultraconserved" lexical items are preserved or change.

3. It is highly likely that as of Out of Africa, that the proto-Eurasians (and likewise the population of modern human ancestral to all living modern human) spoke languages belonging to just one or two language families, so just about any two modern human languages with the possible exception of Khoisan, ought to have more than random similarity in the most highly conserved elements, particularly if the way that these elements evolve linguistically is not random and can happen in parallel. Abundant linguistic evidence suggests that actually the evolution of languages is far from random and happens in certain preferred directions. Failure to account for these factors would make the time depth of a common origin look much younger than it really is. Failure to include suitable controls (e.g. Niger-Congo) casts real doubt on whether the cutoff between some relationship and no relationship arguably seen despite the fact that "Posterior support at internal nodes of the tree is low" is meaningful.

4. This is basically a study using the data that gave rise to a hypothesis to test the same hypothesis, rather than looking to see if any other data can confirm it.

Nathan said...

I'm surprised that Semitic isn't included , but then that would invariably have to include Afro-Asiatic which seems more likely to have originated in N.E. Africa.

Re. Dravidian and ASI
I think it is more likely that both ANI & ASI are each made up of multiple diverse ethnolinguistic groups.

Austro-Asiatic is supposed to have preceded Dravidian but there are atleast 2 other cultures that have not been shown to have affinities with Austro-Asiatic ; the Veddah of Sri Lanka and the Negritos of the Andaman islands.

ANI must lump together more than
*1* linguistic group, ditto for ASI.

Ned said...

two comments:
a) 'The common ancestry of Inuit-Yupik with Chukchee-Kamchatkan lends some support to the idea of Old/New World contacts postdating the initial colonization of the Americas' - this completely misses the fact that Yupik is spoken on both sides of the Bering straits. No relationship with Chukchee is required to postulate old / new world contacts - one ethnic group spans the divide!
b) any statistical work like this has to address the very real problem of loan words. If a detailed set of sound-change laws has not been worked out for the group it is impossible to know whether any word correspondence is due to a common heritage or borrowing from neighbouring languages.

Grognard said...

"It really astounds me how biologists are ready to buy stuff that 90% of linguists regard as crackpottery."

You can say that again.

If you find a pygmy driving a ford in the congo and his neighbor driving a studebaker, you wouldn't assume that this was where automobiles were invented. But that's not only accepted but assumed when it comes to nearly all these ancestry studies.

Nathan said...

My hunch is that linguists are going to poke holes big enough to drive a bus through.

Will be watching to see what Language Log contributors and Prof. Witzel have to say.

WarLord said...

Nonsense. You can't just connect Mongoloid and Caucasoid groups that have been isolated from each other until the end of the Ice Age. Actually, they were isolated until the expansion of the Tocharians some 5000 years ago! I understand that searching relationships among distant languages is a very hard work, but at least, we should remain within the limits of common sense.

Modern genetic studies can actually give us a much better idea about the relationship of today's language families. So we can connect e.g. Basque (R1b) with Burushaski (R2) and subsequently with the Ket language (Q) etc., although such connections depend on the language transmission via males.

shenandoah said...

http://archive.org/stream/originespatricia00hampuoft#page/n21/mode/2up

http://archive.org/stream/originespatricia00hampuoft#page/18/mode/2up

I find this very interesting and possibly relevant to the discussion. The author made the connection between European Titles of Nobility (or, the Royal Houses of Europe) and their Asiatic origins... c1846.

Lathdrinor said...

It is not necessary for language and genetics to be deeply connected, especially at the level of Y haplotypes. Without resorting to linguistic minutiae, it is easy to find examples of rapid and enduring linguistic spread without demographic replacement in ancient, medieval, and modern times - from the Roman Empire, to the Turkic expansion in Central Asia and the Near East, to the British Empire in India. Thus, population genetic differences are not in and of themselves an obstacle to this theory.

The time depth for divergence also does not pose an insurmountable problem due to the isolate-then-spread process, though it is at such times that I wonder about the linguistic methodology used. To what degree do linguists state that the divergence of languages is due to drift, and to what degree do they think that it is due to interaction with speakers of other languages? Is it not conceivable that the time depth of divergence was artificially inflated by new speakers acquiring and changing the language to suit their previous lexical and phonological systems?

I myself fall into the camp of people who think that we simply lack the solid theoretic understanding for making this sort of fanciful macro arguments. I imagine that's why the careful linguists are keeping their distance. But hey, you don't make the headlines by writing treatises on the minutiae of obscure African languages. You make the headlines by saying that all the languages of northern Eurasia come from a single source 15,000 years ago.

Average Joe said...

Somewhat related research:

http://www.plosbiology.org/article/info%3Adoi%2F10.1371%2Fjournal.pbio.1001555

eurologist said...

You can't just connect Mongoloid and Caucasoid groups that have been isolated from each other until the end of the Ice Age. Actually, they were isolated until the expansion of the Tocharians some 5000 years ago!

WarLord,

That's factually incorrect. While southern Siberia probably was no longer contiguously populated during a brief 1,000 - 3,000 year span around LGM, significant populations persisted in large areas throughout LGM [e.g., Y. V. Kuzmin, J Archaeol Res (2008) 16:163–221, and references therein]. These certainly had contact with Gravettian before LGM, and as genetic studies indicate, there is a far NE Asian element in European autosomal DNA (plus NE Asian uniparental markers, likely having arrived throughout much of the time period).

However, occupation during cold times was in just a thin band, with seasonal migrations largely in the N-S direction. As such, any genetic or cultural diffusion would have been very slow.

From the archaeological record, I have no problem connecting Uralic, Altaic, and far NE Asian languages. Kartvelian and IE are another story, and Dravidian seems highly speculative. A common origin of Uralic and PIE is probaly difficult to prove, given that they more or less lived next to each other for at least 5,000 - 6,000 years.

Rokus said...

Very interesting study that finally offers a new perspective on the origin of those wide spread basic words on the northern hemisphere, other than "sustained prehistoric contact". However, already in the case of Uralic I always suspected the influence of substratum for the transmission of such words (compare English, whose basic vocabulary was never affected by Normandic domination), what indeed can be supported by the pre-Uralic presence of Corded Ware.
If this principle holds, and ultraconserved words are actually substratum words, this origin issue may become quite complicated now this extended substratum possibly "passed over" to both Neolithic and Mesolithic populations. Some enormous prehistoric expansion event must be the origin of this substratum, and maybe even these newcomers didn't constitute the first population layer. The inclusion of Kartvelian opens the door for the inclusion of Basque and Chinese into the same potpourri. Of course, language change happens much faster in contact situations, thus a post LGM expansion may be utterly exaggerated. What about the Mesolithic?
What triggers me is the purported southern European origin of this (substratum?) population. So far the motivation for this exact location is unclear to me, though ultimately I think this will all be in favor of a completely new perspective on the spread of Indo-European languages as well, and more attention for the role of Mesolithic populations.

valeryz2001 said...

>they were isolated until the expansion of the Tocharians some 5000 years ago!

Lokomotiv site near Lake Baikal is dated as early as 6ky bc: some skeletons appeared to have western mtdna

http://dienekes.blogspot.ru/2005/04/mtdna-of-lokomotiv-siberians-from-lake.html

Jim said...

"he inclusion of Kartvelian opens the door for the inclusion of Basque and Chinese into the same potpourri."

Wrong Caucasian langages, Rokus. The Dene-Caucasian proposal includes NW Caucasian*, Yeneseian, ST and Basque. Not Kartvelian.

You'll see this as "North caucasian" in some places, but North Caucasian is is not much more solid that Dene-Caucasian as a proposal.

Roy said...

Teo, Linguistics can hardly be described as a science. If its doyen Noam Chomsky is anything to go by
it's mere wordplay. I'd rather go with the biologists.

Teo. said...

@Roy

Nonsense. Comparative linguistics, the topic of this discussion, has next to nothing to do with theoretical Generative linguistics apart from a few shared descriptive tools. Otherwise, historical comparative linguists couldn't care less about Chomsky. I'm not even sure what you are trying to get at.

You also completely missed the point. Nobody is pitching linguistics vs. biology. I merely pointed out the sad fact that research widely considered sloppy and unfounded by linguists (as the Tower of Babel website fanciful reconstructions) is taken as representative of historical linguistics by many biologists and then accepted uncritically as a starting point when they want to talk about e.g. proto-languages.
That's not going to work. The rule applies: garbage in, garbage out - no matter how bona fide the effort.

War Lord said...

"These certainly had contact with Gravettian before LGM..."

Certainly? Based on what? If there were any migrations in the Eurasian steppes, they concerned the expansion of the R1-lineages ca. 40 000 years ago from Central Asia to Europe, and repeated waves from the Altai region to NE Asia (probably post-LGM). As far as I know, the potential contact between Central Asia and Europe during the last Ice Age was blocked by a glacial lake. Furthermore, considering that Mongoloid and Caucasoid populations are not genetically close related, their language relationship would have to stem from some "steppe lingua franca". But this is practically impossible, due to the huge geographical distances.

War Lord said...

I don't mention the theory of the authors about their common origin 15 000 years ago, because this is an utter insanity.

Lathdrinor said...

For those looking for a Y explanation, the bulk of the languages in question, including Dravidian and Eskimo-Aleut, are spoken by bearers of K. The division between LT and KxLT is one of the deepest genetic division within that group and that is analogous to Dravidian's split from the other languages. The sub group of NOP constitute a branch of Uralic-Eskimo-Aleut-Indo-European and to a lesser degree, Altaic, speakers. Thus, at the surface level we have a validation of the Y-language spread hypothesis.

This hypothesis runs into a major obstacle, however: the other side of KxLT - M, O, and S. The primary bearers of these haplotypes do not speak a language connected to the aforementioned Uralic-Eskimo-Aleut-Indo-European-Altaic group, nor is this simply explained by postulating that MOS form an early split, because these haplotypes are each of them separate branches of KxLT. O poses a specific problem because not only is it a huge and diverse haplogroup - not suffering from the isolation experienced by M and S, for example - but it split from NO at a shallower level than the split between NO and P - and thus the split between N / Q+R. With N being a lynchpin of the Uralic-Eskimo-Aleut-Indo-European-Altaic formulation, the whole idea begins to flounder.

One way to work around NO is to say that P was responsible for the proto-language, and that segments northward moving NO simply joined later down the line. We have evidence of this in the form of there being no NO found in native American populations - ie P and NO were geographically separate. In this scenario, a group of KxLT bearers moved south and east, developed into NO, N, and O, all the while another group of KxLT developed into P, evolved the aforementioned proto-language, and stayed in the north and west. Subgroups of N and O then moved north and west into the territory of P, joining the language party. This otherwise elegant solution fails to explain LT, however, and thus Dravidian's inclusion in the party. We then require the extra work around that is the hypothesis that Dravidian was not LT's language, but a language brought to LT populations by members of the P diaspora.

Another explanation is simply that O, M, and S are deviants - groups that diverged from proto-language K by virtue of geographic and demographic isolation. In such a scenario, groups of K moved into the south, became isolated from other members of K, and thereafter evolved into speaking very different proto-languages all the while becoming M, O, and S. Yet such a postulation requires we explain why the same did not happen to N, Q, and R, which managed to preserve the K proto-language. This boils down to a cultural contact/interaction sphere argument, in which case, how do we even know whether it's a Y genetic effect in the first place, given that there are plenty of haplotypes - ie C, D, and I - which are not branches of K but are members of the language party?

I think I've shown my thinking process when it comes to the split vs. diffuse debate - ie the areal diffusion effect is of greater importance when it comes to language level connections. Northern Eurasia is a wide open space which we know to have facilitated distant contacts between Europe, Central Asia, North Asia, the Far East, and beyond to the Americas. It then becomes logical that peoples dwelling across this expanse have a few deeply shared cultural and linguistic concepts. Whether this was the preservation of an earlier proto-language/culture versus the result of a latter day phenomenon - say, the spread of a potent cultural force across the trans-steppe high way - is not easy to answer. The 15kya timeline given by the paper for Eurasiatic's expansion is a starting line, one that has to be traced via archaeology through the Paleolithic and the Neolithic for it to yield fruit.

Belenos said...

This is truly dire stuff. We already know:

a: There are similarities between languages.
b: Those similarities are usually greater in geographially close languages.
c: There are many reasons that a language can have for being similar to another language, NOT ALL OF THEM CONNECTED TO LINEAR HEREDITY.
d: Languages do not change at a constant rate.

How on Earth can any serious investigator be aware of C and D and then come out with this kind of dross?

What his results show is there are similarties between Eurasian languages, which we already knew, it tells us absolutely nothing about WHY, as all the author has done is make up some numbers and feed them ino an equasion he's also made up.

No scientific validity whatsoever, it makes me weep that intelligent people should discuss it with anything but scorn.

Rokus said...

"The Dene-Caucasian proposal includes NW Caucasian*, Yeneseian, ST and Basque. Not Kartvelian."

Though your statement is correct, Pagel's study is on a lexical level. Grammatical/phonological subdivisions are of a different category, and apparently more prone to contact or indicative to shared development than the basic lexicon. Kartvelian certainly shares a basic lexicon with the Dene-Caucassian supergrouping, though may have been more isolated.

The problem is that lexicon similarities never received as much recognition. The number of words without clear etymology that can be found in languages all over Eurasia is much more amazing than presented in this study, albeit probably not always as consistent. For instance the word 'cot' of NW European substratum origin (Nordwestblock), that is also substratum in some Uralic branches and Chinese. Pagel's study certainly isn't exhaustive since it only included 7 language families. I did some shopping myself and in a quickscan in Tsung-tung Chang's severely underestimated 1988 thesis "Indo-European Vocabulary in Old Chinese" I found cognates of the class "among four or
more Eurasiatic language families", in Chinese words such as Who, What, Old, To give, To flow.

andrew said...

Lots of linguistics is good science. This is just a bad example. A linguist posting at language log explains why this particular study is utter bunk, focusing on the weakness of the protolanguage reconstructions relied upon:

http://languagelog.ldc.upenn.edu/nll/?p=4612

Jim said...

Roy, linguists developed the principles - shared innovations as evidence of monophyly, etc. - that paelontologists and presumbaly biologists rely on.

"If its doyen Noam Chomsky is anything to go by..."

Oh, sorry - didn't catch the irony the first time.

DDeden said...

Roy, right on.
Ruhlen's books seem clear.
Pukul (Malay: beat (gong))
Pugilist (English: boxer)

Jim said...

Rokus,
"For instance the word 'cot' of NW European substratum origin (Nordwestblock), that is also substratum in some Uralic branches and Chinese."

Then this would be an argument for a substratum of say, Urlaic in Germanic, not of an unltraconserved etymon inherited form PIE. That etymon happens to be missing in Celtic and Italic, and probably everywhere else in IE.

And thanks for the URL to that paper. I knew that IE material in Chinese has been identified. This is where to go look.

terryt said...

"One way to work around NO is to say that P was responsible for the proto-language, and that segments northward moving NO simply joined later down the line".

Taking note of the doubts expressed as to the validity of the Eurasiatic theory but accepting it for now, I have to agree with you that 'P was responsible for the proto-language'. And also 'that segments northward moving NO simply joined later down the line'. But that would eliminate your postulated Dravidian origin:

"The division between LT and KxLT is one of the deepest genetic division within that group and that is analogous to Dravidian's split from the other languages".

KxLT doesn't really fit the theory. But we can be fairly sure that P moved west through South Asia soo after its formation, so it could have carried Dravidian's ancestral language as it went.

"We then require the extra work around that is the hypothesis that Dravidian was not LT's language, but a language brought to LT populations by members of the P diaspora".

As I mentioned above, I think this is probably the correct expanation.

"ie P and NO were geographically separate".

Exactly. P moved west, NO moved north and M and S stayed where the individual ancestors of all four originated. Or perhaps M and S moved slightly east, but not far.

"another group of KxLT developed into P, evolved the aforementioned proto-language, and stayed in the north and west".

I don't think so. MNOPS was a single haplogroup so must have first coalesced in a single region. That leaves no place for just P to have coalesced very far to the north and west unless MNOPS had expanded hugely before the separate haplogroups developed on the expanding MNOPS margin.

Mark Duin said...

I am very surprised to know that this language does not have semantics. Nice and very informative blog.


Thanks,
Mark Duin

Motivational Speaker

eurologist said...

War Lord:

"As far as I know, the potential contact between Central Asia and Europe during the last Ice Age was blocked by a glacial lake. "

I thought this has long been debunked by both geophysical and archaeological studies. It was a hypothesis, but all field studies I can find claim there was no such thing. But then, it's Russia - so you'll never know (almost ;)).

"Furthermore, considering that Mongoloid and Caucasoid populations are not genetically close related, their language relationship would have to stem from some "steppe lingua franca". But this is practically impossible, due to the huge geographical distances. "

I would claim this is exactly likely, given the long time frame of steady contact and the genetic evidence (far NE proto-Mongolian autosomal DNA in all Europeans, and many E Asian uniparental markers present in Northern Europeans (both ancient and today, and at any rate of deep time history).

Gravettian started about as far East as you can get in Europe - so it's not like you need to connect Beringia with Portugal - you only need to connect the Altai with the Russian plains, and the Altai with far NE Siberia. And there is archaeological and genetic evidence for both.

Va_Highlander said...

With regard to Chinese and Indo-European, Alexander Lubotsky's, "Tocharian loan words in Old Chinese, chariots, chariot gear, and town building", may be relevant.

Rokus said...

With regard to Chinese and Indo-European, Alexander Lubotsky's, "Tocharian loan words in Old Chinese, chariots, chariot gear, and town building", may be relevant.

Having IE cognates for wheel and chariot, and a potential shared Eurasian substrate word for horse, Chinese may be the ultimate argument against a Bronze Age Indo-European expansion. Apparently Bronze Age culture introduced a new vocabulary within an extensive territory that vastly exceeded the IE group. So why this linguistically cohesive group should have expanded only together with a derived vocabulary if all IE people could have acquired their Bronze Age vocabulary in just the same way as the Chinese? That Pegal et al. doesn't even mention that unscientific Kurganist ideology increases their credibility about an ancestral dispersal of related populations in Eurasia. So now in 2013, what is this specific Chinese - Indo-European thing all about?

Ah, BTW, the Basque as well as Chinese (or better: Na-Dene?) lexicon seem to cluster even better with Germanic. In the Eurasian hypothesis PIE probably equals nothing but a poor attempt to unify linguistic populations that even in the Bronze Age already had a long history within a Sprachbund.

terryt said...

"Gravettian started about as far East as you can get in Europe - so it's not like you need to connect Beringia with Portugal - you only need to connect the Altai with the Russian plains, and the Altai with far NE Siberia. And there is archaeological and genetic evidence for both".

And if the authors' dating is too recent that would provide a good fit for such a 'language superfamily' as they suggest.

Slumbery said...

Eurologist
""As far as I know, the potential contact between Central Asia and Europe during the last Ice Age was blocked by a glacial lake. "

I thought this has long been debunked by both geophysical and archaeological studies. It was a hypothesis, but all field studies I can find claim there was no such thing. But then, it's Russia - so you'll never know (almost ;))."


According to this study the Caspian Sea level between the LGM and the end of the Ice Age was about 50-80 m higher than nowadays. That is enough to flood the the area west from the South Ural. At the other hand the article says the sea level during the LGM was even lower than now, so the "block" was not here during the entire Ice Age.
How much it actually blocked the way is also a question. The South Ural had no ice cap, so the water was nowhere close to the polar ice. Still, it was pretty much in the way.

Julien said...

In the context of this paper, it may be noted that we also can reconstruct Palaeolithic protomyth by using phylogenetic tools : I have tried to do that in some peer-reviewed articles. See for instance : http://nouvellemythologiecomparee.hautetfort.com/archive/2013/01/20/julien-d-huy-polyphemus-aa-th-1137.html; http://academia.edu/3226058/Un_ours_dans_les_etoiles_recherche_phylogenetique_sur_un_mythe_prehistorique._-_Prehistoire_du_sud-ouest_20_1_2012_91-106; http://www.academia.edu/3045718/PREPRINT_A_Cosmic_Hunt_in_the_Berber_sky_a_phylogenetic_reconstruction_of_Palaeolithic_mythology._Les_Cahiers_de_lAARS_15_2012_
Myths are much better than language to go back in time, because versions of a same myth differ often less in great area than different languages. Moreover, some mythological versions that belong to one family could be found both in Eurasia and America, which mean that they passed from one continent to another with at a time when the Bering land bridge linked Eurasia and America in one same continent (have a look at the work of Yuri Berezkin for instance)

Roy said...

@ Teo
"I'm not even criticizing the paper yet"
Why the evasion? Let's hear it, so that we can assess this more rationally.

Grognard said...

Very interesting, Julien.

Va_Highlander said...

Rokus:

"Having IE cognates for wheel and chariot, and a potential shared Eurasian substrate word for horse, Chinese may be the ultimate argument against a Bronze Age Indo-European expansion."

China presents other problems for the standard Kurgan narrative, as well. The story goes that the horse and chariot provided overwhelming tactical advantage on the battlefield, a vehicle, both literally and figuratively, for an inferior number of Indo-European elites to overwhelm and dominate otherwise superior peoples.

But consider the Shang. Their chariot burials are the most splendid the world has seen, yet there is not a shred of evidence suggesting that the Shang elite were of Indo-European extraction or spoke an Indo-European language. And though the chariot certainly served as a command platform, there is no evidence that the chariot had any other tactical role throughout the Shang era. It seems to have been indicative of elite status and little else.

So, if the Kurgan interpretation of chariotry is the rule, then why is China such a glaring exception? A model, at least in any scientific sense of the word, should be predictive. That this one fails to predict should give us pause.

Jim said...

"Their chariot burials are the most splendid the world has seen, yet there is not a shred of evidence suggesting that the Shang elite were of Indo-European extraction or spoke an Indo-European language."

This understates the case. There is pretty clear evidence of what the Shang language was like and there is nothing remotely IE about it.

Why is it so hard for people to accpet that culture can pick up one or maybe two technologies and nothing else from another culture?

Rokus said...

So, if the Kurgan interpretation of chariotry is the rule, then why is China such a glaring exception?

I don't object to a certain Bronze Age IE vocabulary, but obviously this vocabulary was easy to accept as part of a Bronze Age package, by populations having a varying degree of Eurasian relatedness. This suggests the arguments for a Bronze Age PIE being per definition congruent to the primary IE expansion are invalid. Rather, shared "PIE" Bronze Age vocabulary was nothing but a Bronze Age layer on top of an older vocabulary of western groupings that had remained "connected" since earlier expansions. I gather those initial IE expansions were Mesolithic and at least partly reminiscent of the earlier expansions meant by Pagel et al. This way, China is not a "glaring exception" to some global IE Bronze Age expansion but rather indicative of a Bronze Age cultural potpourri being strong enough to constitute an important agent for Bronze Age IE convergence among already related people. Even Mallory referred at such a "broader Mesolithic homeland" hypothesis, albeit as a straw man to ridiculize the most obvious archeological alternative against his own pet theory. This is all about evolving linguistic concepts, and those considered by Mallory were just too immature.

Va_Highlander said...

Jim:

"Why is it so hard for people to accpet that culture can pick up one or maybe two technologies and nothing else from another culture?"

Because some very popular narratives depend upon the assumption that one may begin with a historic people speaking a known language and then trace that language back through time via its material culture. At the risk stating the insultingly obvious, that's why.

Rokus:

"This way, China is not a 'glaring exception' to some global IE Bronze Age expansion but rather indicative of a Bronze Age cultural potpourri being strong enough to constitute an important agent for Bronze Age IE convergence among already related people."

I'd put special emphasis on the word, convergence.

I think the least likely scenario is one in which PIE expanded once, and only once, from a single geographic location, neatly branching through time at a reasonably constant rate. Terry, if memory serves, mocks these as "Garden of Eden" scenarios and rightly so. They make for easily understood story lines but inevitably become a Procrustean bed. I do not necessarily deny the possibility of a Bronze-Age linguistic expansion out of south-west Russia. I just don't see why we must focus on this potentiality and ignore population movements resulting from the spread of agriculture, for instance, or economic and cultural interaction along prehistoric trade routes.

The story of conquering horsemen thundering across the steppe is a powerful image, both psychologically and socio-politically, and its linguistic consequences are easy to explain. The linguistic consequences of a mosaic of peoples interacting locally, with their immediate neighbors, while simultaneously embedded in trade networks extending over thousands of kilometers for a thousand years or more, that is a much harder problem, obviously, but one that cannot simply be ignored.

Of course, agriculture itself may not have had a single point of origin, either, and it certainly did not spread uniformly or instantaneously. The same may be said for other technological innovations, as well. This is rather the point.

Nathan Paul said...

IS there a study between South Indian Languages and other languages. for First person and 2nd person there is some similarities.

Ex: you,me, father , mother, sister etc.

Some languages might have spread with mothers also.(Mother tongues?)

Mt haplo M spread areas can be looked for common words between ASI, East Asia, Some Catalan words, some Saami, Finnish words.


eurologist said...

According to this study the Caspian Sea level between the LGM and the end of the Ice Age was about 50-80 m higher than nowadays. That is enough to flood the the area west from the South Ural. At the other hand the article says the sea level during the LGM was even lower than now, so the "block" was not here during the entire Ice Age.

Slumbery,

Thanks for the link. I think the article actually supports my claim. The larges "Khazarian" extension of the Caspian was actually way earlier, in the Eemian ~130,000 ya. This much later and smaller extension coincided with the quick warming after LGM, <~16.8 - 13.7 kya calibrated [my conversion], that is, coinciding with the Bølling-Allerød interstadial. And it was only about to 50m height (absolute, above sea level!), or ~50 degrees northern latitude. That's just north of Volgograd, and leaves plenty of space in the plains south of the Urals (apart from the fact that the northern portion would have been frozen in the winter, and as such passable).

As I have stated before, though, I think the then-existing waterway (spillway) to the Pontic must have been an important fishing and migration route.

Richard Wordingham said...

I think the paper has been grossly misrepresented. The only statistical result I can actually see is that matching roots are reconstructed for the meanings most likely to preserve their words. This could simply result from reconstruction work being concentrated on meanings most likely to yield long range matches if the languages are related!

I suspect doing a proper analysis of matching roots when there are multiple roots per meaning was simply too difficult.