Linguistically, that is. According to this Linguist List report from Edward Vajda, Johanna Nichols, and James Kari:
A long-sought connection between Siberian and North American language families has been demonstrated by linguists from Washington and Alaska. Professor Edward Vajda of Western Washington University (Bellingham), a specialist on the Ket language isolate spoken by a shrinking number of elders living along the Yenisei River of central Siberia, combining ten years of library and field work on Ket and relying on the earlier work of Heinrich Werner on the now-extinct relatives of Ket, has clarified the dauntingly complex morphology and phonology of Ket and its Yeniseic congeners. At a symposium held Feb. 26-27 at the University of Alaska, Fairbanks and a panel to take place Feb. 29 at the Alaska Anthropological Association annual meeting in Anchorage, Vajda shows that the abstract forms of lexical and grammatical morphemes and the rules of composition of the Ket verb find systematic and numerous parallels in the Na-Dene protolanguage reconstructed to account for the modern Tlingit and Eyak languages and the Athabaskan language family (whose daughters include Gwich’in, Koyukon, Dena’ina and others of Alaska, Hupa of California, and Navajo of the U.S. Southwest). The comparison was made possible by recent advances in the analysis of Tlingit phonology and Tlingit-Athabaskan-Eyak presented at the same symposium by Prof. Jeff Leer of the University of Alaska, Fairbanks, and by earlier work by Prof. Michael Krauss of UAF on the now-extinct Eyak language and on comparative Athabaskan, and on Athabaskan lexicography and verb stem analysis by symposium organizer Prof. James Kari of UAF. Working independently, Vajda and the Alaska linguists have arrived at abstract stem shapes and ancestral wordforms too numerous and displaying too many idiosyncratic parallels to be explained by anything other than common descent. The comparison also shows conclusively that Haida, sometimes associated with Na-Dene, is not related.
The distance from the Yeniseian range to that the most distant Athabaskan languages is the greatest overland distance covered by any known language spread not using wheeled transport or sails. Archaeologist Prof. Ben Potter of UAF reviewed the postglacial prehistory of Beringia and speculated that the Na-Dene speakers may descend from some of the earliest colonizers of the Americas, who eventually created the successful and long-lived Northern Archaic tool tradition that dominated interior and northern Alaska almost until historical times.
Vajda’s work has been well vetted. In addition to Na-Dene specialists Krauss, Leer, and Kari, who have reacted favorably, the symposium was also attended by historical linguists Prof. Eric P. Hamp of the University of Chicago and Prof. Johanna Nichols of the University of California, Berkeley, both of whom announced their support for the proposed relationship, and Bernard Comrie, Director of the Linguistics Department, Max Planck Institute for Evolutionary Anthropology, Leipzig and professor at UC Santa Barbara, endorsed Vajda’s method.
As George Bryson says in his (remarkably thorough and accurate—kudos, Mr. Bryson!) Anchorage Daily News story, “Establishing that two such far-distant language groups are closely related is both demanding and rare in the exacting field of historical linguistics.” (Thanks for the links, Patrick!)
Over 40 years ago Columbia University Professor of Linguistics, Robert Austerlitz gave a paper linking Paleosiberian Giliyak (Nivkh) with a native American language spoken somewhere (as I recall) in Washington state. I heard the paper, but don’t recall the details.
Is it possible to come up with an approximate date range for the split between the two groups, or does historical linguistics not work that way?
Very exciting stuff.
Last year I was poking around for OSV langauges. I only found twelve, including Haida, which is featured in this not-bad website.
Cool.
This is one of the reasons language death is lamentable — any language that disappears, at least without being thoroughly recorded and documented, eliminates an opportunity to make this kind of connection that sheds light on language development and the movements of ancient populations.
Vajda’s paper itself is here:
http://www.uaf.edu/anlc/docs/vajda-2008.pdf
See also:
http://www.uaf.edu/anlc/docs/leer-2008.pdf
and other links here:
http://www.uaf.edu/anlc/dy2008.html
Nostratic alert.
Wiki:The Yeniseian languages have highly elaborate verbal morphology, to an extreme found elsewhere in Eurasia only in Burushaski and, to a lesser extent, in Basque and the Languages of the Caucasus. (All of these languages are ergative as well.)
And the paper LH cited drags in Nivkh, the Na-Dene, and the Athabaskans…..
Whereas the Dravidians….. etc.
Nostratic alert.
Wiki:The Yeniseian languages have highly elaborate verbal morphology, to an extreme found elsewhere in Eurasia only in Burushaski and, to a lesser extent, in Basque and the Languages of the Caucasus. (All of these languages are ergative as well.)
And the paper LH cited drags in Nivkh, the Na-Dene, and the Athabaskans…..
Whereas the Dravidians….. etc.
I too am wary of Nostraticism, but this is different – these are cautious scholars, and there are few people who know more than Leer and Krauss about the languages of Northwest America. I happen to know Jeff Leer personally, and in private he’s actually less cautious – he believes in a Sinitic-Athabaskan connection that can be seen in morphosyntactic areas (Jeff, forgive me if I’m misrepresenting your beliefs.) This is striking because, if anything, it’s hard to think of two languages more morphosyntactically distinct on the surface than Navajo and Chinese, so his hypothesis must be based on something deep (I don’t know the particulars.) Also, it’s not based on mere lexical similarities. This is what sets these Vajda and Leer apart from the Greenbergs, Starostins pere st fils, and Ruehlens (not to disparage the above named scholars and their intriguing and provocative ideas, which I believe merit serious consideration, even if I may not hold by them.)
I’ve met Vajda at a conference speaking on Nivkh, and he was very pleasant. Nivkh sounds like the most complicated language ever, and apparently it’s getting more complicated. I suggested that the opposite process to creolization was happening — maybe all speakers of Nivkh are native speakers and the children and grandchildren, etc., of native speakers, so that none of the decomplicating motives and processes are ever at work, whereas complicating processes operate freely. Sort of like a secret language developed over generations.
I’ve read that in New Guinea, small linguistic groups want their languages to be secret, for paranoia and witchcraft reasons, so language evolution is accelerated as once-mutually-intelligible languages deliberately become mutually-unintelligible.
File under “wild speculation”, but not right next the the Dravidian origins theory.
I’ve met Vajda at a conference speaking on Nivkh, and he was very pleasant. Nivkh sounds like the most complicated language ever, and apparently it’s getting more complicated. I suggested that the opposite process to creolization was happening — maybe all speakers of Nivkh are native speakers and the children and grandchildren, etc., of native speakers, so that none of the decomplicating motives and processes are ever at work, whereas complicating processes operate freely. Sort of like a secret language developed over generations.
I’ve read that in New Guinea, small linguistic groups want their languages to be secret, for paranoia and witchcraft reasons, so language evolution is accelerated as once-mutually-intelligible languages deliberately become mutually-unintelligible.
File under “wild speculation”, but not right next the the Dravidian origins theory.
Leer’s and Starostin’s conclusions seem compatible enough: Dené-Sino-Caucasian
What appeared on LinguistList is the press release prepared by Vajda and some of the other scholars in order to emphasize the seriousness and validity of the proposal and to prevent idle speculation. Most people who have proposed cross-Pacific relationships have done so on the basis of not very systematic resemblances of words, often using just one language but not the family that the language belongs to, but what sets Vajda’s work apart is that he has looked at the resemblances in morphology and also at the systematic correspondences in sounds. Right now only the Ket language is still spoken from the Yeniseian family but there also exists a fair amount of documentation of some languages of the same family which have now died out. Vajda has taken into account all of that documentation, as well as what is known of the vast Dene language family. No other person had done such wide-ranging and meticulous comparison. It is truly a spectacular achievement.
p.s. The Dene and the Athabaskans are the same. “Na-Dene” refers to the group including Eyak (now extinct, once spoken on a small territory in Alaska), Tlingit along the Panhandle, and the large Athabaskan family which extends from Alaska, through Northwestern Canada and is also found in small pockets in the US, from Oregon to Arizona (the best-known of these languages is Navaho). Sapir thought that Haida belonged to the same group, but more recently this has been considered very unlikely, and Vajda’s work makes it even more improbable.
Also, it’s not based on mere lexical similarities. This is what sets these Vajda and Leer apart from the Greenbergs, Starostins pere st fils, and Ruehlens.
Yay, I’ve been waiting since yesterday for the obligatory upcoming jab at the Greenbergs, Ruhlens, and Starostins, and here it is. Well, Starostin-fils at least has something to say about it! I might as well start adding “demythologizer” to “historical linguist” in my job description these days.
Anyway, there’s this common perception that theories like Nostratic and Sino-Caucasian are really “based on lexical similarities” the same way that Greenberg’s mass comparison is (actually, I’m not going to talk about Greenberg, but as a side note, I’d like to mention that Greenberg’s work is based on grammatical similarities almost as much as it is based on lexical ones). Nothing could be further from the truth, and this misperception is usually carried over like an infectious disease by people who know little on the subject, but are satisfied with quoting “experts” who think they know more on the subject, but can’t even read the original works because they’re in Russian.
Nostratic and the original Sino-Caucasian are based not on lexical similarities, but on the application of the comparative method to reconstructed proto-languages. This means regular correspondences, sound laws, and etymological work on all the comparisons. In terms of “cautiousness”, both theories are light years away from the evidence presented by Vajda, which is not so much “cautious” as very much “inconclusive” – in particular, his morphological comparisons, which he claims are the key to the whole thing, are based on a very odd and questionable interpretation of both synchronic Ket morphology and the Proto-Yeniseian system (this I can swear on my father’s grave, since I actually worked quite a lot on the Ket verb in the previous decade). In any case, the myth about morphology being more stable than the lexicon is a very harmful one for comparative linguistics and needs to be disspelled. Any disbeliever is welcome to look at English and Russian and see whether the two languages have more common retentions in the lexicon or in the grammar – and this is far from one exceptional case, I assure you.
All in all, it’s just one more sad piece of evidence for me that “acceptance” of one’s theory depends not on the merits of that theory, but on your ways of presenting it. It’s who you know that counts, not what you know. I don’t know, maybe the Moscow school of comparative linguistics should start sending vodka crates to the New York Times staff. 🙂
Cavalli-Sforza relies on Ruhlen’s work and correlates it with genetic information in order to come up with a rough theory of the pre-Indo-European migration and settlement patterns in Eurasia. His book is mammoth and depends on methodologies that I don’t understand, and I’m not sure how well the genetic information corroborates Ruhlen’s linguistic theory (though the Basques are indeed genetically distinct).
A quick and dirty (and possibly over-creative) summary would be that the Basques, Caucasians, and possibly also the Burushaski, Kets, Nivkhs, and Dene are all survivals of Gimbutas’s “Old Europeans”, and that the Semitic, Indo-European, Turkic, and Finnic peoples came along (expanded) later. The Dravidians belong to the newer group, as I remember, and were originally found in Iran.
I like to joke about far-flung theories, and I’m always running into totally implausible ones, but neither I nor most other people here (I think) is totally closed to the possibility that supergroups might eventually be found. But the Basque-Caucasian-Burushaski group really does look like opportunistic lumping; it’s plausible because these are the only isolate languages in that whole geographical area.
John,
it would look like opportunistic lumping, if it were really based on nothing but the bare fact that “these are the only isolate languages in that whole geographical area” (which, by the way, is not completely true, but it doesn’t really matter). But it isn’t. I won’t bore you with material, but what do you make of at least these forms:
1st person pronoun: Burushaski dz’a – Nakh-Daghestanian *zoo, Abkhaz-Adyghe *sa;
2nd person pronoun: Burushaski un – Nakh-Daghestanian *uoo (including forms like Lezghian wu-n, Archi u-n!) – Abkhaz-Adyghe *wa.
Don’t tell me this can be coincidental – and there’s plenty more evidence like this in the personal, deictic and interrogative pronoun systems, not to mention all the other comparisons, of course. Pronominal paradigms like *(d)zV-/*w(U)- don’t just spring up independently all over the world. Oh, and pretty much the same system is also there in Yeniseian.
George, in the long run I think that many people here are persuadable. Wajda’s work helps. But we do have our priors, and we also frequently run into wild-eyed amateurish (often nationalist) linguistic theories.
Don’t take our jokes too seriously, especially not mine. Cavalli-Sforza’s use of Ruhlen, if valid, would give us tremendous insight into prehistory between about 15,000 BC and 3,000 BC or so. But at the same time, these are enormous and very difficult questions.
Oh, I can assure you no one here is looking for easy answers to difficult questions. And you don’t have to tell me how hard it is to distinguish scientific long-range comparison from crackpot theories: I have to shake all kinds of cranks off my back almost every month. I know this kind of feeling.
But the other kind of equally bad feeling is to feel yourself ignored and shunned for reasons that are objectively unjust, simply because people have not bothered to look more closely at what you are doing, and then went on ignoring or despising your work simply because they don’t have the guts to say they were wrong about their initial attitude.
I mean, no one has to accept Nostratic, Sino-Caucasian or any other long-range theory at face value. After all, the first one is fifty years old and the second one is thirty years old, and there’s about three or four people doing serious work on each; for comparison, it took Indo-European linguistics about a hundred years and many dozens of specialists to get it into real good shape. But just as the early works by Bopp and Rask and Grimm were good enough to make other people work out the particulars, so is the early work by Illich-Svitych, Dolgopolsky, Starostin, and others good enough, and the accumulated evidence serious enough to make this into a respectable field of study that warrants further work. Yet what hope is there if most people still don’t feel like noticing the difference between Illich-Svitych’s and Greenberg’s methods, no matter how hard we try to tell them?
And what really infuriated me about this latest symposium on Dene-Yeniseian is that Vajda’s approach, while moderately legitimate, is far from the best approach to long-range comparison. (I have written a long letter to the MLTR mailing list on Yahoo where I explained why this is so – if anyone here is interested, I can share). But just because the evidence has been dressed up so as to satisfy the requirements of Johanna Nichols (resulting in an odd mix of interesting and reliable comparisons with lots of wiredrawn speculation), and no mention has been made of the accursed Russian work on long-range connections of Yeniseic languages, it now happens to be “accepted”. Now this is what I would really call “opportunistic behaviour”. No hard feelings towards Vajda, though – he is just playing by the rules of the game.
George, I really appreciate your presence here and willingness to discuss these things with both good humor and evidence. For what it’s worth, you’re making me more willing to consider the long-range comparisons you mention.
To languagehat – you are very welcome, and on my part, thanks for hosting a blog that’s so friendly and open-minded. And I am always ready to explain, share opinions, and show evidence if necessary. The best people to evaluate it are those who are not afraid of looking at the actual data – and there’s plenty of it lying around (yep, consider this as hidden advertising of the Tower of Babel site, although, of course, there are other sources as well).
“Oh, I can assure you no one here is looking for easy answers to difficult questions.” But those should always be the first things to look for. You settle for difficult answers only if easy answers don’t work.
Ed Vajda teaches at Western Washington University, so it is not hard to find out about him and his work. He speaks Russian very fluently and has been to Siberia many times.
As for him “playing the game”, if this means following the principles of comparative-historical linguistics, what is the problem? As for the suggestion that “the evidence has been dressed up so as to satisfy the requirements of Johanna Nichols”, I find this insulting to both Vajda and Nichols, both very talented, insightful and hardworking people with independent minds (and who both read and speak Russian).
“I’ve read that in New Guinea, small linguistic groups want their languages to be secret, for paranoia and witchcraft reasons, so language evolution is accelerated as once-mutually-intelligible languages deliberately become mutually-unintelligible.” (John Emerson above)
John, interesting — that make a lot more sense to me than the often-repeated inference that it has something to do with biodiversity (ie., to some extent places with a lot of species of flora and fauna also have a lot of language diversity). Since biological evolution and language evolution obviously happen over very different time scales, it makes no sense they would be somehow related.
Dear Marie-Lucie,
I know Vajda, know much about his work – arguably more than he knows about research on Dene-Caucasian – have corresponded with him both in English and Russian, and only two days before have also accepted his personal apologies for pushing all previous research on the external connections of Yeniseian under the carpet in his presentation. Like I said, I have nothing against him personally, but I do have something against the general environment which he has to fit in in order to be able to work on long-range comparison.
My remarks were not meant to be insulting, but to be truthful. If you consider them insulting, then tell me how was I to consider this statement in Vajda’s article:
“I will not, however, generally attempt to offer
Proto-Yeniseic forms, which have not yet been worked out systematically; in fact, at present we still lack a firm reconstruction of the basic Proto-Yeniseic sound inventory. Instead, I
will use cognate forms in the attested Yeniseic daughter languages to illustrate the points I am trying to make.”
From any point of view, this statement is FALSE, because in Sergei Starostin’s long and detailed article from 1982 (which you can find available on our site) he has, for the first time, offered a systematically worked out reconstruction of the basic Proto-Yeniseic sound inventory. Like every reconstruction, it can be questioned on some counts, but it is a thorough reconstruction with plenty of material and a system of rigorous phonetic correspondences. So, what is it again that we “lack”? Furthermore, Vajda certainly knew about this article. Yet not only are its results not made use of in his paper, even the very existence of it is not mentioned.
This, and several similar cases, is the “dressing up” that I mentioned. But I can elaborate further. It is a well-known fact that Johanna Nichols insists that paradigmatic evidence has to be presented in a proposal of genetic relationship if it is to become a valid hypothesis. (She never really said why, but that’s another question). So, according to those rules of the game, Vajda places the major accent on paradigmatic evidence. But there just isn’t enough GOOD paradigmatic evidence. What does he do then? He offers BAD paradigmatic evidence – based on grammatical markers in Proto-Yeniseian that either don’t exist (like the TAM *-Ga- morpheme) or whose meanings are established on one or two examples on a whim (like the spatial prefixes, the meanings for which Vajda actually changes from paper to paper).
As for “following the principles of comparative-historical linguistics”, well… The proper application of these principles, as every handbook of comparative linguistics will tell, is as follows: (a) reconstruct Proto-Na-Dene, (b) reconstruct (or use an existing reconstruction of) Proto-Yeniseic, (c) establish a system of regular correspondences between the two, (d) demonstrate the system on a sufficient number of cognates. In Sino-Caucasian (without Dene so far), all of these principles are followed. Vajda’s work in its current form completely bypasses points (b) and (d), and makes only partial use of point (c). Of course, this is only the beginning for him, and I sincerely hope that his work will be continued and refined. But so far, it certainly has only moderately followed the requirements of the method.
Also, on the New Guinea phenomenon:
this is very wild speculation indeed. To prove that native tribes intentionally accelerate language change, one would have to have historical records of New Guinean languages that are really unavailable. I can understand the intentional “concealment” of certain lexical items as a taboo phenomenon, but that’s not really the same thing as intentionally accelerating language change on all levels (and besides, tabooed lexics as such does not always go out of the language – taboos can be and often are lifted).
I would think the simplest explanation for New Guinean diversity is just the fact that it’s been populated so long ago, as well as the tendency for Guinean tribes to become isolated from each other due to landscape particularities (and maybe other factors).
A fascinating topic and an interesting discussion thread, how could I resist jumping in?
1-I’d suggest, if the relationship is finally accepted by the mainstream, that this new family be called AMERASIAN, since it is the first accepted language family with separate member languages found on both continents.
2-The Ruhlen-bashers out there should look at the written (March 3 08) version of Vajda’s paper: in footnote 2 he points out that while 75% of Ruhlen (1998)’s Ket/Na-Dene are coincidental similarities, he did actually detect the genuine cognates for HEAD, STONE, FOOT, BREAST, SHOULDER/ARM, BIRCH/BIRCHBARK, OLD, BURN/COOK. Despite this, members of the linguistic establishment will doubtless praise Vajda and (continue to) bash Ruhlen (and others). It’s an interesting example of a point Sheila Embleton once made, that historical linguists today consider failure to detect an existing relationship far more acceptable than mistakenly claiming a non-existing one to exist.
3-I disagree with George Starostin: in my experience morphology, especially inflectional morphology, is remarkably stable. Granted that English and Russian have little morphology in common: yet the morphology of both languages is chiefly inherited from Proto-Indo-European, and English and Russian differ in that most of what the one language has kept the other has lost. Comparing Russian to Spanish or (Modern) Greek, for example, the morphological similarities are far more visible to the untrained eye (person-marking affixes on the verb, for example) than the lexical cognates. As I have mentioned on another thread, I am working on a paper wherein I attempt to reconstruct Indo-European using data from Modern Indo-European languages only, and I assure you that a HUGE deal of Indo-European morphology can be reconstructed.
4-I do agree with George Starostin that the linguistic diversity of New Guinea is more likely related to the landscape that to deliberate language engineering: after all, cryptolects and secret languages are a widespread phenomenon, but their long-term impact on a language seems moderate to non-existent. Martin: you are quite right, the CAUSES of biological and linguistic diversity can’t be same, because of the (vastly!) faster pace of linguistic change: but the same factors which PRESERVE biological diversity (a hot and humid climate and a highly rugged landscape) will also preserve linguistic diversity, or contribute to creating it, *given enough time*. The latter factor is why Madagascar, a hotbed of biological diversity, only has a single language: human beings have not lived there long enough for the language the first settlers spoke (Malagasy) to diversify.
historical linguists today consider failure to detect an existing relationship far more acceptable than mistakenly claiming a non-existing one to exist.
Well, of course! How can you take any other attitude if you call yourself a scientist? I think those scientists who for so long refused to admit, say, the possibility of continental drift have far less reason to be embarrassed than those who fell for phrenology, spiritualism, or Lysenkoism. One can argue about the point at which it makes sense to consider suggestive new evidence, but the default approach has to be conservatism. And “75% of Ruhlen’s Ket/Na-Dene are coincidental similarities”—you wonder why people don’t take him seriously? If you throw enough vague similarities against the wall, some of them are bound to stick when a careful scientist looks at the evidence. That doesn’t make you right.
i>To prove that native tribes intentionally accelerate language change, one would have to have historical records of New Guinean languages that are really unavailable.
What I had was an observation of present behavior, its extrapolation into the future, and a speculation that perhaps the past was like the present. The author observed that tiny languages seemed to be maintaining themselves rather better than theory would expect.
This is the end of what I can say since I can remember my source. I am just responding to the italicized criticism.
To Etienne:
Your work on reconstructing Indo-European from modern languages sounds very interesting – I’d very much like to see the results once it’s done. But let’s see, Spanish vs. Russian (let’s take a related verb, ‘see’):
‘to see’: vide-t’ – ve-r
1sg: ve-o – vizh-u
2sg: ve-s – vidi-sh
3sg: ve – vidi-t
1pl: ve-mos – vidi-m
2pl: vei-s – vidi-te
3pl: ve-n – vidya-t
That’s, what, 3 evident similarities out of 6 (1sg, 2sg, 1pl). Now let’s take Finnish:
tiedae-n ‘I know’
tiedae-t ‘you (sg.) know’
tiet-aeae ‘he knows’
tiedae-mme ‘we know’
tiedae-tte ‘you (pl.) know’
tietaev-aet ‘they know’
Hey, look, 3 evident similarities out of 6 again (all of the plural forms look remarkably like Russian)!
This is just to show that morphological evidence alone is a poor guide if you’re not dealing with closely related languages. I know that in the end one would succeed in scraping together somewhat more grammatical evidence between Spanish and Russian than between Russian and Finnish, but only barely so (and yes, I’m aware that in reality all of the above Spanish endings are related to the corresponding ones in Russian, but without help from the ancient languages it’s pretty hard to see).
Also, from a purely logical standpoint what I was meaning to say is that morphology CAN be stable over a certain period of time, but it also CAN be very highly unstable or even just collapse altogether, as it did in English or in Chinese or in the Romance nominal declension system (verbs do fare a little better in these languages). This means that if there is plenty of lexical evidence for relationship, but little morphological one, this cannot be qualified as lack of relationship. And in languages with such complex morphology as Dene-Caucasian (pretty much all of its members are very complex in that respect, except for Sino-Tibetan) it will almost certainly undergo lots of restructuring over short periods of time. Heck, Ket and Kott are two very closely related languages and it’s still awful hard to reconstruct their verbal morphology. Suffice it to say that the former has primarily prefixal personal markers and the latter suffixal ones…
Hat: there’s a difference between “falling for” such things as phrenology and refusing to even consider the possibility of new theories, and the Americanist establishment definitely is guilty of the latter sin today. I’d like to ask a seemingly provocative question: what is the difference between Ruhlen and his 1998 Ket/Na-Dene proposal, with three-quarters of his etymologies wrong according to Vajda, and Sapir’s original claim that the Ritwan languages are related to Algonquian? Sapir thought fewer than half of his proposed etymologies would prove to be correct, and Ives Goddard later wrote that this was “wildly optimistic”. I have the impression that if Sapir had proposed his Ritwan-Algonquian link today he would have been tarred and feathered (in print!), and nobody would subsequently have touched the theory with a ten-foot pole. To the detriment of science, since the Ritwan-Algonquian theory, happily, was in fact carefully examined by later generations of scholars, and ultimately shown to be correct.
Professor Starostin: I agree with much of what you say, except for your claim that data from ancient languages would be needed to show that the Spanish and Russian 3sg, 2pl and 3pl endings are cognate. I would maintain that data from Modern Romance and Modern Slavic would be quite enough to show the kinship. To take a single example: the Italian form VEDE-TE, Romanian VEDE-TI, and Galician VE-DES, would suffice to indicate that the Spanish 2pl ending must originally have had a dental stop, and indeed the regular voicing of intervocalic stops in Spanish, Portuguese and Galician (preserved in Italian and Romanian: compare Italian VITA, Romanian VIATA, and VIDA in all three Iberian languages) would clearly indicate an original *te(s) ending. Italian and Romanian, incidentally, would also point to Spanish VER as having originally had an intervocalic -*d-, bringing it that much closer to Russian. Crucially, in the case of Finnish, Fennic and Uralic data won’t make the forms appear much more similar than they presently are.
Etienne: yes, you are right, Romance verbal morphology, when scrapped and assembled together and reconstructed, will yield more in common with Russian (which is a very good sign of relationship – the older you get, the more parallels you have).
But when we are dealing with higher level relationship, including long-range hypotheses like Dene-Yeniseic, the situation is different. However old Proto-Yeniseian and Proto-Na-Dene may be, they’re clearly more distant from each other than, say, Proto-Germanic, Latin, Ancient Greek, and Proto-Indo-Iranian. It’s rather as if we had, say, Russian, Hindi, and German and were pressed to demonstrate their relationship based on morphological paradigms. Well, we COULD have bits and pieces of paradigms, but they’d hardly be much better than the stuff I showed above for Russian and Finnish.
On the other hand, binary lexical comparison would be much better. For Russian and German (or Russian and Spanish, for that matter), we’d have a hard time filtering out shared borrowings, but we would still end up noticing they follow different patterns from that of the inherited lexicon. We would even be able to establish many non-trivial correspondences, e. g. prove with several good examples that serdce and corazon ‘heart’ go back to the same root.
This is why I do not think that morphology should play a particularly special role in long-range comparison, or that lack of morphological evidence should put a stop to all further kinds of such comparison. I’d rather see grammatical morphemes judged on more or less the same level as basic lexicon entries: relatively stable language entities which should be used for comparison as one – but only one – of the high priorities. Certainly we would be out of our minds not to pay serious attention to pronominal elements, for instance (see the Sino-Caucasian pronominal pattern above). But with a caveat: since they form a connected subsystem of their own, they have an annoying habit of sometimes going out of use all at once – or coming in all at once, which is something that almost never happens with basic lexicon.
Professor Starostin-
I certainly didn’t wish to imply that morphology should have priority over the lexicon: indeed, I had some good thread discussions on this blog (February 18, “Crackpottery and credulity”, and December 14, “Two blogs”) with Marie-Lucie on the topic of the demonstration of genetic relationship, where I argued (inter alia) that lexical comparison alone could certainly be adequate.
And I only half-agree with what you say about morphology: it seems to me one should distinguish morphological change/loss triggered by sound changes versus change triggered by other factors. Hindi, English and Russian are so morphologically different chiefly because of morphological loss triggered by sound change: if one compares phonologically conservative Indo-European languages –say, Spanish/Portuguese, East Baltic and Modern Greek, which are practically the only modern national Indo-European languages which preserve final Indo-European *o, *a, *os and *as–the morphological similarities are more impressive than the differences.
Finally, Slavic comparative data would also strengthen the Russian-Spanish comparison: comparison of the Russian 3rd plural ending /jat/ with Polish nasal /e/ (a regular sound correspondence, NOTA BENE!) would make a reconstruction of the Proto-Slavic ending as a nasal vowel + /t/ quite natural, and since Romance-internal evidence would point to Spanish 3pl /n/ having originally been /nt/, the historical relationship between the two morphemes would be hard to miss.
Etienne:
I think if you half-agree with what I say about morphology, that should satisfy the both of us because I never said there aren’t, or cannot be, morphologically conservative languages. Certainly reconstructing Indo-European morphology on the basis of Spanish, Lithuanian, and Modern Greek is a less hopeless endeavour than the same on the basis of English, Hindi, and Russian. (Even so, you wouldn’t get real far with nominal morphology, though). My only point is that stability of morphology should not be taken by linguists as some sort of universal norm, and that demonstrating genetic relationship on the basis of numerous solid lexical correspondences is preferable to squeezing out fragmentary bits and pieces of possible former paradigms just because you’re always expected to back your propositions up with paradigmatic evidence.
And please don’t call me Professor Starostin, that’s too formal. Just George will do.
I am not a professional linguist,but I
have heard that some linguists have also found
striking similarities between the Na Dene
languages and Tibeto-Burmese languages.
The Tuvans,aTurkic people living on the
border of Mongolia,have been found to be
genetically very close to native Americans.
Some may be descended from Turkicized
Kettic peoples.
Interestingly, DNA just proved several back-migrations of the ancient North Americans into NE Asia (at least twice, about 4,000 and 1,500 years ago). As well as a Jomon back-migration to the Asian Mainland (well, the latter may be a less dramatic journey than I just described if the Jomon peoples also lived on Korean peninsula).
https://www.cell.com/current-biology/fulltext/S0960-9822(22)01892-9
Note the Siberian definition of “Neolithic” by pottery rather than agriculture.
Thanks. I don’t really understand why the Altai, the Primorskiy, and the Kamchatka genomes are treated in the same paper. They are treated completely independently — naturally, since they seem to throw no light on eachother.
Altai: Very interesting that the Altai HG population is a likely source for the non-Afanasievo element in the two Dzungaria genomes mentioned here, and that the admixture was as recent as within 1-5 generations.
Primorskiy krai: One added specimen from the Russian far east, some 100 km east of Vladivostok and 250 km SW of the somewhat later Devil’s Gate cave. The interesting thing is that it shows additional genetic affinities with Jomon, in a proportion already seen in one outlier from the coastal Boisman site in the same region. The Jomon association could be important for understanding the history and early dispersal of pottery, which may have developed in the aquatic harvesting cultures in this region, maybe in Japan, maybe on the continent.
Kamchatka: Three new specimens. They’re tagged “500uncalBP” as a group, but the dates of the individual fragments range from ~1100uncalBP to ~1700uncalBP, whatever that may mean. It’s these specimens that allow the authors to estimate the time and impact of back-migration from America. As Dmitry says, the ancient back-migration is dated to ~5000 BP based on admixture in ancient populations. Recent populations close to the Bering Strait have less but more recent North American ancestry, and the time of admixture is estimated to 1500 BP. As I read the paper, this is not interpreted as a separate admixture event at 1500 BP but as a result of the ancient admixture being continuously renewed with gene flow across the Bering Strait. I’ll add that this may be refined with more ancient data from the region. I’d think that an old admixture in populations close to the Bering Strait may turn up as a more recent admixture in populations further away, or in recent peoples whose main genetic ancestry is to be found elsewhere.
I guess the authors figured it was better to have one Current Biology paper than three Least Publishable Units – though if we’ve really reached the point that a bunch of ancient genomes is a single LPU just because they’re from about the same time & place, something has probably gone wrong.
It’s very customary to collect and analyze all available ancient bones from a wide region and to analyze it together (and to compare with a lot more ancient and contemporary DNAs from prior studies). Especially in the regions of unrecorded history, I’d say that it’s completely legit, because it’s hard to make bulletproof hypotheses, which groups developed into which later groups and how the migrations spread. That’s why the previous Eastern Siberian ancient human DNA paper from 2019 also spanned the whole vast area and tens of millennia time-wise. The Eastern Steppes papers did pretty much the same. The African paper tried to get everything available of the sparse bone DNA record there, too. Of course some of the publications deal strictly with one dig site / one culture, but it’s just as typical to go for a broad picture from as many samples as possible.
I know. I think what I mean to say is that this paper didn’t give much in the way of broad picture, but instead three or four separate local studies gathered in one paper. Each of them is very interesting, but the assembly is somewhat random.
The unifying topic is, “we asked for the ancient bones from Russia’s East and that’s what we received”