zompist bboard

I been hearing a lot about Proto-Nostratic lately, but to me it seems to be quite . . . wacky to say the least.

So, how credible is Proto-Nostratic ?

Is it a controversial topic like recognising Korean and/or Japanese as being an altaic language ?

Short answer: not very.

Long answer: Well... It all hinges on how well you think our modern methods of reconstruction work when applied at that sort of time depth, and how well the method itself is applied. I've seen criticisms of Nostratic that point out how poor their method actually is.

Has anyone really been far even as reconstructed to use even go want to do sound more like?

Polka Dot wrote:I been hearing a lot about Proto-Nostratic lately, but to me it seems to be quite . . . wacky to say the least.

So, how credible is Proto-Nostratic ?

Is it a controversial topic like recognising Korean and/or Japanese as being an altaic language ?

There almost certainly are tree-like relations linking some known eurasiatic families into superfamilies.

Is Nostratic one of them? Well, since 'Nostratic' seems to have been used for just about any jumble of European families you could imagine, I guess some form of it must be fairly near accurate.

I guess there are two explanations for how this could occur. One would be that Nostratic was a Siberian family. Early Siberians expanding after the Ice Age brought the language into the eastern european steppe, where the Yamnaya (PIE) developed. Other Siberians took the language east, giving it to the ancestors of the various east asia families - uralic, tungusic, mongolic, turkic, japonic, korean. We know the ancestors of the tungusic peoples have been there for tens of thousands of years. Alternative, early PIE travellers may have brought the language to the east, though that seems rather too late if we really want it to be so widely developed there (there were Yamnaya in the Altai steppe, historically, presumably speaking an Indo-European language).
Why would Kartvelian be part of that family? Well, there were Siberian migrations into the near east. So it's theoretically possible that some of their languages ended up in the Caucasus, although genetically the Kartvelians appear to have been there for a very, very long time.
In this hypothesis, we may also expect to see this 'Nostratic' be an ancestor of sino-tibetan and of various American languages.

Dravidian would be much less likely to be a member. Yes, they probably migrated from the west, but they don't have a lot of Siberian ancestry and their migration would probably be too early.

Afroasiatic would almost certainly be right out. We know that there was an African migration into the middle east after the development of farming there. It seems likely that Afroasiatic was African, and that that migration was the arrival of the Semitic subfamily in the middle east. The Siberian hypothesis would have to entail those africans coming to the middle east, adopting a siberian language, migrating back into africa to spread the language excessively (and north africans have basically no siberian ancestry), then migrate back as the semitic languages (and then migrate back to africa again in historical times as the arabs). This seems implausible.


Alternatively, nostratic could be associated with early caucasian farmers. Kartvelian would be the main descendent. We know, however, that there was massive migration from the caucasus both into south asia (where they could develop the dravidian language) and into the steppe (where they combined with the europeans to produce the Yamnaya (i.e. PIE). So that would link Kartvelian, Dravidian and Indo-European. Afroasiatic is still less likely, but at least now feasible - a very early Nostratic spread could have put the language into Africa, where it could have spread as Austroasiatic, BEFORE the migration into the middle-east. However, in this version, it becomes much, much less likely that the uralic and altaic languages would be related. [It would be more likely, however, that the Neolithic European languages might have been related].

Any such hypothesis, however, would seem to involve a lot of groups adopting foreign languages, while other neighbouring groups keep their own languages. [Like: how come north caucasian families didn't adopt Nostratic?] I'm not really sure what the logic motivating the hypothesis is.

----

More generally, while it may not be a terrible idea to bear in mind general patterns of apparent relatedness, I think worrying about Nostratic is counterproductive at present. The macrofamily involves a large number of relations between seemingly unrelated families, and it would surely be more productive to focus on demonstrating one of those constituent relations first. When we cannot say safely that Tungusic and Mongolic are related, how can we possible hope to demonstrate a connection between Tungusic and Afroasiatic? Yes, an attempt to demonstrate Tungusic-Mongolic, or Indo-Uralic, or Kartvelian-Dravidian, might want to keep the idea of an earlier proto-language in their peripheral vision, to inform the reconstructions they do. But trying to focus on all parts of the purported family at once would seem to be likely to drown the investigator in false positives and generalisations.

In my opinion, Nostratic is unproven, but not entirely baseless, at least regarding the group called by various names such as "Northern Nostratic", "Mitian" or "Eurasiatic": IE, Uralic, Altaic (which in itself may or may not be a family) and a few others such as Eskimo-Aleut. These languages seem to have some morphological elements such as pronoun roots in common which are hard to explain by either coincidence or borrowing, so a common ancestor seems likely. Alas, no regular sound correspondences have been established yet (though proposals exist). However, I consider the inclusion of Kartvelian very doubtful, and that of Afrasian and Dravidian utterly unfounded.

The more vaguely you define "Nostratic", the more credible it is, of course. "At least some of these dozen language families are related" is, intrinsically, a lot of more likely than "all of these dozen language families are related, and here's a list of 1253 word roots and 183 grammatical endings from their common proto-language". (But, of course, the former formulation is also a lot less useful for anything at all.)

I would agree that sub-proposals like Eurasiatic are, well, "more likely" seems too bold, but at least more explorable. Any links to Afroasiatic are probably premature, since reliable etymological work within the family as a whole has barely been even developed yet. A look at the data presented from AA often reveals a lot of material that's actually based just on Semitic, perhaps even just a small subset. Dravidian is geographically a bit far off as well, and suffers from similar problems: there is much too little work on Proto-Dravidian proper anyway, and most people just seem to use Old Tamil as a stand-in, meaning that they're about 150 years behind IE studies in their understanding. Many Dravidian minority languages have not even been documented in any kind of consistent detail. Same goes also for many Chadic or Cushitic ones.

And it's only etymological work that can hope to get us anywhere. Looking at genetics and migrations proves nothing, since assumptions about what this or that Neolithic population spoke are almost pure speculation. We might be able to put together plausible scenarios for the parents of a few older language families — PIE or Proto-Semitic — but e.g. the reasonably well-known history of the Mongolic languages only stretches about one millennium back; archeologically-based extrapolation could add maybe another millennium. But go four or five millennia back, and pre-proto-Mongolian of the time could have been just about anywhere from the Caspian Sea to the Bering Strait…

Tropylium wrote:
And it's only etymological work that can hope to get us anywhere. Looking at genetics and migrations proves nothing, since assumptions about what this or that Neolithic population spoke are almost pure speculation. We might be able to put together plausible scenarios for the parents of a few older language families — PIE or Proto-Semitic — but e.g. the reasonably well-known history of the Mongolic languages only stretches about one millennium back; archeologically-based extrapolation could add maybe another millennium. But go four or five millennia back, and pre-proto-Mongolian of the time could have been just about anywhere from the Caspian Sea to the Bering Strait…

Well, sure, anything could be anything and nothing can ever prove anything. But I don't think that's very meaningful or important. Even the comparative method does not actually prove anything, it only demonstrates that either there's a real genetic relation or a really big coincidence. But of course, since any language can be derived from any other via regular sound changes, providing the sound changes are complex enough, etymological work can only offer us a sliding continuum of probable to improbable.

Which is exactly what scientific study can offer us in this case too. True, looking at the question scientifically - through archaeology and biology - won't tell us anything for certain. But it can tell us things probabilistically. Like the fact that it's almost impossible to construct a plausible scenario in which the same proto-language can lead to both Japonic and Afro-Asiatic (but not to a whole range of other eurasiatic and african languages).

Similarly, it seems exceptionally unlikely that pre-proto-Mongolian could have been spoken near the Caspian Sea at that point in history. First, you'd have to assume that some of the population in that area spoke PPM while some spoke PIE. Given how few people there were in that area at that point, and how genetically homogenous they were, that seems unlikely. Not impossible, sure, but unlikely. Then you'd have to say that the PIE migration into the Altai was actually PPM, while leaving no apparent trace of the PPM population in any of the other PIE migrations - so the parent population splits in two in a way that just happens to reflect the linguistic division but not any genetic division. That's all unlikely. Then the PPM speakers in the Altai have to give their language, but none of their genes, to a genetically Mongolic population. That's unlikely too. It can happen - Hungarians are genetically mostly IE despite having adopted a Uralic language. But you'd have to assume a further invasion from the Altai into Mongolia proper, significant enough to change the dominant local language, but so insignificant that it's left no cultural or archaeological or genetic (either from period remains or in the modern population) traces. And this would have to happen not in the scenario of a nomadic band conquering a populous settled peoples, but between two pastoralist groups with low population densities.

This is all theoretically possible. And if there were a particular reason to suspect that Mongolian were a very close sister-language of PIE, that's the sort of scenario you might look at (although even then it would be more reasonable to suspect the migration had happened in the opposite direction). Fans of Indo-Uralic, for instance, might point the finger at those eastern "PIE" populations, arguing that PIE proper was instead born in Europe, and that the original kurgans were speakers of Indo-Uralic, and that the eastern, Uralic branch was gradually assimilated by a genetically Uralic population.

But in the absence of any actual evidence for such an Indo-Mongolic hypothesis, we have to rationally conclude that it is, prima facie, highly implausible. Not, of course, absolutely physically impossible - but highly implausible and irresponsible to assume when far more parsimonious models are available. Likewise, Mongolian could have been brought by space aliens. We can't emphatically rule it out. But in the absence of any actual reasons to believe it, it's also disingenuous to suggest that we have no non-etymological reasons to disfavour the space-alien hypothesis.

…If I say "anywhere from the Caspian Sea to the Bering Strait" , that obviously does not mean that I think that we should assume a flat probability distribution across that area, or that the vicinity of the Caspian Sea would be an especially likely candidate. It means I've chosen it as a (somewhat arbitrary) cutoff point circa "improbable, but perhaps still just barely possible enough to consider".

That said, all data I've seen has indicated that the Yamna culture stuck to the northern side of the Caspian Sea, and Khorasan was not IE-ized until the expansion of Indo-Iranian (the Tocharians most likely took a more northern route on their way to the Tarim Basin). I do not think we have good default hypotheses available on what, exactly, people spoke around there before II. You could indeed try to dump in immediate sister languages of IE, but that is but one option among many. We could also consider e.g. pre-Burushaskis, NE Caucasians, para-Dravidians, para-Sumerians… or just something unknown and extinct altogether.

I also don't think we can just assume the genetics of earlier Central Asian populations either. The recorded history of the Eurasian steppes shows repeated expansions and population replacements every few centuries; this can gradually wash away original genetics even without total population turnovers. Already the Mongols themselves have demonstrated this a couple of times, seeing e.g. how Pre-Proto-Mongolic circa 0 CE is usually located in Manchuria (as the Donghu), and only expands westward to replace the earlier languages in today's Mongolia (perhaps the Xiongnu, whomever they were exactly) some centuries later.

(Of course, if there has been actual paleo-DNA extracted and sequenced from the time and area we're talking about, and you're going off of those results, feel free to inform me about that…)

Another fact about the sociohistory of the steppes is that, yes, historically attested nomadic confederations were often polyethnic (Mongolian-Turkic, Hungarian-Khazar, whatever.) Some degree of language shift is basically trivial under these conditions, depending on relatively minor changes in various groups' prestige, and iteration can again relatively quickly lead to a language's disappearence. The complete disappearence of once-major languages such as Hunnic, Khazar, Avar, Alan, Sogdian, Khwarezmian, Bactrian over a period of no more than 1500 years clearly does not require assuming that every case involved the catastrophic total destruction of the ethnic group speaking the language.

So, to stress my original point, which perhaps went too implicit — on the steppes and their neighboring areas in particular (including also the sparsely populated tundra zone), we clearly cannot take the representation at time X, and assume that the people and their languages at time X-1000 or X-5000 would have most likely still been in the same locations.