Nostratic discussion

From: Piotr Gąsiorowski
Message: 40
Date: 1999-09-24

----- Original Message -----
From: Piotr Gąsiorowski
To: Alexander Stolbov
Sent: Thursday, September 23, 1999 12:57 AM
Subject: Odp: Discussion

----- Original Message -----
From: Alexander Stolbov <astolbov@...>
To: Piotr G?siorowski <gpiotr@...>
Sent: Wednesday, September 22, 1999 12:40 PM
Subject: Discussion

> *I had been wrighting this letter for a few days, that is why it isn't directly connected with our current topics.
> Dear Piotr,
> I do hope, that we are able to get a lot of useful things for both of us from our discussions, while we are very interested in
> the same subjects but are standing on quite different (if not to say contrary) positions. And this is great! I think, we can
> succeed if we will not try to prove each other that he (you or me) is wrong but take into consideration arguments of the
> disputing side in order to improve (or correct, or deny at last - who knows?) our own conceptions. Sorry for the sentences,
> which might look selfevident.
> If you don't mind, let us clarify our principal positions.
> A) Existence of super-families.
> As far as I know, you don't believe in existence of the Nostratic super-family. What about other super-families - Austric
> (Austronesian + Austroasiatic + Tai), Sindsch (Niger-Congo + Nilo-Saharan) etc.?

From a scientific point of view, the question is not whether one believes in superfamilies or otherwise but whether their reality can be demonstrated. If I saw really convincing arguments in favour of, say, Nostratic or even Proto-World, it would be stupid of me to refuse to accept them; I'm not a fanatical anti-Nostraticist. The trouble is that what I've seen so far leaves me unconvinced.
Lists of similar roots may impress laymen, but as a historical linguist I know too well how easy it is to find sets of good-looking correspondences between completely unrelated languages. R.L. Trask has recently done so for "Proto-Nostratic" and Basque -- not to prove that Basque is another Nostratic language, but merely to show how easily spurious cognates could be produced. His comparanda are at least as good as many of those cited by Dolgopolsky and other Nostraticists.
The problem is all the more serious for the fact that Nostraticists routinely compare hundreds of languages at the same time rather than doing pairwise analysis in the first place. (It is true that in doing so they at least use the comparative method, albeit in a slightly suspect way. They are therefore more catious than Greenberg or Ruhlen, who use completely unconstrained mass comparison to define Amerindian and other macrofamilies.) It is particularly easy to hunt for lookalikes if you have so many lexicons at your disposal and if you practise "reaching down", that is, allow one language to stand for a whole branch or family, so that e.g. a Mari word otherwise unattested in Finno-Ugric or Samoyedic may nevertheless represent the Proto-Uralic lexicon, and a little-known Omotic dialect stands for Afroasiatic because it happens to have a word which seems to match something in Georgian and Malayalam. What does Dolgopolsky say in such cases? "There are cognates in three families, so the word is in all likelihood Nostratic."
Morphological matches would be much more convincing than lexical correspondences, but then Dolgopolsky claims that Nostratic was an analytic language, basically without inflections and with very little word-internal morphology. That's rather unfortunate, isn't it?
Uralic is a well-established family, yet there are only about 140 securely reconstructed Proto-Uralic roots. I haven't counted convincing IE word reconstructions, but I don't think the numbers amount to more than a few hundred (Pokorny's dictionary is a respectable work but contains a lot of worthless rubbish). For a more obscure grouping like Nostratic one would expect a lower number; however, Dolgopolski claims he has collected about 2300 reconstructible roots. The number is impressive, but the equations are not as good as in the case of Uralic or Indo-European.
 > I believe, that super-families are just "normal" units of linguistic classification (taxons), like families, branches,  groups,
> subgroups etc.
Superfamilies are not "normal" taxa. One of the reasons is that they are almost by definition controversial and many linguists do not accept them. In general, "familes", "branches" and other taxonomic ranks are somewhat arbitrary, I mean there is no objective definition of a branch etc. We use such terms for convenience, but in reality they may stand for various genetic groupings of languages of different size, time depth and so on. Some are widely accepted, others are controversial. There are many "Altaicists" who do not accept Altaic (even without Japanese and Korean) as a valid genetic grouping. Is Altaic a family, then, or a phylum, or a macro/super-family?
My point of view is as follows: any genetically based taxon (whatever we call it) must be defined like the clade in biology -- as a single language plus all the languages descended from it. So Indo-European consists of PIE plus all its offspring, Germanic is PGmc plus all its descendants, etc.
The requirement that a taxon should be monophyletic is important: e.g. Romance is not a valid taxon if it doesn't include Latin. A superfamily is a "normal taxon" if it can be shown that it is an entity of that kind. Nostratic is a well-formed taxon if it can be shown that IE, Uralic, Yukaghir, Mongolic, Turkic, Tungusic, Japanese, Korean, Afroasiatic, Dravidian and Kartvelian go back to a single protolanguage, and that the taxon includes all the languages that derive from that most recent common ancestor. When I see a proof of that, I'll start regarding Nostratic as a family. Illich-Svitych only brought to attention hundreds of lexical similarities among these languages. In my opinion, most of them are coincidental, some may result from universal sound symbolism, and some (not too many) represent widespread loanwords. None of those that I've seen strike me as likely cognates. I'm looking forward to see the long-promised Nostratic Dictionary, but for the time being I find it reasonable to remain sceptical.
So far there are various mutations of Nostratic theory (Illich-Svitych & Dolgopolski, Shevoroshkin, Ivanov, Bomhard, Greenberg & Ruhlen), not always compatible with one another. They can't all be true at the same time, and to a disinterested observer it may look strange that Nostratic sometimes excludes Afroasiatic and sometimes includes Eskimo-Aleutian.
> Moreover, I believe, that in time (every a few thousands years) the rank of every unit increases. For instance, if we examine
> the
> situation of 3000-4000 before present we may consider Proto-Germanic and Proto-Iranian as languages belonging to the same
> linguistic group ("Indo-European group" of "Nostratic family"), i.e. the difference between them is as big as between, for
> example, Polish and Russian. And vice versa, if dialects of Polish and Russian would develop free for several thousands years,
> their descendant languages would belong to related but different linguistic groups ("Polish group" and "Russian group" of
> "Slavic
> family" of "Indo-European super-family" of "Nostratic hyper-family'  ;-)

Exactly. This results from the cladistic nature of all correct genetic classification. There is, however, an important difference between biology and linguistic. We know that all living creatures are members of a single clade -- that is, they are ultimately related. The unity of the genetic code guarantees that. There is no analogous proof for languages. The unity of the innate language instinct only shows that we are naturally predisposed towards a universal TYPE of communicative code, and that there are universal constraints on human language (as a class of systems). Still, there are lots of actual codes satisfying those constraints and this may always have been the case. Unless we all descend from a single small group of early humans, there is no reason why our African ancestors shouldn't have lived in many such groups, practising occasional or regular exogamy. Each group may have had a language of its own (many such codes could develop parallelly from pre-linguistic forms of communication), but every child had the evolutionarily acquired advantage of being able to learn ANY such language -- something that we still possess. The instinct was determined genomically, but of course specific languages had lexicons consisting mostly of arbitrary forms and were flexible enough to follow their own, non-biological courses of evolution. It is this very flexibility of communicative codes that has made language such a powerful tool, easily adaptable to different environments and applictions.
> B) Results of interactions of Mesolithic and Neolithic societies.
> If I understand you right, you don't see big difference between Mesolithic and Neolithic tribes from the point of view of their
> development prospects.
> To make our discussion more sharp I'll try to defend the following position (probably too categorical):
> From the anthropological point of view people of hunting-gathering societies are real Homo sapiens sapiens, but from the
> ecological
> point of view they can be considered just as public animals (maybe the only difference - fire). That is why practically in every
> competition peoples of a society, which passed the "Neolithec revolution", must win. However, it does not mean that they are
> better
> or worse, than Mesolithic people. They just use different rules of the game.
I have already cited examples of mesolithic peoples who didn't lose the game but joined the winners. Perhaps you underestimate their resourcefulness or overestimate the fierceness of the competition and the dramatic character of the neolithic revolution. Many societies acquired only selected elements of the neolithic -- e.g. Uralic and Turkic speakers adopted nomadic pastoralism and pottery, but no agriculture. This didn't lead to their extinction. What is advantageous to a given society depends on the ecological conditions. Agricultural technologies developed in the Middle East may not have worked very well in the forests of central Europe. Perhaps people who brought them to Europe were not greatly interested in colonising hostile environments and hung on to what they found familiar. But the local hunter-gatherers who had learnt (e.g. through trade contacts) the use of crops and animal husbandry were ideal colonisers. How do you like this scenario?
> C) Role of farming in spreading linguistic families (super-families).
> As far as I can judge, you belive that agriculture could be just one among many factors, which simulated spreading of languages.
> It seems to me, that farming was the key reason, which led to such a wide spreading only of languages  belonging to 6-8
> super-families. People speaking other languages occupupy only places not suitable for agriculture or herding.
> I'll try to defend the following scheme:
> one center of domestication - one unique set of "initial" crops (first of all cereals) - one super-family.
But the time-depths suggested for the macrofamilies, as well as linguistic palaeontology as practised e.g. by Dolgopolsky, make Nostratic society late paleolithic or early mesolithic -- certainly not neolithic! If so, what material advantage allowed the Proto-Nostratics to expand at the expense of other societies? Fishing nets? Bows and arrows? Fire? Everybody had them at that time! It is easy to imagine how the great FAMILIES like IE or Afroasiatic may have spread thanks to neolithic technologies, but I find it a priori unlikely that a MESOLITHIC population, one among many, could have dominated an area extending at least from western Anatolia to the Amu Darya (I'm relying on Renfrew's maps here). The same holds for other proposed macrofamilies. What seems to be typical of known pre-neolithic areas is great linguistic differentiation and fragmentation, with dozens of small-sized families and isolated languages. Perhaps you have different views on the dating of Proto-Nostratic (not 15000-10000 BC, as held by its chief proponents, but in the early neolithic). But with a few millennia less, wouldn't our chronology become too shallow for the linguistic changes expected of the various daughter families of Nostratic? Those who reconstruct Proto-Afroasiatic speak of hypothetical time-depths of about ten thousand years. Proto-Nostratic was surely considerably older than that.
> D) The principal feature(s) of Proto-Indo-Europeans.
> I have given my position in the SIEM: horse - light wheel -chariot.
> Yours?
I know more about the principal LINGUISTIC features of PIE than about the "PIE culture" (if there was such a thing) as reconstructible from its vocabulary. I'm rather distrustful of linguistic palaeontology in general. The items you mention are culture terms, thus potential loanwords. Gamkrelidze and Ivanov claim Altaic connections for the English word mare. Wild horses were still relatively common throughout Eurasia in PIE times, so a common IE proto-word for the species would not be surprising even if the horse was domesticated after the IE dispersal. After the domestication of the horse, horse-related vocabulary became very important culturally and may have spread from language to language together with wheels and wagons. The fact that many modern IE languages have similar words for "lemon" or "telephone" doesn't imply that PIEs had them as well.
All Slavic languages have a common term for "king": korol', król, kral, kralj or the like. It would seem certain that these words derive from something like PSl *korlj@ if we didn't know that the word is in fact Charlemagne's name. He died in 814, when the geography of the Slavic subbranches already resembled their present distribution. This doesn't mean that the word in question is irrelevant for studying the history of Slavic, but it certainly isn't a Proto-Slavic lexeme despite the appearances.
> E) Correlation of genetic complexes (as a particular case - racial type) with ethnical and linguistical groups.
> I think that any genetic complex can be correlated with endogamic populations, and human racial features isn't an exception. Any
> such a group can be described statistically. "Racial types" themselves are not somethimg special. Theoretically any combination
> of genes could be taken as a type. The question is in degree of variation inside the group. That is why it is convenient to use
> such terms as "Cro-Magnon", "Negroid" or "Uralic" when describing pecularities in appearance of different groups. They can (but
> don't must) correlate with big linguistic branches. From my point of view, typical early Indo-Europeans can be described as
> people of the Europeoid race.
> As far as I know you disagree. Your position?
Agnostic. In historical times IE languages have been spoken by different "racial types", some of them in areas where distinctions are not quite clear between white and non-whites populations. On the other hand, there have always been more or less "Europoid" populations speaking non-IE languages. What about the other Nostratic families, if their speakers derive from the same ethnic stock as we Indo-Europeans? (I have dark hair and brown eyes.)
> F) "Ironcladness" of laws of cultural (ethnic, linguistic) evolution.
> Of course, there are very many factors, influencing any process in the Universe, for instance, the movement of the Earth around
> the Sun. Still, it is possible to indicate a few the most mighty factors (Kepler's laws in this case). So I hope that it is
> possible to find a few main regularities in cultural etc. evolution, which allow describe the general picture. Naturally,
> exceptions exist, and other factors influence, but any such a particular deviation should be considered and explained.
> Your opinion?
Not a good analogy. Kepler's "laws" are not fundamental but derive from more general gravitational equations which MUST be obeyed by all physical bodies, tolerate no exceptions, and take into account all significant external disturbances as well. This is why it is possible to simulate planetary systems numerically and calculate the courses of stars and planets with incredible accuracy.
Such deterministic predictions do not work in the social sciences, as far as I know. An apparently doomed population may overcome its cultural conservatism, adopt the advanced technologies and institutions of its neighbours and manage to survive against heavy odds. What "main regularities" do you allude to? The laws of history are usually deduced from highly idealised models of social and cultural evolution which may look convincing on paper but which rarely seem to work in real life. The history of Marxist historical determinism is a sad but instructive example.
> Well, a nice protocol of divergences, maybe not completed yet...
Of course not.
> I think we have 2 possibilities:
> 1. To say "I'll never understand him, he'll never understand me" and ignore conceptions of each other, discussing only naked
> facts but never interpretations.
> 2. To be "scientific schizophrenics", considering every new fact not only from the own point of view, but also from the point of
> view of the opponent, and trying to find arguments not only "contra", but also "pro" his conception.
> As to me, I'd prefere the second variant. And you?
We can also disagree politely and tolerantly, trying to make the opponent see our point of view and keeping an open mind just in case the other party makes a convincing point.
To tell the truth, I was once quite enthusiastic about Nostratic theory, deep reconstruction and linguistic palaeontology. When I learnt more about the rigour of linguistic research, about the evaluation of etymological equations etc., I realised how many pitfalls await an amateur armed with a battery of etymological dictionaries, and how important it is to stick to the reasonable. Even great scholars sometimes succumb to romantic speculations based on meagre evidence. You can read in popular books and articles about linguistics that, say, Basque or English are distantly related to each other and also to some native American languages (Wow!), or that it has been positively proved that all languages derive from a common source. You have a well-known linguist's word for it; as a layman you cannot assess the evidence on your own. You are told that the same personal pronouns, or words for "milk", "finger" or "mother" are found all over the globe, which shows that ... . A non-specialist will buy such "proofs" without great difficulty, especially if assured that the very same (or less advanced) methods were used to establish respectable language families like IE.
My own expectations are a lot more exacting, not because I am particularly obstinate or have little faith. I'm just exercising caution, leaving the burden of proof to the proponents of ambitious theories. Meanwhile, since in our discipline nobody can claim to have a complete monopoly of the truth, discussing things with people of a different theoretical inclination always broadens one's mind (and I enjoy it very much, too). I find that such exchanges often lead me to improve and clarify my own position.
> The very best regards,
> Alexander
Don't you think we should publish this correspondence on Cybalist, so that other people might join our discussion? I suppose some of the other members would find it interesting.