Talk:Dené–Yeniseian languages

From Wikipedia, the free encyclopedia
Jump to: navigation, search
edit·history·watch·refresh Stock post message.svg To-do list for Dené–Yeniseian languages:

Here are some tasks awaiting attention:
  • Other : Summarize the morphological evidence from Vajda
    Compare this with proposals by Starostin et al.

This article has been flagged as not following Wikipedia guidelines for neutrality[edit]

This grouping "Dené–Yeniseian languages" has not gained a wide enough acceptance among the relevant specialists to merit calling it a language family without any qualification. To my knowledge (which I admit is not exhaustive on this topic) there is only one well-known specialist (George Vajda) who is an active proponent of this family. I have tried to add qualifying words such as "proposed" in front of the name for this family. I have also mentioned that one of the major scholars cited in this article (Lyle Campbell) disputes much of Vajda's evidence, and Campbell states that the other scholars mentioned in this article do not offer their unanimous support, contrary to what is stated here in this Wikipedia article. All of my work mentioned above has been removed with little or no explanation. — Preceding unsigned comment added by S. Neuman (talkcontribs) 20:01, 9 September 2011 (UTC)

Probably we need more data and quotes and less interpretations and less "advices" to the reader. For instance, Campbell does quote that they supported: "several well-known linguists have declared their support, though often with caution, for example in DYC Hamp, Comrie, Fortescue, Kari, and Nichols." The key point is hat I bolded.
As the article is now, it is clear that there is no unanimous support; but to show more linguists against we need concrete references.
On the other side I will trust more in experts on any of the branches and Campbbell has disappointed me in this article by his dubious use of non-linguistic considerations as if they were evidence, in both senses. So, one of his PRO is that Vajda is a "serious linguist" (and what has to do this with the linguistic evidences?); in the AGAINST some era very odd: geographical, genetical, archaeological? Please, apply these to Hindi and Norwegian as arguments against their relation. (And following the example, apply the geographical alleged impossible distance to the close relation between Romani and Hindi. When the alleged serious arguments against are so feeble.....
THat said, his linguistic arguments do are interesting, though I find them sometimes a litlle forced.
But that's MHO, the question is rather of wording than of content (which do is),as the wikipedia is full of hypothetic linguistic families which almost nobody believe (and definitely this one is not the case).
There is another problem with some of the "criticism" editions on this article, some of them have a too editorial and subjective bias, more as discussing than explaining. The wording problem again.
On the other side we may rely on sources, and these sooner or later will be clearer. But the last time I looked at it, mostlinguist were favourable to that "hypothesis". The percentage seem better than in the case of Uralic, for instance.
So in my oppinion the discussion on the "wording" is wellcome, but should be on concrete data.
ALthough, I repeat, probably it is only a waste of time as surely in the next two years there will appear more "quotable" sources.
That said, I have no problem with adding the adjective "proposed" to the definition while we wait for new sources.
Dumu Eduba (talk) 21:18, 9 September 2011 (UTC)


1) Kari, Leer and Krauss are experts in Na-Dene but not in Yeniseian 2) Werner is a Yeniseist but not an expert in Na-Dene 3) The others are simply historical-comparative linguists

Now, the hypothesis still needs a response from the two other important Yeniseists: Georgiy Starostin and Stefan Georg. Both of them seem to be preparing a critique of their own. They are both going to criticise Vajda's interpretation of the Yeniseic data. Starostin will probably defend the Dene-Caucasian hypothesis at the same time, insisting Vajda hasn't brought anything new in fact, while Georg will in all likelyhood consider the Dene-Yeniseic idea untenable. —Preceding unsigned comment added by Petusek (talkcontribs) 09:44, 26 June 2008 (UTC)

Why do these considerations disqualify the backers of Vajda's theory? There has certainly been no comparable outpouring of support for the elder Starostin's theory and methodology. --April Arcus (talk) 01:54, 7 July 2008 (UTC)
So far Americanists are quite happy with the methodology and results of the Dene-Yeniseian group. Starostin's methodology and results, however, are not widely accepted in Americanist circles. It's unlikely that any criticism on his part will have an impact on Americanist thought in the matter. (Taivo (talk) 21:17, 1 January 2009 (UTC))


An actual problem I see is that the section of the article discussing lumping vs. splitting, Dene-Caucasian, etc. is poorly written and surely confusing for someone not already familiar with the topic. There's also some phrases and sentences that make no sense, like "Some of the evidence for this relationship resembles less rigorous proposals school for a Dené-Caucasian language family, which adds to the proposal Burushaski and the Sino-Tibetan and North Caucasian language families" (so grammatically incorrect that I'm not positive that my interpretation of it is even close to accurate).

While a discussion of the Dene-Caucasian proposal; the Dene-Yeniseian family in relation to it; previous attempts to group Na-Dene with Yeniseian; etc. are all important and deserve space in the article, I don't think the discussion as it currently stands is very useful. And my feeling is that a lengthy discussion of the philosophical differences of "lumping" vs. "splitting" is unnecessary. All that a reader really needs to know is that Na-Dene and Yeniseian have been linked in the past, as part of a larger proposed grouping (Dene-Caucasian), but using methods that most linguists consider unreliable and invalid; and that the current proposed link has no direct bearing on whether the other proposed Dene-Caucasian languages (Sino-Tibetan, Burushaski, etc.) are connected as well. --Miskwito (talk) 22:09, 4 January 2009 (UTC)

Old and New Worlds Connexion[edit]

Isn't it contradictory to state that this is "the first accepted linguistic connection between the Old and New Worlds" and to follow immediately with "Yupik languages [exists] on both sides of the Bering Strait" i.e. in both the Old and New Worlds? I do not want to delete one of the two statements myself, first because i do not know the subject, and second because re-wording it is certainly possible without having to delete anything. Emaskask (talk) 13:12, 1 September 2009 (UTC)

The statement "first accepted linguistic connection" does not include Yupik because the Yupik presence in Siberia is a recent migration from Alaska back into Siberia, not a prehistoric migration from the Old World to the New. It is on the same level in linguists' eyes as the occurrence of western Indo-European languages in the Americas--a later migration and not part of the prehistoric peopling of the New World. (Taivo (talk) 14:06, 1 September 2009 (UTC))

Scholarly consensus?[edit]

It seems that there is no scholarly consenus See Ket_language#Classification. Andries (talk) 05:06, 27 May 2010 (UTC)

"Consensus" doesn't mean unanimity. There are still some who reject Sino-Tibetan and Niger-Congo. But when a collection of the foremost historical linguists (Hamp, Leer, Krauss, Nichols, et al.) accept the grouping, then that is a consensus. With that group behind the proposal, none of whom would be classed as a "lumper" in any definition of the term, then most other historical linguists would fall into line if queried. As far as North American linguists are concerned, then the grouping has been proven and a consensus exists. --Taivo (talk) 05:43, 27 May 2010 (UTC)
Thanks for your reply. The two articles now suggest different things about scholarly consensus or lack there of. They should be aligned. Andries (talk) 19:32, 27 May 2010 (UTC)

Cognate list; revised as A Selection of Synonyms[edit]

Just a heads up that I've added a cognate list to the article. Everything can be verified with and any Navajo dictionary. — Stevey7788 (talk) 22:37, 17 September 2010 (UTC)

Please note that the "cognate list" provided here in September 2010 is actually a selection of synonyms. The section has now been retitled in this way. It is misleading to call such a selection of words "cognates." Also when someone consults the link to the Swadesh Lists for DY, he will find a broader and still developing selection of synonyms for Ket, Tlingit, Navajo, Dena'ina, Hupa and other languages that may be added there. Nuqelnegen4 (talk) 19:08, 25 December 2010 (UTC) — Nuqelnegen4 (talk) 17:06, 25 December 2010 (UTC)User:nuqelnegen4

You don't have "synonyms" between languages. You have "similarities" which might be "cognates" if there are regular sound correspondences that stand up to scrutiny. --Taivo (talk) 07:33, 10 September 2011 (UTC)

One significant exception[edit]

"One significant exception"--i.e., Campbell--is just that--one significant exception. Even he does not describe the hypothesis as false, and his splitter tendencies are well know. Changing the tenor of the article on the basis of the opinion of onenon-specialist (of NaDene or Yeniseian) is hardly supportable. μηδείς (talk) 05:15, 14 September 2011 (UTC)

No, Medeis, Campbell is not alone. Read the IJAL articles. Vajda himself, in his response to Campbell, said that he had not fully proven the issue yet and that there are still holes in his arguments. Several of the authors in the collection of papers also expressed doubts, although hoping that the doubts could be resolved. The consensus among Americanists is that the connection has some intriguing possibilities and Vajda is not a fly-by-night linguist, but it has not been conclusively demonstrated yet. There's not an Americanist who will say that it's proven. It is still just a proposal and not a firmly established language family. --Taivo (talk) 05:21, 14 September 2011 (UTC)
I concur completely with Taivo. Linguistic Science (talk) 05:31, 14 September 2011 (UTC)
Me too.·ʍaunus·snunɐw· 10:47, 14 September 2011 (UTC)
Taivo, didn't you say something quite different above, only last year? Have you changed your stance without acknowledging so?
That said, Wikipedia should not fall into the recentism trap. My impression, as well, is that it is a serious, intriguing or even promising, and well-received proposal, but cannot be portrayed as fact yet. Unfortunately, other articles are doing exactly that currently. This was also criticised by a contributor on Talk:Paleosiberian languages. I will change the article Paleosiberian languages accordingly. --Florian Blaschke (talk) 13:33, 23 October 2011 (UTC)
The initial consensus among specialists was, indeed, favorable. But further analysis and publication has revealed some of the flaws and that initial consensus has evaporated. I don't need to review the history of my thinking every time I make a comment where I have modified my view based on current scientific consensus on an issue. --Taivo (talk) 15:40, 23 October 2011 (UTC)
Well, you have to admit the contradiction must appear surprising to people not intimately familiar with the latest news in Dené-Yeniseian studies, such as me. --Florian Blaschke (talk) 12:17, 26 October 2011 (UTC)
The field of Native American linguistics is a relatively small one and, unlike fields with thousands of active practitioners, it can move quickly from one place to another. One of the Holy Grails of the field is any link between the Old World and the New, so a proposal by a well-respected and methodologially-sound linguist will always raise expectations. But we're not above admitting that our initial favorable response was premature. It is a credit to Vajda that he also recognizes that the proof isn't quite there yet. --Taivo (talk) 13:40, 26 October 2011 (UTC)
I agree, but this proposal is still presented as if it were undisputed fact in various other articles (I've just edited Burushaski language to reflect its status) and countless infoboxes. Obviously it was an error to treat the proposal as orthodoxy so early in Wikipedia, as it was to expect that the initial enthusiasm was premature and easily overrated by the Wikipedian linguists. I, for one, was never comfortable with its presentation as "word of god" on Wikipedia, not only because it's a long-range proposal, but because it was simply to recent to assess its endurance. Groupings other than those that are virtually universally accepted and found in the relevant handbooks have no place in infoboxes, or at least not without heavy qualification with question marks and footnotes to underline their status as proposed language families, and that this classification should not be relied upon.
That said, personally, I could do entirely without Altaic (to name a particularly notorious example) in the infoboxes, as well, because the added value for the reader is negligible. It suffices for the reader to know that a language is Turkic, Mongolic or Tungusic, groups which are highly recognisable even for the layman. The number of language families in Asia is not overwhelmingly large, so the lay reader should not be overtaxed, either, by the exclusion of macro-phyletic groupings and long-range comparisons. While we could introduce areal groupings as replacement for macrophyla in infoboxes, they would break the genetic principle and are not truly needed either: Maps, or at least mention of states (or federal states, in the case of North America), allow the reader to place a language geographically if the name of the family (that is, the top-level genetic unit displayed) doesn't enable the reader to do that already. Otherwise, I cannot think of a potential benefit to the reader arising from the information that language A has been tentatively assigned to macro-phyla B, C and D, as it often tells the reader very little about what the language (not only grammar/typology, but also the bulk of the vocabulary) is like. To know that Japanese has been associated with the Altaic grouping by many researchers, or that Navajo is a Dené-Yeniseian language per Vajda's proposal, tells the reader next to nothing about grammar and lexicon of the language (should the reader happen to be familiar with other Altaic or Dené-Yeniseian languages in the first place), and is just trivia for linguists and language geeks. I think that while discussions of macrophyla and long-range comparisons are fine in the relevant articles, they are entirely expendable in articles on individual languages and low-level groups.
Do you think I should bring this subject up in a more general environment (WikiProject, or something)? Is there a realistic chance of changing the policy on infoboxes and ledes in articles about individual languages and low-level groups? --Florian Blaschke (talk) 16:31, 12 November 2011 (UTC)

(outdenting) If the merits of this proposal are to be more widely evaluated by non-specialists, we'll need to say something about the lumper vs. splitter conflict. Unfortunately, this conflict has taken on the role of what in comp. sci. is often called a "religious issue" -- something which cannot be proven, but nonetheless (or perhaps as a result) people feel emotionally committed to their side to such an extent that they can't step back and try to see things neutrally. I actually think that Vaida has tried very hard to be neutral, which is to be commendable.

As a simple example, I don't think it's remotely controversial to say that Lyle Campbell is one of the most extreme "splitter" partisans, who is quite skeptical of any proposed language connection whose evidence isn't gold-plated. This is abundantly obvious both if you read his books on historical linguistics and his review of Vaida's work. Yet there is no mention either on this page or on Campbell's own page of these tendencies. Nor can you find much discussion of the lumper vs. splitter conflict in linguistics anywhere in Wikipedia. Contrarily, Merritt Ruhlen probably represents the most extreme of "lumper" partisans. However, you *can* find this mentioned on his page -- primarily, I think, because current orthodoxy in historical linguistics leans in the splitter direction. (Interestingly, current orthodoxy in anthropology probably leans in the other direction.)

If I were to take a somewhat jaded perspective, I might say that Campbell's review of Vaida mostly consists in throwing dirt at the proposal to see what will stick. At the very least, it doesn't look to me like he makes any attempt to really evaluate the proposal. Rather, he simply goes down his very long list of potential objections to language proposals and tries to eliminate every piece of evidence that meets any of his objections. "This one might be a borrowing -- ok, gotta take it out. This one might (conceivably) be onomatopoeia -- ok, gotta take it out, too. This one might be an imperfect semantic match (he even considers "sun vs. sunlight" to be imperfect) -- yup, that one goes out, too." (Nor does he make any attempt to admit his biases, but simply presents his views as if they represent consensus.)

My perspective in these, and most similar, conflicts is to be agnostic. As a result I think it's important to call out the biases in both sides and let the reader decide. Campbell's bias needs to be mentioned; similarly, the strong splitter bias in the Amerind linguistic community as a whole (at least, to the extent that it exceeds the overall historical linguistics splitter bias, which it certainly does). My reading of Mithun's book indicates that she also is clearly a splitter. I don't know whether Comrie or Nichols have strong lumper/splitter views. Benwing (talk) 23:43, 29 July 2012 (UTC)

The point is that the splitters are always on the safe side with their tiny uncontroversial families, and it is methodically correct to use the absence of a proposed but undemonstrated connection as null hypothesis (just as in subgrouping, the null hypothesis is that a proposed subgroup is absent and the branches in question are separate). Lumping, i. e., premature acceptance of proposed groups, is a much more serious methodological error than splitting, i. e., the sceptical stance. If we had no pre-modern evidence of the IE languages, splitters would only recognise Germanic, Celtic (or even two separate units Goidelic and Brythonic), Romance, etc. as genetic units, effectively treat Greek, Albanian or Armenian as isolates, and attribute all similarities to contact, I suspect, but this factual mistake would be methodologically sound and actually more acceptable than just attributing the similarities to a genetic relationship without doing any proper reconstruction in a serious attempt to demonstrate the connections as far as the evidence allows (which may allow it, if we are fortunate, or not). It was also sound to treat Baltic and Slavic as separate branches of IE in the past even though evidence that it is a valid subgroup is now mounting. Linguists need not excuse for errors of judgment when they apply the null hypothesis to undemonstrated proposals. Only refusing crystal-clear evidence as valid would be an error, and your claim that Campbell commits this particular error is a serious accusation which you have the responsibility to bolster. At least Campbell has the sincerity to admit that he cannot judge all proposals equally well, while (as far as I'm aware) neither of us has studied either family in question. If Vajda's evidence were really rock-solid and obviously valid, I don't think there would be any issue, but there's also the principle expressed in the adage "extraordinary claims require extraordinal evidence" (which I suspect Vajda is just trying to handwave away by his claim that morphology can (!) remain stable for more than ten millennia), and Campbell is therefore right to proceed really strictly and eliminate all but the most convincing evidence. After all, Dené-Yeniseian seems to have a time-depth far beyond any widely recognised family (even the African macrofamilies, which are themselves seriously questionable, or such a proposal as Austronesian-Ongan – to me the best-looking case among the more extreme long-range proposals I have come across so far: I'd never take it seriously if the evidence didn't look promising; moreover, it's utterly unexpected and surprising, in fact so crazy, obscure and bold and counter to all intuition that it almost has to be true, and symptomatically ignored by the long-range comparison fans), up into the last glacial period, and therefore it is unacceptable to be lenient in this case and ignore even seemingly minor semantic mismatches. While "sun" can metonymically stand for "sunlight", it's not a trivial difference. Of course you cannot expect that many meanings remain stable after fifteen millennia, but that is exactly why long-range proposals are so difficult to establish, and rightly so. --Florian Blaschke (talk) 12:02, 1 August 2012 (UTC)
OK, so you are a splitter. I have no problem with that, but you must understand that not everyone agrees with you. Whether lumping is a "much more serious methodological error" than splitting is simply a matter of opinion -- obviously, the lumpers will disagree. The issue here is that, like many splitters, you stereotype the lumpers as blythely grouping random families together on little or no evidence, using unscientific principles. From an unbiased perspective, I would say the reality is rather that lumpers and splitters simply disagree on what constitutes acceptable evidence, on which scientific principles are acceptable as the basis for evidence, and on how much evidence is required. Orthodox splitters like Campbell typically accept only the comparative method as acceptable principles and only lexical evidence as acceptable evidence, and throw out large categories of potential evidence because it might be non-genetic (possible onomotopoeia, possible sound symbolism, possible borrowings, etc.). It is similar to the difference between the legal standards of "proof beyond reasonable doubt" and "preponderance of evidence". Neither standard is objectively "better" than the other.
In this article I think we owe it to the reader to simply present the evidence and the arguments in favor and against in a factual and neutral fashion, which is what I've tried to do. Benwing (talk) 04:42, 15 August 2012 (UTC)
You are wrong in your assertion that neither is better than the other. A conservative approach is simply a requirement for the conclusions having any degree of scientific validity. This is not just about opinion as in a courtroom, but about making claims about the past - claims that can be speculative or well founded. For historical linguistics to maintain claims to status as a science it must stick to making well founded claims, and maintain the rigor to do so. I don't think it is the case that there is a group of "lumpers". As far as I know there are just the Moscow school and MErrit Ruhlen. And yes their claims are utterly pseudoscientific. But in any case yes we should include both views - but in our classification scheme we have to maintain the most conservative stance. The family is not established untill it is fully accepted in the literature.·ʍaunus·snunɐw· 13:19, 15 August 2012 (UTC)

Ket & Navajo word pairs[edit]

I don't think that section is particularly useful. As far as I am aware, Navajo is not even a particularly conservative Athabaskan language (to say nothing about Na-Dené), and Ket is not particularly close to the Yeniseian parent language, either. A comparison of Proto-Yeniseian and Proto-Athabaskan or even Proto-Na-Dené reconstructions – including several competing proposals in addition to Vajda's – would be much more instructive, and help laypeople and linguists alike get an impression of the nature of the evidence employed in Vajda's argument. I seem to recall seeing a table comparing Proto-Yeniseian reconstructions somewhere, but can't remember where that was. --Florian Blaschke (talk) 19:50, 12 November 2011 (UTC)

I agree. It would be a lot more useful to have a list of proposed sound correspondences. And yes also reconstructions instead of word pairs.·ʍaunus·snunɐw· 03:57, 29 March 2012 (UTC)

Update, please?[edit]

Maybe it is time to update this webpage. Here is the URL for an article that provides evidence that this language family is true:

Anonymous96.226.22.43 (talk) 18:30, 14 March 2014 (UTC)

Note that this article is already cited in the lede, but detailed information from the article has not been incorporated in this Wikipedia article. Paulmlieberman (talk) 12:54, 21 March 2014 (UTC)
Also mentioned at [1]. Dougweller (talk) 15:58, 6 April 2014 (UTC)
The Dene-Yeniseian hypothesis is victory of Edward J. Vajda (2008). But, writer Joseph Stromberg not used to name "Vajda"; very interesting! Mark A. Sicoli and Gary Holton's phylogenetic research is exceptional evidence for this hypothesis. --Kmoksy (talk) 16:57, 6 April 2014 (UTC)
As pointed out in the article, the hypothesis precedes Vajda; what he did was finally make a serious attempt to actually demonstrate it. Unfortunately, he has failed to convince specialists completely so far; they remain cautious and so should we. According to this article, it was actually Greenberg himself who endorsed the idea (which appears to go back to Trombetti), although I wonder if Ruhlen followed him or if it was the other way round. Also, from the article, there appears to be no clear genetic evidence pointing into the same direction (a connection between Ket and Athabaskan speakers), contrary to what I recall being claimed (that genetic evidence validates the hypothesis; in fact it cannot do so in principle, it can only provide a suggestive parallel connection).
I am sorry to say that I see little possibility to judge the proposal until new, high-quality reconstructions of Proto-Yeniseian and Proto-Athabaskan(–Eyak) are published, preferrably by sceptics of the proposal. This could easily take decades, although Vajda's work can provide valuable incentive (especially for grants) to properly get started.
All a linguistically trained person who is not a specialist in either family could try is to take the competing reconstruction schemes for each family and compare them with each other as well as with the reconstructions for the other family, but it is hard to judge the accuracy and plausibility of the reconstructions if you're not really familiar with the attested material. That's why there isn't much left than to wait for the specialists to finish their work, unfortunately. (Things would be easier if there was a sceptical scholar who happened to be a specialist on both Yeniseian and Athabaskan, but unfortunately such a person does not seem to exist.)
The back-migration idea sure sounds interesting, although it appears not to agree very neatly with the three-waves model. If there had really been three waves of immigration, the Amerind one, the Na-Dené one and the Eskimo–Aleut one, this implies that Na-Dené is intrusive and Yeniseian autochthonous in Asia, while the back-migration idea seems to suggest the opposite.
That said, if the common region of origin is Beringia, this may solve the apparent contradiction. In fact, the typological deviation of Kamchatkan from Chukotkan as well as reconstructed Proto-Chukchi-Kamchatkan suggests the former existence of a quite different (substratum) language (or family) on Kamchatka, one that had ejectives and tolerated extensive consonant clusters, reminding more of the languages of the Pacific Northwest (or the Caucasus, of course), including Athabaskan languages, than Siberian languages. That's certainly interesting! --Florian Blaschke (talk) 17:48, 16 May 2014 (UTC)

Dené ¿or? Dene[edit]

Why Dené–Yeniseian languages? See Na-Dene languages. Also see The Dene-Yeniseian Connection of Alaska Native Language Center. --Kmoksy (talk) 21:04, 4 April 2014 (UTC)

Don't know if this was my doing, but half the time I move articles around to make sure we have the necessary rd's, and where they end up may be almost random. We can use whichever spelling is more common. — kwami (talk) 20:32, 6 April 2014 (UTC)