Talk:Dené–Caucasian languages

Cleanup on this talk page
I'm deleting or reducing the following:

1. "Restructuring" - a large part of it. Much of it has already happened. 2. "Suggested structure" - redundant. Although the sentence Objections are appreciated very much, too! will always be valid ;-) 3. "Tables of Sound Correspondences" - that discussion isn't productive anymore, is it? ;-) 4. "In preparation" - not useful any more. I used it as a notebook only. 5. "Word-lists" - heavily reduced, except for a few points that still need to be discussed. 6. "Phonology" - not useful anymore... 7. "Salishan languages" - useless... --Pet'usek [petr dot hrubis at gmail dot com] 00:34, 19 April 2007 (UTC)
 * You should rather archive it.·Maunus· · ƛ · 07:11, 19 April 2007 (UTC)
 * Oops...I didn't know it was possible. If it is still possible, can somebody do that, please? --Pet'usek [petr dot hrubis at gmail dot com] 12:16, 19 April 2007 (UTC)

Restructuring

 * Description of the hypothesis - I've got some more information to add.
 * The reconstructed morphology - Some other information has to be added.
 * Special section devoted to the ancient languages (PDC roots)
 * The Dené-Caucasian Sound Laws - A lot of work still ahead...
 * a. A table (=Summary) (incomplete, separate fields for different positions)
 * b. The roots should contain information on the rules involved, the rules should have some lexical comparisons.

PLUS

Anything you might suggest...--Pet'usek [petr dot hrubis at gmail dot com] 00:12, 19 April 2007 (UTC)


 * I think there should be a == Criticism == section. RMW Dixon surely has said some things which could be mentioned there. 80.126.161.176 06:57, 6 September 2007 (UTC)


 * Such as? David Marjanović (talk) 23:15, 16 February 2008 (UTC)

12-word list
Finished transcription into IPA.* Except for the mysteries; for example there's a ŋ with something below it in the Na-Dené column, and I can't figure out what that is.


 * Hmm...have I sent you the varios Na-Dené papers? I'm sure there will be some explanation for that in those PDF's. I'll have a look.

We also need to explain what /H/ and /A/ are, and the dialect abbreviations for Basque and Burushaski. Was I right in interpreting every G as ? What tones do the Yeniseian languages have (superscript numbers are not IPA)?


 * That's right. We have to do it with every table...hm, or perhaps, if the meanings of the more general symbols don't vary too much, we might create a more general explanatory table at the bottom of the article...but we can always do that. That can wait. By the way, we might (sooner or later) make several separate articles (one devoted to PDC phonology, another one to PDC morphology, one more for PDC lexics) and only briefly quote them in the main article. We have plenty of time for that, of course, as the article isn't that long yet.


 * The article is quite long, actually. David Marjanović 11:58, 26 February 2007 (UTC)


 * Alright then. I'll think of it for a while and try to propose their future structures (i.e. I'll look at the other articles, like PIE, for example, and steal the scheme from them :-D) --Pet'usek [petr dot hrubis at gmail dot com] 17:40, 26 February 2007 (UTC)

* Not just because it is Wikipedia standard, but also because it is misleading not to distinguish orthographies from scientific transcriptions! Those brackets and slashes have a purpose. I have left the orthographies in italics (and will add more – most languages have one these days), but not the transcriptions.


 * Exactly. Excellent point. My strategy has been to add the data in a more or less acceptible form first of all, and only then take the time to go through it and make the final arrangements of the type you suggest.

I have deleted all breve diacritics. Starostin put a macron or a breve on every vowel; if he doesn't, it means he wasn't able to reconstruct the length. In IPA, shortness is assumed by default, and the breve is the rarely used "extra-short diacritic"; maybe more importantly, you can only stack so many diacritics on a letter – to distinguish a with breve from ä with breve, I had to open the edit page; otherwise I'd never have guessed that the latter existed. The nasal vowels with high or low tone still hardly come through. David Marjanović 20:44, 25 February 2007 (UTC) (logged in at last! :-) )


 * You're right, of course. I haven't had time to do all that. I suggest that we mark vowels of an uncertain length with the ordinary IPA length symbol, put in brackets, e.g. . What do you think, David? Anyway, thanks for the tremendous help!


 * Good idea with the parentheses, I should have got it myself. Can you please do it, now that I've erased the evidence and you have the originals? :-] David Marjanović 11:58, 26 February 2007 (UTC)


 * I've just come home, so I can look at it ;-) --Pet'usek [petr dot hrubis at gmail dot com] 17:40, 26 February 2007 (UTC)

Sound correspondences
I just started that section, using the plain and ejective non-"laryngeal" plosives. The purpose of not putting everything online at once is to avoid losing everything in the next Internet Explorer crash (strange things happen today). Please check for strangenesses! I have no idea if I have correctly interpreted the Proto-Sino-Tibetan and  as consonant clusters. Are voiced aspirated plosives reconstructed for PST?

Please dig up more Basque correspondences, and the Na-Dené ones! Man, am I forgetful.

Also... we should agree on using "Dene" or "Dené" (the accent being high tone).

David Marjanović | david.marjanovic_at_gmx.at | 21:37 CEST | 2006/9/28


 * "Dené" is more correct than "Dene", hence we should use "Dené"
 * --Pet'usek [petr dot hrubis at gmail dot com] 18:23, 12 October 2006 (UTC)


 * OK. I'll do that in a minute. However, someone else will have to change the name of the article. I can't do that – I still don't have an account.
 * David Marjanović | david.marjanovic_at_gmx.at | 15:00 CEST | 2006/10/13

I added the voiced plosives. The old sound correspondence table makes clear that – like in the Americanist notation – is, not  (which would be a major weirdness in the system), so I changed this in the phonology table.

David Marjanović | david.marjanovic_at_gmx.at | 00:56 CEST | 2006/9/29


 * Well, in the Excel table I sent you, David, I'm putting "=" between the differing notations (one is used by Starostin, the other is used by Bengtson - really confusing) --Pet'usek 03:12, 6 October 2006 (UTC)

Added the remaining plosives and a few affricates. I'll change the PST clusters mentioned above to voiced aspirated plosives (and affricates) at the next opportunity for reasons of symmetry, but I'd still greatly appreciate evidence... Edit: They are reconstructed for Middle Chinese, anyway.

David Marjanović | david.marjanovic_at_gmx.at | 23:45 CEST, 2006/9/29 | edit 18:05 CET, 2006/11/4


 * So would I. I haven't collected enough materials yet. I should be kept in mind and stressed in the article, that the hypothesis is really in its beginning. The necessarily and inevitably repetitive tasks of making etymological dictionaries/databases, revealing new evidence, abandoning wrong views and making them obsolite is all ahead of us...--Pet'usek 03:12, 6 October 2006 (UTC)


 * Here I'm merely talking about Proto-Sino-Tibetan (or probably rather Proto-Tibeto-Burman...). I know nothing about it; I'd like to know if it's supposed to have voiced aspirated plosives and affricates, because Starostin and Bengtson don't distinguish [h] from or.

Done that, added the remaining voiceless affricates and /j/ (because it doesn't require entering any special characters).

David Marjanović | david.marjanovic_at_gmx.at | 23:24 CEST | 2006/10/1

Added the remaining affricates, the fricatives, and the nasals.

David Marjanović | david.marjanovic_at_gmx.at | 00:11 CEST | 2006/10/4

Now only a few consonant clusters are missing anymore. When I'll have finished, I'll delete the old correspondence tables which are still at the bottom of the page. Please add citations!

David Marjanović | david.marjanovic_at_gmx.at | 23:03 CEST | 2006/10/4


 * You mean citations in general or any specific, concrete citations? :-) --Pet'usek 03:12, 6 October 2006 (UTC)


 * See below.


 * xcellent work, David! You're faster than the wind! ;-) By the way, getting deeper and deeper in the hypothesis, I'm reavealing many little flaws of it. But until I have the proper dictionaries and papers (many have not been written by anyone yet), I have little evidence. Actually, the evidence would not contradict the hypothesis, it would only alter the reconstructed phonological system, some of the correspondences and developements.--Pet'usek 03:12, 6 October 2006 (UTC)


 * I'll finish the table today. Wikipedia wants to be an encyclopedia, not primary literature, so we can't include anything that isn't published (...except on the talk page...). This is one of the reasons for why we should make reasonably clear what comes from where.


 * Ok. I'll go through the literature I have. Anyway, most of the text needs to be rewritten. I'm working on it. As for the vocabulary lists, it's an enormously time-consuming matter. I might be a bit repetitive, but I'm still working on them :-))).--Pet'usek 12:43, 9 October 2006 (UTC)


 * Then I'll try to open your .rar files and look for a few neat sample etymologies.


 * David Marjanović | david.marjanovic_at_gmx.at | 20:35 CEST | 2006/10/6

The correspondence table is finished. (Accordingly, I deleted its old equivalents.) Please check it against the sources, for typos as well as for misinterpretations (I have ignored all parentheses, slashes, and tildes that are in the "DC Wikipedia.doc" file you sent me). In the long run, it would also be nice to have footnotes that refer to something other than Basque.

David Marjanović | david.marjanovic_at_gmx.at | 21:26 CEST | 2006/10/6


 * I'm working on that ;-)--Pet'usek 12:43, 9 October 2006 (UTC)

David, I found the following passage at the end of page 28 of the Preface to SC by S.A.Starostin: "PSC *p yields aspirated ph in Bur. and PST, and p in PY and PNC. Note that the PNC *p, just as all voiceless consonants, must have also been aspirated." Maybe we should mention that in the table as well....? --Pet'usek [petr dot hrubis at gmail dot com] 18:23, 12 October 2006 (UTC)


 * I just put it in. Please tell me what you think about it.
 * Well done. Thank you.--Pet'usek [petr dot hrubis at gmail dot com] 00:21, 15 October 2006 (UTC)
 * I have kept forgetting to ask you something important: The old correspondence tables show reflexes of PDC voiced fricatives. I have simply ignored them. What should we do: add those reflexes to the corresponding PDC voiceless fricatives, interpret the situation as a disagreement between Starostin (no phonemically voiced fricatives) and Bengtson (phonemically voiced fricatives),
 * There is no disagreement between Starostin and Bengtson on this. Bengtson states (page 20 of his Materials for a Comparative Grammar...): "...Proto-Dene-Caucasian probably had several fricatives corresponding to most of the positions of the occlusives: *s, *, *, *x, *χ. There are some indications that these fricatives may have had voiced allophones. (Cf. the conditions in Proto-Athabaskan, e.g. Krauss & Leer, 1981. Athabaskan, Eyak and Tlingit Sonotants)..."
 * Since the allophones have distinct reflexes, they should be given the right :-). It's just as with the other phonemes at differing positions.--Pet'usek [petr dot hrubis at gmail dot com] 00:21, 15 October 2006 (UTC)
 * or assume voiced allophones as certain, assume that they have become phonemic in some daughter-languages (check out the reconstruction of the Proto-Athabaskan sound system here on Wikipedia!), and therefore make two lines for the phonemes in question, one for each allophone?
 * Yes, exactly. Since there is probably no minimal pair available among the PDC reconstructions that would prove the opposite, the easiest thing - at least at the current state of knowledge - is to assume that there was no opposition of voice in the fricatives. Anyway, if you look at the reconstructed system, there are no ejective fricatives either. The three-way contrast was a matter of plosives and affricates only. Moreover, I have long suspected that at least some of the abundant affricates might have evolved from biphonematic groups of occlusives and fricatives (Note the weird opposition between the ST-like groups and TS-like affricates - Bengtson himself doubts there was a difference, the contrast is based only on PY in fact).
 * And yes, in some daughter languages, the differing fricatives may have become separate phonemes.--Pet'usek [petr dot hrubis at gmail dot com] 00:42, 15 October 2006 (UTC)
 * Good, will do, sometime. Good point about the affricates...
 * David Marjanović | david.marjanovic_at_gmx.at | 22:50 CEST | 2006/10/15
 * Done. I just wonder what evidence there is for the existence of and  – they don't appear in the correspondence table you sent me. Are they just assumed for reasons of symmetry?
 * David Marjanović | david.marjanovic_at_gmx.at | 22:11 CEST | 2006/10/17
 * I doubt it was just to make the system more symmetrical. I will provide you with some etyma if you give me some time. Anyway, I'm working on a database of semantically organized DC lexemes PLUS a database of etyma to exemplify the listed sound changes. The latter, I'm almost sure, will increase the attractivity of the article :-).--Pet'usek [petr dot hrubis at gmail dot com] 18:17, 19 October 2006 (UTC)
 * Yes yes yes yes !!!! Take all the time you need. :-)
 * David Marjanović | david.marjanovic_at_gmx.at | 00:17 CEST | 2006/10/23
 * That said… don't make a too big list. If you want to exemplify every sound change in the article, the article will become three times as long as it already is!!! Better look for examples that are attested in most or all descendant branches.
 * Yes, of course. I will send you the whole database, though, and we can discuss which etyma to include in the article. Sorry for the continuing delay - I've had some problems with my laptop :-(
 * David Marjanović | david.marjanovic_at_gmx.at | 18:11 CET | 2006/11/4
 * BTW, I recently added the supposed pitch accent from the cited Bengtson paper to the bottom of the correspondence table.
 * David Marjanović | david.marjanovic_at_gmx.at | 14:58 CEST | 2006/10/13

Haplogroups
It seems that Bengston is a genius! His split of Dene-Caucasian matches the haplogroups. His way indicates that R1a was NOT the Indo-european marker but the North-Caucasian and probably the IE's of Anatolia borrowed the word for horse from North-Caucasian not the other way around.

Current view of J. D. Bengtson, as yet not dated by glottochronological analyses:


 * 1. Dené-Caucasian
 * 1.1. The Macro-Caucasian family (R)
 * 1.1.1. Vasco-Caucasian (R1)
 * 1.1.1.1. Basque (R1b)
 * 1.1.1.2. North Caucasian (G)
 * 1.1.2. Burushaski (R2)
 * 1.2. Sino-Tibetan (O)
 * 1.3. Yeniseian (P)?
 * 1.4. Na-Dené (Q)

Your comments. Please.--Kupirijo 07:14, 9 March 2007 (UTC)


 * Sounds very interesting, except that cases where languages and genes don't match up have to be expected, so I bet a closer look will show it's not that neat. David Marjanović 01:51, 10 March 2007 (UTC)
 * To answer myself… I just had a look at the articles, and it's nowhere near that neat. For example, haplogroup Q is found in almost all Native Americans, not (just?) Na-Dené speakers. Most, if not all, of the haplogroups are far too old to correspond to a Dené-Caucasian branch.
 * Let's come back in 5 years and look what will have been found :-)
 * David Marjanović 02:01, 10 March 2007 (UTC)
 * What do the others speak? I know there are also mongoloid haplogroups in America like C. Isn't that true? And what do you mean old? All isolate languages must come from the Palaeolithic stock like the Rs? That is why they evolved further apart. For example Basque from North Caucasian. --Kupirijo 05:49, 10 March 2007 (UTC)
 * I still think that there are correlations between ALL haplogroups and ALL languages. It is just difficult to decipher and deconvolve, because of the mixing of populations. Humans ALWAYS had a language. That is what makes them human. And if we all descend from one Adam (and one Eve) that Adam spoke ONE language. --Kupirijo 05:56, 10 March 2007 (UTC)
 * Also let me remind you the quote from the Na-Dene article: "This proposal (of Dene-Caucasian) also is supported by biological studies (Rubicz 2002) showing greater genetic differences between Na-Dene speakers and other native Americans than between speakers of Na-Dene and peoples in Eurasia, particularly speakers of Yeniseian languages.". --Kupirijo 06:34, 10 March 2007 (UTC)
 * And also just because Q3 is found in other Native Americans does not mean that Na-Dene speakers as new-comers did not adapt to local customs and languages. You can have people that speak a completely different language adapt another language. A shocking example are the R1b that actually represent most of the Western Europeans whose ancestors used to speak Basque and then Indo-european. Something similar happened in Eastern Europe but the linguists are reluctant to accept it for various different reasons. As a biologist I think genetics is going to shine light to many areas of "evolutionary linguistics" that sometimes can give the wrong impression by looking at reconstructions that are only on paper and disagree with reality, but also genetics will enforce "weak" theories for larger scale relationships that linguists are reluctant (or afraid or superstitious) to touch upon or dismiss other "weak" theories that were due to geographical contacts or mixing. I probably insulted all the linguists here but as an arrogant biologist and an amateur linguist I see things differently (more deterministically I would say). I apologize if I offended anybody. :) --Kupirijo 06:34, 10 March 2007 (UTC)
 * Don't worry, you didn't.
 * What you say is interesting, indeed. Nevertheless, I still think we have to consider both, DNA analyses and linguistic analyses, and that the former alone cannot prove the case without the latter. And, of course, archeological findings. Imagine the following scenarios:


 * Invaders X occupy group Y. X adopt the language of Y. (or most of it)
 * Invaders X occupy group Y. Y adopt the language of X. (or most of it)
 * Also, you can multiply these options by the following:


 * X outnumber Y.
 * Y outnumber X.
 * Or:


 * X are technologically more advanced.
 * Y are technologically more advanced.
 * How would you distinguish between all those possibilities if you only depended on biology?
 * --Pet'usek [petr dot hrubis at gmail dot com] 12:28, 11 March 2007 (UTC)
 * Agreed. Are you a linguist by training? --Kupirijo 08:03, 12 March 2007 (UTC)
 * I'm a university student of linguistics and phonetics.--Pet'usek [petr dot hrubis at gmail dot com] 21:41, 27 March 2007 (UTC)


 * The whole Dene-Caucasian hypothesis is very doubtful - if not an utter nonsense. There exists a genetic support for the relationship between Burushaski and Ket language, as well as a relationship of these languages to some languages of American Indians, Eskimos and Nivkhi. However, this relationship goes 40 000 years back! The position of Basques is uncertain, because their dominant female lineages are of different (Gravettian) origin than Aurignacian R1b. But let's say that the bearers of Badegoulian (Pre-Magdalenian, i.e. R1b) imposed their language upon Gravettian Solutreans and Basque is thus distantly related to Burushaski. An affinity of some Caucasian language to Basque is not excluded, however, because Aurignacians (R1b-people) penetrated from Europe to Anatolia and they dwelled in today's Georgia during the Ice Age (together with G2). In any case, the membership of Chinese-Tibetan among these languages lacks any support in genetics. In fact, Chinese-Tibetan very probably belongs to a large south-east Asian language phylum that diverged during the glacial maximum (20 000 BP) into Chinese-Tibetan, Dai, Hmong, Austroasiatic and Austronesian. The branching of south-east Asian haplogroups well correlates with this language branching! These languages may be distantly related to Dravidian, Burushaski, Ket, Basque, Amerindian and very probably to languages of Papuan highlanders. Linguists should at first learn something about genetics and follow the haplogroups - instead of constructing bizarre theories. 82.100.61.114 (talk) 13:31, 29 November 2007 (UTC)
 * No, it's not that old. Among the native americans, the theory uses the 2 last imigration waves in the American continent, which is probably 4000-7000 years old interval. It´s not that old, at worst case, if it is in the older level of the interval, it wouldnt probably also be possible to make correlations with Indo Arian languages with the Uralic family, assuming a similar rate if divergence. Considering the genetics, you have different distributions for different markers. Which one you can choose? Daniel de França 17:36, 30 November 2007 (UTC)

Explaining my Edits
Hi. Sorry for not explaining m edits here, I was cut short of time on that particular day and later forgot.

1. I moved the notes to the bottom of the page because firstly that is the normal way of doing it on wikipedia and because having a long list of notes in mid aticle is extremely disruptive. When I first read the article I thouhgt it consisted only of the lead and notes! Also the notes are not standardized in their format for exaple by using the tags, which is ok, but makes them a lot harder to ead and makes them look even more disruptive to the article flow if appeared in every section.


 * You're right about the excessive length of the notes to the lead. Most or all of them should be turned into normal text – I'll try to make a history section for them.

2.As for the NPOV the neutral stance is mentioning that the hypothesis is not accepted by most historical linguists, because it isn't.


 * Fine.

I doub't you can find one linguist specialising in native american historical linguistics who support it. Lyle Campbell in his 1997 authoritative classification certainly doesn't. Not even Joseph Greenberg did. As is the article presents the hypothesis as if it is almost proven, and has a solid grounding and acceptance in linguistics- that is not the case. If you are only aware of one linguist (Vovin) who is not convinced then you should read up on mainstream historical linguists.


 * Wait, wait. I said I can only think of one who has had a long, hard look at the evidence and rejects DC. (I haven't read either Vovin's paper nor the responses to it, so I don't know how long and hard his look really was, but let's assume it was for the sake of the argument. For that matter, I don't even know what Vovin's current opinion is – his paper is from 2002, when a lot less was published on DC than is now…)
 * BTW, I don't think I should care much about whether Greenberg accepted it. Greenberg only used mass comparison, which is good as a method for generating hypotheses, but not for testing them. The DC proponents use the comparative method.
 * Campbell, on the other hand, is apparently a Wikipedian: User:Miskwito. Feel free to direct him here!
 * Miskwito is an acquaintance of mine and he is certainly not Lyle Campbell.·Maunus· · ƛ · 11:28, 16 April 2007 (UTC)

Merrit Ruhlen who is also quoted is an extremely controversial linguist of the american tradition and generally none of the theories that he accepts as probable are accepted by the general linguistic community.


 * The article doesn't exactly rely on Ruhlen. The only research on DC he has done was an attempt to relate the ejectives of Na-Dené to the glottal stops of Proto-Yeniseian, and we don't even cite that because Starostin has a better explanation for the latter (see bottom of correspondence table).

The list of references IS extremely tendential: it contains 5 names all of whom represent fringe historical linguistics. The article needs references and viewpoints from mainstream linguists.


 * Well, I've been looking for some, but haven't found (m)any...--Pet'usek [petr dot hrubis at gmail dot com] 03:52, 18 April 2007 (UTC)


 * Isn't that a bit circular? As in: only fringe linguists accept DC, therefore all who accept DC are fringe?
 * Does it really make sense to call Starostin or Chirikba "fringe"?
 * Of course, more references are always good. I just added the Trask references John Bengtson suggested in an e-mail. David Marjanović 20:44, 28 April 2007 (UTC)

Try looking up the proposal in a handbook or introduction to historical linguistics preferably of american or european origin (not because russian material isn't good but because the american tradition is more representative of the mainstream).


 * As far as I know, most of those people haven't looked at most of the evidence, thinking it's just yet another early-20th-century attempt to relate Basque to Georgian, and/or being suspicious because so many "isolates" are proposed to belong to DC, and therefore refusing to simply withhold judgment. Besides, I fully agree acceptance of DC isn't mainstream, but isn't your definition of "mainstream" a little circular nevertheless?
 * That doesn't matter it should be evident from the article that only a handful of linguists takes it seriously enough to even bother to comment on it - at also shouln't make that fact look like it is an unfair judgement, only time can tell whether the proposal will gain wide acceptance, and untill it does an encyclopedic article needs to say that it is not generall accepted.
 * only a handful of linguists takes it seriously enough to even bother to comment on it is a totally nonsensical argument (if it is an argument at all). It looks as if all the other people rejected it a priori, and as if the "bothered handful" commented on it without really considering it seriously. :-)))--Pet'usek [petr dot hrubis at gmail dot com] 03:52, 18 April 2007 (UTC)

3. for the citation needed tags I just reccommend making a reference to a reference by name publication year and page number.e.g. (Starostin 1996:45) but a clickable note would also be ok. I would really prefer if you changed the notes system to the format and made all of the notes end notes with a. If you check there, you can find the same work available on official journals, although not the final, wording order, chapter or title names. But generaly updated differently from the final version of the magazine, after publication, to reflect the peer revision.Daniel de França 16:58, 1 November 2007 (UTC)
 * I also don't think we can demand anyone to have knowledge about where preprints might be found on the internet, and while making preprints avaliable is maybe common practice in some academic fields it is not in all. A title and page number of a peer reviewed article is sufficient reference even if that means that we will have to find the articles on our own.·Maunus· · ƛ · 16:13, 1 November 2007 (UTC)
 * He stated that this theory is not even worth examine. So how can this be different, on practice, then bluffing for the sole presure of trolling. I came to know this theory on wikipedia. If this article was deleted, I would never learn of this and neither researched many topics, starting from here. I wouldn´t like that a few people to be oblivious of any knowledge. As for the repository of fringe theories, I expressed my self bad. I inteded to be ironic for the things we don´t accept or dont believe because we don´t like. Forgive me for the blunt words.Daniel de França 17:00, 1 November 2007 (UTC)
 * No offense taken. We are not discussing whether the Dene-Caucasian hypothesis should be on wikipedia - it certainly should have an article, but we are discussing how to present the material in way that doesn't trick a reader into believing that the hypothesis is unproblematic and close to be proven, or that any large amount of linguists believe in it. Because if the reader were to get this image from the article we would have mislead him.·Maunus· · ƛ · 08:58, 2 November 2007 (UTC)

I'm an amateur linguist myself, and my primary field of research are Vasco-Caucasian languages of Western Europe. They include not only Basque, but also the poor-attested Iberian as well as substrate languages whose existence can be inferred from loanwords. Basque itself has borrowings from many languages, including Vasco-Caucasian ones, so Bengtson's "simple" approach is a reductionist one, stating (among other things) that every word in Basque with a Vasco-Caucasian etymology is native (unfortunately, most of his proposed cognates are wrong). My aim will be to provide a more elaborated theory about Basque and its relatives (which I expect to publish in the near future), as well as correct Bengtson's mistakes.

The Occitan version of this article contains a table of the phonological system of PDC as well as another one of correspondences between branches (North-Caucasian, Sino-Tibetan, Yenisseian, Burushaski and Basque). Petusek (Peter Hubris), which a Bengtson's follower, also handed me a copy of his own table of correspondences between PDC and Bengtson's Proto-Basque. These ones could be edited and added to the English version. --Talskubilos (talk) 13:56, 2 April 2009 (UTC)

Some problems of "criticism" (sic) of DC linguistics
Before commenting on some of the issues raised in the discussion above, please let me introduce myself. I am George Starostin, son of the late Sergei Starostin and also a comparative linguist, continuing some of the directions of his work. This does not mean that I have always agreed with every etymology or correspondence he produced, but it does mean that I am probably more qualified on explaining and defending his views (those which I agree with of my own free will) than anyone else. I am not a major expert on Dene-Caucasian, but I have worked on both comparative Yeniseian grammar and Sino-Tibetan (most particularly Old Chinese) issues, as well as translated the lengthy preface to the North Caucasian etymological dictionary (now available online at http://starling.rinet.ru) into English. So much for my credentials.

Now, to the matter. Frankly speaking, some of the near-animal hatred towards Dene-Caucasian linguistics (and macrocomparative linguistics in general) as espoused in select comments above and in a lot of other places on the Web is a source of constant astonishment for me, especially considering that this starkly negative attitude usually - not always, but very frequently - comes hand-in-hand with a nonchalant ignorance of the real facts. I will try to prove this by analyzing the three paragraphs written by user KeliLank under the previous heading - each of which is rife with errors and misunderstandings.

''Do you realise the extreme complexity of PDC's phonological inventory when compared to the daughter languages? There are so many unusual affricates and ejectives that just about any sound change will fit in.'' Starostin's latest version of the DC (SC) reconstruction involves 50 consonants. That's almost twice less than in Ubykh, a real attested language, and about ten or twelve phonemes less than Abkhaz and other West Caucasian languages. The number per se is large, but not unusual. In particular, I have not heard any specialists in Caucasian languages raise THAT as a prominent issue - because they, of course, are quite familiar with the extreme complexity of some of the modern languages in that area. I would suggest a crash course in Archi, one of the most complex - and, as even a brief perusal of the North Caucasian comparative dictionary will demonstrate to anyone, the most phonetically archaic - Daghestanian languages, as the perfect cure in this particular case.

Starostin (Bengston as well) suspiciously includes typologically unusual sounds like affricates d͡ɮ and t͡ɬ to virtually all of his reconstructed proto-languages, thereby deriving everything from s to k to l to t to ʃ and so forth, through extremely restricted (and often ad-hoc) phonological rules. This is one of those unfounded blanket statements that I would deem completely unacceptable even in a routine Internet exchange, let alone a real scientific discussion. Starostin has worked on the reconstruction of quite a few proto-languages, and if my memory serves me well, he never reconstructed anything like that either for Proto-Yeniseian, or for Proto-Kiranti, or for Proto-Japanese, or for Proto-Altaic, or, in fact, for any language where such a reconstruction was not demanded by hard evidence. The only family where he did reconstruct lateral affricates although no attested language had preserved them, was Sino-Tibetan, where this reconstruction was based on the correspondence "a specific type of Old Chinese initials that belong to the lateral series, yet structurally behave like stops : Tibetan initial clusters like ld-, lt-, lc-", which, in my opinion, does beg for an original lateral affricate situation. As for "deriving everything from s to k to l to t to ʃ and so forth", this is again a blanket statement that foregoes careful evaluation of particular cases in favor of a "oh, this is so complex it can't be true" exclamation. Granted, it is certainly easier to produce such an exclamation than, for instance, to read the detailed explanation of the Abkhaz-Adyghe reconstruction of the system of lateral affricates in S. A. Starostin's lengthy review of V. Chirikba's book on West Caucasian (oh, but it's been published in "Mother Tongue", so I guess it isn't worth reading. Well, a shorter version is still available in the introduction to the North Caucasian dictionary).

''I mean, what can you make of roots like *bHǝrxkV, *t͡ɕʼaːd͡ɮwV or *bʕáːɫho? What is the probability of any human language having so many highly complex roots in its basic vocabulary? Such complexity is obviously due to an artifact from their attemps to make unrealistic phonological changes fit. And even so they remain problematic.'' Whatever this is, this is not a scientific argument. "This cannot be possible because it is impossible". This has nothing to do with probability, because proto-languages are not reconstructed according to the overall probability of phoneme so-and-so occurring in that language: they are reconstructed based on regular phonetic correspondences between their actual inventories. If language A has 50 phonemes, shows regular correspondences with languages B and C that have 25 phonemes, and shows no sign of historical complementary distribution for any of its 50 phonemes, then the proto-language for A, B, and C will be reconstructed with 50 phonemes, and language A will be deemed the most archaic of the three. This is the ABC of comparative linguistics. North Caucasian phonetics is indeed very complex, and no one has ever been able to demonstrate that this complexity is the result of recent innovation, thus, for the moment, it has to be projected over the entire DC system - and this is further confirmed by the fact that different parts of this complexity find confirmation in various other members of the DC family (e. g., Na-Dene languages do preserve the "unusual" lateral consonants that user KeliLanK is so suspicious of). As for specific examples above, I fail to see what it is that exactly makes a root like *bHǝrxkV that much less realistic than the firmly attested Khosrekh Lak daughter form baIrh, or a root like *bʕáːɫho that much less realistic than the regular Ingush reflexation bʕal. Let us not judge languages from the perspective of English or Hawaiian.

''It's actually very easy to come up with "regular" sound changes this way. And even more so when you: a) are willing to relate words with glosses such as 'blood' and 'stone of fruit' (loose semantics is another issue)''

Another blanket statement. Loose semantics is occasionally to be found in any etymological dictionary of any family, and comparative Sino-Caucasian studies are no better and no worse in that respect than any other comparative studies. I don't see any particular problem with the change 'blood' > 'marrow' > 'pith, stone of fruit', but pecking on occasional semantic latitude while ignoring all the good semantic matches (of which there are plenty, as anyone perusing the Sino-Caucasian database will be able to see) is hardly productive. I would also like to remind that Sino-Caucasian, unlike many other macro-comparative theories, is propped by lexicostatistical confirmation, which means there is a significant number of exact semantic matches between the compared languages on the list of the most basic lexical items.

''and b) is “reinterpreting” by yourself the Proto-languages under comparison all the time (“Vasconcic”, “Proto-Caucasian”, “Proto-Na-Dene”, “Proto-Burushaski”… come on, where else can you find the sources for these ‘Proto-languages’? Even the classification of Macro-Caucasian and Athabaskan-Eyak-Tlingit/Haida is disputed!)''

This is not only a blanket statement, but also one that does not make much sense; in fact, it reeks of ignorance. Sources for Proto-Vasconic (not "Vasconcic") can be found in the works of L. Michelena, for Proto-North Caucasian (not "Proto-Caucasian") in the works of N. Trubetzkoy, for Proto-Na-Dene in the works of Krauss, Pinnow et al., for Proto-Sino-Tibetan in the works of Benedict and Shafer. Starostin, Bengtson, Peiros and others have based their work upon previous tradition, with all due respect to their predecessors but feeling free to improve on their work where better material has become available or where not all the correspondences had been properly taken into account.

I simply fail to understand how any serious students of linguistics would take this sort of 'comparison' seriously (ah, it turns out that they don't.) Apparently user KeliLank feels himself free to represent the opinion of all serious students of linguistics in the whole wide world. I certainly do not, but I AM familiar with serious linguistic research in the former USSR, and I can safely vouch that there are numerous "serious students of linguistics" in this part of the world who either agree with the basic premises of Dene-Caucasian (having, of course, consulted the evidence) or are willing to consider it as a valid, serious linguistic hypothesis. This is obviously due to the fact that Starostin, Nikolayev et al. have had much more opportunity to expound and present their views in this part of the world rather than in the States, and that a large part of the literature on the subject is only written in Russian. However, briefly discarding all of that in one sentence is a sign of close-mindedness. Another issue, already raised in the previous posts, is that it is not clear how anything can be taken or not taken seriously if it is simply ignored. There has not been ONE detailed critical discussion of Starostin and Nikolayev's "North Caucasian Etymological Dictionary" - the cornerstone of the SC/DC theory as well - in the whole Western world. Instead, there is either ignorance of its existence or a few blanket condemnations expressed in a couple sentences (e. g., Johanna Nichols' "thumbs down" that has not been accompanied by any particular arguments). Apparently, the general position is as follows: "Ignore it, and it will go away". Well, it won't, not until someone at least takes the time to explain why exactly it is to be ignored.

Unfortunately, Wikipedia seems to have been hijacked by a couple of 'Mother Tongue' fans (yes, that odd little fringe journal that linguists are otherwise happy to ignore).

I am a linguist, and I have published twice in Mother Tongue, so I am not happy to ignore it. Neither are or were such respected specialists as Michael Witzel, Harold Fleming, Paul Benedict, Larry Trask, Edward Vajda, Heinrich Werner, Winfred Lehmann, Igor Diakonoff, George van Driem, Robert Blust, to name but some who have not deemed it below their dignity to publish articles, reviews, and participate in discussions in that "fringe journal" (I am intentionally not mentioning Greenberg, Ruhlen, Bengtson and other "long-rangers par excellence" here, to avoid potential ensuing generalizations like 'oh, all of them are crackpots'). Of course, if the current state of American linguistics is such that research on macro-comparative issues is explicitly banned from "mainstream" journals, regardless of who is publishing what on whatever issue, I can only express my sorrow at this state of affairs.

Now, to the main issue: considering my views on how "Dene-Caucasian" should be treated in Wikipedia.

First, a lot of references have been made to "fringe" and "mainstream" and how anything that is not "mainstream" should be either ignored or at least, well, relegated to the "fringe area" of Wikipedia. I believe that, before doing that, one has to carefully consider the meaning of the word "mainstream". If it is to be understood as "the general attitude on various areas of knowledge as currently enshrined in major US universities", fine, delete the article for all I care. If, however, Wikipedia has even the slightest intention of broadening its scope and not filtering all the current linguistic issues through the one and only MIT paradigm, the article should remain, because in those parts of the world where DC linguists have been simply given more opportunity to present their views, the general attitude ranges from "cautious acceptance" to "respectful skepticism" rather than "rejection of an obviously pseudoscientific conception".

From a certain point of view, DC linguistics is jammed into a stalemate situation. "It's a lousy theory because mainstream specialists reject it." "And why do mainstream specialists reject it without even discussing it?" "Why, because it's such a lousy theory it doesn't even deserve discussion!" "And how do you know it's such a lousy theory it doesn't even deserve discussion?" "Why, because that's what mainstream specialists tell me!" And the wheel rolls on.

To conclude this lengthy passage, my current suggestions are: a) Merge the "Proto-Dene-Caucasian language" and the "Proto-Dene-Caucasian roots" articles into one. The second one does look a little like an attempt to spread the knowledge and convince the unconvinced, which is not really what Wikipedia is all about. If a few roots are given as examples in the "PDC language" article, I would think it sufficient.

b) On the other hand, it might be reasonable to keep the "Dene-Caucasian languages" and "Proto-Dene-Caucasian language" articles separate, as this would closely mirror the situation with other comparative theories, such as "Proto-Indo-European language" and "Proto-Uralic language" vs. "Indo-European languages" and "Uralic languages". The former should, as it now does, relate the history of the hypothesis, the current views and classifications, while the latter may incorporate tables of correspondences etc.

c) Whatever critical publications on DC linguistics there are, one should feel free to incorporate them into the references and bibliography, making sure, however, that these are PRIMARY sources and not quotations of the "everyone knows that DC does not exist, see works by so-and-so" type.

d) Regarding the concurrent theories - by all means, an article on Sagart's Sino-Austronesian is welcome (although how exactly it is "more accepted" by the mainstream than DC evades me; the only reason it may have received more "scholarly attention" than DC is due to the fact that Laurent Sagart has frequented more international conferences than Starostin). Whoever will be doing it should, however, pay particular attention NOT to base it on the link provided by user KeliLanK; it refers to a very old and obsolete version of Sagart's theory, back when he was insisting that Chinese language is most closely related to Austronesian but not to Tibeto-Burman (thus his main argument was with Sino-Tibetan rather than Sino-Caucasian linguistics). He has since then changed his mind on the issue, now regarding Chinese as Sino-Tibetan first and Sino-Austronesian only second, and it is his newer publications on this theory that should form the basis of the potential article.

I would wait with "Proto-Asian", though, since the theory currently exists in but one or two articles and is certainly much less elaborated than either DC or Sino-Austronesian. In particular, Larish offers no phonetic correspondences for his comparanda, meaning that it is not even structured like a true comparative-historical theory.

I will be happy to participate in further discussion, provided we do not run around in circles.Gstarst 13:09, 16 November 2007 (UTC)


 * In my view the best argument one can make against the DC-hypothesis can be made without thoroughly assessing its data: namely that several of the subgroups which it intends to unite are still shrouded in some measure of doubt as to their validity. If a linguist could first build on commonly accepted language groupings and their reconstructed proto-languages that would be the first step towards making a long range hypothesis credible. Na-dene is a contested grouping which doesn't currently have a sufficiently reconstructed protolanguage. So is Sino-Tibetan and Caucasian. It is necessarily the case that a macrofamilies can only be as solid as the family grouping they are resting on. This alone would make Dene-Caucasian one of the least solid macrogroupings. In my eyes nothing of what has been presented in these pages give me the feeling that the DC-hypothesis is not simply another attempt to lump as many unclassified languages into a group as possible - while treating actual data and linguistic methodology with utter disrespect. Sound linguistic methodology dictates that one works from the bottom up starting with the data connecting it bit by bit and then forming the relations and then the hypothesis - it is quite obvious that DC has been formed the other way round. I am aware that this cannot go in the article unless it is referenced to a publication but it is probably the main reason why "mainstream linguistics " or "American university linguists", or "MIT-school linguists" or whatever grouping you want to believe runs the agenda of the main journals of linguistics do not lend the DC-hypothesis its attention. My view about how the article should be shaped is this: it should document the existence of the hypothesis and the main stages of its development. I think that probably it would be justifiable to mention that the linguistic methods that have been used to investigate the hypothesis are closer to the methods of Mass lexical comparison than to a strict application of the comparative method. And it should be clearly stated that the hypothesis is not considered to be proven or have been made probable. I don't think the article should contain data outside of a few examples serving only the purpose of illustrating the scope of the hypothesis, but not to make it seem as if the data is particularly convincing or unconvincing - or as if it has factual value. I think the most important part of the article is the bibliography which allows the interested reader to assess the hypothesis for himself.·Maunus· · ƛ · 21:58, 16 November 2007 (UTC)


 * With all due respect, I must again disagree with the statement that "it is quite obvious that DC has been formed the other way round". It is true that Starostin's first (short) publication on SC was published before the comparative dictionaries of North Caucasian, Sino-Tibetan, and Yeniseian, but this was due to technical factors (not being allowed to publish extensively before the collapse of the USSR). In reality, Sino-Caucasian was reconstructed exactly the way you describe it: first North Caucasian (preceded by all the intermediate subgroups), then Yeniseian (a detailed article on Yeniseian reconstruction is available with all the correspondences), then Sino-Tibetan, and only then was the SC comparison effectuated. For the record, Starostin never brought Na-Dene into serious comparison EXACTLY for the reason that there wasn't a good Na-Dene reconstruction available, although he thought the connection probable. The sound linguistic methodology that you are speaking of is exactly the one that has always been advocated by the Moscow school in all of its publications. For proof, one needs but actually read some of the publications (preface to the North Caucasian dictionary, article on Yeniseian reconstruction, the monograph on Old Chinese reconstruction etc. - how many of the anti-Sino-Caucasianists have done that?).


 * The statement "the linguistic methods that have been used to investigate the hypothesis are closer to the methods of Mass lexical comparison than to a strict application of the comparative method" is objectively untrue. Mass lexical comparison eschews any kind of reconstructions, ignores regularity of phonological correspondences and has no formal methodology for testing its hypotheses. Sino-Caucasian is based on a complex system of intermediate reconstructions, for each of which a system of correspondences is available in published form, and has been formally tested through lexicostatistics. One may take issue with particular correspondences and etymologies, but certainly not with the methodology involved, which does not differ one iota from the classic comparative method. Saying that Sino-Caucasian is based on mass comparison is thus akin to saying that Indo-European is based on the works of Marcus van Boxhorn rather than Bopp, Grimm, and Brugmann.


 * What may be misleading the American public is that the theory is more frequently associated with early publications by Merritt Ruhlen and John Bengtson on the subject - publications that are, indeed, characteristic of the mass comparison method. Since they were all written in English and easily available in the States, the prevailing opinion is that the Sino-Caucasian/Dene-Caucasian theory is based primarily on these works. But Ruhlen has never really elaborated the details of Dene-Caucasian, publications like these being essentially a 'side interest'; and as for Bengtson, he has long since moved much closer to adopting the comparative method (his later works are far more rigorous in content).


 * A few sidenotes: 1) Like I said, Russian comparative linguistics is far more benevolent towards Sino-Caucasian, which would certainly not be the case if it were but one particular application of "mass comparison"; after all, the Russian comparative school is older than the American one and much closer in spirit to the classic German version of historical linguistics.


 * 2) Na-Dene and Sino-Tibetan are not contested groupings. In the case of Na-Dene, the only problem is with Tlingit, which is just one language; you can just as easily deal simply with Eyak-Athapaskan (no one denies that kinship). I know of no one who denies Sino-Tibetan (Sagart used to, but he no longer does). In addition, even if North Caucasian is debated, the situation there is such that the Abkhaz-Adyghe protolanguage is almost fully derivable (in terms of correspondences) from Nakh-Daghestanian (an uncontroversial grouping), meaning that within Sino-Caucasian linguistics, it is permissible to directly compare Proto-Nakh-Daghestanian to Proto-Sino-Tibetan. Meaning that some of the controversies represent "side" problems and do not necessarily have to be resolved before we allow ourselves to move to a higher level. And, just for the record, Sino-Caucasian is objectively far more solid than some of the macrofamilies tacitly recognized or, at least, never explicitly rejected by the "mainstream" - such as, e. g., Nilo-Saharan, where the situation with data handling and intermediate reconstructions is far worse.


 * Besides, fine - if one does not want to assess the merits of Sino(Dene-)Caucasian before assessing the validity of the individual subgroups, why not start with assessing the validity of the individual subgroups? Where IS a detailed assessment of the North Caucasian reconstruction? Why does Johanna Nichols think it permissible to write a paper called "Consonantal correspondences in Nakh-Daghestanian languages" in which she negligibly refers to Nikolayev-Starostin 1994 with ONE sentence and then proceeds to act as if that dictionary never existed? (By the way, if you want a great example of true "mass lexical comparison", be sure to check out that piece of work - because that's what it is exactly). All the data are available, the etymologies and correspondences laid out and explained so that one does not even need to be a trained Caucasologist to check upon them. (In Russia, mind you, the dictionary has been given high praise by some of its most well-known Caucasologists, such as M. Alekseyev and Y. Testelets; the only explicit "thumbs down" - but, again, without a detailed explanation - was from the late G. Klimov).


 * 3) I think it IS explicitly stated in the article as it now stands that the hypothesis is controversial and not generally considered to be proven, and neither I nor any Sino-Caucasologist have any problems with that. My point is that the article should stand, should give a detailed overview of the current state of the theory, and should not incorporate false accusations.Gstarst (talk) 09:20, 17 November 2007 (UTC)


 * In large part, I must agree with George. I could add a few points concerning the alleged complexity of the PDC phoneme inventory from the typological perspective, but any serious student of linguistics (which I was said not to be) should be aware of some well-known facts (like, for example, that elsewhere, we have even more complex systems, such as those of the Khoisan languages in Africa - !Xũ has 47 non-click and 48 click consonants = 95 consonants, 24 monophthongs and 22 diphthongs, i.e. 141 phonemes & 24 distinctive features; that many Athabaskan languages, for instance, have series of coronal affricates which may be unaspirated, aspirated, or ejective + interdental/dental, alveolar, postalveolar, or lateral, i.e. [t̪θ], [t̪θʰ], [t̪θ’], [ts], [tsʰ], [ts’], [tʃ], [tʃʰ], [tʃ’], [tɬ], [tɬʰ], and [tɬ’]; the correspondences between velars and lateral affricates are well-known among Caucasologists - we can imagine a series of changes like the following: [tɬ] > [tx] > [kx], or [tɬ] > [kɬ] > [kx] etc. - and Johanna Nichols, a "mainstream" expert in Nakh-Daghestanian linguistics, reconstructs lateral africates for Proto-Nakh-Daghestanian, too; the other changes (to [t], [l], [ʃ] etc.) are a matter of usual processes like lenition, fortition, etc. and are paralleled by similar shifts within Caucasian, as well as other non-DC language families, including Afrasian, for example, and so on, and so forth).
 * As for the article on DC roots, I began to write it when I was rather new to Wikipedia and didn't know much about the standards of its editing yet. However, there WERE reasons for its creation: 1) the series of articles on (Proto-)Indo-European as (sort of) a model (notice there are lots of sub-articles on Indo-European, such as PIE noun, PIE verb, PIE pronouns, etc.), and 2) the existence of Swadesh (and other types of) word-lists elsewhere on Wikipedia. If, however, it is so problematic to present the most stable parts of DC lexicon here, then I don't really mind its deletion. I don't need that stuff - I just wanted to inform others, give them some basic data to test, etc. I might have a suggestion for a compromise - the Proto-Indo-European root-list has been moved to the sister project of Wikipedia, called Wictionary. Could we just transwiki the PDC roots article rather than delete it (and totally ignore the time spent on it :-))? Thanks for your opinions!
 * As I mentioned in my reply (deleted by KeilanK for formal reasons), I agree there should be a larger critical section. A question arises here what its format should be, which papers to cite, what arguments to include (some should be transferred to the articles on historical linguistics, on multilateral comparison, on long-range linguistics etc.). Any suggestions are welcome!
 * --Pet'usek [ petr dot hrubis at gmail dot com ] 18:43, 17 November 2007 (UTC)


 * Добро пожаловать, Георгий Сергеевич! :-)


 * I am not a student of linguistics, but I am serious, so I'd like to add my 0.02 €. Please excuse the disorganized presentation.


 * First of all, Miskwito (way above) implied that "pseudoscience" is an insult and subjective. The latter, at least, is not the case. Science has a strict definition; in short, as long as you can answer the question "if I were wrong, how would I know?", you're doing science, if you can't, you're not. Ideas that are dressed up to look sciency (e. g. with jargon) but fail the principle of falsifiability are pseudoscience. The DC hypothesis may be completely, utterly, resoundingly wrong, but if it is wrong, we can find out that it is wrong; thus, it is science, not pseudoscience.


 * On the other hand, lamenting the fact that this hypothesis is not proven is misleading. Science cannot prove, only disprove. The IE hypothesis is by far the most parsimonious explanation for a vast range of data, and it is proven beyond reasonable doubt, but it is impossible to define "reasonable" – after all, it's not even possible to disprove solipsism. Proof is for mathematicians, truth is for philosophers.


 * I agree that the latest version of Sagart's Sino-Austronesian hypothesis (and probably even "East Asian") deserves more mention, in the DC article, in the Sino-Tibetan article, and in an article of its own. Obviously, Sino-Austronesian and Dené-Caucasian cannot both be real (in anything resembling their present incarnations at least), so they are serious competition to each other. I will try to start fixing this over the next weeks.


 * Due to the fact that all reconstructed Proto-DC roots are available not just in primary literature but also at starling.ru (which is currently down), I agree we don't need them in Wikipedia. Moving them to Wiktionary is probably a good idea, though.


 * On the other hand, the separate article on Proto-DC should not be deleted. To the contrary: IMHO, all proto-languages for which reconstructions exist should have a Wikipedia article.


 * What is unusually complex about roots that contain clusters of two or three consonants? We clearly shouldn't compare reconstructions of proto-languages only to English or Hawaiian, but consider English words like strange… a three-consonant cluster in front, a diphthong in the middle, and a cluster consisting of a simple consonant and an affricate at the end, all in one syllable. A "highly complex root" is something different.


 * The reconstructed phoneme inventory is IMHO surprisingly regular. It's clearly at the other end of the scale from PIE, which famously had
 * voiceless, voiced, and voiced aspirated plosives, but no voiceless aspirated ones – a combination known today from one single language spoken somewhere in Indonesia;
 * common /p/ and, while /b/ was very rare or maybe absent, which is what would be expected (if at all) of /p/ but never of /b/;
 * no or almost no trace of /a/ or for that matter any other open or near-open vowel.
 * All three of these features, which have not survived in any attested descendants, have been contested, sometimes for decades, and yet all three are nowadays more or less universally accepted (a few glottalists notwithstanding) because they are simply the most parsimonious explanations for the widest range of data and because there are no reasons to think they are not just unusual and unstable but really impossible. (And incidentally, at least two of them, maybe all three, can be explained if we start from AFAIK any non-Bomhard reconstruction of Proto-Nostratic, but I digress.)


 * Now, epiglottal consonants are unusual, but lateral affricates are much more common worldwide than, say, the of my native southern German. Coming to think of it, even epiglottals are AFAIK present in a greater number of languages than !


 * While "just about any sound change" might fit into the Proto-DC inventory, not every one does. Check out what is reconstructed to have become of the lateral affricates in Basque and Burushaski.


 * It's not true that language phylogeny must proceed strictly bottom-up. Again, take IE as an example: What is the closest known relative of the Celtic languages? Between the obvious families (Germanic, Romance/Italic, Celtic, Slavic etc. etc.) and IE itself, very few groupings are more or less universally accepted (Indo-Iranian, Balto-Slavic; Kentum, Satəm; everything except Anatolian; maybe everything except Tocharian), yet nobody doubts the reality of IE on these grounds. To the contrary: serious work on IE started before even Celtic, let alone Tocharian or Anatolian, was recognized as IE! Analogously, it would be nice to have an unassailable and universally accepted reconstruction of Proto-(North-)Caucasian or Proto-Na-Dené before starting to work on Proto-DC, but it's not strictly necessary. Of course it would help; of course testing of the PNC reconstruction should continue, and of course testing and elaboration of Pinnow's work on Na-Dené including Haida should continue; but it's not necessary, any more than a Proto-Italo-Celtic (or whatever) reconstruction is necessary for reconstructing Proto-Kentum.


 * (Incidentally, it has been suggested that the phoneme inventories of languages such as !Xũ are actually much smaller, with many of the "phonemes" actually being consonant clusters. But, firstly, I can't judge that, and secondly, the phoneme inventories of Caucasian languages clearly cannot be analyzed away this way.)
 * David Marjanović (talk) 15:59, 12 January 2008 (UTC)


 * Corrections: it's starling.rinet.ru and works just fine; and the language in "Indonesia" with the PIE-like plosive system is in Malaysia, on Borneo, and called Kelabit. Also, when I wrote "Satəm", I meant just Indo-Iranian + Balto-Slavic (see Rexová et al. 2003). David Marjanović (talk) 14:51, 3 February 2008 (UTC)

accepted relation
Okay, a lot of the work on DC has been sloppy, but it looks like we finally have some evidence good enough to convince linguists working on these languages: I added a quick blurb to the article, but one of you who's been organizing this might want to work it in better.

Of course, this is a binary comparison (proto-Yeniseian + proto-Na-Dene), so it doesn't address branching depths, or whether the rest of the hypothesis has any validity, but it should be able to act as a core against which other approaches like Karasuk can be evaluated. kwami (talk) 21:36, 4 March 2008 (UTC)


 * P.S. This should be added to the articles on the two families as well. Doubt I'll get to that. kwami (talk) 21:37, 4 March 2008 (UTC)


 * Good. I was going to incorporate this information when I saw that you'd done so already. :-) Fine. I'll have a look at that, of course, but I'm still reading the materials presented at the Dene-Yeniseic Symposium. Anyway, thanks! --Pe t 'usek [ petr dot hrubis at gmail dot com ] 12:31, 5 March 2008 (UTC)


 * While the Dene-Yeniseic connection is certainly claimed as a part of the DC proposal, I feel like it's really inappropriate to claim that Dene-Yeniseic is "the first element of this hypothesis to be well received" - DY has been advanced by entirely different authors, and through entirely different (and superior) methods of analysis. This sort of statement is equivalent to claiming, for example, new evidence for the unity of Algic as evidence for a larger-level family, like Almosan. The unity of Dene-Yeniseic has no consequences for any of the other languages involved in DC. --April Arcus (talk) 17:17, 7 March 2008 (UTC)
 * I agree with April Arcus. Just because a blind chicken also finds a grain now and then that doesn't make it any less blind.·Maunus· · ƛ · 17:24, 7 March 2008 (UTC)


 * But it is factually correct. DC is a proposal. It hypothesizes that several established language families are related. Two of those families are Yeniseian and Na-Dene. A probable connection between Yeniseian and Na-Dene has now been established to the satisfaction of specialists in those families. I don't know why you're reading more into it than that. Nowhere does it state or even imply that a connection between Yeniseian and Na-Dene is evidence that Basque is included as well.


 * I agree that it would be improper to argue for Dene-Caucasian as a consequence of this finding, in the Na-Dene or Yeniseian articles. However, this is an article on DC, and Dene-Yeniseian is evidence for one facet of DC, just as van Driem's Karasuk is another (although given that the methodology and reliability differs). Maybe the wording could be changed, but very often this is how things happen. The story of plate tectonics starts with Wegener, even though people thought he was a crackpot, and the evidence that ended up convincing everyone came from different people, using different methods, and the results differed drastically from his model. Nevertheless, they are considered to have vindicated his vision. Similarly, it may be that none of the rest of DC ever will demonstrated, or that if other families are ever included, they will be Iroquois or Ainu or something else never anticipated by the DC crowd. That doesn't change the fact that one aspect of DC has now been substantiated. — kwami (talk) 22:12, 7 March 2008 (UTC)

Relevance of Dene-Yeniseic hypothesis
While no doubt Dene-Caucasian has motivated Na-Dene scholars to explore the Yeniseic languages more closely (and vice-versa), it is not clear that the Dene-Yeniseic hypothesis provides support for D-C. If anything, should D-Y hold up to scrutiny then it would essentially necessitate removing Yeniseic from the D-C proposal. That is not to say that a revised D-C minus Yeniseic might still be tenable, but just that D-Y cannot be considered one part of D-C. I suggest removing subsection on D-Y and instead referring to D-Y as a distinct hypothesis. Gholton (talk) 16:26, 17 November 2008 (UTC)

I agree. Currently, it's as if confirmation of D-Y is being used by Wikipedians supporting D-C to try to justify their unfounded belief on this Starostinian fabrication. D-Y had already been proposed independently by many other researchers before D-C appeared, and its acceptance by mainstream linguists is of no consequence to the credibility of the D-C "hypothesis".KelilanK (talk) 22:31, 17 November 2008 (UTC)


 * It is not confirmation, and it is a distinct hypothesis, but it is also relevant. And no, the validity of DY would not remove Y from the DC proposal; it would only mean (assuming there's anything to DC, which I agree is dubious) that the relationship of those two branches is different than their relationship to the others, which is a merely a matter of internal classification. The validity of Balto-Slavic and Indo-Iranian do not mean that those families cannot be part of IE. kwami (talk) 22:57, 17 November 2008 (UTC)

first element of this hypothesis to be well received by specialists of the languages in question
Is this true? Hasn't some of the work on Yeniseian, Caucasian, Burushaski been by Starostin and other proponents of Dene-Caucasian? --JWB (talk) 18:27, 30 August 2009 (UTC)


 * Yes, but it hasn't been accepted by others. Dene-Yeniseian is the first link that other linguists find convincing. kwami (talk) 19:01, 30 August 2009 (UTC)
 * I mean, haven't some of those linguists worked at a specialist level on those individual languages or families, apart from the question of long-range comparison. For example, Starostin's North Caucasian Etymological Dictionary. --JWB (talk) 20:43, 30 August 2009 (UTC)


 * I think the point is that they haven't been able to convince anyone else who isn't a long-ranger. People who work just on Caucasian aren't sure about Starostin's Caucasian etymologies, let alone anything broader. But DY is different: it's convincing to just about anyone who looks at it, including people who really know the languages in question and could spot the kinds of errors that caused the academic community to trash long-range proposals like Amerind.


 * Or to put it another way, a long-ranger may come along and propose some intriguing connection, but it then falls to real scholarship to justify it. Starostin was trying to do just that w Altaic and North Caucasian, and may have convinced some people, but even so, North Caucasian is already considered a unit in DC. DY is the first established connection between two branches of DC. kwami (talk) 21:00, 30 August 2009 (UTC)


 * My point was not to compare D-Y vs. D-C. I'm talking about what "specialists of the language in question" means. Is Starostin for example disqualified from being a specialist in Yeniseian simply because he has also done long-range comparison, while someone who had done exactly the same work on Yeniseian but not long-range comparison would qualify as a specialist? --JWB (talk) 22:34, 30 August 2009 (UTC)


 * No, of course not. But AFAIK he hasn't specialized in any one area. Maybe I'm wrong here. Starostin would certainly seem to be the most credible advocate of these proposals. Maybe the text needs to be reworded. (Reconstructing a proto-language does not mean you know anything about the languages. Plenty of reconstructions are based on rooting through dictionaries. Not that I'm accusing S of that.) The point is that there have been people, such as Starostin, working on these proposals, but they've largely failed to convince anyone else, including specialists in these languages who are not working on these proposals. That changed with DY, when for the first time outsiders were convinced. kwami (talk) 22:48, 30 August 2009 (UTC)

"disputed by some linguists"
I changed this to "nearly all". There are very few who support Nostratic, DC, Austric, and the like. Even those who think aspects of them are promising seldom agree in toto. — kwami (talk) 08:04, 17 May 2011 (UTC)

Aegean/Minoan, Hurro-Urartian, Sardic languages
According to some hypotheses, these languages are related to North Caucasian or Basque. We might mention that somewhere, but is it also reflected by any Dené-Caucasianists? --Ivadon (talk) 18:40, 6 November 2015 (UTC)

The Andis & Tseses in History of Hypethesis
"...North Caucasian and that their separation from it must postdate the dissolution of the Northeast Caucasian unity (Avar-Andi-Tsezian), which took place around the 2nd or 3rd millennium BC." Sorry,Why are they here Andis & Tsezes? There are only a maximum of 7-9 villages. Why they should be referred in Wikipedia? They do not have any of his literary language, there is no newspaper, no radio and television, they do not have anything in the cultural plane. Who are they interested in? All of them so so-called "originality" from their deep cultural underdevelopment and economic backwardness.How do they relate to the topic? Who is interested in them, when they broke up, etc.? What about them so often mention in Wiki? — Preceding unsigned comment added by 109.172.58.15 (talk) 13:30, 26 October 2016 (UTC)

This is a non-professional article. It's necessary to delete
This is a non-professional article. It is necessary to delete. The article in the encyclopedia cannot begin with such words in an aggressively non-professional spirit. 1.A serious encyclopedic article cannot begin with such words: "is a largely discredited" (!?); 2.In the article a lot of space is given to the vulgar criticism of S.A. Starostin; 3.Obviously, the author of the edits is very willing to argue with Starostin, and he lumps together with famous names in small linguistics for the names of small dwarf singe-scientists (unknown to anyone in the scientific world).This article cannot consist of a primitive (and illiterate) criticism of an outstanding and generally recognized scientist in the world.If the author of edits so desires to criticize S.A. Starostin, then let him do it on a separate page, and not here;4."...the dissolution of the Northeast Caucasian Unity (Avar-Andi-Tsezian)..." Sorry...What it is? Here we are talking about the Nakh-Daghestanian group.The most numerous representatives of this group are Chechens and Avars. After them there are other relatively numerous peoples: Darginians, Lezgians, Ingushes, Tabasaranians, Laks, but the so called "Andians" (Avar name for "Kwan/Khuani/Hoan") and Tsesians are not any nations at all, but the inhabitants of two or three mountain villages, they have no writing literary language, no own culture and literature.And the inhabitants of two or three mountainous unwritten villages each time should not be mentioned next to the name of the numerous Avar people, whose name is known to the whole world.These quans are a small mountain subgroup within the numerous Avar people.Why should these Andians and Tsezes be recorded with the Avar/Abarshahr nation through a dash? --Wrkan (talk) 08:09, 7 March 2019 (UTC)


 * Wikipedia is an open encyclopedia. Please feel free to make constructive edits to improve the accuracy and neutrality of the article, if there are any issues. Btw, the belittling of ethnic groups – however small they may be – will not contribute to the improvement of the article. -- Austronesier (talk) 10:15, 7 March 2019 (UTC)
 * You are not focusing on where you need to emphasize. The article states "The North Caucasus Group" and next is "Avaro-Ando-Tsezian". Thus, it turns out that the North-Caucasian family consists only from "Avar-Ando-Tsezian". And secondly, Ando-Tsezian cannot be indicated on an equal footing with the Avar people, because they are not nations, but ethnic groups that are simply invented by the Communists living in two or three auls. And you do not need to point them all the time next to the Avar people.I repeat once again that the article should be completely removed. It is written in an aggressive style and begins with false aggressive statements--Wrkan (talk) 09:33, 11 March 2019 (UTC)
 * This article more or less correctly sums up the scholarly consensus that "Dene-Caucasian" never existed. If you have issue with that, take it up with the linguistic community. The purpose of Wikipedia is to report the general viewpoint of those deemed as having the most expertise and eminence in the field, and they agree that it is a largely discredited theory.110.45.29.54 (talk) 09:05, 24 March 2019 (UTC)
 * Note:110.45.29.54 is banned user WorldCreaterFighter. Also, I have undone the edits of the sockpuppets of WorldCreaterFighter. Alexlatham96 (talk) 06:29, 2 December 2019 (UTC)