A couple of years ago I posted about Edward Vajda’s discovery of the link between Ket (spoken in Siberia) and the Na-Dene languages of America. Now the Ottawa Citizen has a story on a new collection of articles by Vajda and other experts, describing one of Vajda’s insights:

Vajda, a linguistics professor at Western Washington University, told Canwest News Service in 2008 how years of research with the Ket culminated with a dramatic insight involving words associated with the canoe.
He found that the few remaining Ket speakers in Russia and the Dene, Gwich’in and other Athapaskan speakers in North America used almost identical words for canoe and such component parts as the prow and cross-piece.
“Finally, here was the beginning of a system that struck me as beyond the realm of chance,” Vajda wrote at the time. “At that moment, I think I realized how an archeologist must feel who peers inside a freshly opened Egyptian tomb and witnesses what no one has seen for thousands of years.”

It’s a good piece in general; as John Cowan, who sent it to me, says, “Big news! A language-based story in the [mass media] that gets all the facts right.”


  1. marie-lucie says:

    LH, thank you for linking to this article. Most of the information in it is not new: the book called The Dene-Yeniseian Connection: Bridging Asia and North America (published by the U of Alaska) is a multi-authored work arising from the conference held in Fairbanks and Anchorage in 2008, at which Edward Vajda presented his results in front of several respected historical linguists who approved his work. Other presenters talked about relevant resemblances in mythology and other disciplines.
    I attended the conference too and while I think that the hypothesis of the Dene-Yeniseian connection (which had been proposed earlier by several scholars, for instance Sergei Starostin, though not generally accepted) is correct overall (although – unsurprisingly – details will probably take a long time to figure out), I am skeptical of the claim that the linguistic link shows that Athabaskans must have crossed over from Asia more than 10,000 years ago, at the end of the last Ice Age. This implies that they would have been the earliest inhabitants of the New World, yet the languages are only found in the parts of America closest to Asia (mostly in Alaska and Northern Canada, with a few other spots in some of the Western states), and they are a still a very distinctive and homogeneous language family in the Americas. I think that the link with Asia may well be thousands of years old, but not as many thousands as supposed: the short stretch of water (thoroughly frozen in winter) known as the Bering Strait cannot have been an insuperable obstacle to human travel for more than ten thousand years.
    The Wikipedia article on Dene-Yeniseian gives a short comparative word-list in Ket and Navaho, which is a poor choice as Navaho is the farthest Athabaskan language from Asia and there are not many resemblances – it would have been better to use words from Athabaskan languages closer to Asia (Vajda considered a large number of those languages, not just Navaho). Besides, word comparison can be very misleading: crucially, Vajda has also made careful comparisons of the unusual grammatical structures found in the two language families.
    Vajda and some other scholars were worried about how their announcement would be received by indigenous Athabaskans in Alaska, but the article says that they have established contacts with the Ket and a delegation is on its way to Ket country.

  2. Athel Cornish-Bowden says:

    This is interesting, and certainly points to some contact between the two cultures, but does it really provide evidence of a relationship between the languages? Aren’t “canoe and such component parts as the prow and cross-piece” analogous to “telephone” in the modern world? The fact that essentially the same word is used in English, Turkish and Indonesian tells us something about the diffusion of technological artefacts, but not much about the languages.

  3. You have a point about the diffusion of technological artefacts, Athel, but there’s more to the “relationship” than just the words for canoe parts, as the Ottawa Citizien article says:

    The collection of articles by Vajda and other experts details a multitude of clear connections — nouns, verbs and key grammatical structures — between the language spoken by the Ket people of Russia’s Yenisei River region and dozens of languages used by North American aboriginal groups.

    In fact the first sentence of article reports only “similarities”, not “relationships”:

    An obscure language in Siberia has similarities to languages in North America, which might reshape history, writes Randy Boswell.

    By the by, that last sentence is poorly formulated: is it similarities between languages, languages as such, or North America that might reshape history ? I think all three can be said already to have shaped history, but what is meant is presumably that the discovery of similarities might reshape our assumptions about past events (but not about “history”, whatever that may be).
    To take your “telephone” example a step further, but replacing Indonesian by German: suppose it were determined that the word “cellphone” is now used in English and Turkish, but not in German. Would that suggest that cellphones had not made it to Germany ? It might if you didn’t know that the cellphone is there called das Handy.

  4. michael farris says:

    I’m sure the canoe similarity didn’t play a huge role in forming the conclusions. But it is unusual enough to inspire a scholar considering the sheer physical distance between two groups with no record of contact or intermediaries – Siberia is a big place and the Ket seem to be physically closer to European Russia than the Bering Straight.
    It also is a very striking narrative of the kind the media looks for to ‘humanize’ stories. It’s so nice for the mass media can get things relatively right with a languge related story that I’ll forgive the writer a little (fact based) dramatization.

  5. But shouldn’t this be setting the alarm bells ringing? I honestly have no opinion on the merits of the Yeniseian-Dene connection (I have heard both ringing endorsements and out-of-hand dismissals of Vajda’s work from card-carrying historical linguists), but surely taking nearly “nearly identical” words at face value, especially at these time depths, is precisely the same error that we warn undergrads in hist ling courses against on day one? Does anybody have access to Vajda’s book, and to the proposed etymologies?

  6. aqilluqqaaq says:

    Hmm. Are the doubts and suspicions purely a priori and methodological? My baseline assumption is that when it comes to Ket, kī-ɾ hīɣ áb-ìl áqtȁ dlóvèɾavet. The only Ket word for ‘canoe’ I know is díltìj-. Is that the one?

  7. marie-lucie says:

    I guess you could call me a “card-carrying historical linguist”, at least a practicing one. Read my comment above. Some historical linguists insist on everything being totally sewn up before they will even consider a hypothesis. This is unrealistic. Others are willing to entertain a possibility (and keep up with the progress of an investigation) even if they are not totally convinced or disagree with some of the details. I belong to the second group and have attended several of Vajda’s presentations over the years, including the ones at the 2008 conference. There is a place between “ringing endorsements” and “out-of-hand dismissals”, especially if those come from people unfamiliar with the languages in question.
    Comments above are based on the words of a newspaper article about the topic, not on Vajda’s own writings. Of course an article for the general public, written by a non-linguist reporter, will concentrate on reporting striking items, and will not necessarily be technically accurate: “nearly identical words” might be the reporter’s interpretation, where “showing regular phonological correspondences identified in a number of other words” would be opaque to most people. And as I wrote above, I think the time depth which is assumed is grossly exaggerated.
    The “canoe” cluster of words is only one of many resemblances, which are certainly not limited to a few items of vocabulary. Diffusion might be an explanation if resemblant words existed in languages in geographically intermediate locations, but that is not the case. (Wikipedia under “Dene-Yeniseian” shows a map with the locations of the languages – there is quite a distance between Dene and Yeniseian).
    See a description of the book, with the table of contents, at:
    The book sells for $40.00.

  8. But shouldn’t this be setting the alarm bells ringing?
    No. There’s all the difference in the world between “Hey, these words look similar, therefore the languages are related!” (the undergraduate error) and “Hey, these words look similar, therefore it’s worth doing further investigation to see what’s going on!” (which is the case here). The canoe words are just one tiny fragment of the case for the relationship; it just happened that they are what inspired the researcher to look further, which makes a nice jumping-off point for a newspaper story. Is the reporter supposed to ignore anything colorful for fear of inspiring kooks? Believe me, kooks don’t need inspiration; they generate their own.

  9. marie-lucie says:

    Hat refers to his earlier post in his first sentence above. Those interested in the topic should (re)read both that post and the comments. Hat included the press release from the conference, prepared jointly by Eric Hamp and Johanna Nichols (both internationally known historical linguists) together with Ed Vajda.

  10. marie-lucie, in the description of the book you linked I find the expression “robust noun-forming instrumental suffix”. Do you happen to know what “robust” is supposed to mean here ? I would expect it to be more than a blurb word, say as applied to tennis shoes. There is the unexceptionable expression “robust theory” – yet “robust” here is being applied to a suffix.
    Is it a claim that is being described as “robust”, i.e. a claim that the formation in question is indeed a “noun-forming instrumental suffix” ?

  11. marie-lucie says:

    Grumbly, I could be wrong, but I think that it means that the suffix in question is well-attested, so that its form and function are identifiable with confidence: it might occur in a substantial number of words, along with words that are identical except for this suffix, for instance (eg English -er in teacher/teach, etc). An apparent affix (prefix or suffix) occurring in only a few words, or without affixless counterparts, might be misinterpreted (eg is this an affix, or part of the root? and if it is an affix, what is its meaning and function?), so it would not be “robust”.

  12. aqilluqqaaq says:

    I’m not sure if this is an answer to your question, but in Ket (Vajda, 2004) there are both the case-suffix -(à/a)s, which marks the instrumental, and the derivational morpheme -s, which is a nominalizing suffix. It may be (seems likely?) that the two are etymologically linked. The latter, at any rate, is ‘robust’ in the sense that it’s productive, takes case endings, and is (apparently, etymologically) a free root. So, e.g. óbdàs (‘one/something belonging to father’) from ōb- (‘father’), -da- gen. suf., and -s nom. suf. (cf. óv-às (animate masc. sg. root + instr. suf.).

  13. marie-lucie says:

    The “canoe words”: the article does not say that they were Vajda’s point of departure or first inspiration, in fact finding this set of resemblances occurred “after years of research”. Vajda’s first presentation years ago, which I attended, was not on vocabulary items but on the structural resemblances in the verb (a very complex structure in both language families). The 2008 Alaskan conference was preceded by a two-day workshop in which Vajda presented the details of his work to linguists such as Eric Hamp, Johanna Nichols, and also Michael Krauss and Jeff Leer who are the foremost specialists in Dene historical linguistics. Krauss had earlier been very skeptical, but he declared himself convinced by Vajda’s presentation in this workshop.

  14. A few points–
    1-A distinction should indeed be made between the date when the unity of Yeniseic and Na-Dene was lost and the date when Na-Dene entered the Americas. On the latter point, Proto-Na-Dene might have lost its unity in Siberia: perhaps (early) Tlingit, Eyak and Proto-Athabaskan each entered the Americas separately (or perhaps Proto-Eyak-Athabaskan and Tlingit each did).
    2-Conversely, there is no reason to think Ket (and its extinct kin belonging to the Yeniseic family) has always been in its present-day location: in fact, proto-Yeniseic-Dene may well have been spoken in far Eastern Siberia (or even Alaska), with Yeniseic being further removed (geographically) from the Urheimat than Na-Dene. This might explain why Ket is so unlike other Siberian languages, typologically (whereas a language like Haida, no longer believed to be genetically related to Na-Dene, is typologically very Na-Dene-like).
    3-The Ket-Dene similarities involving the word for “canoe” and related technologies might indeed be due to diffusion (from one another, or from some third language. If similar such words were to be found in languages not in contact with Ket or Na-Dene today, it might give a clue as to the location of the Urheimat).
    4-Diffusion would definitely prove to be the case if the “canoe” words do not conform to the sound laws which comparison of more basic Ket/Dene morphemes will establish.
    5-Obviously, if we could find other genetically related languages we would have a much better idea as to the proto-language as well as to where its Urheimat was located. Which leads me to ask you, Marie-Lucie: did anyone at the Conference mention the attemps made to link Yeniseic and Burushaski? George Van Driem, in his (excellent) LANGUAGES OF THE HIMALAYAS, seemed to find this work very convincing: I mention this because he is otherwise quite cautious about “long-range” relationships.

  15. aqilluqqaaq says:

    there is no reason to think Ket (and its extinct kin belonging to the Yeniseic family) has always been in its present-day location
    But good reason to think it spread from the south Siberian forest-steppe to its present location in north central Siberia.

  16. A comparison of Dene-Yeniseian and Dene-Caucasian came up with Burushaski here some years ago.

  17. marie-lucie says:

    1-A distinction should indeed be made between the date when the unity of Yeniseic and Na-Dene was lost and the date when Na-Dene entered the Americas.
    Good point.
    2-Conversely, there is no reason to think Ket (and its extinct kin belonging to the Yeniseic family) has always been in its present-day location: in fact, proto-Yeniseic-Dene may well have been spoken in far Eastern Siberia (or even Alaska), with Yeniseic being further removed (geographically) from the Urheimat than Na-Dene.
    Ket is the lone survivor of a larger Yeniseian family, several members of which are attested from old Russian documents. It seems that languages of this group were once spoken in a much larger area to the South of their historical location, as shown by names of more Southern rivers and other place names. I don’t think that a Far Eastern location was mentioned.
    … a language like Haida, no longer believed to be genetically related to Na-Dene, is typologically very Na-Dene-like…
    I don’t know enough about the details of either Na-Dene or Haida, but Michael Krauss, a noted Athabaskan specialist, wrote that as soon as he took a serious look at Haida, he found it very different from any Athabaskan languages.
    3-The Ket-Dene similarities involving the word for “canoe” and related technologies might indeed be due to diffusion (from one another, or from some third language. If similar such words were to be found in languages not in contact with Ket or Na-Dene today, it might give a clue as to the location of the Urheimat).
    As far as I know, those words are not found in other languages.
    4-Diffusion would definitely prove to be the case if the “canoe” words do not conform to the sound laws which comparison of more basic Ket/Dene morphemes will establish.
    I don’t think that Vajda would have used this example if the words had turned out not to conform to the sound laws he had discovered using other words.
    5- … did anyone at the Conference mention the attemps made to link Yeniseic and Burushaski?
    I don’t remember anyone doing so, since the conference concentrated on the Asia-America link. Last year, I attended the Athabaskan languages conference in Berkeley, where Vajda presented another paper about his findings. George Starostin and John Bengtson also presented papers, and I don’t think they mentioned it either.

  18. marie-lucie says:

    Ed Vajda is not the only person who has linked Yeniseian and Na-Dene: Sergei Starostin (George’s late father) and some other Russian linguists had proposed a link earlier (and George is also working on documenting it), but it seems that Vajda has gone farther than any others in documenting the link to the satisfaction of North American Athabaskan specialists and historical linguists (the latter tend to be stricter, or less bold or imaginative, than some of their European counterparts)(both strictness and boldness have their uses, and are not necessarily incompatible).

  19. I might have come across as more hostile to Vajda’s work than I intended to.
    There is a place between “ringing endorsements” and “out-of-hand dismissals”
    “nearly identical words” might be the reporter’s interpretation, where “showing regular phonological correspondences identified in a number of other words” would be opaque to most people.
    Indeed, which is why I asked if somebody could quote the exact words (or word clusters). As I stated, I really have no opinion either way. I am perfectly willing to accept that the connections are legit. My main concern was the presentation, that’s all.

  20. Bademantel says:

    reshape our assumptions about past events (but not about “history”)
    I looked twice at the assertion about “reshaping history”. But if you accept that “history” is basically mankind’s attempts to describe and interpret the past, and not actually the past itself (whatever that might be), there is nothing wrong with talking about “reshaping history”. Much of what people call “history” isn’t “what happened” at all; it’s what people say happened. Two different things.

  21. marie-lucie says:

    PI: My main concern was the presentation
    For an article in the daily press, “resemblances of verbal morphology” would not have attracted many readers (or reporters), but “nearly identical words for canoe” could strike their imagination. Verbal morphology was indeed the beginning of Vajda’s observations about Ket and Na-Dene, and vocabulary came later.
    Vajda used to have several relevant papers available online, including what became the introduction to the recently-published book, but they seem to have been removed in order not to compete with the book.

  22. I like all these language related stories coming out in the media these days. Public interest is giving me encouragement about future job security. 🙂 #justagradstudentkindofsentiment
    But to be more serious, I think I’m going to have to pick up a copy of this APUA volume. There is no doubt that this has to be an old language family, but I must agree that this must bring into question our thoughts on time-depth in Na-Dene and the Dene-Yeniseian macrogroup.
    How exciting!

  23. Marie-lucie: if we are talking about the same Michael Krauss article, then I believe he was referring to Haida being dissimilar from Athabaskan in terms of lacking cognate morphemes.
    However, TYPOLOGICALLY, Haida does not “stick out” from its neighboring languages the way Ket (or Basque) does. This does suggest that Ket (and Yeniseic as a whole?) has moved to its present location comparatively recently.
    Perhaps this migration was from further South, as some others have pointed out on this thread. This possibility makes it especially worthwhile to pursue investigations into a Yeniseic-Burushaski link: even if they turn out to be genetically unrelated, there are typological similarities which might be due to both languages having once been in contact.

  24. … not actually the past itself (whatever that might be) … Much of what people call “history” isn’t “what happened” at all; it’s what people say happened.
    Exactly my point, Bademantel ! That may seem to be a justification for the 2001 French legislation saying that there was a genocide carried out on Armeniens in 1915. Here is the entire text of this peculiar law:

    Article 1

          La France reconnaît publiquement le génocide arménien de 1915.

    Jacques Chirac

    Par le Président de la République :
    Le Premier ministre,
    Lionel Jospin

    There are no sanctions on anyone claiming there was no genocide, according to -this->:

    Bien que la France ait reconnu le génocide arménien, par la loi du 29 janvier 2001, la négation du génocide n’est pas punissable en tant que telle. L’Assemblée nationale a adopté en première lecture le 12 octobre 2006 une proposition de loi permettant de réprimer la négation du génocide arménien en se référant et en complétant le dispositif législatif préexistant[29]. Le Sénat ne s’est pas encore prononcé sur le texte.

    The curious thing is, the idea that “history” and “the past” are not something objective, but are “what people say happened”, has brought the French government to within a flea’s blink of censorship. And the idea reeks of relativism.
    The natural solution to this dilemma is to continue talking about history, the past and the present as if they were objective things for which evidence can and should be brought, but to freely admit that our understanding of certain things at a given moment may subsequently change when the evidence changes (and hysterical arguments have been settled about whether the new evidence is really evidence etc). In other words, the desideratum is not a relativistic philosophy. but a philosophy that is willing to learn without too many tears having to be shed.

  25. The natural solution to this dilemma is to continue talking about history, the past and the present as if they were objective things for which evidence can and should be brought, but to freely admit that our understanding of certain things at a given moment may subsequently change when the evidence changes…. In other words, the desideratum is not a relativistic philosophy, but a philosophy that is willing to learn without too many tears having to be shed.
    This is well said, and is my approach as well.

  26. marie-lucie says:

    I second LH. Same thing in specific histories, eg of languages, and more generally in science.
    Etienne: Yeniseian and Burushaski
    I found out that my memory was at fault, and that this potential link was indeed mentioned by George Starostin at the 2009 Athabaskan languages conference in Berkeley.
    I am not competent to evaluate either claim, let alone one claim against the other. Besides, the two proposed links might both be right, if all three groups turn out to be (very) distantly related to one another.

  27. Wolfgang Kuhl says:

    The Institute for Bible Translation Moscow (Институт перевода Библий Москва) published a unique multilingual book on the occasion of the second millennium celebration of the birth of Jesus Christ presenting the birth story as recorded in the Gospel of Luke (2:1-20), translated into eighty languages spoken in Russia and the other countries of the Commonwealth of Independent States (CIS). (Рождество Иисуса Христа. Евангелие от Луки 2:1-20 на 80 языках народов СНГ).
    The book includes this passage in 32 languages never published before such as Dolgan, Dungan, Ket, Nganasan, Rushani, Shor, Shugni, Selkup, Tat, Tsakhur, Yazgulyam etc.
    Mentioned below are a few examples of the translation into the ”Ket” language:
    1. Ким долақ закон, туре қясьда и Август баря қан дуббетин дэ’ңнаириңбет быльдэ ба’ңдиңт.
    (And it came to pass in those days that a decree went out from Caesar Augustus that all world should be registered.)
    2. Кире овыльдэ қусям дэ’ңнаириңбет, бу уонь,
    аська Сирия ба’ңдиңт татыңқовыльда буря и Квириний-қый қя ке’т.
    (This census first took place while Quirinius was governing Syria.)
    14. “Ақтам Есьдаңт, туре дэ’ң быльде ба’ңдиңт лювероңавет !“
    (Glory to God in the highest, and on earth peace, goodwill toward men.)

  28. that all world should be registered
    At the biblos site there are many variations on “that all the world should be taxed”, the translation I knew from having the KJV Luke 2:1-20 read at me on Christmas eve:

    New International Version (1984)
    that a census should be taken of the entire Roman world.
    English Standard Version (2001)
    that all the world should be registered.
    New American Standard Bible (1995)
    that a census be taken of all the inhabited earth.
    that all the world should be enrolled.
    Bible in Basic English
    that there was to be a numbering of all the world.
    Darby Bible Translation
    that a census should be made of all the habitable world.
    [I like the lawyerly precision of this one. It explicitly excludes those parts of the world where nobody lives.]
    Weymouth New Testament
    Just at this time an edict was issued by Caesar Augustus for the registration of the whole Empire.

    I find the notes and commentaries at biblos fascinating. Here is an excerpt from Barnes’ Notes on the Bible:

    That all the world – There has been much difficulty respecting this passage, from the fact that no such taxing of “all the world” is mentioned by ancient writers. It should have been rendered “the whole land” – that is, the whole land of Palestine. The “whole land” is mentioned to show that it was not “Judea” only, but that it included also “Galilee,” the place where Joseph and Mary dwelt. That the passage refers only to the land of Palestine, and not to the whole world, or to all the Roman empire, is clear from the following considerations …
    Should be taxed – Our word “tax” means to levy and raise money for the use of the government. This is not the meaning of the original word here. It means rather to “enroll,” or take a “list” of the citizens, with their employments, the amount of their property, etc., equivalent to what was meant by census. Judea was at that time tributary to Rome. It paid taxes to the Roman emperor; and, though Herod was “king,” yet he held his appointment under the Roman emperor, and was subject in most matters to him. Farther, as this “enrollment” was merely to ascertain the numbers and property of the Jews, it is probable that they were very willing to be enrolled in this manner; and hence we hear that they went willingly, without tumult – contrary to the common way when they were “to be taxed.”

  29. J.W. Brewer says:

    So do we need a statute reading “La France reconnaît publiquement le [common origin of Ket & Na-Dene]”?

  30. “what people say happened”….relativism
    How many hundreds of eye-witnesses do you need before “what people say happened” is actually “what happened”? Not to mention photographs and documents. There is no question that the Armenian genocide really happened, any more than there is any question that the holocaust really happened. Ask the Armen who fled to Amman and still live there on Ashrofeeya overlooking the Roman theater. Anyone who says you have to cover up history in order not to “offend” the nation that perpetrated the massacres has very little sympathy from me. Making it illegal to engage in a coverup is not “censorship”.

  31. Nijma, who on earth are you arguing with? Nobody is disagreeing with you, and it would be nice if you didn’t turn this into a shouting match about genocide.

  32. marie-lucie says:

    So do we need a statute reading “La France reconnaît publiquement le [common origin of Ket & Na-Dene]”?
    JWB, the French government does not spend its time legislating about every possible detail of scholarship. This is a serious issue in Europe. The French ruling about the Armenian genocide comes a few years after a harsher one about the Holocaust. The intent is to stop genocide-deniers (Nazi sympathizers, antisemites, etc, including not just skinheads but a few prominent “revisionists” – historians, political writers, etc) from continuing to propagate whitewashed versions of history in spite of the incontrovertible evidence against them. Search for “laws against Holocaust denial”, which exist in several European countries (including Germany), with the intent of acknowledging the crimes of the past and preventing their recurrence.
    The Armenian genocide is denied by the Turkish government, which teaches a different “history” from the rest of the continent. Laws in France and other countries against denying this genocide prevents propagandists from publishing in these countries.

  33. J. W. Brewer says:

    Trying to find something to discuss other than genocide, free speech, and France, I came across this interesting piece , which claims that you wouldn’t hypothesize any historical connection between the present-day speakers of Ket and the various Na-Dene languages based on looking at their DNA. (It also indicates that in the pre-Vajda days the same linguistic hypothesis was being touted by the oft-controversial Merritt Ruhlen.)

  34. marie-lucie says:

    JWB: In spite of the well-publicized pronouncements of Greenberg, Ruhlen and Cavalli-Sforza a few years ago (about the peopling and languages of the Americas), the findings of linguists and population geneticists are often at odds, because people keep their genes (and mix them with others’) while adopting other languages. The non-indigenous population of the Americas, drawn from Europe, Africa and more recently Asia, provides abundant proof of this phenomenon. It is also well-known that the current distribution of languages in Europe, for instance, does not reflect the first human peopling of that continent. But non-trivial linguistic resemblances (= those which are very unlikely to be due to chance) are nevertheless very important in order to trace population movements, because even if the populations have changed over the centuries, languages maintain traces of their origin and relatedness for centuries and even millennia (the recognized Indo-European languages are the best-known example, but there are others).
    In my opinion, “the oft-controversial Merritt Ruhlen”‘s pronouncements need to be taken with more than a tablespoonful of salt. As I wrote above, he was certainly not the first to propose such a link. From what I know of his work, he is a follower, not a leader.

  35. There is no god but Greenberg, and Ruhlen is his prophet.

  36. marie-lucie says:

    That’s it, JC! (depending on your religion, of course – it isn’t mine).

  37. aqilluqqaaq says:

    The ‘canoe’ example from the March ’08 draft of Vajda, E. 2010. “A Siberian link with Na-Dene languages.” The Dene-Yeniseian Connection, ed. by J. Kari and B. Potter, 33-99. (Anthropological Papers of the University of Alaska, new series, vol. 5):

    *k’ʷ / *č’ and / *č’ before front vowels in early Yeniseic → ??? t (Ket/Yugh), (Kott), with following /ə/ and /ɯ/ rather than /e/ and /i/ in Ket/Yugh PA*č’ʷiˑx > ‘canoe’; Ket/Yugh tij ‘canoe’. Note that the original vowel in Ket/Yugh must have been ɯ, which fronted to /i/ between /t/ and /j/.

  38. Trond Engen says:

    IIRC Vajda said after announcing his results that Ruhlen’s correspondences are all wrong. I also seem to remember that he flatly denied the existence of Dene-Caucasian or a Burushaski connection.
    I have the same feeling as marie-lucie about the time depth (but far less to build it on). The sheer size of the Na-Dene group in America and the size of the region with Yeniseian river names in Siberia suggests to me that there at least for some time must have been a substantial population on both sides of the Bering Strait, in its turn suggesting frequent crossing one way or both. I’ve been trying to hunt down a post-glacial climate history of the Bering region to look for less hostile periods, but no luck yet. The interest until now seems to have been focused on the properties of the landbridge.
    Not that one necessarily needs a less hostile period. Eskimo-Aleut crossed the Bering Strait too. Maybe Na-Dene and Yeniseian is a predecessor of E-A, spreading fast with highly mobile hunter-gatherers along the coast of the Arctic Ocean, gaining population and going inland? That’s a much shorter route. And, hey, boat terminology is even core vocabulary.

  39. Nijma, who on earth are you arguing with?
    Not the whatever-deniers, I hope. The best you can hope for when you meet them is that you see them first and can go the other direction before they see you. They’re not the type of people to spend their time singing Kum Ba Ya, or translating the Declaration of Human Rights into esoteric languages, much less take an interest in evaluating scholarly evidence. Anything that doesn’t fit their agenda goes down the memory hole. That’s why my blog and all other blogs with a WordPress subdomain are blocked in Turkey. It’s unfortunate, but there don’t seem to be any nuanced positions in that particular dogfight. And why dismiss eyewitnesses out of hand?–they have been respectable since the time of Herodotus.
    it would be nice if you didn’t turn this into a shouting match about genocide
    I’m not sure why I’m being characterized as shouting; nothing is in all caps, and I surely don’t insist that you or anyone else agree with me. To be sure, typing the word “genocide” is a political act (our president says “events” or somesuch), and I hide behind a nom de guerre to type it, but to say the French anti-hatespeech laws “reek” and characterize them as censoring and relativistic is not exactly apolitical.

  40. marie-lucie says:

    TE: IIRC Vajda said after announcing his results that Ruhlen’s correspondences are all wrong.
    I am not sure if he said “all” wrong, but only a very few seem to be right. “Correspondences” does not mean just superficial resemblances between words – correspondences of sounds must be found to occur regularly (like p/f in Latin pater, piscis, etc vs English father, fish, German Vater, Fisch, etc). Such correspondences are easy to find in closely related languages, more difficult in those which are more distantly related (sounds have a way to change over time so that words that started the same in two languages can become totally different over the centuries). A list like Ruhlen’s can be considered interesting, but not definitive proof of a connection unless the details fit in with other results (this is one of the problems with Greenberg and Ruhlen’s work in general – they jump to conclusions based on superficial details).
    I also seem to remember that he flatly denied the existence of Dene-Caucasian or a Burushaski connection.
    Vajda was studying Ket, the last surviving Yeniseian language. When you study an isolated language or family, you wonder if it could be related to some other known languages farther afield. Several links had earlier been proposed between Yeniseian and other families, mostly by Russian linguists. The Dene languages are a much larger group, but their potential affiliation to other language groups is also disputed. The scholarly climate in North America is definitely against suggesting links between Amerindian languages and those of other continents: for one thing, the mainstream theory that America was populated before tne rising of the sea which created the Bering Strait (an event which is securely dated geologically) precludes finding relatives of Amerindian languages in Asia because of the too great time-depth since the separation of the continents. (Personally I think that all the evidence points to several population influxes in the meantime, but that is a minority opinion).
    In his study of Ket, Vajda read the relevant literature on the connections that had been proposed. He found resemblances with Na-Dene the most convincing and pursued that connection. His results are now considered valid by a number of American historical linguists familiar with both the methods and the languages (although those results, or some of the details, are not accepted by everyone). He did not find the other proposed connections as convincing, but in linguistics it is axiomatic that you can never state definitely that any two languages are NOT related (since presumably all human beings descend from the same original group), you can only find supporting evidence to show that languages ARE probably related: this is easy for closely related languages (eg the Germanic or Slavic languages), more difficult for those more distantly related (eg the Indo-European languages, which only seem easy to identify after the fact), and much more so for those which might be even more distantly related (eg potential “sisters” of the Indo-European group, for instance those in the disputed “Nostratic” hypothesis). People unacquainted with linguistics cannot imagine the amount of work necessary to investigate such hypotheses and have them accepted as valid.
    Vajda and others think that he has found plenty of evidence supporting a Dene-Yeniseian connection (the first American-Asian linguistic connection considered to rest on substantial linguistic evidence). As for those other proposed connections, he did not find the Yeniseian-Burushaski one promising, and “Dene-Caucasian” would be something for Na-Dene specialists to consider (thus far, this hypothesized connection has not found any supporters in America).

  41. aqilluqqaaq says:

    IIRC Vajda said after announcing his results that Ruhlen’s correspondences are all wrong.
    He said: “Merritt Ruhlen’s (1998) proposed cognate sets contain several genuine cognates, among over 75% coincidental look-alikes.”
    I also seem to remember that he flatly denied the existence of Dene-Caucasian or a Burushaski connection.
    He describes such connections as “the linguistics profession’s equivalent to ‘urban legend’”.
    m.-l., do you know Ket?

  42. marie-lucie says:

    No, I don’t know Ket or even Na-Dene (except from what I can glean in various publications), so I am relying on what I heard (at the conferences) and read (some of the conference papers), as well as on my acquaintance with the milieu of Amerindian historical linguistics and some of its practitioners. Here I am trying to describe the situation as I understand it, but I don’t have a personal interest in either side of the controversy. As I said earlier, there is a middle ground between “ringing endorsements” and “out-of-hand dismissals”, and I am not competent to produce either. I am very interested though.

  43. marie-lucie says:

    I also seem to remember that he flatly denied the existence of Dene-Caucasian or a Burushaski connection.
    He describes such connections as “the linguistics profession’s equivalent to ‘urban legend’”.
    aqilluqqaaq, from your name I gather that you study Eskimoid languages, and I don’t know where you are, but in North America an Amerindianist has to tread extremely carefully about “long-range comparison”. The Dene-Yenesian connection is controversial enough as it is.

  44. Trond Engen says:

    Thank you, marie-lucie.
    I am not sure if he said “all” wrong, but only a very few seem to be right.
    I remember him quoted as saying that where Ruhlen happens to have correct cognates, the reason he finds them is purely coincidental. But “IIR” is often far from “C”, so I dug out Vajda’s 2008 paper from my computer (if anyone wants a copy they can send me an e-mail at

    Ruhlen’s (1998) list of 36
    putative cognate sets involving Yeniseic and Athabaskan/Eyak/Tlingit contains at least
    75% complete coincidences, with perhaps six that include a least one genuine Yeniseic +
    Athabaskan/Eyak/Tlingit. All of the Haida words in the same list appear to be
    coincidental look-alikes with the various Yeniseic or AET comparanda. Unless
    languages are closely related, a comparative list of basic look-alike words is useless as
    the sole basis for arguing the case of genetic relatedness. These conclusions can only be
    drawn after sound correspondences are established. As it turns out, most Dene-Yeniseic
    cognates do not readily look like related words in the modern languages. They do not
    jump off the page to beat the researcher about the eyes and face. Only a detailed
    knowledge of the historical development of the sound system of both Yeniseic and Na-
    Dene can succeed in uncovering sound laws and therefore real cognates.

    “Correspondences” does not mean just superficial resemblances between words
    Mea culpa. I chose ‘correspondences’ to avoid using ‘cognates’ for something based on superficial resemblances, but I forgot to consider the interference from its technical meaning.
    (Personally I think that all the evidence points to several population influxes in the meantime, but that is a minority opinion)
    I never have understood why the Bering Strait was unpassable until suddenly the Eskimo-Aleut came paddling across in sufficient numbers to replace whoever lived there before. I think I’ve accepted it as a coincidental barrier supported by archaeological evidence beyond my reach. Now, with the internet, nearly anything is within one’s reach.

  45. Trond Engen says:

    …, but I haven’t looked at the evidence.

  46. aqilluqqaaq says:

    Thanks, m.-l. I don’t know the Na-Dene languages either. I work primarily on the ‘Palæo-Siberian’ languages, as well as some of the associated ‘long-range’ hypotheses.

  47. marie-lucie says:

    aqilluqqaaq: I work primarily on the ‘Palæo-Siberian’ languages, as well as some of the associated ‘long-range’ hypotheses.
    Tell me more! I am interested in Chukchi-Kamchatkan myself.

  48. I’ve never understood what’s supposed to be so impossible about Neolithic-technology people crossing 85 km of strait when further south other people, also with Neolithic technology, crossed 4000 km of open ocean from the Tuamotu Archipelago to settle New Zealand, Easter Island, and Hawaii?

  49. 1-To aqilluqqaq: I must second Marie-Lucie’s request that you tell us more about your research: unlike her I have little detailed knowledge of any New World language/language family, but I know a thing or two about the principles of historical linguistics, if I may say so myself, and would love to learn more.
    2-To John Cowan: the Polynesian expansion took place much more recently than early peopling of the Americas, though I do agree with your basic point (which is also Trond Engen’s). Indeed, I believe it has been theorized that during the last Ice Age North America could have been settled by seafarers from Europe, gradually moving across the “coast” of the ice sheet the covered the North Atlantic. Considering the linguistic diversity of the Americas, the idea deserves to be taken seriously.
    3-To Trond Engen: Na-Dene is indeed a large language family in North America. But all but two languages (Eyak and Tlingit) belong to a single subfamily, Athabaskan, whose expansion (from somewhere in Alaska, according to Michael Krauss) surely postdates Na-Dene speakers’ settlement of the Americas. So looking back, Na-Dene (Tlingit, Eyak, Proto-Athabaskan) originally covered a rather small area of North America (Alaska Pacific coast, basically: there is published evidence –I can supply a reference, should anyone request it– that Tlingit may have replaced related languages/dialects, which makes it likelier that this area was the Na-Dene URHEIMAT).
    4-In an earlier comment I pointed out that Burushaski, even if it is unrelated to Yeniseic, might show typological similarities due to contact. I had a quick look at Burushaski, in my copious free time: its elaborate verb morphology, complete with prefixes as well as suffixes, is quite un-Indo-European- or Dravidian- or Tibetan-like, but rather Yeniseic- or Na-Dene-like. So there is a likeness, and a contact explanation is, like, likely! (Okay, bad pun. *SLAPS SELF*)

  50. aqilluqqaaq says:

    m.-l. and Etienne: by far the Pal.-Sib. language I know best is Chukot (ckt), which I learned through Russian. Unlike Etienne (and perhaps m.-l. too?), and despite my interest in some of the long-range material (e.g. Uralo-Yukaghir), I’m a more philosophical, than historical linguist and focus on how grammatical forms (in ckt esp. incorporation, transitivity alternations, ‘gerunds’, TAM affixes) are related to variations in semantic mood (broadly conceived). (Short version – it’s not a relation of correspondence.) I also happen to think there are some (perhaps not too surprisingly, but nonetheless sadly) underappreciated gems of Chukchi literature: Рытгэв (Юрий Сергеевич Рытхэу) is well enough known (though largely for his Russian works), but try finding anything by Ваалгыргын (Михаил Васильевич Вальгиргин) in your local library; so I’ve been known (by whom?) to do the occasional interlinear.

  51. marie-lucie says:

    3- : Michael Krauss considers Eyak and Proto-Athabaskan to have been “sisters”. Please, do provide a reference about Tlingit.
    4- : When I first looked up Burushaski on Wikipedia (at least a year ago, if not more), there was not much, but right now there is a fair amount on morphology, and indeed the very complex verb morphology does look a lot like Na-Dene’s.
    Russian linguists tend to link Yeniseic/Yeniseian with Burushaski. Yeniseian used to cover a much larger area South of its present location, and Burushaski, spoken in a remote mountain valley in the Himalayas (the typical “refuge” area, as in the Caucasus), could be the remnant of an extinct language family also once spoken in a larger area to the North, in contact with Yeniseian languages (or possibly related to them). But such links are not an either/or proposition: linking Yeniseian to both Burushaski and Na-Dene is not out of the question from a theoretical point of view – I wonder if Burushaski is also considered in the “Dene-Caucasian” hypothesis. Such far-flung links involving long-separated languages are only tantalizing hypotheses at this time.

  52. marie-lucie says:

    aqilluqqaaq: thank you for your explanation of your very interesting research.
    What is Chukot? is it another name for one of the CK languages?
    I know the name of Rytkev (I think that is the transliteration, or is it Rytkew ?), who was the topic of a National Geographic article a few years ago, but otherwise nothing about modern Chukchi literature.

  53. aqilluqqaaq: You should start a blog about that stuff. I will put it on my blogroll and read it regularly (as I’m sure will others here).

  54. marie-lucie says:

    I would too!

  55. aqilluqqaaq says:

    m.-l.: yes, Chukot (Chuchee, Chukcha, Chukchee, Chukchi), a.k.a. Luoravetlan (spoken in Chukotka Autonomous Okrug, Kamchatka Krai, and Magadan Oblast in the Far Eastern Federal District of the Russian Federation), Kerek, Koryak, and Alutor form the northern Chukotkan (Chukotian) branch and Itelmen, the southern Kamchatkan (Kamchadal) branch of the Chukotko-Kamchatkan language-family.
    Рытгэв /ɹətɣeβ/ is known in English by his Russified name Рытхэу, so Rytkheu.
    lh (and m.-l.): funnily enough I did start a Chukot language blog just a few weeks ago, but, undeveloped as it is for now, I’ve kept it private so far. If I get it into shape though, I’ll let you know if you’d like.

  56. marie-lucie says:

    aqilluqqaaq: Thank you for the corrections.
    I hope you go ahead with the blog.

  57. Please do let me know, and I hope you include some of your interlinears.

  58. 1-Aqilluqqaq: do let us know about this blog! I confess I’ve always been a little intimidated by Chukotko-Kamchatkan: one of my teachers, back in my undergraduate days, had done fieldwork on Alutor. Two others had done fieldwork on Burushaski, and whenever we had assignments in our morphology classes involving data from either language we knew we would need more coffee than usual that evening…
    2-The hypothesis I referred to above about early settlement of the Americas from Europe during the Ice Ages is known as the solutrean hypothesis
    3-Marie-Lucie: the reference: Leer, Jeff. 1990 “Tlingit: a portmanteau language family?” In: Philip Baldin (Editor) LINGUISTIC CHANGE AND RECONSTRUCTION METHODOLOGY. Berlin, Mouton De Gruyter.
    Basically, the author argues that Tlingit arose as a KOINE of (now-extinct) related Na-Dene languages.
    4-MEA CULPA: you are quite right, Marie-Lucie, Eyak has always been accepted to be closer to Athabaskan than Tlingit was, so it was quite misleading of me to refer to a three-member Na-Dene family in my earlier comment.

  59. marie-lucie says:

    3- I remember reading that article, first published in IJAL (International Journal of American Linguistics) (the “American” here refers to Native American languages). I did not remember that particular statement though. I will have to reread the article. (The editor of the book is Philip BALDI – known in particular for his book on Foundations of Latin, which you might know).
    4- The division is Eyak/Athabaskan//Tlingit, and the proto-languages are PA, PEA and PEAT, depending on the depth and inclusiveness.

  60. Trond Engen says:

    Since this thread, as one could hope, now is well above my level of erudition, I’ll just shoot in a claim for more than my due share of credit for provoking it.
    I’ll have a look at the WP article on Burushaski myself. I have a trainride coming.

  61. marie-lucie says:

    TE: I’ll just shoot in a claim for more than my due share of credit for provoking it.
    Of course, Trond, you are indeed entitled to it. I am very glad you brought up the topic, which I had not personally considered. I don’t know that much more now, but even a little at a time adds up to the sum of knowledge eventually.

  62. Trond Engen (weather reporter) says:

    I don’t think I brought up anything, hence more than due…
    Some impressions after skimming WP between my duties of snowwatching for the goatherd:
    The noun morphology is almost IE-like.
    The typology of the verb formation is indeed similar to what Vajda describes from Yeniseic and Na-Dene.
    Why use (introduce?) conitive for something simple and straightforward (which I nevertheless can’t remember the established term for)? And why divide the verbal categories into tense/aspect and mood when that distinction seems to be irrelevant to the morphology?

  63. marie-lucie says:

    TE: it’s conative not conitive, and it has to do with purpose, volition, impulse, urges, etc, a collection of meanings for which there is no single term in ordinary usage (see conation). For the rest, you will have to consult something more specialized about Burushaski, which should (one would hope) explain and justify the appropriateness of using those descriptive categories.

  64. Trond Engen says:

    Yes, sorry, conative. To me ‘conative’ sounds fit for a specialized term within the general area of optative, while the Burushaski conative, at least as exemplified in the WP article, seems to be an inchoative or ingressive (now that I start remembering terms).
    I don’t claim to have spotted a misanalysis of Burushaski grammar, though, only a still inadequate explanation in a good article. I should (as you recommend) read the relevant literature — and understand it too — before making sweeping claims.

  65. I should (as you recommend) read the relevant literature — and understand it too — before making sweeping claims.
    Bah, I never do that. I’m a blogger!

  66. marie-lucie says:

    TE, WP is OK to get some sense of what the language is like, but not for close study. The article gives a reasonably long list of sources on the language, though.
    Etienne, was the other Etienne (Tiffou) your Burushaski professor? And can I ask who were the two Alutor people?

  67. 1-To Trond Engen: inasmuch as you actually look up facts, I regret to say that you are too scrupulous to be a typical tenured linguist. Sad but true. As the saying goes, in theoretical physics all you need is paper, pencil and a wastebasket: most linguists use the same tools, minus the wastebasket.
    2-To Marie-Lucie: Yes, it is BALDI. Just a dumb typo on my part: I had read some of his work, though FOUNDATIONS OF LATIN wasn’t one of them.
    And yes, Etienne Tiffou and Yves-Charles Morin were the two professors who introduced me and my classmates to the joys of Burushaski phonology and morphology: my glimpses of the structure of Alutor I owe to Igor Mel’cuk, whose semantics course was most memorable.

  68. in North America an Amerindianist has to tread extremely carefully about “long-range comparison”. The Dene-Yenesian connection is controversial
    why is it controversial?

  69. Rytkheu was a popular writer in Soviet times, but struggled to find a publisher in post-Soviet times until Abramovich financed new editions of his work.
    Here is a youtube video of ‘When Whales Leave’, a 1981 film based on Rytkheu’s short story. Not that it adds to the linguistic side of this discussion, but I think the imagery associated with his work is representative.

  70. I’m a blogger!
    …socially inadequate, pimpled, single, slightly seedy, bald, cauliflower-nosed…
    Did you hear about that? But Marr’s own blog (programme) Start the Week is very good.

  71. Trond’s weather monitoring was scrupulous.

  72. marie-lucie says:

    Sahura: The Dene-Yenesian connection is controversial … – why is it controversial?
    Because even though it has been endorsed by some well-known historical linguists, it goes against many principles (or prejudices) common among Amerindianist linguists and anthropologists (sometimes observed in public although doubted in private, such as the belief in the Bering Strait as a barrier cutting off America from Asia for millennia). Also, some linguists may agree with the plausibility in general, while not convinced by the details of the proposal, and others will reject it completely for that reason.
    Apart from the Na-Dene specialists Michael Krauss and Jeff Leer, the other top linguists present at the workshop where Vajda’s work was considered in detail (just before the actual conference) were Eric Hamp and Johanna Nichols, some of whose work outside of their respective specialties (Indo-European, Chechen-Ingush) is considered controversial. Outside of America, Russian linguists are more inclined to link Yeniseian with Burushaski or some Caucasian languages.

  73. marie-lucie says:

    Etienne: Igor Mel’cuk, whose semantics course was most memorable
    I saw Igor Mel’cuk in action a few years ago when he was the guest speaker at our regional conference – memorable is the word! Students would not get bored at his lectures, on any topic!

  74. Bering Strait as a barrier
    meaning there are those who don’t believe that Americas, at least partly, were peopled via migration from East Siberia to America? I thought it was a long established view.

  75. marie-lucie says:

    Sashura, of course there must have been a migration from East Siberia to America, but the point of contention is whether there was a single migration before the opening of the strait, while the two continents were joined by a 1000 mile wide “land bridge”, or whether there were one or more later migrations occurring after the waters had risen and separated the continents: any later migration would have had to be by boat, which the early migrants are not supposed to have had (the much more recent Eskimoid migration is assumed to have been on foot on the ice in the winter). The single-migration hypothesis is still the one generally accepted (although there are some dissenting voices), and if so, the Dene-Yeniseian connection would have had to date from before the opening of the strait. However, as I said earlier, the degree of resemblance between the two families makes a 10,000-year-plus (probably more) period of separation, far too long for linguistic plausibility. If the linguistic connection is indeed valid, then the separation of the languages, and therefore the entry of Dene speakers into America, has to be much later than that (and as Etienne observed, the two families may have separated on the Asian side before the Dene moved into America – that would solve the resemblance problem, but not the date of entry problem). Personally, I have never believed in the single-entry-before-the-Bering-Strait hypothesis, so I am not bothered by what seems to be a non-problem. (I don’t know much about genetics, but deriving all indigenous Americans from only 70 “founding” invididuals seems to me hardly more plausible than deriving all human beings from a single female ancestress, “mitochondrial Eve”).

  76. I just signed up to your blogs rss feed. Will you post more on this subject?

  77. I see, thanks, Marie-Lucie.

  78. Marie-Lucie: If we believe, as most people now do, that modern humans all have a common descent, then “mitochondrial Eve” is not a hypothesis but a fact. If you trace the maternal line of all living persons, you will inevitably reach a single woman, because the population gets smaller as you look back into the past.
    The term “Eve”, however, is quite misleading: there is no implication that the single woman in question was the only woman in existence. There may have been, and almost certainly were, other women at the time who either have left no descendants, or whose descendants pass at some point only throughs paternal lines. One thing we do know about “mitochondrial Eve” is that she had at least two daughters, because if she had only one, then it would be the daughter who was the most recent common ancestor of all maternal lines, and therefore herself “mitochondrial Eve”.
    Similarly, “Y-chromosomal Adam” is the ancestor of all living persons through their paternal lines. However, he lived only 75 ± 15 KYBP (kiloyears before the present), whereas “mitochondrial Eve” lived about 200 KYPB.
    The most recent common ancestor (MRCA) through both lineages of everyone living today probably lived only 3 ± 2 KYBP ago, due to the pervasive genetic mixing of the last 500 years, although before that period the MRCA was someone who probably lived about 40 KYBP ago, when Australians and New Guineans separated from the rest of humanity. That degree of relatedness turns out to be quite unexpected: nobody thought until very recently that the common ancestor of all of us might have lived within historic times.
    In addition, there is a point in time when everyone alive was either an ancestor of no one alive today, or an ancestor of everyone alive today. This point is thought to be about 5 ± 1 KYPB, and is called the “identical ancestors point”, because everyone’s family tree before that point is composed of the same individuals.
    All these arguments depend on a single point of logic: the number of notional ancestors of an individual doubles in every ascending generation (we have 2 parents, 4 grandparents, 8 great-grandparents …), but the number of persons who constitute these ancestors steadily shrinks as we go backwards, as explained above. Therefore, we quickly reach a point at which the number of notional ancestors is so large, and the number of actual ancestors so few, that everyone’s ancestors must be the same.
    For example, within a vanishingly small probability, all Europeans are descended from Charlemagne, including you and me. 1200 years, or 40 generations, have passed since his day, which means both of us would have had 2^40 = more than a trillion nominal ancestors then, enormously more than the estimated 25-30 million who actually lived in Europe in that year. So since it is known that Charlemagne actually has descendants (there is a group called “The Order of the Crown of Charlemagne in the United States of America” who can prove their descent through records), we of European ancestry must all be among them. Of course, the other 25-30 million are also ancestors of ours, or else ancestors of nobody.
    (I should point out that because someone is your ancestor does not mean that you share DNA with that person. After long enough, the contribution of a particular ancestor is lost by random chance, because you pass only half your genes to your children.)
    I myself will be one of these ancestors of nobody, as my only daughter is adopted. But what do I care? My genes are replicated over and over worldwide, though none of them in the unique combination that made me. What’s one more hypothetical reshuffle of that enormous mix? I’ll live on anyway in the memories of some of my survivors, and perhaps in blog comments like these, lovingly preserved at the Internet Archive if nowhere else.

  79. @John Cowan: If we believe, as most people now do, that modern humans all have a common descent, then “mitochondrial Eve” is not a hypothesis but a fact
    So if a person believes, then what he believes is not a hypothesis but a fact ?? Or is the conclusion dependent on the numbers of believers ? Back in the 60’s, a bunch of people sat for hours in front of the Pentagon, believing that if they concentrated hard enough they could cause it to levitate. This was one hypothesis which did not turn into a fact.

  80. Come, come; obviously he means that for them it is a fact. I’m as far from a postmodernist as you can get, but even I know it’s silly to talk as if “facts” were magically apparent at all times and places. Many things that we now take to be facts will be seen otherwise by future humans (if humans are lucky enough to have a future); we can either avoid talking of facts altogether or use the term as it comes naturally.

  81. marie-lucie says:

    JC, I think I understand the arguments for the so-called Eve and Adam (not the biblical ones) you mention, but as you say, they belonged to larger populations: “Eve” did not spend thousands of years waiting for “Adam”. Similarly the 70 supposed ancestors of all American natives must have belonged to larger populations: studies of very small populations (both animal and human) have shown that such populations are not sustainable – they have to find mates elsewhere or they will die out in a few generations – this can happen on small islands, or even in remote valleys from which people emigrate to larger centers, without being replaced. On the other hand, if after some centuries or millennia the multiple descendants of the 70 demonstrable ancestors (and of many of their fellow members of the same group) were joined at times by other very small groups (a few castaways, refugees or adventurers from Asia, for instance), those people’s DNA, at first mixed with the existing population, would by now have disappeared, even if they had a continuing cultural impact (for instance, improved boat technology, or – if the original group included both sexes – a language at first limited to the new small group) . So it seems to me that the lack of recognizable DNA from later Asian sources is not necessarily an argument against the possibility of occasional new influxes of population, providing those new populations were quite small. As an example, IIRC “Kennewick man” was at first determined to be Ainu-like, not related to the present population of the area where he was found.

  82. Trond Engen says:

    Stu: Replace “believe” with “accept the axiom”.

  83. Grumbly, my wife was there, and she says it moved just a bit; who am I supposed to believe, Isaac Newton or my wife?
    But seriously, Abbie Hoffman and the others who organized the stunt had no such beliefs. The point of the stunt was to request a permit to levitate the Pentagon 300 feet off the ground. They actually obtained such a permit, with the “300 feet” changed to “10 feet”, no doubt for reasons of safety to the occupants and bystanders. Good enough, said Abbie, and they went for it. The mere existence of such an activity, and with a permit for it yet, was enough to make the Masters of the Garden of War look just a little bit silly.

  84. Trond: Not “axiom” but “position”. Similarly, I believe, or accept the position, that the Earth is a rotating, slightly oblate spheroid (and not, say, an overgrown carrot with the continents on the butt-end), and if you accept that, the existence of a North Pole is quite certain, whether you yourself (or even anyone else) have seen it or not.
    Marie-Lucie: Exactly so. Genealogical arguments are not founded in genetics, but in history and applied mathematics. So the 70 DNA-provable ancestors of the Indians may reflect a much larger number of genealogical ancestors whose genes have not been sufficiently “reinforced” by descent on different lines, and mixed in with them a huge number of people without present descendants.

  85. marie-lucie says:

    I think that someone in the Pentagon had a sense of humour and enjoyed playing along: insisting on a reduction from 300 to 10 feet ‘for safety’ sounds like the reaction of someone pretending to take the request seriously.

  86. Trond Engen says:

    John C: I’m an engineer with no training in formal logic. Is there a technical definition of ‘position’ that I don’t know (and Wikipedia won’t disclose)? Anyway, I’ll admit that ‘axiom’ wasn’t necessarily correct. The fact that all humans have a common descent is probably itself deductable from axioms and observable facts. But not very wrong; I suppose that as long as at least one of its premises is axiomatic, it may be considered an axiom within some subset of, eh, the logical system.
    marie-lucie: Clearly.

  87. Thanks for your explanation, John Cowan. I’d never quite understood how that worked. I must be related to many of you, and may come to stay.

  88. The guest room is cluttered at the moment, but the futon in the living room is quite comfortable, if you don’t mind cats.

  89. marie-lucie says:

    We must all be related to each other, as the great-great- (repeat as needed) -grandchildren of “Eve” and “Adam”.

  90. marie-lucie says:

    JC: So the 70 DNA-provable ancestors of the Indians may reflect a much larger number of genealogical ancestors whose genes have not been sufficiently “reinforced” by descent on different lines, and mixed in with them a huge number of people without present descendants.
    The conclusion is also skewed by the small number of samples, and also by the historical fact that huge numbers of native people were wiped out by epidemics (whether or not introduced by the invaders) as well as by wars and various forms of hardship, whether caused by other humans or by famine, etc. (mass deaths are also well-attested from the Old World from similar causes – we are all descendants of the survivors).

  91. Marie-Lucie: Mel’cuk was always memorable, but often quite intimidating, and I actually dropped out of his class after I asked him a (simple and reasonable) question about semantic primitives and was given a twenty-minute answer which (inter alia) involved medieval Arab poetry, which he insisted on quoting in the original: to a shy and insecure young undergraduate it was all a little too intimidating.
    Anyway, back to this thread: my knowledge of genetics is minimal: however, I do know that there are a great many instances of a mismatch between the closest genetic relatives of a given population and its linguistic relatives. Andalusians, Romanians and Walloons, for example, are all Romance speakers, but are genetically much closer (respectively) to their North African, Bulgarian and Dutch neighbors than to one another.
    This leads to a question I’d like to ask this crowd: could the *linguistic* diversity of the Americas be due to an *earlier genetic* diversity which is no longer detectable today? I ask this because there does seem to be a complete mismatch between the (extreme) typological and genetic diversity of Western Hemisphere *languages* and the lack of detectable *biological* genetic diversity among their speakers.

  92. marie-lucie says:

    Etienne: the (extreme) typological and genetic diversity of Western Hemisphere *languages*
    I think that the “extreme” diversity of those languages is exaggerated: the language families recognized by “mainstream” Americanists are approximately the same as 150 years ago when the languages were first systematically recorded (up to a point) and classified, and most of them are on the order of Germanic, Slavic, Romance, Semitic, etc which are largely obvious groupings, quite recognizable by untrained persons. Few are on the order of Indo-European, and none on anything remoter such as the (controversial) Nostratic or (even more controversial) Euroasiatic (Greenberg’s). Even the Dene-Yeniseian connection (assuming it will be confirmed) seems to be on the order of Indo-European, or even closer. More inclusive groups had been proposed several decades ago, for instance by Sapir, Whorf and Swadesh (separately), but there are problems with those and most of them were never accepted. Greenberg’s “Amerind” was an attempt to group together all the languages remaining after setting aside the very distinctive Eskimoan and Na-Dene – what he did (by his own admission) was to accept just about every larger group that had been proposed, even those which had been demonstrably demolished, which include some groups which are as distinctive as the first two (but lesser known or limited to smaller populations). His “technique” of multilateral comparison was not at all sophisticated, and its application was riddled with errors.
    This does not mean that all those proposals were wrong in their entirety: I think that eventually a smallish number of superfamilies will be recognized, along with a few true isolates, but that will not be done while superficial comparisons of vocabulary remain the main technique for comparing and classifying language families.
    “Biological diversity”: I wonder why on the one hand, every person has a unique DNA “signature”, and on the other hand, groups which are recognizable by their appearance do not (as far as I can gather) have such a common signature. I am not talking just about skin colour, which might be the first thing that Americans think about: in Europe, there are often distinctive “looks” prevalent among rural populations in some regions, even in entire countries, and the same is true among Native Americans: Dene people are generally quite different in appearance from their neighbours on the North Pacific Coast, for instance.

  93. aqilluqqaaq says:

    Some rough and ready figures on the degree of Amerindian linguistic diversity as things stand: at a quick glance I see 421 languages in 59 groups for North America, plus another 556 languages in 101 groups for South America. By comparison, for Siberia (in the very broadest sense), 51 languages in 9 groups.

  94. marie-lucie says:

    aqilluqqaaq: 421 languages in 59 groups for North America
    The number of groups is probably greatly exaggerated, because (as I wrote earlier) those groups are mostly very obvious ones, comprising closely related languages, and the number is approximately the same as that determined 150 years or so ago, when the first comprehensive surveys of US languages were done. This means that there has been hardly any progress in 150 years! But once more links are found between those language families (and they will be found), the number of groups will be greatly reduced, and more in keeping with what is found on the other continents. (There won’t be a single pan-American group like “Amerind”, into which Greenberg dumped all the 57+101 groups remaining after excluding Na-Dene and Eskimoan).
    Johanna Nichols, who has done very original work on the distribution of languages and of particular language features throughout the world, estimates that if the current classification of Amerindian languages is correct, then it would have taken a good 50,000 years to get to that degree of diversity, a very improbable figure given that definite traces of human occupation in the Americas have not been found earlier than 12 or 14 thousand years (at any rate, after the melting of the huge glaciers).

  95. Marie-Lucie: as I recall Nichols, in her 1992 LANGUAGE article (LINGUISTIC DIVERSITY AND THE SETTLEMENT OF THE NEW WORLD is the title, if memory serves), claimed that genetic diversity was just half the problem: she argued that even if some version of “Amerind” was correct it would be impossible for a single ancestral language, or even several typologically similar proto-languages, to yield the *typological* diversity we find in the Americas if it broke up 12 000-15 000 years ago.
    Hence my question above: and what I like about the Solutrean hypothesis is that it might help solve the problem: if the ancestral languages of the Americas originally came from Siberia as well as Europe, perhaps this “mix” from two different language areas accounts for the typological diversity. And indeed the Yeniseic-Na-Dene connection might aid in its resolution as well: considering how typologically unlike most Northern Eurasian languages Ket is, perhaps Na-Dene contributed to making Western North America more typologically diverse than it otherwise would have been…
    And I think you do Greenberg an injustice: one thing about LANGUAGE IN THE AMERICAS I found interesting was his discussion of Oto-Manguean (a language family of Mexico and Central America), which Grenberg had originally thought might be a separate non-Amerind family: it was only after consulting Rensch’s reconstruction of Proto-Oto-Manguean that Greenberg found that it indeed fit into his “Proto-Amerind”. Meaning a language indeed had to fit certain criteria to be considered “Amerind”.
    Now, I was just a silly undergraduate pup at the time, but I remember telling one of the profs who was willing to put up with my questions and comments (most avoided me like the plague) that Greenberg was really lucky: the *one* language family belonging to Amerind where sound changes had eliminated any trace of their “Amerind” ancestry in the attested daughter languages had been reconstructed, unlike all those hundreds of other attested languages, most of whose proto-languages had *never* been reconstructed, but none of which was no longer visibly “Amerind”. Whew! Greenberg was one lucky linguist!
    (The professor in question told me that keeping my sarcasm to myself might be a sound strategy whenever I run into Americanists).
    Aqilluqqaaq: one thing about classifications of languages of the Western hemisphere which I’ve noticed is that among “language isolates” one finds extinct languages where we’ve so little data that for all we know they might belong to an established family: we’ve no way of knowing. Yet even restricting ourselves to well-known languages (living or extinct), there does seem to be a huge degree of *typological* and genetic diversity within the Americas.

  96. I love cats, but will they like my dogs?

  97. marie-lucie says:

    Etienne: concerning “reconstructed” proto-languages:
    In well-known families such as Indo-European or Finno-Ugric, or in the component families such as Romance, most of the languages are still alive or at least abundantly documented, sometimes over millennia, and dozens of linguists concentrating on language history have worked on tracing the history and reconstructing proto-languages (at various levels) over many decades, so even though there are still a few loose ends, there is a general consensus overall, and it is possible to be quite confident of the essential correctness of the correspondences and the reconstructions.
    In the Amerindian field, the documentation is far from complete, many of the languages are already extinct, and the ones still alive are often no longer the main means of everyday communication (so that many words and irregular forms have been lost), and many of the existing reconstructions, if any, have been done by a single linguist, who may or may not have done so competently (few of the current Amerindianists have had historical training). For instance, in the Sahaptian family (in Washington, Oregon and Idaho) there are two languages which have close correspondences of the type: A: k, q vs B: x, X (k and x are palatal, q and X uvular). I am sure you would assume an unremarkable rule changing original stops (as in A) to fricatives (as in B), but the published reconstructions are *k, *X! That way the linguist is sure of being right in at least one of the cases! It should be pointed out, though, that at one time the bias against historical explanations in American linguistics was so strong that some people made up rules that they knew ran counter to everything that they had learned – I don’t know whether this example is an instance of this intellectual perversion or of the incompetence (or rather the lack of historical training) of the linguist.
    At any rate, I do not trust statements which give a proto-form due to a single linguist, especially if this proto-form is given without also listing the attested forms it is based on. Similarly, apparently confident statements such as “Proto-XYZ had the following consonant inventory” seem to me to be premature in most cases. So I would certainly not rely implicitly on Proto-Oto-Manguean having been correctly reconstructed, let alone on Greenberg having had the background necessary to evaluate the correctness of the reconstruction. As for “Proto-Amerind” (for which Greenberg did not think that the comparative method would be necessary), it would have to be based on a correct identification and subclassification of the languages included in “Amerind” (the criterion for Amerind being basically “unlike Eskimoan or Na-Dene”). And how likely is it that a single linguist (Greenberg), even if supremely gifted and thoroughly trained, would be able to reconstruct the ancestor of hundreds of languages of considerable diversity? (Malayo-Polynesian consists of hundreds of languages, but they are very close to each other, since the Polynesian expansion is quite recent).

  98. Marie-Lucie: I quite agree that more needs to be done in reconstructing proto-languages, in the Americas and elsewhere. But enough descriptive work has been done on enough languages in the Americas to indicate that there is indeed an amazing degree of diversity, typological if not genetic, on the continent. And the contrast between this diversity and the genetic uniformity of Native Americans is what prompted my question above.
    By the way, on a related topic: it was Mary Haas, I believe, who had once suggested that Americanists might profitably try comparing different reconstructed languages to one another: am I right in believing that no Americanist followed up on this suggestion?

  99. marie-lucie says:

    1. This was your question:
    could the *linguistic* diversity of the Americas be due to an *earlier genetic* diversity which is no longer detectable today?
    And this is what I wrote earlier about the possibility of occasional influxes of small populations (some of which, as you say, could have come from Europe as well as from Asia):
    if after some centuries or millennia the multiple descendants of the 70 demonstrable ancestors … were joined at times by other very small groups …, those people’s DNA, at first mixed with the existing population, would by now have disappeared, even if they had a continuing cultural impact (for instance, … a language at first limited to the new small group)
    2. it was Mary Haas, I believe, who had once suggested that Americanists might profitably try comparing different reconstructed languages to one another: am I right in believing that no Americanist followed up on this suggestion?
    In my particular area of research, most linguists (many of them trained by Mary Haas) take it as axiomatic that you should only compare reconstructed proto-languages with other reconstructed proto-languages, not with directly attested languages, and some of them have done comparisons (and further reconstructions) on this basis. To which I reply (for the reasons given above) that one cannot trust those reconstructions implicitly without knowing what forms they are based on (of course, some of those reconstructions are indeed well-done, but by no means all). And it is true that there has been a lot of descriptive work done, but in most cases not nearly enough (often for reasons beyond the linguists’ control, such as the death of consultants).
    For comparing language families, there is also the fact that most existing reconstructions are very shallow, minimally different from the modern forms (because they are based on very similar languages, sometimes barely distinguishable from different dialects). This is OK for the close study of intra-family relationships (identifying intra-family or foreign borrowings, for instance, or the relative timing of various changes), but not always so for inter-family comparison. Imagine for instance that Indo-European existed only as a daring hypothesis, but that some reconstructions had been done, on the level of Proto-Ibero-Romance, or even (but more controversially) Proto-Romance. How much more useful would PIR be than just Spanish in trying to explore the possibility that the Romance languages might be related to the Slavic ones?

  100. I am denesuline and very proud to be part of the dene na dene group. It is easy to surmise that we came across the Barren strait some years ago. But I beg to differ, I believe that we went there. Why do we have such a small group there now as the kets? Should we not have more and if we did come from there, how could it be that the whole group would move enmass to NA. Further, it is now known that the Navajo are a spring off group from my group and that they migrated from he north almost one thousand years ago.
    Just more questions and we may never know the answers.
    Marsi tchogh
    In denesuline boat is ts’i or ts’i chene which is the same name we use for porcupine.

  101. marie-lucie says:

    AA, hello: Of course a movement of people can occur in two directions. But it seems that the Ket people have not always been so few: it is known that languages related to Ket were once spoken over a much, much wider area of Siberia. It does not mean that the original inhabitants would all have left: more likely, they were joined (more or less willingly) by other people whose languages they eventually adopted. Perhaps some of them moved to escape the newcomers. And even a few thousand people migrating into a new place suitable for their way of life can become millions in the course of many centuries.
    “We will never know”: there are many things we will probably never know, but that should not keep us from trying to discover what we can know if we put our minds to it.

  102. Allan Adam: I’m glad a speaker of a Na-Dene language is actually following our discussions. I’d just like to add a little to Marie-Lucie’s answer (Marie-Lucie, if you think I’m misrepresenting your position, just SHOUT).
    You’re right that Navajo (and Apache and others) has been shown to represent a later migration from the North. How do we know this? Because in Northern Athabaskan languages there is more diversity than in Athabaskan languages in the American Southwest. That is to say, Northern Athabaskan languages differ from one another more than (say) such non-Northern Athabskan as Navajo and Apache languages do.
    Because languages, over time, become ever-more different, it follows that the less diversity there is, the more recently the language must have spread.
    This applies beautifully, by the way, to English in Canada or the U.S.: Atlantic Canada and the American East coast have a number of very distinctive dialects, whereas West of the Great Lakes there is little difference between the English spoken in (say) Vancouver and in Denver. Even if we knew nothing of the settlement history of North America, the linguistic data alone would indicate that English in North America had spread from East to West, and not from West to East.
    Now, by the same logic, look at Na-Dene. It consists of Tlingit, Eyak, and all Athabaskan languages. Tlingit and Eyak are spoken in the Alaska panhandle, with some Northern Athabaskan neighbors. This seems to indicate that Northern Athabaskan must have spread from West (in or near the Alaska panhandle) to East. So Dene Suline is almost certainly a language which has spread to its present-day territory from the West, after Proto-Na-Dene in the Americas had become different languages.
    So: the territory where Dene Suline is spoken today is out as the original homeland of Proto-Athabaskan or Na-Dene. By virtue of the discovery that Na-Dene and Ket have a common ancestor, a homeland further West does seem to be called for. Which does seem to point to “somewhere in Siberia” (which doesn’t narrow it much, I agree: in fact I had written above that Alaska could have been the original homeland).
    Historical linguistics can’t say too much about WHERE a given language was once spoken: it can say a lot about how old various elements IN A LANGUAGE are. Thus, TS’I in Dene Suline does indeed look like it is related to the Ket word for “canoe”. This would make it part of what we call the “inherited element” of Dene Suline: that is to say, words that have been in a language for as far back as we can reconstruct. Conversely, a word such as MARSI “thank you” is a borrowed word (from French): there was a point in the history of Dene Suline when this new word entered the language (possibly via Cree, rather than from French directly). Future research will allow us to better tell what are the inherited and what are the borrowed elements, in Dene Suline as well as in all Na-Dene-Ket languages.
    Again, comparison with English is useful here. Suppose, again, that we had no written record for English or any European language. If, however, someone showed that English is related to a number of Northern European languages (German, Dutch, Danish…) it would be possible, through comparison, to distinguish the inherited from the borrowed element in English. Thus, English words such as SING or STONE are certainly inherited in English, because cognates are found in other Germanic languages. English words such as FAITH or ROCK are borrowed (from French in this case): the absence of cognates in other Germanic languages shows that the words are not inherited.
    Now that we know Dene Suline and its Na-Dene relatives have a relative (Ket) in the Old World, we should likewise be better able to separate the inherited from the borrowed element.
    I apologize for the length of this comment, but I hope I have made things a little clearer for Allan Adam, and perhaps for other readers (if they exist…) as well.

  103. marie-lucie says:

    Very clear, Etienne.
    I am sure that there are other interested readers besides Adam Allan and me.

  104. there are other interested readers there are.
    Giving it a slightly different turn, I remember Marx/Engels cited Navajo/Hopi as an example of prehistoric socialism/communalism. What is the current thinking on this?

  105. marie-lucie says:

    How much did Marx/Engels know about Navajo or Hopi culture, as opposed to about the English slums of their time?

  106. I wouldn’t pay much attention to anything Marx/Engels had to say about Native American culture.

  107. marie-lucie says:

    The Navaho and the Hopi live in the same general area, but their cultures are very different.

  108. How much did Marx/Engels know about Navajo or Hopi culture, as opposed to about the English slums of their time?
    Quite a lot, was Engels’ own opinion. In Der Ursprung der Familie, des Privateigentums und des Staats [The Origin of the Family, Private Property and the State], there are more page references to North American Indians in the “Glossary of names of peoples, tribes and gentes” than to any other group – the Greeks come in second place. Engels claims to be basing his thought processes on the works of anthropologists and early ethnologists such as Bachofen, MacLennan and above all the American
    Lewis Henry Morgan. The subtitle of the book is “Following the research of Lewis H. Morgan”.
    This tiresome book is a classic example of the superior-tone-mixed-with-nasty-asides-about-other-writers style that dominated the writing of German Marxist authors. Also, you see there the jargon-level already risen high enough to lap round Engels’ ankles.

  109. You can get Marx and Engels in the cheap blue and red-bound complete editions formerly published in East Germany. My copy of the Ursprung is from the Dietz Verlag Berlin, 1974. I picked it up recently from those theater steps nearby, thinking it would be nice to have something to get annoyed about for a second time. A complete edition of Marx/Engels containing possibly as many as 10,000 pages might set you back a dollar. The ratio of paper to price is hard to beat.

  110. The ratio of paper to price is hard to beat.
    Has this topic never been discussed at Language Hat? What a lapse. Anyway, I propose the yellow pages.

  111. I may have brought that up once last year. It happens occasionally in a bookstore that I find several books I want for their content, but don’t have enough money with me to buy all of them at once. In that case, I buy those with the higher page-per-Euro yields.

  112. The bible is very good value.

  113. How much did Marx/Engels know about Navajo or Hopi culture,
    thanks for the Morgan connection – I didn’t know about that.
    This tiresome book

  114. English words such as FAITH or ROCK are borrowed (from French in this case)

    While rock is fairly unproblematic, faith is quite unexpected. It is from Latin fides with the usual pathway through French fei (modernly foi), which would be expected to give fay in English — and so it did, but fay died out in the 16C. So what gives with the obviously English -th at the end? The OED has two theories, both of them problematic:

    1) It is a rare survival of the Anglo-Norman form foit from before the final /t/ was lost in the 11C. If so, then th was originally written as a mere flourish, but eventually acquired a spelling pronunciation (as in author < OF au(c)tour). But if so, it is the only monosyllabic A-F word in /-t/ to get this treatment.

    2) It is the native denominal suffix -th as in truth and its twin troth, health, strength, width, length, depth, sloth, filth, breadth (from slow, foul, broad), etc. Height, sleight (from sly) also belong here. But this suffix is normally applied only to adjectives.

  115. Huh. If I’ve ever thought about that problem, I’ve forgotten the thought. But I don’t see adding the native suffix as all that problematic. What about tilth?

  116. Weekley, FWIW, mixes both theories: he derives faith from OF feid “(pronounced feith)”, and says, “The survival of the –th sound, due to association with truth, sooth, etc., is unique”.

  117. Tilth is from the verb till, which makes it different, but it’s still not a noun from a basically synonymous noun, as faith would be on explanation 2.

  118. Right, but it’s not from an adjective, which dilutes the “normally applied only to adjectives” problem. The more versatile it is, the better it looks.

  119. Trond Engen says:

    Thanks for digging up this thread. I didn’t find it when I needed it a month ago.

  120. A mon-th, you say?

  121. While moon and month are plainly closely related, this is probably a PIE root extension rather than the Germanic suffix -th in the other words; compare (with different root extension) mensis and with no extension Greek meno-.

  122. Greek me:n- is actually from *me:ns- with the same -s- as Latin me:nsis, but this was lost in or before Proto-Greek by regular sound change (along with most other s‘s).

  123. (though in some dialects the s survives as gemination: Aeolian has a genitive μηννός.)

  124. (or rather μῆννος, with Aeolic accent retraction. O for an edit button…)

  125. The PIE stem was *meh₁-nōs/*meh₁-n̥s-. In Germanic we have a common shift from a minor consonantal declension to a nasal stem in the ‘moon’ word, and an unexpected stem-final *-t- instead of *-s- in the ‘month’ word. This may be a relic of a PIE *-t/s- alternation (possibly also in perfect participles, cf. Goth. weitwoþs), though it’s rather odd that only Germanic should have preserved the *-t- variant. If this explanation is accepted, the paradigm should be reconstructed as *meh₁-not-/*meh₁-n̥s-

  126. Eli Nelson says:

    “Spilth” is attested from 1608, according toy phone copy of Merriam-Webster. Another -th word not based on an adjective is growth. Apparently, it is first attested from around 1550, but there is supposed to be a parallelly formed word in Old Norse. The word birth also is not based on an adjective; my dictionary says it definitely comes from an Old Norse form. An older word is weight, which originally had the same suffix. All of these are from verbs, though: I can’t find any that are derived from another noun.

  127. Eli Nelson says:

    Oh, what about “wealth”?

  128. It’s unclear whether the underlying root is well or weal. If the latter, this would be noun + -th indeed. Both OED and Etymonline say it’s analogical with the rhyming health, and OED points to parallels (not cognates) in Dutch weelde and Low German welde.

    I finally read the OED article on -th, and apparently the verb and adjective suffixes were originally distinct, the latter being marked by umlaut where possible. Here’s the text:

    1. from verbs; in some words, as bath, birth, death, math, oath, Germanic, repr. various Indo-European suffixes, as -tos, -tâ, -tis, -tus, in which the t following the stressed syllable regularly became þ in Teutonic; in others, as growth, tilth, going back to ON. or Old English; in others, as blowth, spilth, stealth, of later analogical formation. In many words Indo-European t remained in consequence of its position, or þ was subsequently changed to t.

    2. from adjs. (rarely ns.), representing Indo-European -itâ, Germanic –iþô, Gothic -iþa, Old English -þu, -þo, -þ, with preceding i- umlaut, forming abstract nouns of state: as filth (Old English fýlþ, Old Saxon fūlitha from fúl foul), health, length, mirth, strength, truth; in Middle English and also in cognate langs., dearth, depth; of later analogical formation, breadth, sloth (cf. Old English slǽwþ), wealth. In some words of this group, þ has, by phonetic causes, become t, e.g. Old English híehþu, Middle English heiȝþe, now height, ON. slægð, Middle English sleiȝþe, now sleight.

    So while it says “rarely nouns”, no actual examples are given, not even faith.

    This suffix retains a small amount of productivity: there are 11 kghits for dumbth ‘stupidity’, but no dictionary definitions, not even in Urban Dictionary.

  129. See this small discussion of -th (adj. -> n.) and some other th‘s in the comments:

  130. marie-lucie says:

    English math? of course this is not the short form for “mathematics”, but what does it mean?

  131. Mowing; familiar now only as part of aftermath. OED:

    Cognate with Old Frisian mēth (neuter), Old Saxon mād– (in the compound māddag mowing day; Middle Low German māt (feminine)), Old High German –mād (in the compound āmād (neuter) aftermath; Middle High German māt (neuter/feminine), German Mahd (feminine); compare also the weak feminine noun from the same base represented by Old High German māda (Middle High German māde) swathe), ultimately < the Germanic base of mow v.1 The quantity of the stem vowel in Old English has conventionally been taken to be long on the evidence of the cognates cited above; however, the subsequent history of the word in English suggests that Old English mǣþ may have had a doublet form mæþ with short vowel: compare Dutch mad, mat (neuter) swathe, and also the possible parallel of two Germanic ablaut grades (one with long stem vowel, the other with short) clearly shown by the cognates of the Verner’s law variant mead n.2 Almost all of the phonological evidence for the word after the Old English period points to the reflex of Old English mæþ (or to an otherwise unparalleled shortening); evidence for the reflex of Old English mǣþ occurs in a few place names, as Meda (1086), Meðe (1175; now Meeth, Devon), la Methe (1249; now Meethe, Devon), and in the 18th-cent. form meath reported in the works of William Ellis (see Soc. Pure Eng. Tract (1945) lxiv. 103).

  132. What is this “Indo-European -itâ” of which the OED speaks? Is it supposed to be related to the (similarly adj.-to-n.) suffix -ta:t- of Latin and Greek?

  133. The exact PIE form of those suffixes (or rather suffix complexes) is sometimes hard to recover. The main source of the Gemanic abstract-noun suffix *-iþō seems to be PIE *-é-tah₂, as in Vedic nagnátā ‘nakedness’. The Latin/Greek variant seems to be “more derived” (*-o-tah₂-t-), as does Slavic *-ota (basically the same, except that there’s no final stop). Cf. Goth. niujiþa < *newié-tah₂, Lat. novitāt-, Gk. νεότητ-, Old Czech, SCr., etc. novota < *newo-tah₂(-t). They are similar and most probably related, but not the same in every detail.

  134. Trond Engen says:

    IInteresting. Is this related to the Proto-Germanic adjective suffix *-ōdaz > ON -aðr? If so, why is *-ōdaz masculine while the bare dental suffix is feminine?

  135. marie-lucie says:

    Thanks LH. Aftermath did not occur to me, and I had had never thought of asking myself what the math in it was.

    The persistence of aftermath long after the disappearance of math alone illustrates two common events in the evolution of a language: the loss of a term associated with a technique which is no longer in general use in a population, and the persistence of a compound which has acquired its own meaning, no longer connected with that of the original word. As to the first factor, Standard English is supposed to have evolved in urban environments, but most of the cognates of math in other languages seem to have been lost too. Perhaps the old word lost out everywhere to other words with similar meanings, partly because of its eventual homophony with other words with unrelated meanings.

    The alternate form mead (itself homophonous with the name of the honey beverage) reminds me of Saint Mary Mead , the English village known the world over to fans of Agatha Christie through its most famous inhabitant, Miss Marple ! Meadow must be an unambiguous derivative, still alive and well.

  136. marie-lucie says:

    Trond: the Proto-Germanic adjective suffix *-ōdaz > ON -aðr? If so, why is *-ōdaz masculine while the bare dental suffix is feminine?

    I don’t know PG or PN, but this type of thing is not rare. It looks like *-ōdaz is made up of two suffixes, and in such sequences the last suffix determines the gender of the word.

    But an adjective by definition agrees in gender with its noun, so why is the PG suffix masculine only? (unless former adjectives with this form of a suffix have become masculine nouns).

  137. Trond Engen says:

    That last sentence didn’t make much sense. Adjectives take any gender, What I mean is that the adjective suffix is used to form ON deverbal abstracts in -naðr, i.e. the adjective suffix is added to the inchoative/perfective -n-. This is easier to understand if it’s the same suffix as the one used to form abstracts from adjectives, but that won’t account for the masculine gender..

    (I have this idea that the ON inchoative suffix -n- is essentially the same morpheme as the denominal adjective and perfect participle ending -inn and the first part of the deverbal abstract suffix -naðr: “become X-ed”, “X-ed” and “X-edness”.)

  138. Mead and meadow actually reflect the same OE noun, like shade and shadow. The PGmc. stem was *medwō < *h₂meh₁-twáh₂, which regularly gave OE mǣd, mǣdwe (as in on mǣdwe ‘in the meadow’). Each form gave rise to a levelled-out paradigm. Mead dropped out of general use (surviving in placenames), outcompeted by the “locative” variant meadow.

    Mead (for the drink) is one of the most venerable culture words in Indo-European. It goes back to *medʰu- (OE me(o)du). The two mead words were not homophonous in OE. They fell together phonetically when the short vowel of Middle English mede became lengthened in an open syllable.

    Old English mǣþ ‘mowing’ survived only in compounds and placenames, which explains the shorthening of the root vowel. Otherwise it would have become Modern English “meath“.

  139. David Marjanović says:

    This may be a relic of a PIE *-t/s- alternation (possibly also in perfect participles, cf. Goth. weitwoþs)

    That sounds intriguing; could you explain more?

  140. The ON denominative adjective suffix -aðr (as in hjalmaðr ‘helmeted’) is the same as OE -od (Mod.E. -ed), Lat. -ātus (barbātus ‘bearded’), Slavic *-atъ, Lith. -otas, etc. The adjective-forming suffix *-tó- was added to a noun whose thematic *-o- (if it was there) was replaced with *-ah₂- (presumably with a collective meaning), forming the complex suffix *-ah₂tó-. The very same *-tó- formed verbal adjectives (in Germanic, past participles of weak verbs)

  141. That sounds intriguing; could you explain more?

    The perfect participle also shows the variants *-wot- ~ *-wos- in the “strong” cases, but *-us- in the weak cases, in the neuter ending *-wos and before the femininising suffix, *w(e)s-ih₂. This has led some (notably Jens E. Rasmussen) to posit a (pre-)PIE dental affricate in these suffixes (yielding *s and *t as positionally conditioned reflexes. Rasmussen actually reconstructed it in all es-stems and in several other important morphemes. According to him, adjectives in *etó- were originally thematicisations of es-stems.

  142. The Latin/Greek variant seems to be “more derived” (*-o-tah₂-t-)

    Isn’t it just *-teh₂-t-? There are athematic forms in both languages, e.g. βραδυτής, maiestās.

  143. Yes, that’s why I separated the thematic vowel from the suffix proper with a hyphen. Germanic also occasionally had *-ðō < *-táh₂ rather than *-iþō < *-é-tah₂ with athematic stems, cf. Goth. junda ‘youth’.

  144. marie-lucie says:

    mead/meadow, shade/shadow

    Thanks PG. I vaguely remember reading that the two members of the shade/shadow pair were originally part of the same paradigm, rather than the one in -ow being derived.

    When first reading your comment I thought that the “locative” morpheme was -we, but I see now that the -w- was part of the stem (lost in absolute final position), and the “locative” element is indeed -e.

  145. To be precise, in both words the original endings of the (the *-wa- in ‘shadow’ and *-wō in ‘meadow’) developed into pre-OE *-u. Then final high vowels were dropped after a heavy syllable but retained after a light one. The result was OE sċeadu (with a short root vowel) and mǣd (with a long vowel). Both had oblique case endings with /w/, e.g. sċeadwe and mǣdwe. Hence the Middle English pairs shade : shadwe and mę̄d : mę̄dwe, which split into pairs of different words each. The difference between them is that shade and shadow managed to specialise semantically, achieving a “division of labour” allowing them both to survive, while mead and meadow remained in competition, so one of them had to go.

  146. PS *-w- was originally the last segment preceding the thematic vowel in both words, so diachronically not part of the case ending. But in the further development of the Auslaut the difference before stem-final elements and inflectional endings proper was blurred. For the same reason we say that the usual ending of the second declension in Latin is -us, though in fact /u/ is a reflex of the thematic vowel (the stem termination), and the ending in PIE was just *-s.

  147. marie-lucie says:

    PG: If Latin 2nd declension -us was *-u-s, with *u part of the stem, what was the original difference with the 4th declension)?

  148. The vowel. *-o-s (m./f.) or *-o-m (n.) in the second, *-u-s (m./f.) or *-u-∅ (n.) in the fourth.

  149. marie-lucie says:

    Thanks PG. I was not aware that the 2nd declension included feminine words. Baldi (Foundations of Latin says that nouns of this class are “typically masculine or neuter” but does not give exceptions, while 1st declension nouns are “overwhelmingly feminine” except for a number of stated examples.

  150. Latin feminine nouns in the second declension. There are 103 of them listed, but most are Greek borrowings, false hits, or names of trees.

  151. Names of trees and other plants are very commonly feminine in Latin (and Greek), e.g. fagus ‘beech’. Names of cities and islands also tend to get treated as feminines, presumably by notional agreement with urbs / insula. And Greek loanwords bring their gender with them, which is why words like methodus are feminine. The most common feminine noun in –us, domus ‘house’, isn’t on that Wiki list, presumably because its forms are a mix of the second and fourth declensions, favoring the latter.

  152. There are also a couple of interesting feminine o-stems denoting humans, with exact cognates elsewhere: nurus ‘daughter-in-law’, anus ‘old lady’. They escaped the non-Anatolian h₂-femininisation of such nouns.

  153. marie-lucie says:

    Thank you all!

  154. There are also a couple of interesting feminine o-stems denoting humans…

    Oops, wrong! I should have said they are original o-stems, *snusó- and *h₂an(H)o- (cf. Gk. νυός, Arm. now ‘daughter-in-law’; Hitt. hannas ‘grandmother’). In Latin, they have been both shifted to the u-declension (IV) on account of their unusual stem ending.

  155. David Marjanović says:

    Schnur “daughter-in-law” survived at least into the 20th century in some central German dialects. It’s a homonym (down to the feminine gender!) of… the size class between a thread and a cord.

  156. Shnur ‘daughter-in-law’ survives in Yiddish too, I believe.

  157. marie-lucie says:

    Latin nurus ‘daughter-in-law’ : In France the Norman dialect (or at least a local variety) has la nore for ‘daughter-in-law’.

  158. Very nice!

  159. I suppose there’s no plausible way of deriving *snusó- “daughter-in-law” from *suH-no- “son”?

  160. (Serbian) Church Slavonic and Old Russian snъxa, Russian сноха, S-Cr. snaha, Bulgarian снаха, Old Polish sneszka (diminutive) of course reflect *snusó- moved to the ah₂-declension.

    TR: There’s no way to derive it regularly or with just a li’l bit of special pleading. Of course words for ‘daughter-in law’ can be derived from ‘son’. For example, Polish has synowa, a substantivised sexist possessive adjective (‘belonging to the son’). But *snusó- is really too strange. Comparison with *sneubʰ- ‘get married’ (of a woman), as in Lat. nūbō, would be a priori more promising (at least the crucial segments are in the correct order), but even here we have just a similarity, not a valid equation.

Speak Your Mind