Linguistic Family Tree.

We’re all used to the idea of the tree as a model of development through time, whether of species or languages, but rarely is it portrayed so strikingly as in Minna Sundberg’s gorgeous rendering (from the site for her webcomic Stand Still. Stay Silent). The only quibble I might have is that it appears (from the connection of the root systems) to support the Indo-Uralic hypothesis, but what the heck, so did my dissertation director Warren Cowgill, so who am I to quibble? A tip of the Languagehat hat to Arika Okrent at Mental Floss.

Comments

  1. The most beautiful language tree I have ever seen, and useful too.

    Quibbles: No Irish; Corsican joined with Sardinian, and Faroese joined with Icelandic, I suspect for geographic reasons, not linguistic ones; Ossetian joined with Pashto. And why does such a wonderful artist not hand-letter the language names, or at least use a more interesting typeface?

  2. Pretty interesting. I didn’t realise that Venetian, Piedmontese, Emiliano, Ligurian, and Lombard belonged on the same branch as French, and on a different branch from Italian, Nap. Cal, and Sicilian.

  3. It’s a lovely map, but missing the other parts of the globe. I suppose she’ll have to create a small coppice of trees for that.

  4. George Gibbard says:

    Pashto and Ossetian do belong together as the two largest Eastern Iranian languages:
    https://en.wikipedia.org/wiki/Eastern_Iranian_languages. These share the innovation of leniting initial voiced stops, as in Pashto and Yaghnobi ɣar ‘mountain’ (I don’t know the Ossetian) = Avestan gairi-, Sanskrit giri-, Proto-Indo-Iranian *grri-.

  5. @Iulia: She’s only showing Indo-European and Uralic because the purpose of the tree is to give the backgound on the languages spoken in Skandinavia. I actually find it quite nice that she goes to all those lengths to show that the characters in the comic speak different languages and to which degree they can understand each other – that’s something one doesn’t see often in comics.

  6. George Gibbard says:

    Not that I know speaker counts for the various languages, I was blindly trusting the artist who seems to have very carefully scaled the size of each leafy clump to match what little I know about speaker numbers.

  7. @Bathrobe: Yeah, it’s customary to place the boundary between Gallo-Romance and Italo-Romance (and also between Western and Eastern Romance) at the La Spezia-Rimini Line, which coincides with a number of isoglosses within Italy (like the use of s-plurals versus vocalic plurals, and the voicing of Latin intervocalic voiceless stops). But from what I can tell, there is some opposition to the idea that the Romance languages can be neatly grouped into branches as this sort of analysis would suggest.

  8. Sorry, forget what I said about s-plurals. Except for Friulian and Ladin, the northern Italian varieties in question use vocalic plurals.

  9. AJP Crown says:

    I’m not sure she’s got the squirrels quite right, unless they’re meant to be cats.

  10. Stefan Holm says:

    Why, Y, do you call Faroese joined with Icelandic a quibble? Both certainly have no closer relatives but each other. They don’t easily communicate orally but fairly well in written form.

    If a tree were to be historically correct the Scandinavian languages should be grouped into three:
    (1) West (Faroese, Icelandic, Norwegian, Norn).
    (2) East (Danish, Swedish).
    (3) Gutnic.

    This was valid in medieval times even if the differences were only minor, dialectal ones, traceable already in runic Norse (earlier monophtongization in East Scandinavian, ‘o’ in East corresponding to ‘u’ in West etc. – in a continuum though not following modern national borders).

    The modern tree would be different because of political changes through the centuries. One could mention the strong medieval Low German influence in continental Scandinavia, the Danish sovereignity over Norway (1397-1814) and the Swedish occupation of Gotland (1645), of Scania, Blekinge, Halland, Bohuslän and Jämtland (1658). Today the grouping would be (just as Minna Sundberg shows it):
    (1) ‘Atlantic’ (Faroese, Icelandic).
    (2) ‘Continental’ (Danish, Norwegian, Swedish).

    It’s true that (spoken) Danish is rapidly drifting apart. But we do understand each other in continental Scandinavia, if both parties just want to and don’t aforehand decide otherwise – there is a psychological factor involved.

  11. Athel Cornish-Bowden says:

    I particularly like the fractal edges at the boundaries, which, to me at least, symbolize the difficulties of deciding where a language ends and a dialect begins.

  12. David Marjanović says:

    I think it’s… fairly obvious that IE and Uralic are closely related. From p. 12 of this pdf (I recommend reading the context, I didn’t want to paste at least two pages here):

    Since both of the Uralic forms clearly represent Proto-Uralic, and, likewise, both Indo-European forms represent Proto-Indo-European (even Proto-Indo-Hittite, if one agrees with the special status of Anatolian languages within Indo-European), it is clear that these “borrowings” have to be attributed to the oldest layer of borrowings from one family into another (Rédei presumes the direction to have been from Indo-European to Uralic and not vice versa, but this is not really relevant for my purpose here). How many are there? Rédei acknowledges seven [Rédei 1988: 651–654]. Seven easily “identified” old borrowings from one proto-language into another that include words for ‘name’ and ‘water’.

    Certainly, cases where the word for ‘name’ has been borrowed are known; likewise, for the word ‘water’; likewise, cases where two or more random items on the 100-wordlist have been borrowed from a single source. But in most, probably even all, such cases borrowing of such basic items has only become possible due to a concentrated “bombardment” of the recipient language by lexical items from the donor language — “bombardment” which, obviously, begins with a large number of technical and cultural terms. A situation under which two ancient languages “meet”, exchange terms for ‘name’, ‘water’, ‘give’, and ‘sinew’, and then part company borders on the ridiculous, and at least requires extra proof.

    Brackets in the original.

  13. David Marjanović says:

    Whether Indo-European and Uralic are each other’s closest relatives is, of course, a different question.

  14. Stefan Holm says:

    The only thing confusing me with the tree is the text: Sizes of the branches represent the recorded native speakers before year 0. This can’t refer to the age of the speakers since nobody speaks before birth. It can’t even refer to ‘year 0’ since that one never existed (1 BC was followed by 1 AD). If it’s a typo and she meant a specific modern year, what then does ‘before’ mean.

  15. “Gallo-Iberian” is a misprint for “Gallo-Romance”?

  16. I think it’s… fairly obvious that IE and Uralic are closely related.

    OK, how can I stand against you and Cowgill? I will give up my hesitation as the mulish conservatism of an old fossil.

  17. Stefan Holm says:

    From what I know of their works the Muscovite nostraticists are professionals. Far from exercising mass comparison the easy way, they examine each item in every detail (phonetics, semantics, morphology) being faithful to the comparative method. Superficial similarities between languages can be seen by laymen like me. But since all languages change by time it would after millennia be odd to judge genetic relationship by just counting the number of look-a-likes.

    It’s rather searching for a system (hidden for the naked eye) in the irregularities, through hard work, digging into every aspect, that counts. I agree with David that Starostin Jr. argues well about the improbability of basic words like ‘water’ and’name’ being borrowed into a language without a whole bunch of less common ones as well. It proves no common IE-Uralic origin but it opens for the possibility and further research.

  18. Stefan Holm: “The only thing confusing me with the tree is the text: Sizes of the branches represent the recorded native speakers before year 0. […] what then does ‘before’ mean.”

    This refers to the events laid out in the first chapter of the story; ‘year 0’ refers to the year when the plague/sickness killed the majority of the world’s population. Throughout the subsequent chapters, years are not given in ‘AD/CD’, but in ‘years since everything fell apart.’

    Likewise, I’d assumed that the layout of the chart was from the point of view of the survivors/current ‘known world’ population, encompassing what they know/believe (the ‘Uralic is a separate base’ thing), and not dealing with what is too far away to concern them (languages/families not represented here).

    SSSS a good story with great art, well worth adding to one’s rss aggregator.

  19. Kind of reminded me, design- and content-wise, of this:

    Dude, that’s the previous page on the same site!

  20. ” I didn’t realise that Venetian, Piedmontese, Emiliano, Ligurian, and Lombard belonged on the same branch as French, and on a different branch from Italian, Nap. Cal, and Sicilian.”

    The Liga Lombarda would set you straight on that in a heartbeat.

    It’s interesting that those varieties are all spoken in Cisalpine Gaul as opposed to [Classical] Italy proper.

    Quibble – she puts Celtic in a very strange place, very far form italic.
    Kudos – She puts Sanskrit as a little side twig, not the main branch all the Indic languages descend from.

  21. I stand corrected on both Iranian and Scandinavian.

  22. I don’t see how this diagram claims IE and Uralic are genetically connected. She’s very precisely ambiguous about it. Are these two trees growing next to each other in the same soil? (probably.) Are they descended from a larger trunk which we cannot see, or share a root system? (maybe.)

  23. Jim, while Sanskrit is close to Proto-Indo-Aryan, it was not the parent language of all languages in the Indian Subcontinent.

  24. Stefan, for a devastating demonstration that the Moscow Nostraticists are using a faulty methodology, and a strong hint that some downright dishonest scholarship has been involved, see Lyle Campbell’s contribution to the book Nostratic: Sifting the Evidence ed. Josephs & Salmons (Amsterdam: John Benjamins, 1998). Alexander Vovin’s later critique of Dybo is also worth reading (and Dybo’s counter-rebuttal is laughable).

  25. Jim, while Sanskrit is close to Proto-Indo-Aryan, it was not the parent language of all languages in the Indian Subcontinent.

    Hence the “Kudos.”

  26. Stefan Holm says:

    Sorry Sanna, for my ignorance. Year 0 for a second gave me some unpleasant associations to the Khmer Rouge. I know next to nothing about cartoons but at least a little about linguistics. From the viewpoint of the latter Minna has produced an excellent piece of art with a factual accuracy questioned only by useless fault-finders like myself (and a few others on this blog).

    Så bara kör på tjejer (gissar jag rätt utifrån era namn så begriper ni svenska både Minna och Sanna)!

  27. Stefan Holm says:

    Christopher, I haven’t read neither Campbell nor Vovin and my views are miles from expertise. It’s the Moscow school’s own description of proper methodology I appreciate (like here in a commemorative of Sergey Starostin). http://altaica.ru/personalia/e_starostin.htm

    By these measures I’m a little hesitant when it comes to Greenberg and Ruhlen.If we ever are going to trace origins of the languages of the earth, I think we have to stick to the methods proposed by the Moscow school. Whether they themselves follow them or not I can’t tell. But the days of Nikolay Marr are hopefully gone. The future’s not ours to see (as Doris Day sang) but here as a teaser is a historic linguistic scheme from the University of Moscow: http://starling.rinet.ru/images/globet.png

  28. “Hence the “Kudos.”

    Yep.

    The part I left out was praise for her courage. I hope the Hindutva goons don’t know about her or this tree.

  29. I think it’s… fairly obvious that IE and Uralic are closely related.

    I would say they are promising candidates, but my chief reason for thinking so is the partial similarity of their pronoun systems and inflectional morphology. I’m not particularly impressed by the “water, name, sinew, salt…” set. Some of the similarities are certainly accidental. For example, PIE ‘sinew’ is morphologically complex and quite evidently derived within PIE from a PIE verb root using productive suffixation. *sneh₁-wr-/-wen- vs. Uralic *sïxni doesn’t look good to me, and some of the other “cognates” may be similarly analysable in PIE (but not in Uralic).

    As for the tree — it’s beautiful, but hardly accurate as a representation of anything. Indo-Iranian vs. “European”? Why only extant languages? And if extant (supposing that Cornish and Sanskrit qualify), where are, for example, the Sorbian languages?

  30. Christopher: From what I understand, Lyle Campbell never thinks any languages are related unless they’ve been well-known to be related for at least a century. (Corrections welcome, in case this turns out not to be the case.)

    Piotr: Accurate does not imply complete.

  31. where are, for example, the Sorbian languages?

    She says:

    (Naturally most tiny languages didn’t make it on the graph, aww. There’s literally hundreds of them in the Indo-European family alone and I could only fit so many on this page, so most sub-1 mil. speaker languages that don’t have official status somewhere got the cut.)

  32. Accurate does not imply complete.

    And I don’t mean it’s inaccurate because it’s incomplete. It presents relationship in a misleading way, mixing up geographical and genetic relationships in some cases (South Slavic, the “zones” of Modern Indo-Aryan), and grouping branches into odd-looking clusters. Romance (Italic) sandwiched between Armenian and Balto-Slavic, but far from Celtic? Hellenic at the base of Germanic? What sort of topology is this? And I really haven’t heard of a “European” clade before. But technical quibbles aside, it’s a really nice piece of artwork.

    LH: Both Sorbian languages are oficially recognised and protected as minority languages in Germany. They fare better than Cornish by any standards.

  33. I wondered about Sorbian, and Nuristani too. Sorbian has ca. 50,000 speakers, more than Scots Gaelic. Nuristani has 25,000, so I could see leaving it out, though as a significantly high-level clade I feel it would have deserved a leaf.

  34. I agree: the Nuristani languages are extremely interesting precisely because of their position within Indo-Iranian but outside {Indic + Iranian}. They should definitely be there.

  35. marie-lucie says:

    SH: I’m a little hesitant when it comes to Greenberg and Ruhlen.

    I am not at all hesitant to describe their “methods” as laughable. Ruhlen has made much of the method of setting “tests” for non-linguists, grouping carefully chosen sets of words in such as a way as to make the results (identifying their families) evident, and then saying “See? you too can be historical linguists!” or words to that effect.

    JC: From what I understand, Lyle Campbell never thinks any languages are related unless they’ve been well-known to be related for at least a century.

    That’s just about it.

  36. P.S. And what’s that Likai language in a sister relationship to Kurdish? Does Minna Sundberg mean Laki (a.k.a. Leki)? It’s really part of the Southern Kurdish dialect network, not a separate branch of the tree. Am I missing something?

  37. Alex Fink says:

    *sneh₁-wr-/-wen- vs. Uralic *sïxni doesn’t look good to me

    Lehtinen has observed that many cases of Proto-Uralic *x, including this one, are unnecessary (and are reconstructed thanks to placing too much weight on Finnic evidence): lengthening of Pre-Finnic *ä *a, into the latter of which *ï merged, was regular before a non-semivowel sonorant plus *-i. See Aikio (2012) for more on this. So it’s *sneh₁-wr vs *sïni ([sɨnə] ?), which is a little better.

    (Though on the general question of Indo-Uralic, at least as something we’ll ever recover, I’ll stand with LH in hesitation.)

  38. So it’s *sneh₁-wr vs *sïni ([sɨnə] ?), which is a little better.

    Yeah, but it’s the suffix (absent in Uralic) that makes the word a noun meaning ‘sinew, cord, vein’ (or the like). Without it, *sneh₁- means ‘spin, twist’. Dammit, even English sinew is just accidentally similar to Skt. snā́van- ‘sinew’, not related to it!

  39. Alex Fink says:

    Dammit, even English sinew is just accidentally similar to Skt. snā́van- ‘sinew’, not related to it!

    Huh, Wikipedia thinks it is. How should we trace sinew?

  40. And how do they explain the intrusive vowel and the missing final *r/n in Germanic? Either of them would be a serious problem in isolation. Together they can’t be shrugged off unless we are back to etymology as une science où les voyelles ne sont rien et les consonnes fort peu de chose. I side with those who etymologise Germanic *sinu- (and the secondary strong feminine *sinw-ō) as *sh₂i-nú-, from the root *sh₂ai- ‘bind, fasten’ (Hitt. ishiyanzi, Ved. sinā́ti with laryngeal metathesis). Cf. OE sīma ‘cord, rope’ < *síh₂-mon- = Hitt. ishiman-.

  41. Trond Engen says:

    Bjorvand & Lindeman compare Germanic *sinu- to Av. hinu. I’d say that FU *sinë- looks like another old IIr loan.

  42. I remember seeing a pretty scathing takedown of Ruhlen’s proposed Basque etymologies by Larry Trask, as well as a refutation of his Proto-World etmologies by Campbell. As an interested layman I’d love to see at least some of the more modest proposals verified, like Altaic or Eurasiatic, but my forays into the literature always leave me confused – who am I supposed to trust on the topic of Turkic-Tungusic-Japonic cognates?

  43. David Marjanović says:

    Oh, I have heard of this “European clade”. It’s defining “innovation” is the “split” of “PIE” *a, “faithfully preserved” by Sanskrit, into *e, *a and *o. That’s right, it’s a long-forgotten 19th-century idea that was founded on nothing more than “but the Vedas are so ancient”. Verner took it for granted in the famous paper on his law.

    It can’t even refer to ‘year 0′ since that one never existed (1 BC was followed by 1 AD)

    Not if you’re an astronomer. 🙂

    From what I know of their works the Muscovite nostraticists are professionals. Far from exercising mass comparison the easy way, they examine each item in every detail (phonetics, semantics, morphology) being faithful to the comparative method.

    There is one problem I can see: not their method, but their data. For PIE, they use Pokorny’s etymological dictionary published in 1959. The reason is that nothing has replaced it as a one-stop source. There are two big problems with it: it’s outdated several times over, almost completely lacking laryngeals – to pick the most obvous issue –, and it consistently errs on the side of inclusion, containing everything Pokorny or Walde could find that might conceivably go back to PIE even if, AFAIK, that required ad-hoc assumptions about analogy or such. Somewhat similarly, for Proto-Uralic they use Rédei’s etymological dictionary from 1988; not only has Ante Aikio alone successfully reconstructed a really large number of Proto-Uralic roots since then, but even the basic dichotomy of Uralic into Samoyedic and Finno-Ugric has come under heavy fire. Proto-Altaic is still shaky, and the state of the art of Proto-Afro-Asiatic reconstruction is in a really bad shape, because large branches of that family are woefully underresearched. And while there are two or three reconstructions of Proto-Afro-Asiatic, each with its own problems, there’s no reconstruction of Proto-Eskimo-Aleut at all, so the Moscow School simply ignores Aleut and resorts to Proto-Eskimo… And so on and so forth.

    The Moscow School Nostraticists know that Pokorny erred on the side of inclusion. That’s why their own etymological dictionaries do the same. If you don’t know this and open one of those books at random, you’re likely to conclude that it’s all garbage, because the robust comparisons only make up maybe half of the total number!

    From what I understand, Lyle Campbell never thinks any languages are related unless they’ve been well-known to be related for at least a century.

    I don’t know much about Campbell; from what little I do know, however, he strikes me a phylopessimist.

    Biology was going in that direction throughout the first two-thirds of the 20th century: people started out looking for “reliable characters” that would let them know the true phylogeny effortlessly, found that one candidate after another didn’t hold up, and resigned to “we can’t ever know anything”. Of course it’s true that everything can evolve convergently – if it can evolve once, it can evolve twice – or be inherited by lateral gene transfer. Sure, the probability of this happening isn’t equal for all characters, but outside the molecular realm the probabilities are hard to impossible to quantify, and they’re never zero. Likewise, every word and every grammatical feature can be borrowed or evolve convergently, and Campbell is by no means wrong in pointing that out.

    What happened next in biology is that a method was (very gradually) adopted which turned phylogenetics from an art into a science. It amounts to drawing a lot of trees, counting the ad-hoc assumptions that are needed to explain the data from each tree, and then by Ockham’s Razor preferring the tree(s) that require(s) the smallest number. I’m running such an analysis right now so I can present the results at a conference next week.

    The funny part is that historical linguistics has had such a method for so long it doesn’t even have a name, it’s just called “the comparative method”. The difference seems to be that, while the principle of parsimony is quite transparently the whole motivation behind the comparative method, its practitioners are less conscious of that and tend not to actually count their assumptions, so they can’t quantify how unlikely an unlikely hypothesis is – and consequently tend to think in absolutes of “proven” vs. “unproven” like creationists, the only other people who still use the p word when talking about science. The founder of phylogenetics in biology had a heavy philosophical bent (to the point of inventing a whole load of terminology just for the fun of it), and philosophy of science already existed as a field by then; it did not when historical linguistics got going about a hundred years earlier.

    I remember reading (about?) Campbell’s critique of Vajda’s Dené-Yeniseian hypothesis. Absolutely, that hypothesis requires more work, for instance because there are currently three pretty different reconstructions of Proto-Yeniseian out there, of which Vajda’s own is just one. But Campbell, IIRC, didn’t go into that. He was more like “this could be a chance resemblance, so it must be a chance resemblance”. For instance, he appeared to insist that all roots that are too short (without quantification) must be ignored, because they are more likely to be chance resemblances than longer ones. Of course they are, but that doesn’t decrease their value to 0! Most PIE roots consist of two consonants sandwiching a vowel that different languages preserve in different ablaut grades. Throw them all into the analysis; as you add data, the signal adds up, and the random noise cancels itself out.

    I’m a little hesitant when it comes to Greenberg and Ruhlen.

    Their method is great at generating hypotheses that are worth looking into. It’s incapable of testing them; and I don’t think Greenberg or Ruhlen have ever pretended otherwise.

    So it’s *sneh₁-wr vs *sïni ([sɨnə] ?), which is a little better.

    Even so, it’s not obvious if the PU *n should be compared to the *n in the PIE root or to the one in the *wr/*wn suffix, and the PU *s to the PIE *s or to the whole cluster *sn, initial clusters not being allowed in PU. And this holds both for common inheritance and for borrowing from PIE to PU.

    Aren’t all three PIE laryngeals supposed to correspond to PU *k (in “the earliest loans”, some or all of which have of course been proposed to be cognates on occasion)?

  44. David Marjanović says:

    The latest word on Altaic seems to be this. I still haven’t read all of it; let’s see how far I get tonight.

    In any case, it postulates a thick layer of West Turkic (Chuvash-Bolgar) loans in Mongolic that the Moscow School has overlooked.

  45. Trond Engen says:

    Also, I’ve been reading the comic the whole evening. Reading the prologue in these days of Ebola demagoguery, I was uneasy with the implied politics, but she somehow managed to make me stop bothering.

    I agree that the rendering of linguistic difficulties is good, but it would be even more interesting to read it in Scandinavian, since she obviously has a clear idea of what the different characters are saying in their respective (or eachothers) mother tongues. This would allow the exploration of some interesting new sociolinguistics in the surviving parts of Sweden, Denmark and Norway. In all three countries the capital and its surroundings were wiped out and new statelets were formed centered in a part of the country with a distinctly different dialect and deep-rooted local pride. In the cases of Norway and Sweden these dialects are even conservative in ways that makes them shear features with Icelandic, the new language of education, trade and public service.

    I’m not too much bothered by the inaccuracies of the tree. It’s a service to readers, most of whom have at best a faint idea of these relationships from before, and they’ll come out of it with a much better understanding, even if some details that they probably won’t notice anyway are wrong.

  46. David Marjanović says:

    …Correction: I had read the whole thing before, and it’s shorter than I remembered.

  47. Dude, that’s the previous page on the same site!

    You see? I have a great eye for these things!

  48. For more on Altaic, I recommend browsing through papers by Alexander Vovin, many of which are on his Academia.edu page. One of them, The end of the Altaic controversy, discusses briefly his conversion from a pro-Altaicist to an anti-one. The level of erudition needed to deal with this sort of material, and which Vovin demonstrates throughout, is dizzying. For example,

    “The Chinese character 牙 ‘fang’ has LMC reading /ŋia:/, so the Altaic hypothesis seems to be vindicated and emerges victorious, but the EDAL authors forget to inform their readers that both of these transcriptions are found in the Jurchen Hua-yi yi-yu which were compiled during the Ming period (1368–1644 C.E.). For this period LMC reading /ŋia:/ is simply anachronistic, and one has to use the EM reading /ya/ for the character 牙 ‘fang’.”

    How is a normal human being supposed to judge the value of Altaic reconstructions without knowing how Jurchen (a Tungusic language) was transcribed into Chinese letters during the Ming period? And then we get into Japanese (with its complex history), Korean (ditto), etc., and the whole thing gets even more absurd.

    The other thing I like about Vovin is his snarkiness, which is very refreshing. “Any kind of a body part term can be compared in EDAL to any other body part, e.g. ‘horn’ is compared to ‘gum’. Proto-Altaic speakers apparently were bestowed with a unique physiological feature unknown to any other types of homo sapiens, with horns instead of teeth growing from their gums.”

    I enjoy this combination of erudition and sarcasm very much. I associate both qualities with Russian and other Eastern European scholars of a certain generation, e.g. Manaster Ramer.

  49. I would appreciate it if someone could enlighten me re some of the comments above on sanskrít, specifically regarding how sanskrít is not the mother of indic languages.

    I always thought sanskrít evolved into the north Indian languages. Sort of like Latin is to French as sanskrít is to Hindi.

    Apologies if this is historical linguistics 101.

    Thank you

  50. The latest word on Altaic seems to be this [link: Stefan Georg]. I still haven’t read all of it; let’s see how far I get tonight.

    The problem with Altaic is that the field is so strongly polarised, with very little middle ground between the two warring camps. One of the best accounts of the history of Altaic studies, the Altaic Wars, and the present state of the battleground, is unfortunately in Polish (“Teoria ałtajska” by Marek Stachowski, 2012). Stachowski is sympathetic towards the Altaic Hypothesis, but takes the arguments of the anti-Altaic camp seriously and does his best to be impartial in his assessment. He also has some very good constructive ideas at the end of the article. Earlier he wrote a similarly incicive but balanced review of Nostratic studies and the Moscow School (2011). It’s top-quality stuff. I’ll try to convince Professor Stachowski that both articles should be published in English as soon as possible.

  51. I remember reading (about?) Campbell’s critique of Vajda’s Dené-Yeniseian hypothesis. Absolutely, that hypothesis requires more work, for instance because there are currently three pretty different reconstructions of Proto-Yeniseian out there, of which Vajda’s own is just one. But Campbell, IIRC, didn’t go into that. He was more like “this could be a chance resemblance, so it must be a chance resemblance”.

    I don’t think it’s a fair summary. Campbell actually wrote a pretty thorough review of the Dene-Yeniseian conference volume. He shows much respect for Vajda and his hypothesis, and though he doesn’t think the evidence is sufficient at present, the bottom line is:

    Vajda’s and the other papers of DYC are fun to read and to contemplate, and the hypothesis undoubtedly will continue to stimulate much discussion about proposals of distant genetic relationships and appropriate means and methods for assessing them.

    In his earlier publications (especially the oft-cited article on the IE an Uralic tree names in Diachronica, 1990) Campbell regards a deep historical connection between IE and U as demonstrated, though he leaves the question open if the correspondences are due to contact, common ancestry, or both.

  52. Oops, I forgot to add a link to Campbell’s review.

  53. David: Oh, I have heard of this “European clade”. It’s defining “innovation” is the “split” of “PIE” *a, “faithfully preserved” by Sanskrit, into *e, *a and *o. That’s right, it’s a long-forgotten 19th-century idea that was founded on nothing more than “but the Vedas are so ancient”. Verner took it for granted in the famous paper on his law.

    It’s true that Verner alludes to “the common European period” once in his article. Note, however, that he was fully aware of Indo-Aryan innovations such as Grassmann’s Law (1863), so invoking the antiquity of Vedic as an argument was already a thing of the past. The secondary character of the Indo-Iranian vowel system (“the Law of Palatals”) was demonstrated at the same time as Verner’s Law, in several separate papers independently (historians of linguistics sometimes refer to 1876 as annus mirabilis), so the idea must have been in the air for some time. The first IE family tree sketched by August Schleicher in the early 1860s had {Balto-Slavic + Germanic} and {Aryo-Greek + Italo-Celtic} as the primary subfamilies, with Indo-Iranian two bifurcations away from the root.

  54. marie-lucie says:

    At the Dene-Yenisei conference (2008) where Vajda presented his work (I was there too), several well-known historical linguists were invited, among them Eric Hamp, Johanna Nichols and Michael Krauss, who afterwards issued a joint statement for the local press (since the topic was of great interest in Alaska and among the native Athabaskan speakers) praising Vajda’s work.

    Campbell regards a deep historical connection between IE and U as demonstrated, though he leaves the question open if the correspondences are due to contact, common ancestry, or both.

    Precisely the question that one would like to see answered.

    Campbell’s award-winning book American Indian languages: the historical linguistics of Native America has a chapter (the last but one) titled “The methods”, which is worth reading about his positions.

  55. David Marjanović says:

    Thanks for the link to Campbell’s review, I’ll read it. I’ll also try to read Stachowski’s papers, though not during the conference next week, when I won’t even have enough time to sleep. 🙂

    The other thing I like about Vovin is his snarkiness, which is very refreshing. “Any kind of a body part term can be compared in EDAL to any other body part, e.g. ‘horn’ is compared to ‘gum’. Proto-Altaic speakers apparently were bestowed with a unique physiological feature unknown to any other types of homo sapiens, with horns instead of teeth growing from their gums.”

    I like snark, too, but it’s really easy to overdo. Semantic drift between “horn” (the material at least), “cartilage”, “tough connective tissue” and the like – or even “gnarly growth on a tree trunk” – is not uncommon in Indo-European.

    A. Dybo and G. Starostin replied to Vovin (and one of the arguments by Georg in the talk I linked to) in this pdf of 140 pages, titled “In Defense of the Comparative Method, or The End of the Vovin Controversy”. I highly recommend it. (It’s much less snarky than the title suggests.)

    G. Starostin has also given a conference talk on Dené-Yeniseian; the much, much shorter pdf of the slides is here.

    He also has some very good constructive ideas at the end of the article.

    Sensibly, he says Altaic is not a question of life and death 😀

  56. David Marjanović says:

    Heh. Footnote 9 to Stachowski (2011), brackets in the original:

    […] indeed, the story is told that several scholars learned sufficient Danish only to be able to read this seminal work in the original. […] I did have the advantage of closely discussing the text and translation with mag. art. Kaare Thomsen Hansen (Director of the Central Asian Institute, University of Copenhagen) […]. I take comfort in the fact that he tells me the author’s style is difficult and unclear at times („Translator’s remarks” w angielskim wydaniu Grønbech 1902)

    (“in the English edition of”)

  57. David Marjanović says:

    Now that’s weird. Only mag. art. is in italics in the original.

  58. Stefan Holm says:

    David: The year 0 never existed unless you’re an astronomer.

    But on the other hand ‘daylight saving time’ doesn’t exist to an astronomer. Hail Vladimir Vladimirovich who this week decided to get rid of that totally unnecessary ‘invention’. (His subjects in the northern polar regions complained about never seeing the summer sun.)

    Next fight will be against dinner, in Swedish middag, (German Mittag), literally ‘midday’. This is now consumed, say, 7.00 p.m. (post midi), ‘after mid day’. In ‘daylight saving time’ though only with six hours astronomic delay. Try to search for whoever came up with that terrible word lunch(eon) (‘of unknown origin’) and send him to Sweden for trial (in exchange for Julian Assange – who deserves to be a free man anyway)!

  59. Amir: I always thought sanskrít evolved into the north Indian languages. Sort of like Latin is to French as sanskrít is to Hindi.

    Sort of. The Romance languages developed from what we call Vulgar Latin — a collective term for the varieties spoken by the Romanised population of the provinces. The Classical Latin of Cicero & Co. was the literary emanation of the everyday language of educated upper-class speakers, artificially standardised and written rather than spoken.

    Classical Sanskrit was the standarised version of Old Indic, defined and codified by Indian grammarians (the culmination of that tradition was the work of Pāṇini in the 4th century BC). It was based on, but not identical with, the Vedic language in which the sacred lore of India had been composed and transmitted for centuries. By the time of Pāṇini Old Indic was the archaic language of the high (oral) literature. Classical Sanskrit became the formal medium used by priests, poets and scholars, but it wasn’t really ancestral to anything. The very point of standardisation was to fix the high language and prevent it from evolving. The spoken Indic languages contemporary with Classical Sanskrit were descended from Old Indic vernacular dialects closely related to Vedic. It wouldn’t be correct to say that Vedic is directly ancestral to Middle and Modern Indo-Aryan (just as, say, Old Church Slavonic must not be confused with Proto-Slavic). Other, unattested varieties of Old Indic also contributed to their development. It is possible, for example, to identify features lost in Vedic but preserved in some more recent (even modern) forms of Indo-Aryan.

  60. Next fight will be against dinner, in Swedish middag, (German Mittag), literally ‘midday’. This is now consumed, say, 7.00 p.m. (post midi), ‘after mid day’.

    Same is true for Russian ужин, etymologically ‘midday meal’ (: юг ‘south’; cf. Slovenian júžina ‘midday meal,’ which Vasmer tells me was borrowed into Austrian German as Jause ‘snack’).

  61. marie-lucie says:

    The meanings of words for various meals seem to be constantly evolving toward later and later times. Pretty soon people will have breakfast before going to bed and dinner upon arising. Eventually after a complete trip around the clock they will return to their original times!

  62. There’s a fascinating (and highly convincing) story of how Latin nona came to mean noon in English (hint: don’t believe what standard reference sources say). I owe it to Magda Charzyńska-Wójcik from the Catholic University of Lublin, who made it the topic of her Habilitation lecture (I was present there as one of the reviewers). Magda had invested a lot of painstaking detective work in it, so you’ll have to wait for the published version.

  63. Stefan Holm says:

    I agree with the opinion charging the Moscow School for relying too much upon doubtful data like Pokorny. But I still admire their (claimed) methodology.

    It is a firm belief of mine that we are on the edge of a breakthrough in historical linguistics. The key is cooperative schoolarship between linguistics, paleontology, genetics, archeology, climatology, biology and other disciplines. Stupid things will appear (and have appeared) but among them some pearls of divine luster will be hidden Behaviourist anti-Chomskyans may complain but I stay optimistic about finding the source of language in our species, the Homo Sapiens.

    Allons enfants!

  64. Matinee and entree (at least in the US) are other words whose meaning has drifted to refer to something later that the etymology suggests.

  65. Magda had invested a lot of painstaking detective work in it, so you’ll have to wait for the published version.

    OK, but you have to promise to let me know when it comes out — I’ll make a post of it!

  66. The single most convincing piece of evidence I’ve seen for Indo-Uralic is the pronouns — not just personal pronouns, but relative (e.g. Finnish jo-ka, cf. PIE io-) and interrogative (e.g. Finnish ku-ka, cf. PIE kw-) pronouns. I believe there’s at least one attested case of a relative pronoun being borrowed (from an Indic language to a neighboring non-IE language, IIRC), but is any language known to have borrowed its word for ‘who?’?

    (But then, I’m no Uralicist, and now someone will doubtless come along to say that the Finnish words can’t really be traced back to PU, or something like that.)

  67. TR: perhaps so, but then the anti-attaicists say that both (proto-?) Tungusic and Mongolian borrowed their pronouns from Turkic.

    On Campbell: although he indeed gets very set in his ways, I appreciated his nuanced review in his American Indian languages (mentioned by m.-l.) of various lumping hypotheses for languages of the Americas. He gives each a “probability” score (from -100% to 100%) and a “confidence” score (0–100%), the latter apparently gauging his own ability to judge the matter. Thus, Tarascan-Quechua rates -90% with 80% confidence, Tlingit-Eyak-Athabascan rates 75% (40% conf.), though adding Haida to these rates 0% (25%). Oddly, at the time the book was written he was much more sanguine about Takelma-Kalapuya (80% / 60%) than LH’s resident Penutianist appears to be at present.

  68. I’ve gotten into a couple arguments over the question of whether long-range relationships, in general, can be assumed to exist – even if they can never be proven. I’ve had someone (a young historical linguist, in fact) tell me that all currently known phyla must be assumed to be unrelated until proven otherwise, and that suggestions to the contrary need to be placed in the Russell’s teapot league of ignorable hypotheses along with things like animal language. But my contention has been: based on things we know other than the direct evidence of the languages themselves – like the fact that all healthy individuals in all known human societies acquire language from infancy, and the extreme rarity of completely new spoken languages arising naturally – doesn’t the idea of so many separate glottogenesis events seem pretty unlikely? Is it really more of a null hypothesis to assume hundreds of such events than to assume a smaller number?

    The analogy that I used was this: imagine that we’re looking at a skeleton of an 8,000-year-old man, and we want to know what his name was. Obviously there can’t be any direct evidence of what he was called, so he’ll go down in the records for all time as unnamed. But don’t you think he probably had some name, even if we’re never going to know it? Don’t you think it would be problematic to equate “he had no name that we know of” with “he probably had no name”? Maybe he came from a society where people actually had no names, but based on what we know about human societies, this isn’t highly likely. In the same way, I think it’s problematic to equate “there are no known relationships between currently attested language families” with “all known language families are probably the result of independent creation”. But sometimes I get the feeling that people are so eager to denounce the more fringy long-rangers that they refuse to even entertain the idea that there might be long-range relationships.

  69. John Cowan, Marie-Lucie: I think you are both too harsh on Lyle Campbell. He does accept the validity of the Algic family (Algonquian + Yurok + Wiyot), which was only proposed by Sapir in 1913, and only came to be universally accepted at a much later date.

    A comparative study of Algic and other proposed long-distance relationships (such as Altaic or Nostratic…), comparing the nature of the evidence presented, the nature of the objections presented, and the threshold required in order for the proposal to be accepted as worthy of further study, and then ultimate acceptance, would probably make for a most illuminating study in the methodology as well as the sociology of linguistics. If any reader of this thread is looking for a challenge, well, go ahead! As a famous scholar once said, “I’d rather read such a book than write it!”.

    Amir: actually, Classical Latin is definitely the ancestor of Vulgar Latin and of the various Romance languages, whereas the relationship between Vedic and Classical Sanskrit on the one hand and (Medieval and Modern) Indo-Aryan languages on the other is slightly more nuanced.

    Piotr: you may be interested to know that a fellow countryman of yours, Witold Manczak, is the scholar whose work convinced me that Classical Latin is indeed the ancestor of Vulgar Latin and the Romance languages.

  70. doesn’t the idea of so many separate glottogenesis events seem pretty unlikely?

    I guess, but so what? I mean, it doesn’t get us very far to say “some unknown and unknowable number of hundreds of thousands of years ago, all people spoke the same language.” It’s like saying (based on whatever calculations) “there must be intelligent life elsewhere in the universe.” Even if true, where does it get us? The latter doesn’t tell us anything about where such life is or what it is like, and the former doesn’t tell us how current families are related. It’s trivial to say “well, they’re all related if you go back far enough”; the question is (in this case) whether Indo-Uralic is a more recent relationship than that, and the assumption of all languages having the same origin is of exactly zero use in answering it.

  71. Oh, I agree that it’s a trivial idea absent proven recent relationships; it’s just that I’ve found that a lot of people react to it with surprising hostility.

  72. marie-lucie says:

    Y: Oddly, at the time the book was written he [Lyle Campbell] was much more sanguine about Takelma-Kalapuya (80% / 60%) than LH’s resident Penutianist appears to be at present.

    There is nothing odd about it, and your resident Penutianist has not changed her mind, which was made up about T and K some time before the book appeared. How much actual acquaintance did Campbell have with either of those two languages, and on what basis does he consider them probably related? Probably according to a 1918 article by Frachtenberg which cites a number of word pairs but is very misleading overall.

    Look at Marianne Mithun’s encyclopedic compilation The Languages of Native North America (which considers all the languages, families and potentially more distant relationships). Read the respective chapters on Takelma and Kalapuya, which mention and quote from a paper presented in 1998 disputing the alleged close relationship. (The paper is quoted with authors “K and T” in one chapter and “T and K” in the other, but it is the same paper and should be listed as “T and K”, which it is in the bibliography). My current work in progress builds on that paper and on deeper acquaintance with the two languages as well as other “Penutian” languages. (Campbell does not believe in any version of Sapir’s Penutian, so he lists a number of those languages as isolates, while I have found very good reasons to consider them related – even though a few of the languages included by Sapir might not make it eventually)

  73. m.-l., Oh, I certainly trust you on that. The ‘odd’ thing to me is that it’s a rare case of the naysaying Campbell being over-optimistic about a putative language family (and “60%” confident about it).

  74. marie-lucie says:

    Yes, that is a difference! But other people have believed in it (and still do), so he was not taking a big chance. Unfortunately, most other people’s positive opinions seem to be based on VERY superficial acquaintance and on repeating generalities rather than delving into the details.

  75. marie-lucie says:

    Etienne: John Cowan, Marie-Lucie: I think you are both too harsh on Lyle Campbell. He does accept the validity of the Algic family (Algonquian + Yurok + Wiyot), which was only proposed by Sapir in 1913, and only came to be universally accepted at a much later date.

    Perhaps JC’s mention of a century was a little exaggerated! But I think that LC likes to wait until he really knows which way the wind is blowing before he makes up his mind.

  76. I mean, it doesn’t get us very far to say “some unknown and unknowable number of hundreds of thousands of years ago, all people spoke the same language.”

    Note that Proto-World (the purported ancestor of all known languages) is logically independent of monogenesis. Even if all known languages do have a common ancestor, it wasn’t necessarily the only language spoken at the time, any more than the most recent common ancestor of present-day humanity was the only human living at the time (particularly since he or she lived only about 5000 years ago by current thinking).

  77. doesn’t the idea of so many separate glottogenesis events seem pretty unlikely? Is it really more of a null hypothesis to assume hundreds of such events than to assume a smaller number?

    Full sign languages are now known to have sprouted completely independently in a great many places. I don’t see why spoken languages couldn’t have done so as well, though the social circumstances for that happening nowadays don’t exist anywhere.

  78. Well, that’s why I specified spoken languages; I would struggle to think of plausible way in which a spoken language would arise totally independent of other languages in the way that sign languages can. A group of pre-linguistic infants abandoned in the wild, but surviving to adulthood? A society becoming mute and later reinventing language? (Or if we’re getting really fanciful, maybe a language taken from some other member of Homo?) But I could imagine creolization rendering genetic affinities unworkably messy.

  79. TR: perhaps so, but then the anti-attaicists say that both (proto-?) Tungusic and Mongolian borrowed their pronouns from Turkic.

    Do they say this about interrogative pronouns, or just personal pronouns? Cases of the latter getting borrowed are indeed well attested, but I don’t know of any case of a borrowed interrogative pronoun in a spoken language (apparently ASL gets some of its interrogative pronouns from English-based finger-spellings).

  80. Etienne: Classical Latin is definitely the ancestor of Vulgar Latin and of the various Romance languages…

    Classical Latin (as the posh sociolect used from the 1st century BC to the 3rd century AD) is doubtless more similar to the common ancestor it shared with Vulgar Latin beacuase of its conservatism. But to say that it is the ancestor of Romance is like saying that Archaeopteryx is a direct bird ancestor, or that modern varieties of English come from Alfred the Great’s upper-crust literary Late West Saxon. Plenty of features attributable to Proto-Romance are anticipated in the epigraphic and low-register varieties of Latin contemporary with CL, if not earlier, but studiously avoided in Classical Latin (to name but a few: the gradual loss of the neuter gender, the elimination of deponent verbs, 1st declension plurals in -as, prothetic vowels in /sC/ clusters, lowering of short high vowels, /ea, ia/ > /ja/ after consonants, etc., etc.).

  81. Lazar: By similar reasoning, one might conclude that there must have been a time when only one human existed; otherwise we would have to assume that multiple humans “arose” independently. Even having a couple of “first parents” would be a problem (solved in Genesis by deriving Eve from Adam’s rib).

    But humans have always existed as a sizeable population. When our distant ancestors diverged from the chimpanzee lineage, they did so as a non-homogeneous population, not as a single proto-human whose brother happened to be a proto-chimp. The same goes for later speciations: we passed through all of them as a normally variable population.

    The language faculty did not pop into existence overnight but evolved gradually from pre-linguistic modes of communication. Languages are not designed or invented: they are collectively shared cultural patterns that emerge, through self-organisation, from repeated interactions within a community. There is no treason why numerous individual languages should not have evolved in parallel in the numerous communities of early humans.

  82. Stefan Holm says:

    a “probability” score (from -100% to 100%)

    Non-scientists shouldn’t be allowed to use the word ‘energy’ and non-mathematicians to use the word ‘probability’. A probability is a number between 0 and 1 – period. Popular you can use 0-100% but there simply is no such thing as a negative probability. This is both descriptive and prescriptive.

    Carefully handled the theory of probabilities is an extremely sharp sword. Modern physics, the most exact science of all, to a very high extent actually relies upon probabilities (elementary particles come in trillions so the law of large numbers make their as individuals absolutely indeterminable behaviour almost equal to exactness as a collective).

    So the cross-border cooperation in order to promote historical linguistics I previously mentioned should definitely include mathematicians. Markov chains etc. should only be used by people who know exactly what they are dealing with, If so it’s a wonderful weapon. In the words of Sir Francis Galton:

    I know of scarcely anything so apt to impress the imagination as the wonderful form of cosmic order expressed by the “Law of Frequency of Error”. The law would have been personified by the Greeks and deified, if they had known of it. It reigns with serenity and in complete self-effacement, amidst the wildest confusion. The huger the mob, and the greater the apparent anarchy, the more perfect is its sway. It is the supreme law of Unreason. Whenever a large sample of chaotic elements are taken in hand and marshaled in the order of their magnitude, an unsuspected and most beautiful form of regularity proves to have been latent all along.

  83. Self-correction: treason => reason 🙂

  84. “We see no reason why gunpowder treason ever should be forgot.”

  85. It’s not quite the fifth of November yet.

  86. (Ha! Pipped at the Post button…)

  87. Heh, the article on pronoun borrowing strays into the Moluccas (Maluku) without once mentioning that they are Papuan 🙂

  88. @ Piotrek: Thanks for your helpful explanation on the relationship to modern Indic languages of Sanskrit.

  89. Piotr: from what I can tell, Stachowski’s account stops just at the beginning of the recent flurry of work by Martine Robbeets, on what she calls Transeurasian languages, a construct which is similar to Altaic. I haven’t read her works in detail. Stefan Georg is critical of her recent work just as he is of older Altaicist work.

  90. David Marjanović says:

    It is possible, for example, to identify features lost in Vedic but preserved in some more recent (even modern) forms of Indo-Aryan.

    I’m dying to know some! 🙂 Is [r] one, as opposed to the [ɻ] used in all except maybe the very earliest Vedic?

    which Vasmer tells me was borrowed into Austrian German as Jause ‘snack’

    Correct – and it may be relevant here that, in the dialects, this word doesn’t end in /ɛ/ but in /n̩/; because this is normal for feminine nouns that end in /ɛ/ in Standard German, the Standard form may well be reverse-engineered, and the /n/ may be etymologically “real” in this case.

    Plenty of features attributable to Proto-Romance are anticipated in the epigraphic and low-register varieties of Latin contemporary with CL, if not earlier, but studiously avoided in Classical Latin

    Somewhere on a wall in Pompe(i)i a teacher cum discipulos suos has preserved himself for posterity. *Sic transit gloria de mundo.

  91. David Marjanović says:

    I’m dying to know some! 🙂 Is [r] one, as opposed to the [ɻ] used in all except maybe the very earliest Vedic?

    To answer my own question, this is about things like the Vedic megamerger of something like six different consonant clusters into kṣ while various Prakrits distinguished at least ggʰ from kkʰ. In Vedic chanting today, at least, [ɻ] isn’t used at all – instead r is a just slightly retracted trill, among other surprises.

Trackbacks

  1. […] Hat links to a beautiful family tree illustration of Europe’s […]

Speak Your Mind

*