Natufian Origin for Afroasiatic?

Greg Pandatshang sent me this post by Razib Khan arguing for a Levantine (Natufian) origin for the Afroasiatic phylum on genetic grounds. Khan himself asks “Why would I have any particular insight?” and I’m inherently suspicious of attempts to link language and genetics, but it’s an interesting topic; Greg says he’d be curious to know what LH readers would make of it, and so am I. The conclusion:

The hypothesis I present is that after the descendants of the Natufians made the transition to farming, some immediately pushed into areas of Africa suitable for farming and/or pastoralism. They quick diversified into the various Berber and Cushitic languages. The adoption of Nilo-Saharan languages, and later Khoisan ones, was simply the process of successive and serial admixture into local populations as these paternal lineages introduced their lifestyle. In the Near East many distinct Semitic languages persisted across the Fertile Crescent, and for whatever reason the various non-Semitic languages faded and Semitic ones flourished.

Any and all thoughts are, as always, welcome.

Comments

  1. Well, just because Colin Renfrew believes it doesn’t make it wrong.

    And for once the timing is about right: we don’t have to worry about migrations that happen thousands of years before the splits in the various language groups, as we seem to have to in IE studies.

    Natufian: Language.

  2. (for the unwashed, like me) wikipedia The possibility of Natufians speaking proto-Afro-Asiatic, and that the language was introduced into Africa from the Levant, is approved by Colin Renfrew with caution, as a possible hypothesis for proto-Afro-Asiatic dispersal.

    Are we overturning the ‘out of Africa’ hypothesis? If not, presumably those who came out of Africa (to the Levant) already spoke a language. Was that proto-proto-Afro-Asiatic?

    Are p-p-AA’s descendants still spoken in Africa (where?) Or did it get wiped out by Levantines spreading/trading back into Africa (again, where?) [wikipedia on Proto-AA is so vague/speculative as to be useless.]

    Is it just that we’re at the event horizon, because it’s all so long ago? We have a hazy emanation from proto-Afro-Asiatic; older than that is just a black hole.

  3. Greg Pandatshang says:

    I believe AA is generally seen as having the deepest time depth of any well-accepted language family in the world, with the possible exception of Dene-Yeniseian. So it wouldn’t be surprising if its connection to its nearest relatives (if any exist) were undetectable.

    However (per my comment over at gnxp) a simple model of Afroasiatic intrusion from the west Asia into north Africa would require that AA languages in Africa should be a single early branch of AA, i.e. Egyptian, Berber, Cushitic, Oromo, and Chadic as a legit “North African” subfamily. That does not seem to be the case. Therefore, a more complex model would be necessary to explain the data, and that risks becoming overly convoluted.

  4. SFReader says:

    This hypothesis makes excellent sense.

    –Are we overturning the ‘out of Africa’ hypothesis? If not, presumably those who came out of Africa (to the Levant) already spoke a language. Was that proto-proto-Afro-Asiatic?

    Natufian evolved from Kebaran culture – c. 18,000 – c. 12,500 BC, which represented break of cultural continuity with preceding Antelian culture (and the Antelians were likely to be descendants of those out of Africa types)

    18,000 BC is a period after end of Last Glacial Maximum, so it fits very well.

    Very advanced Kebarans came from some LGM refugium and overrun the Levant, gradually evolving into Natufian. There was an article a year ago which claimed that agriculture was invented in the Kebaran stage (about 15000 BC), but died out due to later climatic change. Their Natufian descendants had to reinvent it again.

    AFAIK, some linguists argue with accepted dating of proto-Afroasiatic, placing it several millenia earlier. If they are right, proto-Afroasiatic would have been spoken by Kebarans and Natufians would speak some branch of it (which would include both Semitic and proto-Egyptian, I believe).

  5. David Eddyshaw says:

    Berber and Cushitic get a mention, but not Chadic, to which belongs the most widely spoken of all Afroasiatic languages after Arabic.

    Chadic alone is extremely diverse internally, much more so than Indoeuropean, say. The argument from the vastly greater diversity of Afroasiatic in Africa than Asia is not to be handwaved away.

  6. Athel Cornish-Bowden says:

    I love reading about this stuff. I more or less follow Cavalli-Sforza’s genetic arguments, but I’ve long suspected that he was out of his depth when he ventures into linguistics. In other words I share your suspicions about mixing language and genetics.

  7. marie-lucie says:

    Greg: I believe AA is generally seen as having the deepest time depth of any well-accepted language family in the world, with the possible exception of Dene-Yeniseian.

    I am not qualified to comment on (p)AA, but it seems to me that the great time depth attributed to D-Y results from accepting the increasingly doubtful assumption that the ocean rise creating the Bering Strait forever blocked human contact between Asia and America until the arrival of Europeans (only the Eskimo/Inuit being able to travel across it when the winter ice afforded passage on foot or with sleds). In a discipline where “mainstreamers” consider 6,000 years the upper limit of possible demonstration of language relatedness (an amount of time based on the IE case, supported by the archeological data), D-Y is assumed to have more than double that time depth. I find this conclusion implausible. For a number of reasons (linguistic and otherwise) I am convinced that there have been a number of trans-Pacific voyages, both voluntary and otherwise (given the sea currents), although not necessarily involving large numbers of people. This opinion is independently being shared by increasing numbers of scholars.

  8. I love reading this stuff too. There seems to be no limit to how much you can just make stuff up — I mean make the (lack of) evidence fit almost any hypothesis.

    Khan: Among the Masai, who have a clear minor West Eurasian ancestral component, albeit far less than Ethiopians, … So were there people to-ing and fro-ing in and out of Africa all the time?

    Even if we can trace their genes, I’m a lot less than convinced they all took their language with them. You have to make a separate case for that. We recently discussed the Angles/Saxons and Vikings on NW Europe. The Vikings who became Normans seem not to have taken their Norse language with them into Britain, even with Norse left over after the Danelaw.

  9. marie-lucie says:

    The Vikings who became Normans seem not to have taken their Norse language with them into Britain

    By the time they went to Britain they had become French speakers as a result of the all-male conquerors marrying women from the French-speaking population among which they now lived. Only a few retained some knowledge of Norse, taught to them at a special school.

  10. For a number of reasons (linguistic and otherwise) I am convinced that there have been a number of trans-Pacific voyages, both voluntary and otherwise (given the sea currents), although not necessarily involving large numbers of people. This opinion is independently being shared by increasing numbers of scholars.

    Indeed, it would be a miracle if absolutely no such crossing had taken place during the entire early Holocene.

  11. So were there people to-ing and fro-ing in and out of Africa all the time?

    Of course. Movement of individuals, at least, has been going on between the Levant and Egypt for the last five thousand years (I pick this date because Akkadian and Egyptian are both recorded in writing then, and are already completely mutually unintelligible). As another example, the writing system of Ethiopic clearly descends directly from that used for the Old South Arabian (Ṣayhadic) languages: it wasn’t carried across the Suez by birds.

  12. it seems to me that the great time depth attributed to D-Y results from accepting the increasingly doubtful assumption

    It’s also that Athabaskan is said to be about as diverse as Indo-European, so Proto-Na-Dene must have a time-depth greater than IE, since Eyak and Tlingit are much more remote from Athabaskan than the Athabaskan languages are from each other. How much longer ago, then, must it have been before Proto-Na-Dene and Proto-Yeniseian were unified? Of course this argument is not really quantitative (we know that languages do not all evolve at the same rate), but it is suggestive.

  13. “So were there people to-ing and fro-ing in and out of Africa all the time?

    “Of course. Movement of individuals, at least, has been going on between the Levant and Egypt for the last five thousand years (I pick this date because Akkadian and Egyptian are both recorded in writing then, and are already completely mutually unintelligible). As another example, the writing system of Ethiopic clearly descends directly from that used for the Old South Arabian (Ṣayhadic) languages: it wasn’t carried across the Suez by birds.”

    Not just the writing system — aren’t the Ethopic languages themselves transplants from the South Arabian Peninsula? They’re not just Afro-Asiatic, they’re specifically Semitic languages, with verbal systems (and much vocabulary) very similar to Arabic, Hebrew, etc.

  14. I thought there was some back-tracking on the Yenisei-Na Dene hypothesis, but I haven’t been following this very closely. Does the hypothesis continue to be accepted by a substantial number of linguists who are competent to evaluate it? (And recognizing that there are some who insist on very high standards of proof of genetic affiliation, and who will probably never accept it.)

  15. someone mentioned chadic language. curiously, they have a high frequency of a Y chromosomal lineage which is almost 100% eurasian back-flow

    http://www.sciencedirect.com/science/article/pii/S0002929716304487

  16. Greg Pandatshang says:

    About Na-Dene, doesn’t everybody (or reasonably close to everybody) accept that it represents a post-Beringian migration by boat from Asia?

    The only reason I mentioned D-Y as a possible deeper time depth is that I couldn’t remember the details of Vajda’s proposal off the top of my head, so I figured I might as well hedge. I do remember that Vajda in one of his talks gives a very surprising dictum to the effect that Algic was a late Beringian migration (so apparently he believes there were several of those) that is related to Dene-Yeniseian.

  17. marie-lucie says:

    JC: It’s also that Athabaskan is said to be about as diverse as Indo-European…

    Said by whom? I know little about the Athabaskan languages, but I am skeptical of this statement. Their relationship was recognized quite early, especially thanks to their very distinctive verbal structure, some of which they share with Eyak and Tlingit (situated at both ends of the Athabaskan territory in Southeastern Alaska). Perhaps there is a greater difference in vocabulary than in IE, especially between the (main) Northern languages and the Southern ones (Navaho, Apache, etc)? But vocabulary can be greatly influenced by contact with neighbouring languages, so it is not as “probative” as complex morphology.

    Bill W: the Yenisei-Na Dene hypothesis: I have not closely followed the development of the hypothesis, so I don’t want to hazard an opinion.

    (And recognizing that there are some who insist on very high standards of proof of genetic affiliation, and who will probably never accept it.)

    Indeed! “Some of those” are competent enough to deal with obvious families in which languages are very similar (on the order of Romance or Slavic) and expect the same high degree of resemblance in any suggested groupings. They usually rely mostly on lexical resemblances, treating morphology as an afterthought. (I refrain from citing names in order to protect the guilty).

  18. marie-lucie says:

    Greg: Vajda in one of his talks gives a very surprising dictum to the effect that Algic was a late Beringian migration (so apparently he believes there were several of those) that is related to Dene-Yeniseian.

    This is new to me, but again I have not been following the D-Y hypothesis very closely. A late Beringian migration, perhaps, but grouping Algic and D-Y does seem surprising!

  19. Greg Pandatshang says:

    Regarding to-ing and fro-ing between Africa and the Middle East, no one’s denying that there was lots of opportunity for it. There are both Semitic and non-Semitic Afroasiatic languages aplenty in Ethiopia and AFAIK everybody agrees that Ethiopian Semitic arrived recently (i.e. within 3000 years).

    All I’m saying is that the idea that the to-ing and fro-ing resulted in a linguistic map where most of the diversity is in Africa, giving the impression that AA originated there even though it was actually from the Levant, looks pretty ad hoc without additional historical or linguistic evidence for that model. (“Looks ad hoc” is not the same as “impossible”, naturally). For instance, if Semitic and Egyptian constitute a clade, that would be a start, but I’ve never seen that suggested.

  20. David Eddyshaw says:

    @Razib:

    The paper doesn’t have anything at all to do with the Chadic language group. The name “Chadic” leaves something to be desired as the accepted name of the Afroasiatic subgroup, as it certainly invites this very misunderstanding. It doesn’t at all mean “languages of Chad (the country)”

    https://en.m.wikipedia.org/wiki/Chadic_languages

  21. Marja Erwin says:

    1. There’s no such thing as an Indo-European skull-shape. Why should we expect such a thing as an Afro-Asiatic y-chromosome?

    2. There are such things as languages spreading from the periphery into core agricultural areas. At times, Indo-European, Altaic, and Uto-Aztecan. Why not Afro-Asiatic?

  22. David Eddyshaw says:

    I’m not in the least suprised that there are genes in Chad which have returned to Africa from elsewhere. Quite apart from the obvious (the Banu Hilal) there were Jewish tribes in the Sahara prior to Muslim times. And I myself have met blue-eyed black Africans as far south as Burkina Faso.

    There’s no doubt at all that Africa has been far from cut off genetically from Eurasia at any point. But that actually undermines the idea that Afroasiatic originated in Asia, by underlining the fact that genetics really has little to tell us about language prehistory.

  23. The paper doesn’t have anything at all to do with the Chadic language group. The name “Chadic” leaves something to be desired as the accepted name of the Afroasiatic subgroup, as it certainly invites this very misunderstanding. It doesn’t at all mean “languages of Chad (the country)”

    i know. but the R1b-V88 is found among chadic language groups. i only linked to that paper as it confirms there was autosomal admixture and put a specific date on it.

    sort by total %, highest frequency is among chadic:

    https://en.wikipedia.org/wiki/Haplogroup_R1b#R1b1a2_.28R-V88.29

  24. by underlining the fact that genetics really has little to tell us about language prehistory.

    genetics has falsified the initial demic diffusion model of colin renfrew. genetics has also falsified he cultural diffusion model of thai languages into mainland southeast asia. genetics has confirmed that the bantu expansion was almost a total replacement/folk wandering until you hit south africa. genetics shows that lowland PNG pops who speak austronesian languages have some asian ancestry.

    other stuff too.

  25. Etienne says:

    1-A much more serious objection (to my mind) to the thesis of a Levant homeland for Afroasiatic is the absence of any evidence in the Levant of a non- or pre-Semitic Afroasiatic language. Even if no such language had ever been written down, one would expect to find some traces of (one? some?) of them in the form of (for instance) loanwords in at least one of the neighboring non-Semitic/Afroasiatic languages which were written down (Sumerian, Hittite, Elamite…).

    2-I’d always assumed/accepted that the homeland of Afroasiatic had been the Upper Nile valley, and I wonder whether this might not explain the dialect structure of Semitic. Let’s assume Proto-Semitic was transplanted from Africa to the Arabian peninsula, across the Red Sea: presumably, for this to happen, its speakers would probably have been the carriers of a maritime-oriented culture, and thus Proto-Semitic would subsequently have spread along the coast of the Arabian peninsula before spreading inland.

    Now, the basic division in Semitic is between East Semitic (Akkadian and Eblaite) and West Semitic (all other Semitic languages). I always found it odd that there should be such a sharp divide between the two branches of Semitic, as the East Semitic languages were geographically less isolated from many West Semitic languages than many West Semitic languages were isolated from one another.

    But let us assume that from an initial point along the East Coast of Arabia Proto-Semitic spread both North and South, doubtless remaining a dialect continuum for a long time, with the northernmost and southernmost points of the continuum growing ever more dissimilar, even as Semitic continues its coastal spread…with the Southernmost point of the continuum spreading along the coast of Yemen, Oman, then spreading into the Persian Gulf and thence entering Mesopotamia…while the Northernmost would spread into Sinai and thence along the Eastern Mediterranean coast…with, ultimately, the two long-sundered points of the continuum coming into contact again in the Middle East. It would fit the data, wouldn’t it?

    Naturally we would have to assume that West Semitic subsequently expanded Southward into the Arabian peninsula, eliminating all the intermediate (i.e. neither East nor West) forms of Semitic…

    3-David Eddyshaw: for both Berber and Egyptian it has been argued (I’ve relevant references, should anyone want them) that their relative uniformity (compared to Semitic, Cushitic and Chadic) is due to comparatively recent, more local language spreads, which in both instances involved the elimination of a much more linguistically diverse landscape.

    4-Greg Pandatshang: Has Vajda published anything on Algic and its relationship to Dene-Yenisean?

    5-All: In a real sense this work is premature: for a fair number of Chadic and Cushitic languages we have so little data that whatever conclusions we reach today might well need to be wholly revised, if not wholly rejected, because of new data…

  26. Trond Engen says:

    marie-lucie: Said by whom? I know little about the Athabaskan languages, but I am skeptical of this statement.

    Thank you! I wanted to say that, but I have no leg to stand on.

    I think D-Y spread along the arctic coast and then up along the rivers Yenisei and Mackenzie. The distance along the coast is no longer than that of the Inuit expansion. The coastal Post-Proto-Dene-Yeniseians may have been (pick one or more) absorbed/marginalized/inadvertently infected with disease/killed/pushed inland by one or more of a series of arrivals through the Bering Strait, of which the most recent was the speakers of Proto-Eskimoic.

    Couldn’t Afro-Asiatic have spread into North-Africa from the Fertile Crescent in several waves? Chadic would constitute the deepest layer, maybe identical to the Natufian expansion, but was (pick one or more) absorbed/marginalized/inadvertently infected with disease/killed/pushed inland by one or more of a series of arrivals from the Levant, of which the most recent was the Arabic speakers.

  27. Trond Engen says:

    … but I like Etienne’s hypothesis for Semitic.

  28. A much more serious objection (to my mind) to the thesis of a Levant homeland for Afroasiatic is the absence of any evidence in the Levant of a non- or pre-Semitic Afroasiatic language. […] For both Berber and Egyptian it has been argued […] that their relative uniformity […] which in both instances involved the elimination of a much more linguistically diverse landscape.

    By the same token, any non-Semitic AA languages in the Levant might also have been eliminated, leaving only Semitic and the overseas varieties. I don’t seriously argue for this position, but it is good to keep an open mind (though not so open that your brains fall out). As you yourself pointed out, Corsican varieties are the most diverse of the Romance languages, even though the family did not originate in Corsica.

    I always found it odd that there should be such a sharp divide between the two branches of Semitic, as the East Semitic languages were geographically less isolated from many West Semitic languages than many West Semitic languages were isolated from one another.

    Not so odd when you reflect that except for internal reconstruction (which is inherently unanchored) we know nothing about what Akkadian might have looked like before Sumerian affected it.

  29. Chadic would constitute the deepest layer

    Surely Omotic is the deepest, unless you don’t accept that eruv.

  30. Greg Pandatshang says:

    Here’s Vajda’s dictum that I mentioned above: https://www.youtube.com/watch?v=7M0QnAqQUmw, approx. 56:00 to 59:00

    However, it looks like I misremembered part of his remarks. He doesn’t comment one way or the other on a linguistic relationship between Algic and Dene-Yeniseian. He says that Dene-Yeniseians and Pre-Algonquians were geographically closer in Siberia and that they share a lot of folklore.

  31. In a real sense this work is premature: for a fair number of Chadic and Cushitic languages we have so little data that whatever conclusions we reach today might well need to be wholly revised, if not wholly rejected, because of new data…

    Sure, but it’s fun to talk about!

  32. Trond Engen says:

    John Cowan: Surely Omotic is the deepest.

    Well, yes, quite likely, but I didn’t want to go there..If AA came from Southwest Asia, and Omotic is AA, then I think it represents a different mechanism of dispersal, and the greater divergence might e.g. be due to a strong substrate. I also didn’t bring up the complexities of Cushitic.

  33. Trond Engen says:

    Greg Pandatshang: [Vajda] says that Dene-Yeniseians and Pre-Algonquians were geographically closer in Siberia and that they share a lot of folklore.

    This is Berezkin territory again. (The link is from SFReader in the Ancient Indo-European Folktales thread from a couple of years ago).

  34. @Greg P the idea that the to-ing and fro-ing resulted in a linguistic map where most of the diversity is in Africa, giving the impression that AA originated there even though it was actually from the Levant, …

    No I didn’t speculate that (proto-)AA originated in Africa. I said there must have been an ancestor of proto-AA. And that (or its ancestor) must have come ‘out of Africa’. Unless someone is going to argue humans (and their languages) didn’t originate in Africa at all.

    … looks pretty ad hoc … I see nothing other than ad-hockery throughout the thread. Neither could it be any different: there’s just not enough evidence. Most of the languages were never written down.

    @Razib genetics has falsified … hasn’t shown whether the languages did or did not travel, even if the genes did (not).

  35. David Marjanović says:

    A tiny percentage of Neandertal DNA is found all over Africa, so evidently there has been some back-migration. It’s by no means limited to speakers of AA, though.

    the Yenisei-Na Dene hypothesis […] Does the hypothesis continue to be accepted by a substantial number of linguists who are competent to evaluate it?

    Only something like five people in the world are competent to evaluate it. Of those, all seem to agree that D and Y are closely enough related that their last common ancestor is reconstructable to a reasonable extent, but some of them think that other languages are even closer to these than they are to each other (specifically, Burushaski to Y, Sino-Tibetan to D), and the others haven’t weighed in on that.

  36. David Marjanović says:

    …with the Southernmost point of the continuum spreading along the coast of Yemen, Oman, then spreading into the Persian Gulf and thence entering Mesopotamia…while the Northernmost would spread into Sinai and thence along the Eastern Mediterranean coast…with, ultimately, the two long-sundered points of the continuum coming into contact again in the Middle East. It would fit the data, wouldn’t it?

    That would require leapfrogging over the Sumerians. And while their language has a substrate, that looks stunningly much like IE, not like AA at all. (Some words from this “Euphratic” language also seem to have reached Akkadian without going through Sumerian, IIRC.)

    we know nothing about what Akkadian might have looked like before Sumerian affected it

    What is Eblaite like? Is it known well enough to tell whether it has as much Sumerian influence?

  37. Greg Pandatshang says:

    Able was I ere I saw what Eblaite was like …

  38. marie-lucie says:

    David M: the Sumerians. … their language has a substrate, that looks stunningly much like IE

    I don’t know too much about the Sumerian language but I don’t remember reading about a substrate to it, let alone one like IE. I have a vague impression that someone (perhaps you) mentioned an article which I then read, about lexical resemblances with IE in some ancient language, could that have been it? but as you well know, lexical resemblances are not necessarily indications of a “substrate”. Anyway my recollection is rather vague but I would like to know more about this alleged substrate.

  39. ə de vivre says:

    And while their language has a substrate, that looks stunningly much like IE

    There are some plausibly IE words that reached written Sumerian via Akkadian very late, probably via the Kassites, who probably had an Indo-Aryan superstrate. If we’re thinking of the same guy(s), most of the Sumerian-IE comparisons are based on dubious internal reconstructions of proto-Sumerian and a number of words that don’t look like much more than chance. It’s really hard to tell what a native Sumerian word from a loan—especially a non-Semitic loan. Word-derivation isn’t super well understood, syllable structure is probably more complex than the “everything is CV(C)” assumption that used to be standard though no one is sure exactly how, and sign-readings for even common ideograms are often pretty tenuous.

    Some people get worked up about substrates because there are a lot of words related to agriculture and metal-working that are probably non-Sumerian, but since, archeologically, we know that agriculture and metal-working were both imported to Mesopotamia from elsewhere, the more plausible story for me is that (proto-)Sumerian speakers imported the vocabulary along with the technology. You just have to jump through too many hoops to make the case that (a/the) Sumerian language was in (at least) Southern Mesopotamia by the Uruk period.

    What is Eblaite like? Is it known well enough to tell whether it has as much Sumerian influence?

    There’s a decent amount of East Semitic weirdness that isn’t easily attributable to contact with Sumerian: retention of the inherited verb tense system and prepositions with no known Semitic cognates—off the top of my head.

  40. Marja Erwin says:

    I hardly know anything about the Sumerian question, but Whittaker has proposed an Indo-European substrate, calling it “Euphratic.”

    “Unless someone is going to argue humans (and their languages) didn’t originate in Africa at all.”

    I suspect that language, abstract art, and representational art all emerged in the upper paleolithic, and in Africa first. My pet theory, since there isn’t any increase in brain size, is that there was increasing neurodiversity, and that all these things aided communication between autistic and allistic people in the same or neighboring communities. Reliance on “tone,” and “facial expressions” can impede communication.

  41. Greg Pandatshang says:

    AntC, sorry, I didn’t mean to imply that you were speculating about an African origin. However, many scholars have indeed posited an African origin for AA, much more recently than the origin of Homo sapiens out of Africa.

    (How sad it is to imagine that most (between 80%~98%) of the linguistic history of humanity is unrecoverable to us because of the limits of the data we have for reconstruction!)

    Not all ad hoc suppositions are created equal. I would say that the more ad hoc we get the more we need Occam’s razor. The simplest model of the AA urheimat puts it in northern Africa, because that’s where the diversity is. However, it may very well be the case that the simplest reading of the linguistic data is at odds with the simplest reading of the genetic data.

  42. A tiny percentage of Neandertal DNA is found all over Africa, so evidently there has been some back-migration. It’s by no means limited to speakers of AA, though.

    it is found in yoruba, so yeah. david reich told me didn’t see much in dinka though. surprising to me.

  43. hasn’t shown whether the languages did or did not travel, even if the genes did (not).

    ok, so the bantu peoples are shocking homogeneous from cameroon to zimbabwe. so people moved. but we don’t know if the bantu language moved with them? it was always there.

    OK….

  44. m.-l.: For a number of reasons (linguistic and otherwise) I am convinced that there have been a number of trans-Pacific voyages, both voluntary and otherwise (given the sea currents), although not necessarily involving large numbers of people. This opinion is independently being shared by increasing numbers of scholars.

    So far, the only unequivocal linguistic evidence for trans-Pacific contact is *kumara, the Eastern Polynesian name for the sweet potato, a word from a probably Ecuadorean source, brought over together with the plant in pre-European times. A more recent claim, for Polynesian linguistic traces in California, doesn’t pass muster. The physical evidence has not been overwhelming, so far. There’s been a back-and-forth over the genetics of South American chickens, which I haven’t been keeping up with. Matisoo-Smith and Ramirez have described some prehistoric crania from Isla Mocha by the Chilean coast with Polynesian morphology, but so far there are no DNA results from them.

    Piotr: Indeed, it would be a miracle if absolutely no such crossing had taken place during the entire early Holocene. Actually, it is quite likely that it hasn’t. Colonization of small islands shows up very clearly in the archaeological record. Since there is no evidence of pre-Austronesian colonization in remote Oceania, it would be hard to explain a cross-Pacific voyage, difficult under any circumstances, which somehow either failed to notice any intermediate islands or avoided colonizing them.
    The ca. 1,000-year pause in Austronesian eastward expansion, between the settlement of West Polynesia and of East Polynesia, and the 300-year pause between the latter and the settlement of New Zealand and Hawai‘i (and perhaps Rapanui as well) shows that the push to the eastern Pacific was not something that would happen casually, even by people who were capable and interested in finding new islands to colonize.

  45. (Chadic / Athabascan) [is] about as diverse as Indo-European

    I always wonder if claims like these are counted from the modern Indo-European languages, or from Latin vs. Greek vs. Sanskrit. Modern Indo-Iranian languages, for one, are already themselves plenty diverse, though this tends to go unappreciated by philologists who ignore anything without a written tradition. Yet e.g. none of the dozen or so modern Eastern Iranian languages descend from any of the half a dozen attested Old/Middle Eastern Iranian languages.

    Claims like these also tend to be very linked to the state of comparative research on a family. 15-30 years ago, we had people around who used to think that Uralic is “deeper than Indo-European”, and possibly as much as 10000 years old. But by then it has turned out this was essentially impressionistic glottochronology, based on things like poor-quality data on Samoyedic (still a lexical outlier, but not anywhere near as much as earlier research indicated), and in reality Uralic is probably not substantially older than Indo-Iranian.

    Surely Omotic is the deepest

    Surely Omotic is the worst-researched, and this should be treated as the default hypothesis for why it appears so distant, until we have some solid examples of innovations common to all the non-Omotic AA languages.

    In case the point is not yet clear enough… The versions of Nostratic that include Afroasiatic (Dolgopolsky, Bomhard) have an interesting imbalance where Semitic shares more vocabulary with “Nostratic” than with the remainder of “Afroasiatic”; and also their Proto-Nostratic and Proto-AA phonologies look almost identical. I could three-quarters-jokingly suggest that per this data, it’s not so much that AA is a part of Nostratic, as much as Nostratic is a part of AA, and a sister group to Semitic.

    In reality lexical distances aren’t worth much if they aren’t put in context. If you have a 10,000-word dictionary of Arabic, a 10,000-word dictionary of Sanskrit, a 500-word dictionary of Anfillo, and a 500-word dictionary of Sardinian; the latter two additionally not optimized towards Swadesh-list-esque vocabulary; it is essentially guaranteed that you will find more (real or apparent) lexical matches between Arabic and Sanskrit than within the actual genetic groups Arabic-Anfillo and Sanskrit-Sardinian.

  46. marie-lucie says:

    Y, I suppose you refer to the book Polynesians in America, written by a number of scholars, which deals mostly with the South Pacific. This is not (or not mostly) what I am referring to.

    The Polynesian expansion occurred mostly over the enormous expanse of the Pacific Ocean, some of it at its widest, where most of the lands are small islands, and it took impressive navigational skills to reach the farthest ones from Asia. I am talking mostly about the North Pacific, where the situation is quite different. On the Asian side as one follows the coast northwards there are some substantial islands and peninsulas rather than vast distances over open water, and the distance between Asia and America gradually diminishes as one gets closer to the North Pole. The strong Japanese current first flows North, then curves East and South down the North American coast. A boat following this current, or carried by it, will not quite reach the Bering Strait, because of the Aleutian “belt” of small islands which seems to deflect the current toward America. This current still brings Japanese flotsam to Southern Alaska and British Columbia, including not only recognizable fishing gear but sometimes wrecked boats and occasionally human survivors.

    Unlike the Polynesian expansion which spread a single language family and basic culture over a huge maritime environment, the North American West Coast was home to a large number of different families crowded between the ocean and the mountain chains of the interior. Culturally there was almost as much diversity, from hunter-gatherers to former empires of great sophistication. Among these very diverse populations, there are still resemblances of many kinds with a number of Asian cultures, in terms of social organization, beliefs, art forms and motifs, myths and legends, and many other features. The varied nature and geographical extant of these various resemblances makes it unlikely either that they are coincidental or that they result from a single, millennia-old migration on foot, on sleds or in small boats.

  47. m.-l., I was referring to Jones and Klar’s papers, but, as you say, that is a different matter from the North Pacific voyages that you describe. I agree fully that there’s no reason to presuppose a single Beringian crossing à la Greenberg’s Amerind, and that coastal journeys from Asia are at least as plausible as land-based spreads.

    As to accidental drift voyages, although they happen, I have not seen any strong claim for a linguistic or cultural trait transported from Asia to the Pacific NW by such a voyage. Even where there are cultural similarities, I don’t know why they couldn’t be explained through the usual mechanisms of cultural spread from North Asia into North America, rather than by exceptional later voyages such as this.

  48. Y: Actually, it is quite likely that it hasn’t.

    Just to be clear, I was thinking about the “sub-Beringian” coastal route in the North Pacific, not Polynesian-style transoceanic dispersal with advanced seacraft and navigational methods.

  49. David Marjanović says:
  50. The Sahara had a green episode between 11.5 kya and 5 kya, and subsequent to this we find Afro-Asiatic languages along north, south, east and northwest fringes, suggesting an era of early broad expansion followed by multiple isolations after 3000 bce. If AA is demonstrably older than 11.5 ky, the homeland would likely be in one of the fringe areas. It wouldn’t be in the pre-green Sahara itself. Both Chad and the Levant are candidate areas. Chad has the diversity, but weren’t the smoothing effects of politics, writing, etc. already in play in the Levant by the times of Ebla and Ugarit, and certainly ever since? The good news is, we can totally eliminate Norway.

  51. Trond Engen says:

    Phil Jennings: The good news is, we can totally eliminate Norway.

    Gone for å couple of weeks, and LanguageHat becomes the Swedish Riksdag .

  52. ok, so the bantu peoples are shocking homogeneous from cameroon to zimbabwe. so people moved. but we don’t know if the bantu language moved with them? it was always there.

    OK….

    Razib, nobody’s saying it’s impossible for languages and genes to travel together, just that you can’t deduce one from the other. Of course it happens a lot, but that doesn’t mean it’s inevitable.

  53. This is a half-formed thought, but I wonder if there are some patterns of gene movement that would make us more or less confident that language travelled with the genes.

    For example, a rapid, near total replacement of the existing gene pool would seem an almost certain guarantee that the language of the replacing group replaced the language of the former group, but we don’t have to trust intuition-we have well attested examples we can draw from (which, because they’re more recent, also have the advantage of providing plenty of archaeological and genetic material for comparison to the prehistoric). Perhaps these relationships are mediated by archaeologically observable factors that we can quantify. The exceptions would be as interesting as the rules.

    The advantage being that we would then have archaeological and genetic evidence that informs our understanding of what likely happened linguistically (and how confident we can be).

  54. marie-lucie says:

    Thanks Y and Piotr.

    I did not mean to suggest that the cultural resemblances I was referring to had to be the result of accidental voyages, only that navigation in the North Pacific in the Asia-to-America direction was not only possible but favoured by the maritime and geographical conditions. Therefore it is to be expected that there were a number of crossings, whether intentional or not. An intentional crossing is reported in a semi-legendary Chinese tradition of an emperor sending a ship or even a fleet with many young people on board, probably intended to set up a colony or colonies on the other side of the ocean. At least one person came back and wrote a report about the regions visited, with details that seem to correspond to the actual North American coast.

    Of course, not every period of history has been favourable to ocean crossings and explorations. The Viking voyages and parts of the Polynesian expansions took place during periods of warmer climates, not during “Little Ice Ages” when rougher seas and sometimes the presence of icebergs made navigation too dangerous.

    Y, I am not sure what you mean by the usual mechanisms of cultural spread from North Asia into North America, most of which had to have occurred by boat, except in the very Far North on ice over the Bering Strait.

  55. marie-lucie says:

    Here are some traditional details from the American West Coast. I summarize from tales collected by Leo Frachtenberg about 100 years ago, in the Alsea language formerly spoken on the Oregon Coast. Some details of these tales seem to be quite unusual for the region.

    * In one story some fishermen are carried away to sea by a large fish or other animal that they have harpooned. After what they think is 4 days but turns out to be 4 years (a common motif in Japan, at least), they get back home so changed in appearance that their relatives at first think they are strangers from over the ocean.
    * Several tales make references to head-hunting, practiced by enemies only. In one story two boys find their father decapitated and go to a neighbouring enemy village to retrieve the head, which they find tied to the main roof beam.
    * At one point the culture hero comes to a village in which the local people seem to be terrorized. They avoid a house some distance from the village, empty of inhabitants but surrounded by unknown birds that they find threatening. He tells them that the birds are not dangerous but only want to be fed. (The NW cultures did not raise birds, but chickens from the Asian side were taken along in trans-Pacific voyages).
    * In another story the houses “of that time” are described as built on stilts above the water and the canoes stored below.

    These details are not what one would expect from the areas close to the Bering Strait.

  56. Marja Erwin says:

    Euphratic seems awfully early and awfully far south for western Indo-European. Which seems most likely?

    1. Indo-European spread and diversified at an early date, and one branch of Indo-European reached southern Mesopotamia.

    2. Some other macro-Indo-European or northern Mitian languages paralleled phonological developments in western Indo-European, and one branch of these reached southern Mesopotamia. I’m not sure how likely this is– do Uralic or Altaic forms fit? Whittaker’s examples of Indo-European grammar embedded in these loanwords would be a mistake.

    3. Indo-European spread and diversified at the usual dates, such as Anthony’s, and one group of western Indo-Europeans were Isoted/Grantvilled through time and space into late prehistoric Mesopotamia.

    4. Whittaker’s whole theory is a mistake.

  57. Euphratic seems awfully early and awfully far south for western Indo-European. Which seems most likely?

    Does it have to be IE or would resemblances in Northwest Caucasian account for the IE-looking material? It’s known that Northeast Caucasian was more widespread and it’s also possible the same is true of NW Caucasian, with no particular requirement that all the populations be contiguous.

  58. Lars (the original one) says:

    Better link for Whitteker (Though David’s seemed to work for several people, I had to Google and found a link with an extra /old/).

    To Marja’s list I’d add

    5. Before being imposed on a pre-existing Mesopotamian civilization, Sumerian was spoken somewhere in close contact with that extreme of the (Meta)-Proto-Indoeuropean dialect continuum, wherever that was at the time, that later gave rise to NW-PIE.

    I note from Whitteker’s article that none of the proposed Eufratic toponyms has topological reference — something like Gate of the Gods can be reapplied when a new center of worship is built. If a Mesopotamian river name could be interpreted as ‘noisy one’ vel sim in Euphratic, it would be a better argument for locating it there.

  59. Razib, nobody’s saying it’s impossible for languages and genes to travel together, just that you can’t deduce one from the other. Of course it happens a lot, but that doesn’t mean it’s inevitable.

    we’re talking about likelihoods and parsimony. obv we don’t have a time machine.

    1) anyone who has looked at bantu genetic data notices very short genetic distances btwn cameroon and mozambique (admixture with something ‘different’ really gets noticeable in south africa). africans may be genetically diverse, but that’s not true when you look between groups because of the bantus.

    2) the first ancient DNA from malawai indicates zero ancestry in modern malawians from these hunter-gatherers. this is hard to believe, but make sense in light of #1.

    3) you have a bantu language family all across the zone with very low genetic distances

    before genetics i had people tell me that it was probably cultural diffusion re: expansion of this farming toolkit. obv it’s not that. it’s not impossible that the bantu language didn’t spread with this demographically massive expansion from north-central africa to the east and south over the past 3,000 years. but seems unlikely.

  60. m.-l., do the Alsea stories have parallels in Asia (other than the Jaanese 4 days/4 years motif)? Are they in any of the standard indexes of folklore motifs?

    (By “the usual mechanisms of cultural spread from North Asia into North America” I meant substantial population movements, or cultural spreads through things like economic exchanges or cross-marriages; as opposed to casual or accidental contacts.)

  61. marie-lucie says:

    Y, I don’t know about parallel Asian stories, but I was struck by details which do not seem to be Amerindian at all, or even North Asian, but sound more Austronesian. I have consulted other collections of stories in Penutian languages (of which Alsea is one) and I found that in spite of some commonalities with neighbouring collections those unusual details in Alsea stood out.

    I do need to look at indexes of folklore motifs, but I have been more busy with linguistic matters. I am very interested in comparative mythology though.

    «usual mechanisms : Oh, sorry, how stupid of me! I misunderstood the structure of the sentence! I thought you meant “cultural spread from North Asia to North America” as a single phrase, rather than “usual mechanisms of cultural spread”, which of course I would have been familiar with.

  62. See Boas’s discussion in section III, especially pp. 383–385.
    Mythology and folk-tales of the North American Indians

  63. Genetics definitely confirms that today’s Bantu peoples come from the expansion of a common Proto-Bantu people, but then again, that’s a hypothesis that was already formulateable on the basis of the linguistic evidence.

    Research on what I would expect to get eventually called “historical sociology” (with research questions like “what ethnic groups / cultures existed where at what time”) seems to repeatedly rouse conceptual confusion. Linguists often seem to think that it’s a part of archeology; geneticists often seem to think it’s a part of linguistics. I would think it’s best treated as its own field entirely, even though it requires input from all other (pre)historical sciences.

    And the input is largely one way only. If there existed language isolates among the sea of Bantu languages (as we have by now figured there indeed are within some parts of Niger-Congo — Laal, Bangime, Jalaa — or for that matter, the undemonstratedness of the inclusion of Adamawa or Mande), then showing historical genetic continuity would not budge one bit the linguistic conclusion that they’re regardless language isolates. Or consider a language like Tok Pisin: if we want to know what it is exactly in terms of linguistic descent (an English dialect with a Papuan substrate? an English-Papuan creole? an original but purely English-based creole? a descendant of some other English-based creole? a Papuan language with extensive English influence?), then no amount of genetic analysis of Tok Pisin speakers will be able to answer the question.

    Genetics does well indeed when we want to assess scenarios of historical sociology, including those proposed by linguists. But if we do not distinguish that these are hypotheses of sociology and not linguistics per se, then when it comes to assessing theories that are purely linguistic questions (“is X demonstrably related to Y?” “is X plus Y a subgroup?”), genetics tends to not achieve much else than shooting itself in the foot with overconfidence.

  64. Very well said, and I like your idea of a “historical sociology” field.

  65. David Marjanović says:

    what they think is 4 days but turns out to be 4 years (a common motif in Japan, at least)

    Oh, there’s a mountain in Salzburg that does that to people.

    Euphratic seems awfully early and awfully far south for western Indo-European.

    Certainly. But what if the Yamnaya culture is only where the non-Anatolian branch of IE comes from? Just recently a paper on DNA came out saying there was no immigration into Anatolia from the northwest that could explain the origin of the Anatolian languages, as many had thought. So what if Gamq’relidze & Ivanov are right about the IE homeland after all (roughly around Armenia somewhere), Anatolian went west from there, then the feminine gender developed, then “Euphratic” went south while the rest went north across the Caucasus and founded the Yamnaya culture? That could explain why the Yamnaya people have 50:50 ancestry from Eastern Hunter-Gatherers and Caucasus Hunter-Gatherers.

    That said, Razib once said in a blog post that evidence was about to be published that the CHG ancestry in Yamnaya belonged entirely to the female line, which wouldn’t fit this scenario at all.

    I’m not sure how likely this is– do Uralic or Altaic forms fit?

    In short, no. On top of that, geographically Uralic is an even worse fit than IE, and Proto-Altaic should be sought somewhere around Mongolia or so.

    Does it have to be IE or would resemblances in Northwest Caucasian account for the IE-looking material?

    There’s IE morphology in it. Good luck explaining that by something as starkly different as West Caucasian. (Or East Caucasian for that matter.)

    I misunderstood the structure of the sentence!

    Me too, in the same way.

  66. David Marjanović says:

    Very well said, and I like your idea of a “historical sociology” field.

    All seconded.

    If there existed language isolates among the sea of Bantu languages (as we have by now figured there indeed are within some parts of Niger-Congo — Laal, Bangime, Jalaa — or for that matter, the undemonstratedness of the inclusion of Adamawa or Mande), then showing historical genetic continuity would not budge one bit the linguistic conclusion that they’re regardless language isolates.

    Basque is such a case, BTW. On the grand European gradient of ancestry in terms of Early European Farmers, Western Hunter-Gatherers and Yamnaya, the speakers of Basque are exactly where you’d predict from geography alone, and their Y chromosomes are almost all R1b, which was introduced into western and even central Europe from Yamnaya.

  67. Marja Erwin says:

    I know Uralic and Altaic pose worse geographical problems. But Uralic is supposed to be related to Indo-European. I figure such outgroups would help determine whether a specifically Indo-European, a macro-Indo-European, or a generally Northern Mitian model would fit better.

    If true, would the feminine endings point to steppe Indo-European after the split with Anatolian? Would the fish-icon for pes and other sounds point to western Indo-European? I suppose they could simply show that these features are older than previously thought.

  68. “But Uralic is supposed to be related to Indo-European.”

    While a couple of people from Leiden have argued for that in recent years, the vast majority of both IEists and Finno-Ugrists reject such a connection. Of course, Uralic does have a number of quite uncontroversial early Indo-European loanwords.

    As for connecting Na-Dene and Yeniseian, doesn’t anyone else feel this might be complicated by Yaroslav Gorbachov’s thesis that Yeniseian might be traced to significantly further south in Mongolian/China, crossing the West Sayan mountains only in relatively recent times?

  69. David Marjanović says:

    If true, would the feminine endings point to steppe Indo-European after the split with Anatolian? Would the fish-icon for pes and other sounds point to western Indo-European? I suppose they could simply show that these features are older than previously thought.

    That would be simple in the fish case, where it has actually been suggested that that root first applied to one particular species and was only later generalized to “fish”. It wouldn’t be simple at all in the case of the feminine gender, which shows no trace of having ever existed in Anatolian.

    While a couple of people from Leiden have argued for that in recent years, the vast majority of both IEists and Finno-Ugrists reject such a connection.

    They don’t, really. They state that the published evidence is too weak to do much with; but they don’t propose any suggestions for what IE or Uralic could be more closely related with.

    That said, there isn’t much published evidence (and no unpublished one that I know of, of course!) that would indicate that IE and Uralic are each other’s closest known relatives. The Leiden school’s work on Indo-Uralic compares IE and Uralic to each other, but apparently to nothing else; that way they aren’t going to find out if IE is, say, closer to Etruscan or Uralic is closer to Yukagir (an old but still weak idea) or to Altaic or part thereof (as suggested but not elaborated by Marek Stachowski).

    As for connecting Na-Dene and Yeniseian, doesn’t anyone else feel this might be complicated by Yaroslav Gorbachov’s thesis that Yeniseian might be traced to significantly further south in Mongolian/China, crossing the West Sayan mountains only in relatively recent times?

    (Vajda talks about this close to halfway through the video linked to above: July 12, 4:34 pm.)

    Hardly; the distance to Alaska stays about the same. But then, what evidence there is doesn’t really say that Yeniseian and Na-Dene are each other’s closest relatives. Some of the fieldwork we’ll need for a reliable reconstruction of Proto-Sino-Tibetan is currently ongoing…

  70. but they don’t propose any suggestions for what IE or Uralic could be more closely related with.

    I don’t understand. Why are they under any obligation to propose such suggestions? Why isn’t it enough to say “insufficient evidence”? To get somebody declared not guilty of a crime, you don’t have to produce an alternative perpetrator.

  71. Sure. But in the absence of any conclusive DNA evidence, a very good way to reject the hypothesis that Jimmy Carter (poor man) is my biological father is to give evidence that Thomas Cowan actually is, the two being mutually exclusive.

  72. marie-lucie says:

    [those who reject a relationship between IE and Uralic] don’t propose any suggestions for what IE or Uralic could be more closely related with.
    LH: Why isn’t it enough to say “insufficient evidence”? To get somebody declared not guilty of a crime, you don’t have to produce an alternative perpetrator.

    You don’t have to, but many people have spent years on death row (or even been executed), often on flimsy evidence, only to be cleared by the discovery of an alternative perpetrator against which the evidence was compelling.

    Back to language classification, here is another case in the “Penutian phylum”.

    After Dixon and Kroeber classified a group in California as “Penutian”, Sapir, who had been working on the Takelma language of Southern Oregon, was “struck by resemblances between Takelma and Yokuts” (the latter being one of the original Penutian group, in Southern California), but he did not add Takelma to that group, preferring a geographically-based classification in which Takelma was included in a tripartite “Oregon Penutian” group along with Kalapuya in the Central valley and three “Coast Oregon” languages (Alsea, Siuslaw and Coos). Apart from Takelma which he had studied personally, Sapir relied on data supplied by Frachtenberg on the other Oregon languages. Meanwhile F had published a paper comparing some Takelma words with similar ones in Chinook and Kalapuya. F’s paper is heavy on lexical resemblances, but refrains from putting T and K together because of morphological differences. Nevertheless, a hasty reading of the paper (which is rather confusing) gives the impression that he thinks that the two are closely related. Later, after more Kalapuya data had been collected (by Melville Jacobs), Swadesh gathered some of the material for one of his famous lists of basic vocabulary, and concluded that the K and T were indeed closely related and together formed a group he called “Takelman”.

    If you look up major reference words such as Campbell’s 1997 work on the historical linguistics of Native American languages, or Goddard’s edited Volume 7 (Languages of the Smithsonian Handbook of Native American Languages), you will see “Takelman” described as an ‘obvious’ family or words to that effect. What is obvious to someone seriously comparing the two languages T and K is that the authors of those descriptions never bothered to look at the languages themselves. Also, none of them even mentioned Sapir’s observation about Takelma and Yokuts. Indeed, Takelma and Yokuts are very similar, especially in their complex verbal morphology, while Kalapuya morphological structure is completely different from theirs. Lexical resemblances between T and K should be attributed to a period of close contact, at a time when the two languages (not contiguous in the historical period) occupied adjacent territories. Of course, linking one language of Southern Oregon (Takelma) most closely with one of Southern California (Yokuts) is not obvious at first sight, but many other families on the West Coast (and elsewhere) occupy territories which are not close to each other.

    At least one linguist working on Kalapuya (one of very few) concluded that Swadesh must be right, that T and K shared a “special relationship”, so old that the two languages had had time to completely restructure their respective morphologies (while a number of their words were still practically identical). This conclusion does not make any sense either from principles of language classification or from everything that is known about the consequences of language contact (e.g. Sarah Thomason’s work).

  73. m.-l.,
    Takelma and Yokuts are very similar, especially in their complex verbal morphology—more so, do you think, than Yokuts and Utian?

  74. David Marjanović says:

    I don’t understand. Why are they under any obligation to propose such suggestions? Why isn’t it enough to say “insufficient evidence”?

    “Insufficient evidence” is enough for “we can’t tell if they’re each other’s closest relatives”; it’s not enough for “they’re as distantly related as languages can be”. That’s surprisingly often confused in historical linguistics.

  75. marie-lucie says:

    Y: Are Takelma and Yokuts closer than Yokuts and Utian (= Miwok + Costanoan)?

    I haven’t spent much time on that, as I have been concentrating in putting an end to “Takelman” by spending time and effort on Kalapuya. But that is the impression I get. It would make sense that Takelma (which was mostly in Oregon but overlapping a little with California) and Yokuts, at the extremities of a North-South almost-continuum, should be more conservative than Miwok and Costanoan, which are roughly in the middle of the territory. In terms of verbal structure, it seems to me that Miwok (I don’t know about Costanoan) uses consonant gemination where Takelma and Yokuts use glottalization, at least in non-initial position (this is a rough impression only, I haven’t looked carefully yet).

    Kalapuya (actually a family of three languages) should be better known, as there is quite a lot of recorded material, at least on one of the languages, Santiam. Jacobs published a lot of texts, only a few of which have interlinear translation. This seems to have stopped a lot of linguists from looking at it seriously. The phonology is very easy, nouns are not hard to identify, but the verb system is quite complex (in a vastly different way from Takelma’s). Most of the scholarship has remained unpublished, and the various attempts at analyzing verb structure often arrive at quite different conclusions. So, a real challenge!

  76. David Marjanović says:

    A few Kalapuya words were apparently recorded on Francis Drake’s voyage.

  77. Just one, claims the linked paper, <petáh> ‘edible root’ = *pduʔ ~ *pdoʔ ‘wapato’. The analysis is overall not convincing.

  78. David Marjanović says:

    One click away from Vajda’s presentation is another presentation by him (27 minutes) which is cautiously optimistic about Dene-Caucasian and is explicitly completely agnostic on whether there’s an exclusive Dene-Yeniseian subgroup in it. It’s from 2013. Calls for more research on comparative polysynthetic verb morphology; doesn’t mention Bengtson’s 2008 paper on that subject, nor of course Guillaume Jacques’s very recent conviction that the polysynthesis of the Rgyalrong and Kiranti languages is cognate and will one day make it possible to reconstruct such morphology for Proto-Sino-Tibetan.

    Another candidate for a particularly close relative of Yeniseian is… *drumroll*… Hattic. It’s long, but make sure not to skip the interspersed critical comments on the current state of “Sino-Caucasian” reconstruction.

  79. Holy crap, I’m stuck on the first item in the table of contents, Stefan Bojowald’s “Noch einmal zum Personennamen t””’wṯw in Urk. IV, 11, 9″ — the ””’ represents the most amazing pile of backwards and forwards apostrophes I’ve ever seen. I have no idea what it means, but I’m glad I don’t have to pronounce it.

  80. David Marjanović says:

    t””’wṯw

    On academia.edu I see two apostrophes on top of each other, then an inverted one, and then again two normal ones on top of each other. I guess that’s an egyptological transcription. The lack of vowels fits that of course.

  81. tꜢʿꜢmṯw.

    (What looks like two superimposed apostropes is technically “LATIN CAPITAL LETTER EGYPTOLOGICAL ALEF”).

  82. Ah, that’s a much better version.

  83. The alef transliterates the character 𓄿 (looks like a vulture), which represented glottal stop or zero in Middle Egyptian, but apparently [l] or perhaps [ɾ] in Old Egyptian. In Egyptological pronounciation it is generally /a/. Graphically, it is two vertically stacked Semitic-style alefs.

    The ʿ is the traditional transliteration of 𓂝 (looks like a forearm), called “Egyptological ayin” and traditionally assumed to be a voiced pharyngeal fricative like Hebrew and Arabic ʿayin, though [d] in Old Egyptian, and also /a/ in Egyptological pronunciation. Graphically it is a Greek rough breathing.

    You may want to blow up this page with Ctrl + or Command + in order to see the hieroglyphs in this comment better.

  84. marie-lucie says:

    DM, Y: Indeed there is only one Kalapuya word, pdo, probably of the same origin as wapato (but not ‘potato’).

    It is not out of the question that Drake could have gone farther North than California, but in that case he might have encountered speakers of Alsea, Siuslaw, Coos, Tillamook and several others coastal languages, but hardly speakers of Kalapuya who lived in the interior, quite far from the coast.

    Lyon has done a lot of research, but does not seem to be aware of the most recent results. Note the map with “Takelman-Kalapuyan” as a single family (refuted in Mithun 1999). He also mentions Molalla but is not aware of Harris’ 2006 dissertation which is a grammar of the language. On a more basic level, he keeps referring to the words on the old lists as “cognates” of words recently attested in the various languages. But “cognates” refers to words for which a common genetic origin has been demonstrated. Here the words recorded 300 or so years ago could be the same as in more recent lists, or they could be borrowings between languages, therefore not “cognates” of any reported words.

  85. Many other problems with Lyon’s paper. The suggested words matched with the Drake vocabularies are from a grab bag of languages, with some truly farfetched semantic and phonological matches: <huchee kecharoh> ‘sit down’—*háh-tsa ‘basket’ (Kwalhioqua-Clatskanie), or *hats yî´qa tcɪ̄ ʟōwa ́kats ‘just continually there he sat’ (Coos), take your pick; <hióh> ‘king’ is matched with any of *hayu, hiyu, hyas ‘several, many, a group’ (Chinuk Wawa), hīyaᵋ ‘cousin, friend’ (Alsea), or taiyū ‘chief’ (Chasta Costa). These are languages spanning the entire length of the Oregon coast, some 300 miles.
    As far as I can tell, this paper is part of a project exploring the idea that Drake’s 1579 landing place (“Nova Albion”) was on the central Oregon coast, rather than in one of the bays just north of San Francisco. Drake’s diaries are very clear on having sailed north as far as Oregon, but then returned south and anchored at latitude 38°, where the encounter with the Indians happened. The Oregon theory explains that away by claiming that the English deliberately falsified Drake’s records to mislead the Spanish. Um.

  86. marie-lucie says:

    Y, You are right. I read or rather skimmed rather quickly as it was obviously useless to try itemizing the wrong or doubtful stuff. “Much ado about nothing”.

  87. David Marjanović says:

    Post these comments at Panchronica, not here!

  88. Trond Engen says:

    There’s actually quite a lot that is known about Francis Drake’s landing in America .

  89. David Marjanović says:

    In the German “translation” it’s Cornelius Coot who bears the Drake name (Emil Erpel).

  90. Trond Engen says:

    His Norwegian name is (quite boringly) Kornelius Kvakk. I thought of that just a couple of days ago, in the Eider Museum on Vega, having the idea that he might be brought over as an eider duck (Somateria mollissima), maybe a faux Dano-Norwegian Evald Edder-Duun.

    The name ‘Cornelius Coot’ has a touch of Dutch that suits an American city founder, and more so back when the location of Duckburgh was less Pacific.

  91. Your Eider Museum is missing its link.

  92. Trond Engen says:

    The missing link.

    Not that the link was very important.

  93. David Marjanović says:

    But the best remains Cyrano de Donaldac, the one in whose presence you mustn’t dare to mention beaks/bills.

  94. That’s a pleasant-looking little hus.

  95. His Norwegian name is (quite boringly) Kornelius Kvakk.

    I must disagree with “boringly” there. I think you underestimate the pleasure that a word like “kvakk” can bring to those of us languishing within worldviews circumscribed Sapir-Whorf style by West Germanic languages disguised as Romance pidgins.

  96. Trond Engen says:

    David M.: But the best remains Cyrano de Donaldac

    This one? A very good name, although it don’t seem to be attested outside the apocryphal Italian tradition.

  97. Trond Engen says:

    Matt: the pleasure that a word like “kvakk” can bring

    About the same in Norwegian as Quack does in a typical WGLDARP.

  98. In the mirror universe I am chuckling through a goatee at the word “quack” from a hus overlooking a fjord.

  99. David Marjanović says:

    This one?

    No, that’s another adaptation of the same story! But yes, it’s from the Dano-Italian tradition – and to call that apocryphal is a strange thing to do for someone living right next to Denmark. 🙂 It upheld the sacred fiction that the immortal Walt Disney was doing everything on his own well into the 1990s.

  100. Trond Engen says:

    My layman’s understanding is that Scandinavian Donaldism generally discerns several layers of canonicity. The bulk of stories in the Scandinavian tradition are not considered at all. The sacred core is the long stories with both script and artwork by the great Carl Barks in the years roughly 1942 to 1967. The newspaper strips by Bob Karp and Al Talioferro from the same period enjoy a similar status, but due to the nature of the material they are less important for establishing Duckburg’s history and major geneologies. Stories co-authored by Barks are somewhat less canonical. A more recent addition approaching canonical status is the work of Don Rosa. While clearly secondary to the established canon, and even including some elements from apocryphal sources, especially Italian, its apparently deep insights into Duckburg society soon gave it a strong position

  101. Trond Engen says:

    I should backtrack a little on the boringness of ‘Kvakk’. It just occured to me that it’s a play on Kristian Kvart, or king Christian IV of Denmark-Norway, who as the founder of Christiania, Oslo’s repositioning, reconstruction and rebranding after a major fire, has a similar position on a public square.

  102. Greg Pandatshang says:

    Sorry, I missed a bunch of the comments in this thread and am too busy to catch up on all of them. But the basic gist is that Norwegian ducks speak Proto-Afro-Asiatic? I guess I’m not all that surprised, I think.

  103. Trond Engen says:

    It’s the Bell Beaker connection.

  104. Thread win goes to Trond Engen. 🙂

  105. Trond Engen says:

    Razib: 2) the first ancient DNA from malawai indicates zero ancestry in modern malawians from these hunter-gatherers. this is hard to believe, but make sense in light of #1.

    A complete replacement of the previous inhabitants may well be in the Bantu genetic data, but I wonder about the minority groups on the East African savanna. Apart from Hadza and Sandawe, there were apparently one or more waves of Cushitic cattle breeders before the Bantu expansion, while Nilo-Saharan (Maasai) came later and entered into the niche, or the marginal lands, held by Cushitic speakers. Among those were hunter-gatherers who at some point may have taken up their Cushitic language from the dominant neighbours and who are now in the final stage of a shift to Maasai. Several languages presumed to be Cushitic are only known as substrates in Maasai, Very interesting processes, and it would be surprising if there’s no genetic trace of the Pre-Bantu inhabitants — in non-Bantu as well as Bantu speakers in the region..

  106. Trond Engen says:

    (Several waves of Nllotic speakers too, of which the Maasai may belong to the most recent. All the Nilotic languages south of the Turkana seem to have had similar relationships to Cushitic and to indigenous hunter-gatherers.)

Trackbacks

  1. […] Hat links to speculation that the Afroasiatic language family has its origins in the Natufian […]

Speak Your Mind

*