MUSKOGEAN AND LAMB’S-QUARTERS.

While trying to figure out if Muskogean (the language family to which Choctaw and Chickasaw, among others, belong) is considered to be part of any larger grouping (apparently some people take it for granted it’s part of the “Hokan-Siouxan” group while others treat it as independent, Wikipedia calls Hokan itself “a hypothetical grouping of a dozen small language families spoken in California and Mexico” and says “few linguists today expect Hokan as a whole to prove to be valid,” and I’m certainly not qualified to even have a thought about the matter), I ran across an interesting paper (pdf file; abstract here) by Prof. George Aaron Broadwell called “Reconstructing Proto-Muskogean Language and Prehistory: Preliminary results” that’s chock-full of the kind of detailed lexical comparisons and reconstructions I so enjoy. One thing that makes it exotic from the point of view of someone trained in Indo-European (where the inherited vocabulary includes terms for ‘beech,’ ‘birch,’ ‘wolf,’ and ‘salmon’) is the list of “Reconstructable Proto-Muskogean terms,” which includes words for chestnut, chicken snake, chickenhawk, chigger, chinquapin, chipmunk, civet cat (?), clam/spoon, copperhead, corn, cotton, and crawfish, to take only the c‘s (the full list is on pages 15-16 of the paper). But what impelled me to post about it is the point he makes about a common problem in historical linguistics:

How can we reconcile the presence of a word for corn
with the generally accepted archaeological position that corn was not present in the southeast until considerably later, ca. A.D. 700?…
A common approach in dismissing linguistic evidence that does not correlate with the archaeological results is to suggest that the reference of the words has changed through time (cf. Renfrew 1988). For example, the word for corn might have originally referred to some other grain. When corn was introduced to the southeast the word for the older grain might have been applied to the new-comer.

However, it seems unlikely that speakers of all the different languages in the family would have coincidentally decided to call the new grain the same thing. Once a language has split into two mutually unintelligable daughter languages, the speakers do not consult with each other about naming new phenomena.

The unlikeliness of this hypothesis increases when we realise that we must also assume that the words for shucking corn and corn riddle originally applied other actions and objects, and that once again widely separated people have coincidentally chosen the same words for actions and objects associated with the new grain.

I therefore conclude that presence of a word for corn in Proto-Muskogean constitutes a genuine conflict between the linguistic and archaeological data.

I wish all historical linguists were so forthright about the difficulties involved in trying to correlate linguistic and archeological evidence.

Oh, and I learned a new word, lamb’s-quarters (Merriam-Webster’s Collegiate, the copyeditor’s bible, hyphenates it; the AHD does not), the common name for Chenopodium album, a kind of goosefoot that M-W says is “sometimes used as greens” but AHD simply, and unkindly, calls a “weed”; it’s taahwa in Creek and taani’ in Chickasaw. You can get a USDA “plant profile” here, see some more pictures here [link dead as of 2017], and get ideas about collecting and eating it here [link dead as of 2012]. I’d still like to know how it got its striking name, though.

Comments

  1. I was smiling right through that excerpt; I only realised when I was fifteen or so that the “Corn” in “Corn Flakes” implied a specific grain, the old usage having stayed around on the east coast of the Atlantic.
    Of course the development of the meaning in English doesn’t have anything to do with the development of the word for Maize in Muskogean, but using ambiguous language [well, yes, it’s not, really, because it’s clear Broadwell is using American English–still some part of me is aware of the possible confusion] to describe ambiguous data is the sort of approach to saying nothing at all that I would laugh at from Flann O’Brien.

  2. Michael Farris says

    Is this the same paper where he discusses the problems of branches of Muskogean? I think it’s interesting how the four or seven languages of Muskogean are classified in more ways ways by different linguists than you might think possible for such a relatively small and transparently related family.

  3. Yes, it’s the same paper.

  4. Lamb’s-quarters: One theory says that the mature leaf looks like a cut of lamb’s meat––the quarter. An article at http://www.rawfoodinfo.com/articles/art_LambsQuarTreasury.html says another common name for it is “mutton tops”, which supports the looking like a cut of meat theory. Not sure how that fits in with looking like a goosefoot (Linneaus)or with its zillions of other common names: baconweed, cenizo blanco, dirtweed, dirty dick, fat hen, frost blite, hélunjóli, hvidmelet gåsefod, jauhosavikka …

  5. “However, it seems unlikely that speakers of all the different languages in the family would have coincidentally decided to call the new grain the same thing. Once a language has split into two mutually unintelligable daughter languages, the speakers do not consult with each other about naming new phenomena.”
    Not necessarily, cf the problems with the word for ‘whiskey’ in Iroquois languages (all ‘cognate’ compounds meaning ‘fire-water’). Basically the diferent groups decided on the same word, a compound of ‘fire’ and ‘water’, which were both terms cognate between the different languages. That way you get a nice set of regular correspondences of an item that obviously post-dates the breakup of the family. The same is true for “washing machine” in Yolngu.

  6. Does he happen to mention the Dravidian-origins theory?

  7. OED: 1773 J. HAWKESWORTH Voy. III. 442 We also once or twice met with a plant like what the country people in England call *Lamb’s quarters, or Fat-hen. Evelyn (1699) does not mention it (or any of its other common names).
    The meaning, I would guess, comes not from the part of the garden where the lambs hang out, but from some no longer obvious visual similarity between the leaf and the hindquarters of a lamb either living or dead.
    On preview (long lunch), I see Janet has already brought up this theory. To lamb’s quarters, goosefoot, and fat hen, add pigweed for 4 domestic animal common names.

  8. Also, I would take all the archeological dates for corn reaching a particular latitude with a considerable amount of salt.

  9. Yes, corn could have been known before it was widely grown.
    The linguistic time depth estimate is based on Swadesh’s standard rate of change, but I think I’ve heard of cases where this is known to give a very wrong estimate. Also, Johanna Nichols suggests languages with head-marking typology may destroy cognate evidence faster than dependent-marking languages that make up most of the familiar large Old World families.

  10. Chenopodium/goose foot was actually a domesticated plant in the eastern US. It is now a common weed, but 2kya it was domesticated for larger seeds and was used extensively along with marsh elder and others as a major cereal until maize became important.
    Archaeologists only have a few direct data points for the introduction of maize into Eastern US and it is very possible this date may change as new finds get dated. The best evidence is that it arrives circa AD 700, but it could have arrived earlier. At the beginning, it was used as as a minor food crop. Despite its unimportance as food early on, it appears to be ceremonially very important and is found in some interesting ritual contexts. This might be the clue to unravelling the linguistic mystery.
    BTW, our best evidence is that betwen 1000-1200 AD maize becomes an important staple crop and begins to be used for 30-80% of the caloric intake. Our evidence for this comes from carbon isotpe analysis of skeletal remains and it is pretty unambiguous.

  11. Patrick, are those dates just for the South-East US or for elsewhere too?

  12. Edward Sapir, in the 1920’s, connected the Muskogean Indian languages most closely with Hokan-Siouan. Other linguists since then have felt that Sapir’s assesment was premature and have suggested Algonquin and even Mayan as closer relatives instead. I’m partial to Algonquin myself but still, nobody really knows.

  13. Where an item or technology diffuses gradually from one group to the next, it seems quite plausible that the ones to whom it is new will calque the term used by those from whom they have acquired it, since it is very likely that they will understand their neighbors’ language. This process will lead to pseudo-cognates whose actual time-depth is less than it appears to be.
    Also, a comment on Hokan, which is a group that I have studied. Nobody but the most extreme long-rangers takes Hokan-Siouan or any form of Macro-Hokan seriously anymore. There just isn’t decent evidence for them. Even core Hokan is considered unproven, and very likely not a family. Some of the “core Hokan” languages probably are related, but others are doubtful. In particular, it is fairly widely thought that Chumashan does not belong with the rest.
    This is an interesting case in that there has never been any good evidence for Hokan. Hokan was cobbled together with tiny bits of evidence, almost all of it nothing but words that vaguely resemble each other. The sets of putative cognates were mostly tiny, regular sound correspondances non-existant. Hokan is to a large extent a relic of a period of unfettered speculation.

  14. Thanks, Bill — I love hearing from people who are actually au courant with subjects where I have to depend on whatever snippets I happen upon.

  15. Clarie – Those dates are definiely for the southeast, which I know best, and for much of the eastern U.S at least as far north as Ohio and Maryland. It would have been later for the northen part (New England, Great Lakes area) because maize required some genetic changes to adapt to the shorter summers.

  16. Marianne Mithun in The Languages of Native North America (which is a fun book, Hat, you should consider getting a copy) doesn’t mention Hokan-Siouan at all, let alone any relationship between it and Muskogean. Hokan, Macro-Siouan (including Caddoan and Iroquoian) and Gulf (which hypothetically includes Muskogean together with various other Southeastern languages) are listed as “stocks”, in the sense of hypothetical related groups above the level of the family. Mithun is skeptical of all of them, particularly Gulf. Of Hokan she says

    Evaluation of the Hokan hypotheses remains problematic. The antiquity of Hokan would be at least as great as that of Indo-European, if not much greater, but documentation of the languages is considerably more limited. Many of the languages are spoken in contiguous areas, […] The prolonged contact among speakers of many of the languages makes it difficult to distinguish true cognates from early loans. Furthermore, some languages proposed as Hokan seem to share more features with languages considered outside of Hokan than with others within Hokan (Haas 1964b). Hypotheses of a Hokan stock as a genetic unit continue to play an important role in prompting investigation of the historical relationships between these languages, but it should be recognized that Hokan is not yet considered a demonstrated genetic entity.

  17. My mummy told me when I was little that they were called ‘lamb’s quarters’ because they were soft like little lambs. Folk etymology is grand! Lamb’s quarters were called ‘melde’ in Saxon, which is thought to be the root of such place names as Milden and Melbourn (Cambridgeshire). The plant was a staple food in Europe since prehistoric times – according to my well thumbed copy of Richard Mabey’s -Food for Free-, ‘. . . the seeds formed part of the last, ritual gruel fed to Tollund Man.’ The young leaves are quite tasty, rather like spinach.

  18. Long ago Mary Haas connected Algonquian to Muskogean in a grouping she called ‘Macro-Algonquian’ (funnily, it wasn’t called ‘Macro-Muskogean’). No one takes this grouping seriously anymore, especially not Algonquianists. The two families are NOTHING alike typologically, save for being head-marking.
    I thought there were basically TWO theorized groupings of Muskogean — one that has Chickasaw/Choctaw as the first branching, as opposed to the other that has Creek/Seminole as the first branch. Everyone seems to agree that these are the two ‘ends’ of the family, with Koasati/Alabama and Mikasuki intermediate between them.
    The presence of a reconstructible word for corn in Proto-Muskogean doesn’t surprise me much. If archaeology has corn present in the southeast at AD 700, it probably was known a few centuries before that. At 500 AD the Muskogean languages probably weren’t that different yet, and they were probably still geographically pretty close together, so it seems entirely possible that some Muskogean dialect borrowed the word and passed it to all the others. If none of the dialects had undergone any of the relevant diagnostic sound changes yet, this borrowing would be indistinguishable from an inherited Proto-Muskogean etymon.
    A similar analog from Algonquian is that Proto-Algonquian has a cleanly reconstructible word for ‘seal’, as in the animal. However, most people are rather uncomfortable locating the Proto-Algonquians anywhere where there were seals — the only candidate locations would be Hudson Bay, Lake Ontario, or the Maritimes, none of which work well for various reasons. So the compromise position now is that most researchers think the word arose at a later time after PA had broken up (but not by too much), and that the word was simply passed around among different sister dialects/languages.

  19. David: Koasati and Mikasuki at opposite ends of the spectrum? From my readings, Koasati is closest to Alibamu, then Appalachee, and then Mikasuki. Mikasuki is often described as a language descended from Hitchiti. This passage at Access Geneology says http://www.accessgenealogy.com/native/tribes/creek/creekhist.htm
    “The people speaking the cognate Hitchiti and Koasati were contemptuously designated as “Stincards” by the dominant Muscogee. The Koasati seem to have included the ancient Alibamu of central Alabama, while the Hitchiti, on lower Chattahoochee river, appear to have been the remnant of the ancient people of southeast Georgia, and claimed to be of more ancient occupancy than the Muscogee.”
    A former student of Mvskoke, I have had a soft spot for Koasati for decades. It’s one of the most robust of Native American languages – 98% of the tribe in Louisiana still speak it. Hope they weren’t hit too bad by the latest Hurricanes.
    Also,there were outlying languages like Natchez which are classified as tentatively within Muskogean. Natchez was spoken into the 1940s by the Sam family, with whom Mary Haas worked. Watt Sam’s son, Archie Sam, a noted Creek traditionalist and keeper of the flame for Natchez identity, continued searching for speakers of Natchez until his death in 1996.
    Here is a comparative vocabulary of Atakapa, Tunica, Natchez and Chitimacha: http://www.native-languages.org/famgul_words.htm
    Compare to Muskogean: http://www.native-languages.org/fammus_words.htm

  20. Oops! Sorry David, I misread your post concerning Koasati / Mikasuki. Apologies. And my sources equating Natchez with Muskogean are out of date… that’s the problem of living 1000 kilometers from the nearest decent English language library.

  21. As for Muskogean, what I meant was that the greatest distance in the family is clearly between Chickasaw/Choctaw on one hand, and Creek/Seminole on the other. Koasati/Alabama and Hitchiti/Mikasuki are intermediate between them, and share several isoglosses with both.
    Natchez MIGHT be very distantly related to Muskogean, but it’s not ‘a Muskogean language’. It’s typologically rather similar to Muskogean, but that doesn’t necessarily mean anything. It’s very hard to find convincing cognates between Natchez and Muskogean. I personally think Natchez is just an isolate.
    There’s a very nice sketch of Natchez by Geoff Kimball (speaking of Koasati) just published earlier this year in the Scancarelli/Hardy volume ‘Native Languages of the Southeastern United States’. Considering he worked entirely off fieldnotes from seventy years ago, he managed to figure out quite a lot.

  22. Epazote is a Mexican chenepodium closely related to goosefoot/lambsquarters. It’s traditionally added to beans as a pot herb, where it’s believed to be an anti-flatulent.
    If you’ve got a decent mercado in the area, you should be able to pick up some dried epazote year-round. (I’ve always found it in one of those cellophane packets hanging near the dried whole chilis.) Try adding it to a big pot of pinto beans. It’s got kind of a funky/musty taste that’s off-putting at first, and then you start to want more. “Oh, that’s strange,” you’ll say. “I don’t think I like it. Maybe one more taste….”

  23. Yeah, I used to eat epazote when I lived in Astoria, Queens, with its booming Mexican population. Ever had huitlacoche? Sounds so much better than “corn smut”! Tastes good, too, and not like anything else I’ve ever eaten.

  24. Going Dotty in Kansas says

    Hmmm. So far the comments haven’t mentioned the *other* lambs-quarters, Trillium erectum, also known colloquially as bethroot, birthroot, wakerobin, Indian balm, Indian shamrock, squaw root, and ground lily. That such an intriguing common name would be applied to two such distinct plants demonstrates the usual failing of common names. Alas, all my sources are silent as to the origin/association with lambs and quarters.

  25. Gene Fellner says

    Unrelated languages in the New World? What happened to the paradigm of three waves of migration that was published with such fanfare ca. 1983?
    Amerind, Na-Dene and Eskimo-Aleut. (Or was one of those Athabascan?) Blood types, dental patterns, and linguistics aided by computer analysis was said to have made sense out of it and archeologists pinned down the last two migrations to 2000BCE and 4000BCE although the first might be any of several times before 12000BCE. And they all originated in practically the same neighborhood in what is now northwestern China.
    The second wave settled west of the Rockies and the third wave got stuck in the Arctic because the first wave was already well established everywhere else. And Kennewick Man couldn’t possibly be an ancestor of the Indians who live in Washington now because their ancestors didn’t arrive until 5,000 years after he died.
    This was all starting to make nice neat sense, and now you say that there are languages in North America that don’t seem to be related to any others? What happened to the three-migration model?

  26. Gene Fellner says

    BTW, you can grow your own epazote and save a fortune. Talk about a weed. If you can’t get the seeds online, buy the freshest stalks you can find and stick them in pots with Rootone. We live surrounded by redwoods with only a few hours of direct sunlight per day in the summer and the epazote practically took over our greenhouse.

  27. marie-lucie says

    GF: What happened to the paradigm of three waves of migration that was published with such fanfare ca. 1983?
    It made a splash among non-linguists, but is still as bad as before as concerns linguistics.
    Nobody contests the two, later Northen waves (Eskimo-Aleur and “Na-Dene”), which involve very obvious, very distinctive language families which have long been identified as such (except that hardly anyone still includes Haida under Na-Dene), but the people of the “first” migration from Asia (which probably occurred in “bursts” over a long period) were probably already diverse linguistically if not so much genetically: there are still many unrelated languages in China, there were probably many more of them in the past, and most of the migrants were probably not “Chinese”, although intermarriage through millennia would have blurred genetic distinctiveness.
    The “evidence” for “Amerind” presented by Greenberg and Ruhlen is so full of errors of both data and methodology that the results cannot be taken seriously by competent persons. My own feeling is that the currently accepted classification of the languages of the Americas into about 120 separate language families (including Eskimo-Aleut and the reduced Na-Dene) is probably wrong, and that the number of superfamilies comparable to Indo-European will turn out to be much smaller, with “Amerind” split into a number of smaller units. But there is still a lot of work to be done before these results become established on a firm basis.

  28. So much work, so little time, nobody wants to do it, and the main source of data perishes every day. There will come a time when we not only don’t know what went on, we won’t be able to know what went on.
    ~~ gloom, despair, excessive misery ~~

  29. John Cowan: it’s worse than you think. A peculiarity of the field of North American native linguistics which repeatedly struck me is that a disturbing amount of data on various languages is never published because it doesn’t fit into gate-keepers’ presuppositions and theoretical orientation.
    Or, of course, because nobody cares: a senior scholar in the field told me that there exists a reconstruction of the proto-language of an (accepted) North American language family, arrived at during a series of workshops some decades ago, which was never published because no academic press was interested. As a result, copies of the reconstruction were (when I was told this story, which was about three years ago) only in the hands of the three workshop organizers, one of whom was suffering from Alzheimer’s, one of whom was in prison, and the last of whom had mysteriously disappeared soon after his retirement.
    Combine the above with the extreme hostility felt by many speakers of these languages regarding any claim that their languages “came” from anywhere, the racial identity politics and associated games that come with the topic, and you have a disturbingly warped field.
    I heartily recommend a blog, “That moniyaw linguist” (moniyawlinguist.wordpress.com) if you (or anyone reading this: I suspect Marie-Lucie would like it) want to get a better idea as to what the field is like. Its author is an American in Canada who specializes in Plains Cree: if you go through the archives of you not only will learn a great deal about Plains Cree and other Algonquian languages, you will also learn a lot about some of the not-so-pleasant aspects of Academia and race relations in the Canadian West(and yes, I’m the Etienne who comments there).

  30. ~~ if it weren’t for bad luch, I’d have no luck at all ~~
    M-L, a Hokan question. In a soft, impressionistic way langauges like Chimariko and Atsugewi feel like their structures are like Caddoan and Iroquoian, and considering the distnaces, that makes you wonder. Maybe languages have kind of a structual trajectory that carries them in the same general direction even after they separate. And those two Cali languages feel pretty different from Miwokan, well, South Sierra Miwok at least, which have an almost SAE feel.
    I remember and old paper, maybe by Mary Haas or one of her first generation students, that compared verbal manner prefixes between Caddoan, and Siouan, as well as pronouns. Maybe it included Iroquoian. Long time ago. There is oral history to the effect that the Caddoans and the Iroquoians had a common ancestry, as some kind of entity.

  31. Trond Engen says

    Etienne: I like that blog. Thanks!

  32. marie-lucie says

    Jim, I am sorry but I know next to nothing about Hokan, except that “Hokan-Penutian” has always referred to a geographical not genetic grouping (the two groups have many representatives in Califonia). Miwokan is one of the Penutian families, and the Indo-European-like structures of some of these families were noticed by Sapir. I think that Caddoan and Siouan are often considered related, but I don’t know about Iroquoian. Mithun’s book would probably mention these controversies (Iroquoian is her major specialty).

  33. SFReader says

    —My own feeling is that the currently accepted classification of the languages of the Americas into about 120 separate language families (including Eskimo-Aleut and the reduced Na-Dene) is probably wrong, and that the number of superfamilies comparable to Indo-European will turn out to be much smaller, with “Amerind” split into a number of smaller units. But there is still a lot of work to be done before these results become established on a firm basis.
    I am quite sure that Greenberg never claimed that Amerind was comparable to Indo-European in age.
    Given Amerind’s dating to 14000 BP (or even earlier, if we include time in Beringia), it’s actually more than twice as old as Indo-European and should be compared with something more ancient, like proposed Nostratic or Eurasiatic.
    *I don’t believe Amerind theory, but on different grounds – it seems apparent that there were two waves of advance into prehistoric America. One, represented by Clovis hunters from Beringia and second, a bit earlier, from coastal hunter-gatherers. The second group appears related to the Pacific coastal peoples similar to Ainus or ancient inhabitants of Philippines and Southeast Asia.
    If at least some Indian language families are descended from the latter, their divergence from the other Amerindian groups could be measured in tens of thousands of years.

  34. marie-lucie says

    Etienne: a disturbing amount of data on various languages is never published because it doesn’t fit into gate-keepers’ presuppositions and theoretical orientation.
    One problem is that most journals want to publish articles that will support particular theories, especially the fashionable ones deriving directly or indirectly from Chomsky’s work. Graduate students are also encouraged to publish this sort of thing. “Raw” or almost raw data can seldom be made to fit into the theoretical straitjacket, and therefore they are of no interest to the journals, for which the ideal format is a lot of theoretical discussion with a sprinkling of supportive data for some facet of the theory.

  35. If at least some Indian language families are descended from the latter, their divergence from the other Amerindian groups could be measured in tens of thousands of years.
    And any relationship would now be irrecoverable.

  36. marie-lucie says

    SFReader: I am quite sure that Greenberg never claimed that Amerind was comparable to Indo-European in age.
    I am not talking about age. Most of the established families included among the ones constituting “Amerind” are of the order of Romance or Slavic. Larger subgroupings (which would be of the order of Indo-European) are not generally accepted (rightly or wrongly) but Greenberg used them uncritically.
    Given Amerind’s dating to 14000 BP (or even earlier, if we include time in Beringia), it’s actually more than twice as old as Indo-European and should be compared with something more ancient, like proposed Nostratic or Eurasiatic.
    This recommendation might be true IF “Amerind” was a valid construct and the starred forms given by Greenberg or Ruhlen represented genuine reconstructions based on solid data and methodology. None of these assumptions can be entertained given the huge number of errors and the lack of a solid basis for the starred forms, which G and R say are NOT means as reconstructions (but in that case, one wonders what they are meant to represent). The Nostraticists (whatever one thinks of their work) do accept Indo-European classifications and Proto-Indo-European reconstructions and the methodology that led to them, and do not attempt to classify the Nostratic languages and to reconstruct Proto-Nostratic from a haphazard collection of modern forms similar to each other in most of the languages of Eurasia, without regard to phonological correspondences or to things like the strong possibility of borrowings.
    As for “dating” the non-existent “Amerind” (even you admit you don’t believe in it) I don’t think for a moment that migration into the Americas stopped for ever (or would still be impossible if it had not been for Columbus) after the land bridge was flooded over about 14,000 years ago. The Asian coast of the Pacific is full of peninsulas and islands, which would have allowed people with even simple boats to follow the coastline, encountering very few long stretches of open water. The major North Pacific current also curves from Japan along the Aleutians towards Alaska, British Columbia and California, and brings all sorts of flotsam to North America, occasionally including wrecked fishing boats from Asia. There have been fishing cultures with boats for millennia along the Asian coast, and during that period there have been climate fluctuations, the warmer times affording relatively safe and pleasant conditions for sea travel. If the indigenous inhabitants of Taiwan were able to use their boats to go South and ultimately (over a few thousand years) spread all over the Pacific Ocean, is it totally unlikely that some of them might have followed the coast and the currents North and East and entered the Americas, only a few millennia ago? And other peoples could have done the same at different times.

  37. marie-lucie says

    oops: … the starred forms, which G and R say are NOT meant as reconstructions…

  38. SFReader says

    Regarding pre-Columbian contact post-separation. There are two accepted cases – Eskimo expansion circa 1000 AD throughout Arctic Canada all way to Greenland(from Bering sea area, but probably from American side. But they came there from Siberia another millenium earlier)
    And Na-Dene expansion, which is much older, 6000-5000 BC, probably.
    I wouldn’t exclude possibility of others, but so far haven’t seen any evidence

  39. –The Asian coast of the Pacific is full of peninsulas and islands, which would have allowed people with even simple boats to follow the coastline, encountering very few long stretches of open water.
    The problem is that the Aleutian islands were inhabitted continously for 9000 years (and current Aleuts appear to be their direct descendants).
    This would make very hard for any incoming tribes to use this 2000 km long island chain to cross from Asia to America.
    In 18th century, it took about 50 years for Russians who had firearms, artillery and European sailing ships to conquer the archipelago.
    Hard to see how any Stone Age tribal group could have done anything like this

  40. marie-lucie says

    SFR, Why do you need to involve conquest? We are not talking about armies, but more likely small groups of people. A hunting or farming people needs land to provide its livelihood, and existing occupants can be in the way, but a fishing people only needs a small allotment of land as a base from which to go to sea, in a region where fish and shellfish are abundant enough to be more than sufficient for the local population. For examle, for centuries people from Western Europe went to Newfoundland every year to fish for cod, yet the fishers themselves were not the ones to impose governments on the native people, nor did they make war with the natives or with each other for access to the fishing grounds, which in those days seemed inexhaustible. Even in the modern period, the arrival of a few Europeans in Africa or Asia or the Pacific Northwest did not necessarily lead to conquest (at least at first): like the Phoenicians and Greeks in the Mediterranean, the newcomers first established trading posts, some of which later evolved into cities, but they did not try to conquer the hinterland: peaceful relations were essential to commercial success.
    As an example (at a later period) of Polynesians travelling to America, there is the case of the Polynesian-type Chumash canoe, on islands near Los Angeles. An archeologist (Terry Jones) and a linguist specializing in the Chumash language (Kathryn Klar) have established that the distinctive canoe-building technique used in one of the islands, together with a few associated words, must have come from Polynesia. According to them, a single Polynesian individual could have been sufficient to introduce the Chumash to the technique and the words, although that seems less likely than one or more boatloads of travellers. The boatbuilder(s) would have had to live among the Chumash for some time (and to learn the language) in order to have such an impact, but there is no evidence that the interaction was not peaceful. Going back to the Aleutians, a group of Asian migrants could have stayed in a small section of the first island, later moving to the next island, etc, without “conquering” the island chain, instead moving East from time to time until they reached a place on the mainland where they could settle permanently. I don’t mean that one small group would have eventually peopled the continent (where humans were already established), since others could have followed the same route from time to time.

  41. SFReader says

    —Why do you need to involve conquest? We are not talking about armies, but more likely small groups of people. A hunting or farming people needs land to provide its livelihood, and existing occupants can be in the way, but a fishing people only needs a small allotment of land as a base from which to go to sea, in a region where fish and shellfish are abundant enough to be more than sufficient for the local population.
    I’ve read enough 17-18-19th centuries accounts of peoples of far Northeast Siberia, Alaska, Aleutian islands, Pacific Northwest, etc to understand sheer imposibility for these warlike peoples to engage in a benevolent behaviour of the sort you describe.
    Any small group of people arriving to these waters would have been immediately killed or enslaved (or perhaps even worse if some contemporary accounts of ritual cannibalism are believed)

  42. marie-lucie says

    SFR, we are talking about hundreds, possiblly thousands of years ago, in places which were mostly sparsely populated. Why do you think there are so many small pockets of unrelated languages along the West Coast, in places where there are few possibilities for settlement because the coast is so rugged? Those warlike people you are talking about also traded with each other and intermarried, in addition to fighting. Fighting was suspended during the fishing season, sometimes allowing several tribes to use the same fishing grounds (whether freely or in return for some payment). And new slaves with valuable skills making them indispensable can survive in a new environment, eventually blending with the population (perhaps that was the Chumash case). Also, a single fishing boat drifting away from home would indeed be vulnerable, but (like the Polynesian ancestors) deliberate migrants probably travelled in groups, including families, and carried weapons as well as food and other supplies, so instead of being killed by the inhabitants of a new place, migrants arriving at, say, a small cove might themselves have killed the few people they found there, and claimed the place for themselves.

  43. I’ve read enough 17-18-19th centuries accounts of peoples of far Northeast Siberia, Alaska, Aleutian islands, Pacific Northwest, etc to understand sheer imposibility for these warlike peoples to engage in a benevolent behaviour of the sort you describe.
    Come on. Those peoples became “warlike” at least in part because of the behavior of the Russians they encountered, who immediately tried to kill or enslave them. As m-l says, we know nothing about the behavior of the indigenous populations millennia before the Russian conquest, and it is ludicrous to extrapolate from much later historical accounts.

  44. Etienne, I apologize, but I accidentally deleted your comment — please repost it.

  45. Jim: I believe Bill Poser once left a comment on a thread here at Casa Hat to the effect that scholars no longer believe in Hokan as a language family, and regard what (non-coincidental) similarities there are as being due to language contact.
    As for Caddoan, Siouan and Iroquoian: all three have been suspected of being related. Based on what data I have seen (I am a historical linguist, but not an Americanist, so take this with a (modest) grain of salt) the case for a Caddoan-Iroquoian genetic relationship seems stronger than the case for either being related to Siouan.
    SFReader: the Aleutian Islands may indeed have been inhabited for 9000 years or thereabouts, but the very fact that a single language, Aleut, itself part of a larger family (Eskimo-Aleutian), was spoken throughout the Aleutians shows very clearly that the inhabitants of those islands did not remain in “splendid isolation” for 9000 years, and for all we know they may have shifted languages repeatedly. For all we know perhaps they spoke a Penutian language before shifting to Proto-Aleut (I mention this possibility because the distribution of Penutian languages does suggest a maritime spread).
    Marie-Lucie: of course, the migrations which brought Eskimo-Aleutian and Na-Dene to the New World needn’t have been the only later migrations to the New World. But such migrations might not have left new languages or language families: after all, no Polynesian language was durably transplanted to the New World, but this is no argument against the claim of Polynesian-Chumash contact.

  46. Polynesian-type Chumash canoe
    I actually read “Chuvash” for “Chumash” here and was thoroughly croggled. The Chuvash speak a highly aberrant Turkish language and live in Central Russia!
    The Nostraticists (whatever one thinks of their work)
    Nostraticists are probably wrong; Greenbergians are not even wrong. This is, as Mark Twain said in a different context, the difference between lightning and the lightning bug

  47. -the Aleutian Islands may indeed have been inhabited for 9000 years or thereabouts, but the very fact that a single language, Aleut, itself part of a larger family (Eskimo-Aleutian), was spoken throughout the Aleutians shows very clearly that the inhabitants of those islands did not remain in “splendid isolation” for 9000 years, and for all we know they may have shifted languages repeatedly.
    I wonder about mechanism of language change in such remote era.
    Russians conquered Aleuts, reducing their population severalfold in the process, enslaved the men and had sex with their women (so thoroughly that modern Aleut “Y-chromosomes were characterized to haplogroups of mostly Russian, Scandinavian and Western European origin (approximately 85%), which is in stark contrast to the 3.6% of Aleut mtDNA lineages identified as non-Native American, and thus indicating a large degree of asymmetrical gene flow between European men and Aleut women.”)
    And despite all this, Aleuts did not switch to Russian language (though if Russian rule continued for another century, they might have done this)
    So I ask again, how the tribal group speaking proto-Aleut language managed to force their language on aboriginals? Was their conquest even more drastic?

  48. — Those peoples became “warlike” at least in part because of the behavior of the Russians they encountered, who immediately tried to kill or enslave them.
    I disagree strongly. We have plenty of suggestive info about pre-contact indigenous societies – fortified villages, interclan and intertribal warfare (including acts of genocide), competition over hunting, fishing or grazing (in case of reindeer tribes of arctic Siberia) grounds, widespread slavery (and in the Pacific Northwest even something resembling a slave-based class society)
    All this happening without any Russian or European influence.
    —-As m-l says, we know nothing about the behavior of the indigenous populations millennia before the Russian conquest, and it is ludicrous to extrapolate from much later historical accounts.
    That’s true. But I think the opposite assumption
    is even further from truth.

  49. Marie-Lucie:
    All examples you list would not result in transfer of whole languages from Asia to Americas.
    The last one (armed band killing another local small group and settling there) looks possible, but not on the scale necessary (five thousand kilometers of island and coast hopping from Asia to America killing locals on the way – sounds scary, but unrealistic)

  50. marie-lucie says

    Etienne: For all we know perhaps they [the Aleuts] spoke a Penutian language before shifting to Proto-Aleut (I mention this possibility because the distribution of Penutian languages does suggest a maritime spread).
    I totally agree about the Penutian languages, since most of them are strongly correlated with rivers emptying into the Pacific. (“Penutian” is not generally considered a genetically homogenous group, although I believe it will eventually be recognized as such).
    … no Polynesian language was durably transplanted to the New World, but this is no argument against the claim of Polynesian-Chumash contact.
    Polynesian languages are spoken in the Pacific, practically all of them on small islands. They are part of the very large Austronesian family, the cradle of which has been determined to be Taiwan, on which around 15 different (but related) languages are attested. It was from Taiwan that the ancestors of the Polynesians departed in boats, starting several thousand years ago and eventually colonizing most of the lands found there. All the maps of the Austronesian expansion show sea routes leading South from Taiwan, but it is unlikely that no Taiwan aborigines ever ventured into other directions, specifically North and Northeast, at least at times of favourable climactic conditions. If any of these Northerners made it to America and left some kind of linguistic legacy, one would expect this legacy to come from Northern and Northeastern languages rather than from the ancestor of the Polynesian languages, which was presumably in the South of Taiwan.
    The Polynesian-Chumash contact which I mentioned above is not thought to have involved more than a very few men (even just one man has been potulated, but I think this is unlikely), and the only traces of their passage (thus far) are the “Chumash canoe” (which is not typical of all of the Chumash populations, most of which use a different craft) and a handful of words. This contact must have been relatively recent (a few centuries at most) for the Polynesian words to still be recognizable.
    [The book I cited is “Polynesians in America: Pre-Columbian contacts with the New World”, ed. by Terry Jones and others, Altamira Press, 2011. There are 12 contributors, mostly archeologists, and the contacts documented are mostly from South America, plus the Chumash case. The editors deliberately chose to concentrate on factual evidence (archeological and linguistic) of specifically Polynesian contact, and not to discuss more controversial topics.]
    SFR: All examples you list would not result in transfer of whole languages from Asia to Americas.
    A small community migrating (not necessarily all at once), bringing advantages of some kind (technology, stronger social organization, etc) at a time of favourable climactic conditions, could prosper and therefore attract natives to itself. Their language, learned by others, would become altered, but there would still be (at least for a time) traces of its origin. If conditions in the country of origin were such that emigration from one area was common at certain times, there could have been periods of one-way travel bringing more and more migrants along the same route.
    The last one (armed band killing another local small group and settling there) looks possible, but not on the scale necessary (five thousand kilometers of island and coast hopping from Asia to America killing locals on the way – sounds scary, but unrealistic)
    I don’t mean that one small group of newcomers would systematically kill natives along their way (the local population would certainly resist), only that this possibility might have existed in some cases and at least some of the migrants would have come prepared for hostile encounters. Of course, a single small group would not have covered thousands of miles at one go, or subsisted with their language intact for very long, but if there was a succession of small-scale migrations including families (or at least women) coming from the same Asian region, there could have been eventually a large enough core population that they would have extended their territory, and their language could have survived and even spread, although influenced by the local language(s).

  51. Indeed. Consider all the speakers of the Taishan dialect of Cantonese in all the Chinatowns of the world: the result of “small-scale migrations” that “have extended their territory, and their language [has] survived.”

  52. Another curious case of language survival
    The village settled by only two ethnic Russian males who came with their Eskimo/Aleut wives and mixed children. And the entire village ended up becoming native Russian-speakers (until 1950s, anyway).
    But I suspect their experience was uniquely 19-20th century – literacy, school, Russian Orthodox Church, etc.

  53. Jim (sorry, I should have written this in my earlier comment): I smiled at your reference to Miwok as being almost SAE. You will be pleased to hear that none other than Edward Sapir himself, in sketching out the main features of Penutian (to which Miwok belongs), pointed out that Penutian was oddly similar, typologically, to… Indo-European.
    SFReader: language shift in far northern locations is by no means rare. Yakut, in Siberia, spread (quite recently) because its speakers had mastered the art of breeding horses in this cold climate, which gave them a major advantage over speakers of other Siberian languages. Evenki, at an earlier date, had spread throughout Siberia because of its speakers’ mastery of reindeer breeding/riding.
    Social organization can be the driving force too: the expansion of Proto-Cree throughout Northern Canada (from the Rocky Mountains to Labrador: one of the most geographically extensive cases of pre-modern language spread) after 500 AD or so (meaning that Proto-Cree expanded over territory that had been inhabited for millennia) has been argued (I’ve a reference, should anyone want it) to be linked with Proto-Cree speakers’ extensive family/social network, which facilitated long-distance trade: hence Proto-Cree-speakers’ neighbors were quite anxious to marry/have their children marry these newcomers (Proto-Cree speakers) or their children in order to gain access to this trade.
    It would have taken just a few generations for Cree to replace the earlier languages under such circumstances. Especially since Proto-Cree would have been the only common language for all these local groups wishing to trade via the extended trade network created by these Proto-Cree-speaking newcomers…
    And indeed this is how language shift takes place most of the time: not with a group of armed and hated invaders suddenly showing up on your doorstep some unfortunate morning, but with some new group you come into contact/start trading with. Knowledge of these newcomers’ language becomes desirable if not vital for some reason(s) or other, leading adults to ensure that their children acquire it, with the next generation taking the new language and the advantages deriving from its knowledge for granted, leading the next generation to think of the “new” language as theirs, which in turn leads to the next generation having this “new” language as their first language, eventually leading to the original language being forgotten.
    More broadly, “extreme” environments (tundra, desert…) are ones where language shift/spread frequently occurs. I think this is because in such environments establishing/maintaining long-distance trade can be a life-or-death matter, because the local ecology is so limited.
    Hence the spread of Yakut and Evenki in Siberia, of Eskimo-Aleut in the far North of the Americas, of Tuareg Berber in the Sahara, of Uto-Aztecan over the American Southwest/Sonora desert, of Iranian and later Turkic over Central Asia, of Tibetan over the vast Tibetan Plateau…
    Conversely, it is telling that extreme linguistic diversity (indicating that there hasn’t been any language spread in a long while) is found in places like (pre-modern) California, New Guinea, the Caucasus: that is to say, areas with very rich local biological resources. Local groups could and did deal/trade with their neighbors, of course, but tellingly, it never was a life-or-death matter: a small tribe in any of the above environments didn’t *need* outside trade as a matter of survival.

  54. A very interesting analysis; I’ve never thought of it like that.

  55. marie-lucie says

    Etienne, excellently put.

  56. SFReader says

    —some new group you come into contact/start trading with. Knowledge of these newcomers’ language becomes desirable if not vital for some reason(s) or other, leading adults to ensure that their children acquire it
    Do you really need to learn a foreign language for trade?
    Those 18-19th century accounts tell me that no, successful trading is quite possible with both parties speaking not a word in language of a partner.
    Sometimes a trade jargon would develop if contacts were long-lasting (see Chinook jargon), but in most remote places people wouldn’t even need that.
    Now, long-term tribal alliances involving extensive intermarriage – that would create a need to learn another language!
    But I suppose this requires a full tribe suddenly showing up, not small groups of people.

  57. I thought Etienne was describing the spread of English!

  58. And they all originated in practically the same neighborhood in what is now northwestern China.
    This year there was news that the ancestral home of the Native Americans was traced to the Altai Mountains.

  59. Maybe this is a better article.

  60. marie-lucie says

    SFR: a full tribe suddenly showing up, not small groups of people.
    In Polynesia there were exploratory voyages by small numbers of people, who returned home to report, and those early voyages were followed by larger-scale voyages by families bringing with them food, supplies and even animals such as dogs and chickens. Although current thinking about voyaging in the North Pacific tends to imagine only very small-scale, usually involuntary trips, some of the voyages must have involved much larger numbers.

  61. Bathrobe: from what I understand the Altai Mountains aren’t so much the homeland of the ancestors of the first Natives of the Americas as an area whose genetic commonalities with Natives of the Americas is due to both being “peripheral”, geographically, from an Eastern Siberian perspective. In effect both preserve, genetically, what must have been (12 000 or so years ago) the “typical” Siberian genetic profile.
    SfReader: a common language is indeed required if a *long-term trading relationship” is what you are after: extensive negociation and trust-building are rather difficult enterprises without a common language.
    Incidentally, I strongly suspect that the spread of Indo-European followed a dynamic much like what I sketched above. For some reason people believe that, if Indo-European was the language of the first humans who domesticated the horse (which seems likely), it follows that Indo-European spread as the language of a horse-riding warrior elite.
    This always struck me as dubious: in the Americas groups such as the (post-contact) Mapuche or Sioux became first-rate horse-riding warrior cultures, and ultimately succumbed to European expansion because horses were the least of the advantages Europeans enjoyed.
    By contrast, we are expected to believe that Proto-Indo-European was spoken by a warrior culture whose sole advantage over its neighbors was its domestication of the horse. And that it overran everything in Eurasia from Ireland to India, without anyone on this vast territory learning anything about horse domestication before it was too late. The examples of the Sioux and the Mapuche seem to indicate that this is a rather unlikely scenario.
    It seems to me that the spread of Indo-European is better explained if we consider that horses, combined with wagons/wheeled vehicles generally, would have caused an economic revolution wherever they were first introduced, allowing trade on a large scale between groups which would previously have remained comparatively isolated from one another. And in the context of this massive expansion of trade Proto-Indo-European would have been the obvious lingua franca.

  62. Wagons and chariots both, judging from Greek and Irish epic.

  63. Wagons and chariots both, judging from Greek and Irish epic.

  64. marie-lucie says

    Etienne: I like your interpretation of the IE spread. According to what I have read (I don’t remember where), domestication of the horse for drawing vehicles preceded its use for riding, especially riding for military purposes. (Occasional riding, eg by children, must have occurred, but that is different from training horses – and riders – specifically for those purposes).
    The reconstructed vocabulary of PIE has words not only for horse but for cow, sheep, pig, goose, animals which have much more limited mobility than horses. Warriors on horses can travel far and fast, cows can travel far but not fast (witness the pastoral nomad cultures of Africa), and pigs and domestic geese don’t like to travel at all. Mounted warriors could not possibly have taken any of those animals with them, but horses drawing wagons could have walked along with people and cows, and carried pigs and geese on long trips, as well as less mobile people (children, elders, etc) and whatever was intended for trade.
    The ancient Greeks and Irish had both wagons and chariots, but that was long after the breakup of Indo-European.

  65. Marie-Lucie: Nevertheless, words for wheeled vehicles and their parts are reconstructible to PIE; they tend to be about chariots in Greek and Celtic and wagons in the other branches.

  66. Marie-Lucie, John Cowan: just to be clear I am not denying that Indo-European may indeed have been the language of a horse-riding warrior culture (perhaps making use of war-chariots). What I am saying is that this is irrelevant to the dynamics of the spread of Indo-European which I proposed above. That is to say, the introduction of horse-drawn wagons would have triggered an economic revolution which would have favored the spread of Indo-European.
    If my guess is right, then whether Proto-Indo-European was originally the language of a warrior culture or not is unrelated to the history/dynamics of its spread.

  67. Etienne: Oh, I agree. But economics, we might say, is the continuation of war by other means.

  68. The other Polynesian-American contact that’s now pretty much beyond doubt is the sweet potato, which spread from the Americas as far west as New Zealand. It used to be thought that this was natural transmission, but sweet potato seeds don’t survive immersion in salt water, and DNA work shows that the Polynesian varieties descend from the cultivated plant, not from one of its wild relatives. What is more, the Polynesian name kumara (in Pascuan and Maori, ‘uala in Hawai’ian), is suspiciously close to Quechua k’umar(a). (However, the sweet potatoes of the Philippines and other non-Polynesian areas were brought there by Europeans.)

    Nature news item.

  69. Unfortunately, the Polynesian origin of the Chumash canoe has now been thoroughly debunked: the Chumash canoe arose because of the need for seaworthy boats that could reach the mainland from the (California) Channel Islands. There is the possibility that some South American chickens, of which we have only the bones, were Polynesian in origin — but new DNA research has apparently debunked that too: it was never more than marginal.

    So that leaves only the sweet potato as tangible evidence of contact.

  70. marie-lucie says

    Etienne: … Edward Sapir himself, in sketching out the main features of Penutian (to which Miwok belongs), pointed out that Penutian was oddly similar, typologically, to… Indo-European.

    The word “Penutian” was first coined by the anthropologists/linguists Kroeber and Dixon for a group of five language families of California (which had several other such groups), and Sapir’s typological description fits four of those language families. Two features in particular are reminiscent of IE languages: verbs with several stems depending on tense and mood (as in Greek and Latin) and noun declensions indicating cases. As Sapir expanded his classification out of California to other languages, he expanded the name Penutian to more languages, first Northward into Oregon, Washington and British Columbia, and later Eastward along the Columbia River and the neighbouring plateau. But he does not seem to have changed the overall description, which is less applicable to the more Northern languages.

    Sapir’ first inkling that the “Penutian” languages of California were not an isolated group came when he looked at these languages after working with Takelma, formerly spoken mostly in Southern Oregon but also right across the California border, which he had researched with the last speaker (writing an impressive grammar of the language). Indeed Takelma verbal structure is very similar to that of the Miwokan and Yokutsan languages of California, while most of the other Oregon languages Sapir classified (at least tentatively) as Penutian are quite different from them in this respect.

    Much work still needs to be done, mostly on the Oregon languages, which are all extinct. I have been chipping away at it for some years.

  71. David Marjanović says

    Back in 2005…

    Even core Hokan is considered unproven, and very likely not a family. Some of the “core Hokan” languages probably are related, but others are doubtful. In particular, it is fairly widely thought that Chumashan does not belong with the rest.
    This is an interesting case in that there has never been any good evidence for Hokan. Hokan was cobbled together with tiny bits of evidence, almost all of it nothing but words that vaguely resemble each other. The sets of putative cognates were mostly tiny, regular sound correspondances non-existant. Hokan is to a large extent a relic of a period of unfettered speculation.

    The second paragraph is quoted and cited in this presentation, which finds a core Hokan grouping to be about as old as Indo-European. The same author has published this paper on, wait for it, comparative paradigmatic morphology.

    Quotes from the paper:

    “My own position is that the genetic relationship between most languages usually subsumed under Hokan is highly likely, and that the existence of the Hokan family can be taken as a working hypothesis, subject to further proof or refutation.”

    “For the purposes of the present paper, the following languages and families will be regarded as Hokan: Karuk, Chimariko, Shastan, Achumawi-Atsugewi (Palaihnihan), Yana, Pomoan, Salinan, Yuman, Seri, and Oaxaca Chontal (Tequistlatecan). I suspend my judgment with regard to poorly attested languages/families like Esselen, Coahuilteco, Comecrudoan and Cotoname. Another poorly documented language, Cochimi, is generally considered as related to Yuman (Mixco 1978). Following Kaufman (1989), I do not accept a Hokan affiliation of Chumashan. I also do not include Washo and Tol (Jicaque), although these languages might be related to Hokan on a deeper level. Washo was regarded as Hokan by virtually all supporters of the Hokan hypothesis. However, it has so few reliable matches with the rest of Hokan in the basic lexicon that its membership in the Hokan family seems improbable. If Washo and/or Tol will ever be shown to be related to Hokan, it will only be through comparison with reliable Proto-Hokan reconstructions, rather than with isolated morphemes in individual Hokan languages.

    While reconstructing tentative Proto-Hokan forms, I will use the system of sound correspondences proposed by Kaufman (1989: 84–93).”

    Apparently Hokan is not so hokey after all.

  72. I have to say, a lot of those word lists at the end do not look like they show convincing correspondences to me, but of course I know nothing about these languages. I hope someone who does will respond.

  73. David Marjanović says

    Or just ignore the word list at the end of the presentation for the time being and go straight to the morphology in the paper…

  74. Trond Engen says

    Yes, I don’t understand why he bothers with lexicostatistics at all. The morphology is much more convincing. But it’s also five years ago, and you’d think there would be progress (or lack of it) on comparison of lexical roots since then. Maybe there are so few people working on Hokan that new papers could just as well be published by SETI.

  75. Apparently Hokan is not so hokey after all.

    “Not so hokey” is a good assessment. Zhivlov’s papers clarify some things, but they are still a first step. Contact is a huge issue, which he doesn’t address: consider Numic adopting an elaborate system of instrumental verbal prefixes, following its neighbors. The hither/thither morphemes which Zhivlov discusses, the clearest system in the paper, could in part have spread by contact.

    I do find it encouraging that Zhivlov is willing to keep Washo out. Pickiness is a virtue.

    Maybe there are so few people working on Hokan that new papers could just as well be published by SETI.

    That is exactly right. Zhivlov’s papers are the first to be published on comparative Hokan in a long time, though a bit of work is being done on individual recognized families.

    While on that subject, there’s a remarkable new paper (the first of two): Amy Miller, Phonological developments in Delta-California Yuman, IJAL 84(3), 383, 7/2018. Quote:

    I argue below that Kumeyaay is not a single language, nor even three languages, but rather a much larger grouping: a major subdivision within the Delta-California subgroup, encompassing two main branches and at least six languages. I further propose that “Cocopa” is not a single language but a second subdivision within Delta-California Yuman, comprising not just Arizona Cocopa but also Cucapá and one or more languages no longer spoken. I argue that the little-known speech varieties Ko’alh and Kwʔaƚy are the surviving representatives of a third Delta-California subdivision, Kw’ally. The Delta-California subgroup, I conclude, has considerable internal complexity. Long-neglected Delta-California Yuman languages stand in urgent need of recognition and documentation.

    In other words, there is language documentation that has been done very recently and can still be done on these Hokan languages, all within one or two hours drive of the California-Mexico border. Miller finally concludes,

    This suggests that the Yuman family is not broad and shallow but rather narrow and deep, and no doubt much older than the 2,000 years previously proposed.

  76. David Marjanović says

    I don’t understand why he bothers with lexicostatistics at all

    Moscow School: very basic vocabulary (in particular the more stable half of Swadesh’s list of the 100 most stable meanings) is considered stronger evidence than morphology. After all, it works between modern IE languages that have almost no morphology in common, like English and Farsi.

    Zhivlov’s papers are the first to be published on comparative Hokan in a long time

    Other than this reference, I guess:

    Kaufman, Terrence. 2015. Some Hypotheses Regarding Proto-Hokan Grammar. Retrieved from https://www.albany.edu/ims/pdlma/2015%20Publications/Kaufman-some%20hypotheses%20regarding%20protoHokan%20grammar-revd2015.pdf.

    Kaufman’s other cited paper is indeed from 1989.

  77. This is an updated version of his 1988 paper. Like that one, it has reconstructions, but not the source materials for these reconstructions, presumably for lack of space. Zhivlov has gone back to the sources (and some more up-to-date ones) to retrace Kaufman’s steps, among other things.
    Kaufman’s notes are at AILLA, in Texas, and have been getting organized and digitized over the past several years (the guy is enourmously prolific.) Supposedly they will be available online soon.

  78. David Marjanović says

    Yay!

  79. More info here.
    The Hokan notes are for some reason password-restricted.

  80. Kaufman’s revised version of the 1988 paper is here, not the earlier link, which is a separate paper.

  81. David Marjanović says

    From 2017…

    Unfortunately, the Polynesian origin of the Chumash canoe has now been thoroughly debunked: the Chumash canoe arose because of the need for seaworthy boats that could reach the mainland from the (California) Channel Islands.

    Likewise unfortunately, the link no longer works. It leads to a Berkeley site that holds enormous numbers of texts and audio files in and about languages of California, including Chumash, but I can’t find that paper.

  82. David Marjanović says

    Oh! I should have thought of that. #Neuland

  83. I’ve read articles about the idea of South American/Polynesian contacts and the genetic evidence. Only yesterday in reading a reply to a David M comment at Language Log did I learn that the words for sweet potato in Hawaiian and various other Polynesian languages are cognate (or strikingly similar) to those in several S. American languages.

  84. David Marjanović says

    The great big paper on the population genetics of Pacific sweet potatoes is in open access.

  85. Thanks. That includes a brief reference to the Polynesian and Quechua sweet potato words.

  86. The Chumash canoe paper is officially here, and at academia.edu.

  87. To complicate matters, this paper uses genetic evidence to support non-human introduction of the sweet potato into Polynesia.

  88. Thanks @Y

    Discussion is continuing at the Log. From your ref:

    our data strongly suggest that the presence of the sweet potato in Polynesia predates human colonization of the region by thousands of years and consequently is most probably due to long-distance dispersal,

    I think this is the paper that is widely discreditted, as having used idiosyncratic gene analysis and suspicion of contaminated/denatured samples. “Thousands of years” is not plausible.

    @Ryan the words for sweet potato in Hawaiian and various other Polynesian languages are cognate (or strikingly similar) to those in several S. American languages.

    ‘Cognate’ is not at all the same thing as “strikingly similar”. ‘Cognate’ means we know when, where and how a word travelled. In its travels it would be subject to regular sound changes (including the receiving speakers adapting/mis-hearing it to fit their language’s sound patterns). So — almost by definition — it would not be “strikingly similar” in sound.

    BTW the people whose pronunciation is so strikingly similar live a long way up the Andes; there’s no evidence they went anywhere near the coast. So we have to posit Polynesian sailors who sailed orders-of-magnitude further across the Pacific than any other voyages as at that time (c. 1100 CE, when settlement had only reached Samoa/Tonga, maybe Cook Islands) _and then_ climbed orders-of-magnitude higher than any mountain in their homelands _and then_ learnt of the crop and how to cultivate it, and its name _and then_ come all the way back with only a few roots/seeds and a single word. There’s other S.America languages on the coast that have words probably cognate to *kumara, but their pronunciation is not strikingly similar. So now we have to posit sound changes by Polynesians from the littoral speakers’ version that amazingly landed right back at the Andean pronunciation.

    That remarkable alleged sound-alike — and that these similarities are being touted by people who know nothing about language transmission [**] — to me is the best evidence _against_ it being a cognate. (That’s “best” out of an exceedingly poor proffering.) Plus that there is no solid evidence of how the speaker interaction is to have happened. Plus (and with all due respect to the amazing skills of Polynesian sailors/navigators) I know enough about deep-sea voyaging to plain disbelieve anybody got that far without a desalinator.

    Anyhoo, since discussion is continuing at the Log [***]. Let’s not darken Hat’s door with it. (He’s already made his position clear.)

    [**] I include there Prof Mair, who seems to enthusiastically endorse all sorts of crackpottery about long-distance language and cultural exchange.

    [***] My posts there keep being blocked. So far they’ve eventually got unblocked. But my latest (chasing up David M’s cites) not yet. With Hat’s permission, I might post it here if it doesn’t appear tomorrow.

  89. You mean where Mair suggests Polynesian Moa= fowl may be related to Galician Moa=millstone/gizzard?

  90. There’s other S.America languages on the coast that have words probably cognate to *kumara, but their pronunciation is not strikingly similar.

    As I recall, most of the languages of the Pacific coast of South America are long extinct, and several are entirely undocumented. Arguments ex silentio appear rather weak in such a context.

  91. Anyhoo, since discussion is continuing at the Log [***]. Let’s not darken Hat’s door with it. (He’s already made his position clear.)

    No, no, I’m loving it — it’s right in the LH wheelhouse, and I’m not following the Log thread any more (Mair exceeded my irritation threshold). Pray continue!

  92. Lars Mathiesen (he/him/his) says

    exceeded my irritation threshold — me in 2015 or so, that’s why I’m here innit. (I don’t bear many grudges, but those I do I take seriously).

  93. David Marjanović says

    Sweet potatoes growing wild all over central Polynesia for thousands of years?

    You mean where Mair suggests Polynesian Moa= fowl may be related to Galician Moa=millstone/gizzard?

    He doesn’t; he “just” says he doesn’t know what to make of it.

    As I recall, most of the languages of the Pacific coast of South America are long extinct, and several are entirely undocumented. Arguments ex silentio appear rather weak in such a context.

    Also, didn’t the 2013 paper actually find one on the coast of Ecuador?

  94. David Eddyshaw says

    where Mair suggests Polynesian Moa= fowl may be related to Galician Moa=millstone/gizzard

    Obviously both are loans from Western Oti-Volta, e.g. Dagaare noɔ, Kusaal nua “fowl.”
    For the n/m change, cf Kusaal nyɔɔr “nose” = Moba miɛl “nose.” These things happen all the time.
    The ocean-spanning exploits of the Western Oti-Volta speaking peoples are legendary.

    My posts there keep being blocked

    I foolishly attempted to comment a while back on Mair’s post on Kusunda, where he was credulously repeating the BBC claim that Kusunda lacks negatives, and my comment was deleted within minutes. It was subsequently reinstated, and made the subject of another post by Mark Liberman. I’ve no idea what’s going on with that site these days, but it surely put me off any further interaction with it. I can’t take the drama. I think that when Mair (as often) vaguely suggests something without actually outright endorsing it as fact, if you contradict it, he gets upset at your uppitiness while simultaneously reserving the right to bridle at the supposed implication that he actually did endorse it.

    (The Kusunda thing is actually based on a grain of truth, mirabile dictu. Various well informed Hatters commented about it on Liberman’s post, to good effect.)

  95. I like the level of discussion here better than on LL, pardon my boosterism.

  96. That’s OK, so do I.

  97. That LANGUAGE LOG thread simply gave me a headache. Sigh. There are some sane people there, one of whom observed that Japanese “arigato” could all too easily be mistaken for a borrowing from Portuguese “obrigado”, especially since there are known instances of Portuguese loanwords in Japanese (we know for a fact that “arigato” is not one of them). Mair’s reaction to this sensible observation reminds me why I have neither left comments there nor read anything there (except if attention is drawn to a particular post by someone here at the Hattery) in nearly a decade.

    Something nobody seems to have pointed out is that Polynesian languages have an amazingly small number of phonemes and syllable types, so that the odds of finding a coincidental look-alike in a Polynesian language for a word with the same or a similar meaning in some other language are disquietingly high (I think I once pointed out here that the same is true of Basque).

    What I find very suspicious is that:

    A- The similarity of Polynesian and Quechua */kumara/, which has been known/noticed in scholarly circles for some time, does not seem to have spawned any attempt by Polynesian and Quechuan linguistics scholars to seek to reconstruct the various segments of the various proto-forms; and

    B- No other instances of alleged borrowed vocabulary shared by Polynesian and Quechua have been proposed by anyone.

    Now, if it could be shown that the Polynesian forms are exactly what would be expected of a loan from Quechuan at the relevant time period, and/or that other words within this same semantic field likewise fit in with our understanding of Polynesian and Quechuan linguistic (especially phonological) history, then the case for this instance of pre-Colombian trans-continental experiment in trans-cultural culinary cultural appropriation (no less evil for being unlinked to white supremacy-sorry, it is hard not to parody this sort of garbage when you are exposed to it daily) would be strengthened. Considerably.

    Now, the fact that nobody has tried to do so may be indicative of the catastrophic decline of serious historical linguistics, but I wonder whether some scholars (with the relevant qualifications) have taken a look at it (both */kumara/ and possible other cognates) and quietly concluded that it is fatally flawed in some fashion. Since nobody likes to learn that the punch has been spiked, it may be some time before this negative conclusion is published.

  98. Etienne: Sure there is a Polynesian reconstruction!
    I don’t know about Quechua, but the word might be a Wanderwort anyway, borrowed by the putative Polynesians from some coastal language.

  99. David Marjanović says

    …and indeed, cumal is attested in a coastal language of Ecuador, so there’s no need for Polynesians to climb up into the Andes. They could well have landed, found one thing worth taking home, took it and the word for it.

    The 2013 paper linked above – open access! – is not just about genetics; it presents the rest of the accumulated evidence as well.

  100. I second the comments about Language Log; thank goodness for this place!! I will repeat the plea I made there for someone to check out this paper if they can:
    Adelaar W.F.H. (1998), The name of the sweet potato. A case of pre-conquest contact between South America and the Pacific.
    I obviously don’t know exactly what’s in it, but it could well be the place to find answers to questions about dates and forms for the Proto-Quechuan word and its descendants.

  101. Thank you @everybody, then I’ll continue (speaking as the obrigado/aragatō guy. My post at the Log did get unblocked, and DavidM made a sensible response to it.)

    didn’t the 2013 paper actually find one [a language] on the coast of Ecuador?

    Yes, that’s the Cañari language with word *[kumal]. that DavidM says could only have been borrowed as */kumara/.

    Ref Etienne’s question/Y’s chart, once */kumara/ got into Proto-E-Polynesian at CE 1100, the sound changes to the daughter languages are all perfectly regular — including some ‘reverting’ from /r/ to /l/.

    What I haven’t seen is any commentary that *[kumal] fits the sound pattern of the applicable S.American languages at the applicable time. Note it’s only a few S.American languages that have that word (as at today), and all of them have a more common word for the darn potatoes. IOW can we reject that the word travelled _from_ Polynesia to S.America much later (probably by European agency)?

    @Y borrowed by the putative Polynesians from some coastal language.

    _If_ it’s a borrowing, the chief problem I see with the sound-alike is the sounds are suspiciously close, given it’s hypothesised the borrowing was 1,000 years ago.

    And if it’s a borrowing, (why) was it the only borrowing? Of all the wondrous things you might find in S.American cultures at the time, were tubers the only thing you brought back? The Polynesians did/do value meat (pigs, chickens, they were at least partly responsible for killing off the last of the Moa in NZ — that’s a contentious comment, don’t broadcast it). Given how scarce is wildlife on the islands, you’d bring back protein, not starch surely!

    But my chief difficulties are with that ‘If”: There’s not a skerrick of material evidence of Polynesians in S.America. (There’s maybe suspicions of some human gene exchange, but very far from clear whether that’s at the applicable date, or after European contact.)

    And Polynesian expansion was by island-hopping. We know they didn’t get from Cook Islands to Aotearoa/NZ direct — much too far — because Kermadecs named ‘Rangitahua’ = the Stopping-off Place, and archaeological finds there. It’s more than double the distance from Cook Islands to NZ to get to the S.America coast from the nearest settled/visited Polynesian island . (But that was settled only much later/after the kumara got to more central Polynesia.)

    And the Polynesians were superb masters of navigation. Unlike stupid Heyerdahl, they started their voyages _against_ the known prevailing winds, expecting that if they didn’t make landfall, they’d get blown back home, because prevailing ‘innit. To get the huge distance to S.America would need the prevailing winds to not prevail for an extraordinary stretch of time.

    With long sea voyages, the chief constraint is carrying enough fresh water. You can catch fish, so you won’t starve. You can collect rainwater/overnight condensation (Heyerdahl experimented with that). You can stop off at islands with rainwater pools — but there’s no islands en route to S.America. You’d simply die of dehydration.

    Given how all the Polynesian cultures are great oral historians and myth-builders, with plenty of tales of remote ancestor lands and arduous sea voyages, you’d think somebody would mention travelling an order-of-magnitude further East than any settled island, and finding a land mass that clearly wasn’t an island, and bringing back a food staple.

    I’m going to dismiss the sound-alike as coincidence. (Perfectly respectable purely-Te Reo Māori derivation given on the Log. 😉 ) But how do I explain the crop getting West to meet the expanding Polynesians? It seems the tubers/their seeds wouldn’t survive that much salt water/they’d need to be floating on something. Speculation: perhaps the Quechua have something to do with it/that would match the timing: the Humboldt current flows Northwards up the coast of S.America, then Westwards towards the islands. There was trade from Quechua lands up to Meso-America, starting around CE 800. One/some of their balsa rafts got blown off course and drifted on the current, ending up somewhere in the Marquesas/(what is now) French Polynesia — also where Heyerdahl ended up, and where one of the gene-analysis papers guesstimates was the point of landfall. Polynesians exploring Eastwards in advance of settling found them, and spread them back.

    Of course that speculation (with no evidence behind it) doesn’t override my default position: we shall never know.

  102. Aargh! Now the Hattery is eating my posts. Mr Hat would you be so kind …

    @Andy, the Adelaar paper is here, to you only 30 Euros. Tantalisingly in the preview I can see Campbell’s 1997 warning

    Languages with relatively simple phonemic inventories and similar phono-tactics will easily exhibit many accidentally similar words (which explains why Polynesian languages, with very simple phonemic inventories, have been
    [cont p 94]

    @DavidM found one thing worth taking home, took it and the word for it.

    Chiefly, I don’t think Polynesians ever got there (see my reasoning in durance vile). But _if_ they did and interacted with the culture enough to find a word for a thing — only “one thing”? Really?

    Penelope to Odysseus: You go off to sea for months/years! I’d given you up for dead! You’ve met these people and heard their language. You come back all pleased with yourself, and all I get is bloody tubers!

  103. Andy, Thanks for the Adelaar reference (in Janse, ed., Productivity and Creativity: Studies in General and Descriptive Linguistics in Honor of Ε. M. Uhlenbeck).

    A few relevant points from that paper:
    kʰumara is a starchy variety of sweet potato, in contrast to the sweet apichu.
    — Sweet potatoes were grown in coastal areas and in tropical lowlands, not in the highlands.
    — Cognates of the word are used in Quechua varieties of tropical lowland Colombia, Ecuador, and NE Peru, where sweet potato is still cultivated. It is also used in Cuzco, in the highlands of southern Peru, whose inhabitants “entertain regular contacts with tropical areas which they have colonized in the past.”
    — Brand (1971) has previously suggested, apparently without evidence, that kʰumara is not a Quechua word but was borrowed from Cañari. Adelaar dismisses this, for lack of evidence, but now we know that it was in fact recorded in Cañari.
    — Adelaar argues that Quechua was spoken in central coastal Peru in the past.
    — Some Quechua varieties of tropical NE Peru use kumal and kumala. “There is no ready explanation for the use of l instead of r. It may be due to a local substratum, considering the fact that l was not originally found in Quechua.”

  104. Thanks @Y, and … the ‘ pre-conquest contact between South America and the Pacific’ part says?

    Adelaar seems to research only native American languages.

  105. @AntC: Adelaar doesn’t have anything informed to add regarding Polynesian contact. He assumes the dates for East Polynesian colonization accepted at the time, which are now known to be too early. He suggests that it’s more “economical” to assume a one-way trip from the continent rather than a round-trip from Polynesia. He speculates that if kumara were borrowed into Aymara, it would bear an initial stress, and wonders if that explains the initial stress on the word in Māori (it doesn’t: this is regular stress for Māori.)

  106. David Marjanović says

    But _if_ they did and interacted with the culture enough to find a word for a thing — only “one thing”? Really?

    Why not? They were looking for a place to settle and found one that was already settled, so they took some provisions and sailed home. 😐 There’s really no reason to think they stayed more than a few days.

  107. There’s really no reason to think they stayed more than a few days.

    Again, _if_ they ever got there, they’d be totally knackered and close to death. They’d have to spend a long time recuperating and restocking for an almost-as-arduous [**] journey back.

    You seem to think these guys are on some sort of Carnival Cruise. ‘Pacific’ is a total mis-nomer; the place is deadly.

    [**] Going back with the prevailing winds/currents wouldn’t be quite so bad. But you’d probably loop back [Humboldt current] a long way north of where you set out from. There’s 2,000km of just nothing to navigate by. Whereas with island-hopping, although you’re out of sight of land, there’s cloud patterns, bird flight paths, swell formations and that mysterious Te lapa light phenomenon.

    Sheesh! I can see I’m talking to a bunch of land-lubbers. Did Odysseus have an easy time navigating? And he was in a relatively well-frequented bathtub.

  108. Well, the Ioannidis et al. paper suggests that they stayed long enough to get some local passengers to come along for the return trip.

  109. Mr Hat would you be so kind …

    Freed from durance vile; thanks for letting me know.

  110. AntC: What is this supposed Maori etymology? I didn’t find it (but wouldn’t believe it if I did.)

    Words can match randomly, sure. They even more often almost-match. But how likely is it that the only Polynesian cultivar of South American coastal origin happens to have a perfect phonological match with one of only a handful of words for it in the languages of coastal South America?

  111. > Penelope to Odysseus: You go off to sea for months/years, and all I get is bloody tubers!

    Odysseus to Penelope: Gimme a few minutes. [Returns with a skyphos filled with sweet potato fries for their teenaged kids ]. Let’s see if they call them bloody tubers.

    Penelope [watching the kids eat]: All right maybe it was worth it, but still, I did miss you.

  112. More fun: a number of old prehistoric skeletons discovered on Mocha Island, off the coast of central Chile, show a feature called the rocker jaw, very common among Polynesians, rare elsewhere, and especially rare in the Americas. These were first noticed in the early 1900s. A 2010 paper by Matisoo-Smith and Ramirez further investigates them using modern osteometric techniques. I heard around that time that Matisoo-Smith, a pioneer in paleogenetics, was working on getting some genetic material out of those remains, but haven’t heard anything about it since.

    In other news, a 2014 paper analyzed genetic material from skulls of Botocudo people, of Minas Gerais state in eastern Brazil. Two of them showed purely Polynesian ancestry. (The skulls were destroyed in the 2018 fire at the National Museum.)

  113. @Y how likely is it that …

    … the word for — exact meaning — thank-you in Japanese is exactly how a Japanese ear would hear thank-you from Portuguese, especially since we know the Portuguese visited Japan, and left a few words there.

    … given both Polynesian and S.American languages have relatively small phoneme inventories (and neither has consonant clusters), there would be sound-alikes. I’d say highly likely there’d be plenty (see the “more thorough running of the numbers” link on LLog). Then the question is how likely one of those sound-alike pairs would also be meaning-alike?

    I don’t want to be culturally imperialist, but I suspect both Japanese and Portuguese have much larger vocabularies than Proto-Polynesian or Quechua. So the likelihood of meaning-alike is higher than finding a meaning-alike pair in Japanese/Portuguese.

    Then the question is how likely that (one of those) meaning-alike pairs happens to denote a foodstuff that travelled (by means so far unproven) from one language’s area to another?

    And _if_ that travelling was by human agency, how likely is it that the (unknown/unrecorded in myth) travellers would take only one item (and a pretty uninteresting item at that) and only one word, having been in contact with the source language/people enough to be given ?fresh from their fields, presumably having communicated enough that they were setting out on a long/arduous return journey for which they’d need plenty of food.

    Indeed why would the Quechua (supposing their generosity) give kumara? Why would the Polynesians take them? You can’t eat them raw. You can’t cook them on a canoe. They’re starch, not particularly nutritious [**]; you’d have to protect them from seawater; they’d take up space needed for fresh water. They’d just be a durned nuisance.

    How likely Polynesians got there in the first place, so this fairy-tale could happen at all? I’d say hugely unlikely. Even if they got blown off course in the Roaring Forties [***], and washed up on a beach and their canoe was repairable, none of their navigational skills would help/they’d be completely disoriented.

    So I’d like to balance against those unlikelinesses the likelihood of seeds/tubers travelling out by non-human agency. There’s a handy ocean current going from exactly the Quechua areas to islands we know were reached by Polynesians early in the Eastward expansion. And the Quechua were cultivating and transporting kumara at about the right time.

    What is this supposed Maori etymology? I didn’t find it (but wouldn’t believe it if I did.)

    It’s in Edo Nyland style, ‘ kū-umu-māra-mara ‘. No I don’t believe it either. Just an exercise to show how easy it is to find a purely Te Reo sound-alike/meaning-alike.

    [**] And ref Ryan, Kumara are indeed tasty when plied with enough fat. Just no means to kill suckling-pigs and heap the lot in earth ovens, in a canoe.

    [***] Note Shackleton/Worsley’s open-boat voyage through the Forties from Elephant Island to the Falklands ended up smashing the Endurance’s lifeboat. But then they knew it would be one-way; they knew there were islands with settlements out there [and British goddamit!]; they had a chart and a compass and a sextant.

  114. Richard Scaglion (2005) “Kumara in the Ecuadorian Gulf of Guayaquil?”, presenting documentation of the word for “sweet potato” in Cañari, is available here on academia.edu. Very enjoyable!

  115. exactly how a Japanese ear would hear thank-you from Portuguese

    A Japanese would presumably borrow obrigado as /oburigadu/ (in Modern Portuguese) or such, not /arigatoː/. Two phonemic vowels don’t match, plus one syllable is missing. Most of the classic examples of false cognates are likewise only near-matches, not perfect matches—and those were cherry picked out of many meanings in many languages.

    /kumara/ is exactly what you’d expect from a novel item being borrowed, along with its exact name, just like piñata or ramen.

    larger vocabulary…

    The Māori dictionary of Williams is about the same size as a standard dictionary of a European language, probably tens of thousands of words. I read about a project for a project of a monolingual dictionary of Tuvaluan (a language of several thousand speakers living on low atolls), which was to have some 40,000 entries. All languages have several thousand words used by fluent speakers, plus thousands more of specialized cultural and technical terms.

    likelihhod of picking up food…

    That has happened all over the world, especially among agricultural societies. You eat something, you like it, you figure you’d like to grow it yourself. That’s how potatoes and other American cultivars came to Europe. Sweet potatoes are nutritious and suitable for growing in tropical climates, like taro and yam which the Polynesians had already had. So they got some cuttings to take back. It’s speculative, but no more speculative than saying “aw, it couldn’t have happened.”

    How likely Polynesians got there in the first place

    They made it to Easter Island, after covering a long empty stretch of sea from Pitcairn and Henderson. They would be motivated to continue and seek more islands. Europeans took only 300 years since Magellan’s time to reach every island in the Pacific, and they had to come from further away, and had scurvy to deal with.

    On the other hand, South American navigators, while skilled coastal sailors, had no reason to think that there’s anything to explore beyond the horizon. That explains why the Galápagos and the Juan Fernández Islands were never inhabited by humans before the Europeans.

  116. @Y off the coast of central Chile [Mapuche tribes]

    is nowhere near the Quechua/*[kumal]-speaking areas. I agree genetic analysis would provide a much firmer basis of evidence. The osteometric analysis (certainly beyond my pay-scale) seems to be all hung around with ifs and buts.

    For the Botucudo skulls, the paper can’t rule out that the Polynesians arrived via “European-mediated transport”. Minas Gerais is Eastern Brazil. I’m not seeing how Polynesians would get there without European transport.

    I have a big problem with both these papers:

    The Mapuche paper’s cite to The possibility of Polynesian contact with North America, particularly in the area occupied by the Chumash tribe has been discredited.

    … long distance voyaging … [Mapuche paper]
    Computer voyaging simulations conducted by a number of researchers in the 1990s (…) demonstrated that voyaging from Polynesia to the Americas was feasible and during an El Niño event such a trip could be much faster.

    It has been established that the Polynesian Pacific expansion from Southeast Asia covered distances of thousands of kilometers, reaching New Zealand, Hawaii and Easter Island … [Minas Gerais paper]

    It’s as if they’re saying crossing the Pacific from anywhere in Polynesia to anywhere in the Americas is like crossing the English Channel or the Mediterranean. Get a map and look at them to the same scale. The Pacific is a third of the surface area of the whole globe, more than the earth’s whole land area, wider than North America, wider than Africa, wider than Eurasia.

    It is utterly misleading to talk about “long distance voyaging” as “thousands of kilometres”. Yes it’s thousands of km from Cook Islands to New Zealand. No Māori did not make that journey in one step. There’s no leg of the journey as far as 1,000 kms, let alone the 2,000 – 3,000 of open sea to get to S. America.

    If you get somewhere ‘much faster’ [than what? in what timescale? how much fresh water do you need to stay alive that long?] that means you’ll go back much slower. And without weather satellites you can’t predict when the weather systems will turn round.

    Or are these researchers accusing Polynesians of being as reckless as Heyerdahl and setting off on a one-way journey to possible oblivion. How would they know there’s a nice big North-South continental coastline to catch the canoe of corpses if nobody’s ever come back from it?

    I don’t think I’m being culturally imperialist and underrating the navigational skills of ‘primitive’ peoples. I’m giving them huge credit for being prudent and knowledgable, and respectful of the ocean. (Like my hero James Cook. Any genetic evidence Polynesians are related to bluff Yorkshiremen?)

  117. @AntC, Y: Cheers re the paper!!

  118. Computer voyaging simulations conducted by a number of researchers in the 1990s (…) demonstrated that voyaging from Polynesia to the Americas was feasible and during an El Niño event such a trip could be much faster.
    [from Mapuche paper]

    I’ve now chased up the cites elided there. (Or at least close enough: Irwin et al 1990 is behind a paywall, Irwin 1989 is in effect a preprint )

    Irwin & Flays 2015 has some sailing-nerd calculations, particularly re cross-wind navigation. Note no Polynesian exploration sets out down the prevailing wind. Hence they didn’t reach Australia.

    The attribution in the Mapuche paper is downright misrepresentation. Irwin says you _do not_ set off in an El Niño event, because you don’t know you’re in an event ’til it stops blowing. (The lack of weather satellites, again.) Since you don’t know when an El Niño might arrive, he concedes you might get caught up in one/be unable to escape/end somewhere unexpected. This is an accident, not a repeatable strategy.

    We of course don’t know how many Polynesian exploration voyages never returned. Irwin elaborates carefully a strategy for exploring in the most prudent directions first; learning about seasonal patterns whilst doing so; then gaining confidence in exploring more risky directions — especially by catching the tails of a passing weather system that’s blowing contra the prevailing wind. Since you’ve observed the pattern/time of year of those systems, you know they’re short-lived, you’ll be able to get home afterwards.

    Archaeological evidence appears to require that the sweet potato be carried from South America to eastern Polynesia in time to be transmitted to the margins …
    Or, less probably, there could have been an inadvertent voyage by a canoe trapped, rather than assisted, by an episode of El Niño.
    [Irwin 1989]
    If ever a Polynesian canoe reached South America, … [it could probably get home ok] . It is the voyage from Polynesia to South America that appears to present a much greater challenge.
    [Finney 1994]

    Both authors seem to me to be using weasel words: forced to speculate how Polynesians might have reached S.America, because kumara. This is not evidence.

    So to recap:

    * Yen 1974 (not a Philologist) in a botany paper makes a throwaway remark about a soundalike.
    * Nobody disputes the botany: it’s the same lineage in S.America and Polynesia — before Europeans.
    * A bunch of other non-Philologists take that soundalike remark as proof of human transmission of the plant.
    * Nobody makes any analysis of the likelihood of chance soundalikes;
    nor of possible pronunciation drift over ~1000 years;
    nor of a purely Polynesian genesis of the word.
    * Maritime historians are faced with a fait accompli to account for Polynesian navigation to S.America.

    I think the alleged evidence does not yet allow us to pass go. Do not collect $200. Certainly do not throw the dice to see if you can sail to S.America.

  119. For those no longer following the LLog thread (and no reason you would be), the tale has gotten more tangled.

    */kumal/ is a reconstruction in the Cañari language. That language is extinct and poorly attested. What informally gets called Cañari is sensu strictu Cañar Quechua, spoken on the coast and lower mountains pre-conquest. Quechua (various spellings, with shades of meaning) is the ur-language of the Inca empire of the highlands; original Cañari is reconstructed from a supposed substrate of the present-day coastal language.

    ‘Pre-conquest’? There were two conquests: C15-th16th Inca overran the Cañar civilisation — who’d been defending themselves valiantly for a few hundred years; the poor blighters got very quickly thereafter invaded by the Spanish.

    So inconveniently for our purposes, the Incas/Quechua got involved too late for the kumara-to-Polynesia story. I’ve tried looking for reliable vocabulary lists of pre-both-conquests Cañari. All paywalled. This might be helpful.

    So we know there’s two Quechua words for sweet-potato:

    apichu for the sweet varieties and cumara for the starchy. …
    The Quichua [**] language seems to have no inclusive term that can be applied to all kinds of sweet potatoes.

    — which to my mind rather rules out that either word is originally from Cañari [***], but all the maybe-evidence is paywalled.

    So presuming our hypothesised Polynesian visitors didn’t climb the mountains to get their root crops and word for them, I see no evidence that */kumal/ would be a word they’d encounter on the coast.

    [**] That ‘Quichua’ spelling is maybe-significant: here means the Ecuadorean variants.

    [***] or perhaps the starchy variety grows better on the coast, so the Inca didn’t adopt the thing/the word until they’d conquered the Cañari? And they regarded them as different vegetables: turnip vs swede/rutabaga, yam vs ‘sweets’.

  120. The Cañari word is recorded directly; see the reference Xerîb linked to.

  121. Thanks @Y (and Xerib) I see nothing there contradicting what I said.

    The key evidence is a report from a priest 1582. By then the Cañari people on the coast/lowlands spoke Cañar Quechua. I’d be looking for stronger evidence the priest was distinguishing pre-Quechua Cañari vocab vs strictly-Quechua than

    It is entirely possible that …
    … therefore presumably makes a special attempt …

    Furthermore this ref doesn’t seem to acknowledge the two words seem to refer to different varieties. “language seems to have no inclusive term” above. (Unless, as I hazarded, the different varieties grew in different places.)

    The priest does report [translated]

    As for vegetables there are great quantities of amaranth, cabbages and other vegetables. There are some small onions that the indigenes eat that are called zarayuyu.

    Now if I’d been at sea for many weeks, I would have killed for onions — and indeed any sort of green vegetables — never mind your tubers. Furthermore, onions travel well, amaranth as a seed also travels well. Why wouldn’t you take those? AFAICT all today’s Polynesian words for Alliums derive from English.

    I see that paper’s Figure 4 ‘Worldwide distribution of pre-Columbian sailing rafts’ [ref 1971]. This is complete fantasy beyond Peru-Ecuadorean waters themselves. Oh, and the wreck of Kon-Tiki. Polynesian rafts weren’t constructed of balsa (on account of the lack of balsa trees — or large trees in general), weren’t to Ecuadorian designs, weren’t seaworthy for more than paddling within atolls/small island groups, and would have had very poor upwind performance (not clear if they were wind-propelled at all). Polynesian construction for their great voyages was from two shaped/planked hulls lashed together — canoes, not rafts. The platform between the hulls is raised well out of the water, to try to keep precious stuff dry. (Kon-Tiki had the waves flowing through/between the logs. Nothing was dry.)

    I see the ref to Heyerdahl, along with TBH utter bollix about similarities of ‘primitive’ sails. wp:

    The lateen [sail] originated in the Mediterranean as early as the 2nd century CE, during Roman times,

    So it was the Romans who sailed to both S.America and Polynesia? No: if you’re going to sail upwind or even effectively cross-wind (as Heyerdahl didn’t), and you need to be able to raise/lower/shorten sail, you’d end up with something like lateen rig. It’s compelled by the ‘design requirements and constraints’.

    So I contend this paper is bending the ‘evidence’ over backwards to try to support Heyerdahl’s theory of Pacific exploration from S.Am. Weird: by 2005 Heyerdahl had been rejected.

  122. @Y Europeans took only 300 years since Magellan’s time to reach every island in the Pacific, and they had to come from further away, and had scurvy to deal with.

    This is a somewhat unfair comparison:

    The Europeans took Polynesian navigators with them, so they knew they were heading for an island, and how far away it was.

    The Europeans travelled mostly with the prevailing winds/they generally weren’t making return journeys — or rather they were returning to Europe by continuing downwind round the globe. Not, for example, attempting to travel both ways between Easter Island and the central E.Polynesian islands.

    Not that I’m belittling Cook and the other early explorers: you can count me out of rounding Cape Horn in a coal barge.

  123. While I’m clearing the decks …

    another preposterous tale of Polynesian seafaring — this one to Antarctica. Drawn from a perverse translation/European record of oral history.

    (Apologies if I posted that already.)

    The crucial word is huka, a general Polynesian word for foam/froth/white stuff on the sea. There not being a lot of snow in the Polynesian homelands, Māori in need of a word adapted ‘huka’ for that, and for hail/ice generally.

    Then a colonial surveyor/ethnologist willfully translated ‘huka’ on the sea to ice = icebergs.

  124. You can count me right out of any debates about pre-Columbian navigational feasibility, but if pre-Columbian contact were to be ruled out, I will say that the scenario of an early post-Columbian borrowing into Polynesian languages mediated by Spanish ships or the like strikes me as far more plausible than the idea that the Polynesians happened to come up with the same name for the same originally South American species as Quechua speakers purely by coincidence.

  125. I’d be looking for stronger evidence the priest was distinguishing pre-Quechua Cañari vocab vs strictly-Quechua

    This is a cogent criticism of Scaglion’s paper.

    In the passage from the Relaciones Geográficas de Indias, the author also mentions “otra raiz que se llama racacha”. This must be arracacha (Arracacia xanthorrhiza), or in Quechua raqacha. He goes on to say, “Hay unas cebolletas que los naturales comen, que se llama zarayuyu”. This term can be understood as a Quechua “weed of maize fields” or the like : Quechua sara ‘maize’ and yuyu ‘weed, herb, green plant’.

    (I wonder what these cebolletas were, exactly. I have read that the genus Allium has very low diversity in South America, or was even absent before the introduction of cultivated members of the genus. Some related plants of the genus Nothoscordum are edible and used locally like onions in South America, apparently. Perhaps it is one of these. I would be interested in learning about pre-Columbian Allium cultivation in South America.)

  126. Lameen: There’s archeological evidence for early (pre-Columbian) sweet potato cultivation in East Polynesia. Linguistically, *kumara came in early enough to participate in regular sound changes, specifically *k>ʔ in Tahitian and Hawaiian and *r/l>ʔ in Marquesan.

    AntC, Xerîb: fair point about Cañari but I think Scaglion’s essential point was that the word was used near the coast, which had not been shown before.

  127. I think Scaglion’s essential point was that the word was used near the coast, which had not been shown before.

    Agreed. I was wondering, for example, whether the coexistence of Quechua kumar, etc. “starchy and less sweet, white-fleshed sweet potato variety”, and apichu “sweeter sweet potato variety, sweet potato in general” might reflect that fact that the first word was a borrowing from a coastal language as a term for variety more recently introduced to upland areas. (For the meaning distinction, see here, p. 405.) Apparently, the words apichu and tuctuca are used in Aymara for “sweet potato”.

  128. David Eddyshaw says

    I just this minute came across the word doku “poison” in the Waama language of the Atakora in Benin. It is clear that this must be the source of the totally identical Japanese word. I told you that the seafaring exploits of the Oti-Volta peoples were legendary …

  129. I will say that the scenario of an early post-Columbian borrowing into Polynesian languages mediated by Spanish ships or the like strikes me as far more plausible than the idea that the Polynesians happened to come up with the same name for the same originally South American species as Quechua speakers purely by coincidence.

    I’m gonna have to sentence you to read through a list of such coincidences until you cry uncle.

  130. @Y (and Xerib) I think Scaglion’s essential point was that the word was used near the coast, which had not been shown before.

    It was used near the coast as at 1582, in an area by then heavily influenced by Quechua. So some 70 – 100 years after the Inca had defeated the Cañari.

    What we need is a use near the coast as at ~1000 in ‘pure’ Cañari. A lot of language change can happen in 500 years in a war zone.

    I appreciate evidence for that’s gonna be impossible, due to the absence of priests who could write anything down. I couldn’t find much of (reconstructed) word-lists for Cañari — especially enough to see if its sound-pattern differed from Quechua in ways that would tell whether */kumal/ was more likely Cañari or Quechua (or Cañari substrate under Quechua).

    Or I would entertain evidence */kumal/ wasn’t used in highland/Inca Quechua. Perhaps the */kumal/ variety didn’t grow so well in the mountains.

    What might tell us is the paywalled Urban (2018, Wiley online library), ref’d from wp on ‘Cañari language’, link in my post way above.

    My main point remains: I just don’t believe any Polynesian craft could or would get across that much open ocean (especially at such an early date). And my secondary point remains: the Humboldt current and prevailing winds from Ecuador would bring anything floating nicely into (what is now) French Polynesia/pretty much where Kon-Tiki ended up — so within reach of Eastward-exploring Polynesians at around the right timing to pass into the daughter languages.

  131. What might tell us is the paywalled Urban (2018, Wiley online library

    Matthias Urban (2018) “The Lexical Legacy of Substrate Languages: A Test Case From The Southern Ecuadorian Highlands” is an inconclusive attempt to establish Cañari as a Barbacoan language. It does not mention sweet potatoes and—as far as I can tell from a quick read—does not otherwise touch on topics directly related to the origin of the Polynesian word for “sweet potato” or the diffusion of the crop in the Pacific. “Sweet potato” is not one of the 30 or so possible substrate words in Cañar and Azuay Quechua that he attempts to compare to words in Barbacoan languages.

    Among scanty materials documenting the Barbacoan languages that I can access, Bruce R. Moore (1966) Diccionario castellano – colorado, colorado – castellano gives len as the Tsafiqui (Colorado) word for “sweet potato”. Randall Q. Huber and Robert B. Reed (1992) Vocabulario comparativo. Palabras selectas de lenguas indígenas de Colombia give ũth as the Páez word. These are not obvious comparanda to the form comal ~ cumal from Cañar, of course. (Besides these, John N. Lindskoog and Carrie A. Lindskoog (1964) Vocabulario cayapa just give camute (doubtless from general Spanish camote) for the Cha’palaa (Cayapa) language, and Huber and Reed (1992) also give pattatta (doubtless general Spanish batata) for Awa.)

  132. David Marjanović says

    And my secondary point remains: the Humboldt current and prevailing winds from Ecuador would bring anything floating nicely into (what is now) French Polynesia/pretty much where Kon-Tiki ended up — so within reach of Eastward-exploring Polynesians at around the right timing to pass into the daughter languages.

    Well, that explains how the return trip was possible if there was enough drinkable water on the boat. If getting caught in an El Niño event explains the trip in the other direction…

    We’re not talking about coconuts. The tubers can’t just float in the sea on their own.

  133. Meanwhile I delight in learning that Red is a Barbecue language.

  134. I remember wondering, circa 1990, why we were reading excerpts from Kon-Tiki in middle school language arts classes. I knew at that point (although I’m pretty sure none of my classmates did) that Heyerdahl’s thesis had been totally debunked.

    In terms of the spread if species, there really does seem to be a huge amount of chance involved. For example, I was actually discussing with one if my sons recently how rats spread all over the world. Two successive species of rats have become economically (and medically) important vermin only as a result of human activity. For millennia, black rats and brown rats were confined to south Asia, in spite of their being no barriers other than mountains keeping them there, until Indian Ocean trade spread them to the rest of the world.

  135. The tubers can’t just float in the sea on their own.

    Seeds can survive, as long as they are on top of something to float them, and they would end up in Central East Polynesia. That would require sweet potatoes growing on the beach in America, seed capsules landing on something to float them, sprouting on the beach on their destination island, and surviving as a wild population until people arrive.

  136. I’m gonna have to sentence you to read through a list of such coincidences until you cry uncle.

    If you have forms in two languages which match perfectly in form and meaning, where the forms are not too short or sound-symbolic, where the term is for an introduced material or cultural item, where one or both forms don’t have a native etymology, and where there’s a plausible historical explanation for a contact—then borrowing is the default hypothesis.

    (And since Lameen’s bread and butter is contact linguistics, I would defer to him on that.)

  137. I’m gonna have to sentence you to read through a list of such coincidences until you cry uncle.

    Bring it on then: how many such coincidences can you come up with simultaneously involving:
    – 5 or more perfectly matching successive segments (3 of them consonants, even!)
    – perfect semantic match (at least at the species level)
    – a referent known to be novel to one of the speech communities involved, and long-familiar to the other?

    You don’t find anything near that level of specificity in Mbabaram:English dog, let alone Spanish mucho:English much or any of the other standard examples. If it were a monosyllable I’d already be a lot more open to coincidence as an explanation.

  138. And,
    – with no known etymology in the speech community where it is novel, whose historical linguistics is well understood.

  139. OK, fair enough. But I’m still willing to assume coincidence unless I see a more convincing historical chain of borrowing.

  140. What do you mean by “a chain of borrowing”?

  141. I mean clear means of transmission, like Latin-speakers coming to Britain (known fact) and sharing their words.

  142. Well, here you have an American cultivar showing up as a Polynesian cultivar. It had to come from one to the other. The only thing in the way is whether you think human transport is much less likely than accidental transport. I think the former is far more likely.

  143. January First-of-May says

    Did the native Ecuadoreans use balsa wood rafts back in the 12th century? Perhaps there really was a raft with sweet potatoes (and/or seeds thereof) on board (for whatever reason, maybe for trade up the coast) that ended up blown off coast due to a storm or something, approximately repeated Kon-Tiki’s path, and landed in what is now French Polynesia where (possibly much later, if the landing went well enough) a band of Polynesian explorers happened to be visiting. There could even have been living Ecuadoreans still there as well who could have explained what it was that they were carrying…

    Certainly sounds more likely that a Polynesian canoe somehow making it all the way to Ecuador (or even Chile) against the winds, and then all the way back again. Of course either way involves lots of coincidences but I think this direction has less of them.

  144. @Y with no known etymology in the speech community where it is novel,

    Is it novel? Do we have prior etymologies for any Polynesian words as at a thousand years ago?

    As I pointed out in the LLog thread (see at ‘Nyland’), ‘kū-‘ is a known prefix; ‘māra’ and ‘mara’ are known words. ‘umu’ is a known word: earth oven where you cook kūmara. Also ‘uma’ is a known word, ‘ara’ is a known word, ‘rā’ is a known word (sun AMOT) and particle (as in ‘haere rā’ = goodbye). So *[kūmara] perfectly fits expected sound patterns AFAICT (IANAP), no need to posit a borrowing.

    I’m pretty sure I could find soundalikes for all those subsequences in Cañari (if we had a full set of vocab, thanks @Xerib for looking) and/or Quechua. So out of all those chance soundalikes, what’s the probability one of them is also meaning-alike?

    Ok this is probably saying no more than that the Polynesian sound inventory is relatively small, open syllables, no consonant clusters, yada yada. And S.Am native languages similarly, it seems.

    There is a Māori name ‘Raukūmara’ for a range of mountains in the North Island. None of my resources will give me an etymology. The word also means a specific shrub, endemic to NZ. ‘Rau’ is a word in itself or can be used in compounds/wordbuilding.

    … Oh and my third point is there’s no Polynesian mythology/oral history about an epic Eastward journey to a land mass that clearly wasn’t an island. With people already on it; and who restocked their wāka and spoke a strange language. Polynesians are not reticent about vaunting their travels, nor reciting their lineages back to ancient heroes. Now kūmara aren’t so important to northern/tropical Polynesians, but Māori value them hugely because they’re one of the few starchy crops that grow successfully in temperate climes. They would have preserved the myth.

    @David M The tubers can’t just float in the sea on their own.

    The plants are usually propagated by stem or root cuttings, says wp. But they also produce seeds. “True seeds are used for breeding only.” — not being a horticulturalist, I’m not sure what that’s telling me.

    So birds eat the seeds(?) And follow the prevailing winds and poop them out on some island. (See for example the wide distribution of Frigatebirds even unto S.Am. west coast and all across the South Pacific, several species now extinct — whose flight Polynesians followed to get to Rapa Nui. ” frigatebirds drink freshwater when they come across it, by swooping down and gulping with their bills.” — so a puddle in a kumara field.)

  145. A bird that doesn’t poop for 2,300 miles?

  146. Stu Clayton says

    As long as it doesn’t poop out on its journey, there should be some seeds left over at the terminus.

  147. Why wouldn’t it poop on its journey? Not saying I know, but that seems extraordinary. Downwind as the crow flies, this is a 100-hour flight for a frigatebird, but there’s no reason to think they’d up and decide to fly straight there.

    As for coincidence, I think the bigger coincidence is frigatebirds never managing to propagate the wild plant across the Pacific, then suddenly dropping cultivar seeds in the same millennium that the Polynesians began their voyages.

    I’m also skeptical that a balsa could make its way across the Pacific on its own without submerging its cargo in saltwater by the end. Do kuumara thrive after submersion? Do they grow within reach of a wave toss?

  148. David Marjanović says

    a Polynesian canoe somehow making it all the way to Ecuador (or even Chile) against the winds

    That’s where El Niño would come in handy.

    ” frigatebirds drink freshwater when they come across it, by swooping down and gulping with their bills.” — so a puddle in a kumara field.

    Way too small, I’d think.

  149. Stu Clayton says

    I wrote “as long as it doesn’t POOP OUT”. That means “doesn’t run out of steam”, not “eschews pooping”.

    How Often Do Birds Poop? (SURPRISING Answer! + FAQs)

  150. Polynesians didn’t travel long-distance by rafts, balsa or otherwise. They traveled in wooden canoes, carrying people, chickens, pigs, and dogs. The Hōkūleʻa did many 1000-mile trips (some with chickens and pigs), as a revival of an old tradition. Back when everyone traveled in such boats, and many builders and navigatos were around, 2000-mile trips would not be such a reach. And Polynesians could tack back and forth against the wind.

    @AntC: That is not an etymology. If you insisted that ramen is not a Japanese loanword because ram and men are English words, that would not be an etymology either.

    All this is reinventing the wheel. People have looked at this word very hard for a century now, including Māori and Hawaiian scholars, as well as navigators and ship builders.

  151. @Ryan Hahaha

    the bigger coincidence is frigatebirds never managing to propagate the wild plant across the Pacific, then suddenly dropping cultivar seeds in the same millennium that the Polynesians began their voyages.

    Under the bird-vectored hypothesis, we don’t have to posit they “never” managed to propagate before some date. They could have propagated several centuries before Polynesians got to wherever. And the kūmara self-sowed from then on. S.Am cultures were trading along the coast (and presumably growing crops for trade) from about 600 CE, so that puts an earlier limit: there weren’t such cultivars before then. (This hypothesis also explains why kūmara didn’t get further westwards to meet the humans earlier: the birds were all pooped out before they got to Fiji/Samoa/Tonga.)

    @Y 2000-mile trips would not be such a reach. And Polynesians could tack back and forth against the wind.

    Just no! 2000-mile trips were achieved by island-hopping at distances usually less than 1,000km. Getting so far as Rapa Nui was extraordinary (about 2,000km not miles from Pitcairn), but by then Polynesians knew what they were doing and knew how to read the natural signs. And had kūmara cultivation well understood.

    The coast of S.Am is an order of magnitude further than any known hop, and no land at all, let alone somewhere to replenish fresh water. Furthermore by 1000 CE, Polynesians had not mastered the techniques for travelling (and returning reliably) over 1,000km.

    Have you — or anybody else proposing this preposterously long voyage — ever tacked ‘back and forth’ against a prevailing wind? It takes 3 to 4 times longer than the same distance-made-good (as we sailors call it) downwind/crosswind. Furthermore the best those wāka could achieve was about 75° sailing angle to the wind. Modern monohull ocean-going yachts achieve about 45°; those terrifying things on stilts in the Americas Cup maybe 25°. Do the trig.

    No I wasn’t positing an etymology. I was pointing out *[kumara] is easily within the sound pattern of Polynesian languages/we don’t need to look for a borrowing. Also that Polynesian languages exhibit wortbildung (and indeed abbreviating longer words to plug them into a build), so we don’t have to look for an etymology for the whole thing /that having a soundalike for all three syllables isn’t much evidence.

    (Yes we are by now just rehashing prior points. So you’d think these scholars would by now have come up with something more convincing than a stray comment by a non-philologist/botanist.)

    as well as navigators and ship builders.

    Yeah. I note none of them has volunteered to try to recreate this hypothesised voyage — unlike Kon-Tiki. And nowadays we have forecasting that tells us when a El Niño is arriving.

  152. My point is that typically wild breeds are more likely to self-propagate than cultivars, which often adapt to cultivation, losing traits that allow them to survive and reproduce in the wild. Presumably wild I. batatas plants and frigatebirds have existed for time measured in many hundreds of thousands of years. Yet across those ages, frigatebirds don’t seem to have brought sweet potato across the Pacific to self-propagate until the very millennium in which Polynesians made their crossing. Your accidental crossing / self-propagation theory depends on an amazing coincidence. And your independent development of the word kuumara depends on an amazing coincidence. What are the chances of both?

  153. I take back what I said about tacking, having read Ben Finney’s classic Anomalous Westerlies, El Niño, and the Colonization of Polynesia. He does agree that sailing to South America would have been challenging (and expands further here), but as he says, “it might be wise to follow the dictum ‘never say never.’ All kinds of ocean voyages have been made over the world’s oceans with a wide variety of craft. Perhaps some particularly tenacious Polynesian explorers did stumble across the Equatorial Countercurrent, and by design or accident followed it all the way to the western coast of the Americas. If so, however, they would have had to have carried a good supply of food and water, and/or have been expert at fishing and wringing every drop of water from passing showers, in order to survive the many months such a voyage might have taken.”

  154. Yet across those ages, frigatebirds don’t seem to have brought sweet potato across the Pacific

    That we don’t know: perhaps they did, but the plants didn’t survive for long. So this particular I. batatas got lucky that Polynesians turned up in time to foster them.

    Frigatebirds don’t go inland at all. Perhaps sweet-potato is usually a highland/inland crop (I don’t know). It wasn’t until humans brought them down to the coast that the interaction happened.

    independent development of the word kuumara depends on an amazing coincidence.

    No that one’s the least improbable amongst these coincidences. There’s plenty (well, plenty more than zero) around the world’s languages of soundalikes-meaningalikes. Ref our host. The Oti-Volta languages seem to be highly productive of examples. (More seriously: linguists don’t go looking for soundalikes-meaningalikes, because they’re tiresomely brought up by non-linguists and crackpots. So I suspect a thorough survey — which I’m not volunteering for — would produce heaps. Especially amongst languages with small phoneme inventories, open syllables, no consonant clusters, etc.)

  155. @Y good on you for reading Finney. I thought he’d be too sailing-nerdy for the audience here.

    The trouble with positing a El Niño is they’re unpredictable. Whereas Polynesian strategy was all about repeatable/returnable outings. (Not surprisingly.)

    There’s a possible round trip (Finney hints at) going out in the Forties from RapaNui, north at the continent in the Humboldt, North-West from Ecuador to French Polynesia, then revert to island-hopping to the Eastern main islands. You’d need an amazingly encyclopaedic knowledge of all the constellations and island/wave patterns to even recognise you’d got back to familiar-ish seas. Yes many months, as Finney says. And little opportunity for fresh water/supplies.

    But more important: how would you know in advance all those currents/tradewinds/the continent itself were there? And at which times of year they’re reliable? RapaNui was unknown at the time of discovering kūmara. If many months, you’re going to get caught in the change of seasons where all the meteorological/sea-current ‘action’ switches from N. to S. hemisphere.

    The Europeans had compasses/sextants, knew the earth was a sphere and roughly how big (except Columbus), took local guides with them in unfamiliar seas, and were equipped for several _years_ at sea. And still plenty of them perished. Columbus was both spectacularly wrong, and spectacularly lucky. (And the Atlantic is tiny compared to the Pacific — especially at the squeeze point from West Africa.) So indeed ‘never say never’.

  156. @Y The Cañari word is recorded directly; see the reference Xerîb linked to.

    Is this the best we have? The priest is not a phoneticist. Today’s pronunciation in Cañar-Quechua might be over-influenced from 400+ years of Spanish [**].

    comales (que quiere camir camotes), … [1582 priest’s report]

    The term comales appearing in the above extract would have been a hispanicised plural; the singular would be comal. It is entirely possible that the term cumal reported amongst Quechua-speakers …
    [2005 Scaglion commentary]

    Fair enough to reverse engineer the Spanish plural back to singular. Can we reverse engineer what the priest wrote back to what was spoken? How well does ‘comal(es)’ fit C16th _Spanish_ sound patterns? Does Cañar(-Quechua) have word-stress patterns/rhythms? Do they match Polynesian stress patterns? [***]

    Scaglion is an Anthropologist. How sharp is he on phonology?

    [**] Today’s pronunciation of Te Reo Māori is known to be over-influence by English — and that’s happened in less than 200 years. What the (mostly German-trained) phoneticists recorded as ‘wha-‘ appears to be modelled on the initial consonant in Old English ‘hwæt’. And there’s a separate phoneme recorded as plain ‘w-‘ (‘whiri’ minimal pair vs ‘wiri’). There might have been some regional variation in pronunciation: ‘hwa-‘ varying to ‘fa-‘ varying to ‘wa-‘ (aspiration lost).

    Colonial administrators and the bloody priests again were not so assiduous in their recording: the River Wha-nga-rei running through the town gazetted as Wangarei — pronounced wong-a-ray by red-blooded residents (and there’s a lot of them).

    Anyhoo nobody these days says ‘hwa-‘/ɸa-/ — except pedantic Anglo-Saxons like me. Some trying to be politically correct say ‘fa-‘, most pronounce same as ‘wa-‘.

    [***] I hear Māori word-stress patterns as lacking lexical stress, like French. ‘māra’ vs ‘mara’ are consistently different words — the macron denotes lengthened vowel, never reduced.

  157. David Eddyshaw says

    pronounced wong-a-ray by red-blooded residents

    I have heard old-style RP-speaking relics-of-empire Brits refer to the Burkinabé capital city of Ouagadougou as “Wogger.” This struck me as unfortunate. Happily, by far the most old-style relics of empire in those parts are actually Francophones.

  158. Happily, by far the most old-style relics of empire in those parts are actually Francophones
    If the British had got that part of North Africa that was French, we could have had “Wipers to Wogger” instead of “Paris – Dakar”.

  159. David Marjanović says

    But more important: how would you know in advance all those currents/tradewinds/the continent itself were there? And at which times of year they’re reliable?

    So perhaps somebody set out on an exploration or even settlement voyage, got caught in El Niño instead, and ended up in the Blue Desert without knowing how that happened. It’s not like we have any idea how many such voyages got lost and never returned; maybe here’s one that did return.

    If he could have, Columbus would have demanded the monopoly on luck before he set sail. But he didn’t.

  160. Can I ask the assembled brains trust …

    The whole brouhaha here presumes the pronunciation of this dratted word didn’t change in the 500 years from ~1000 CE (hypothesised contact) to the priest recording it 1582. This despite the Cañari getting overrun by the Quechua-speakers nearly a century before the Spanish arrived. (I’ll leave aside that we have not a skerrick of evidence it was a Cañari word ever. And we certainly have no means to reconstruct any sound changes in Cañari.)

    Now it’s true the reconstructed Proto-East-Polynesian *[ku:mara] is the same pronunciation as in Te Reo Māori today. But the pronunciation in other E.Polynesian languages has drifted. Furthermore the pronunciation of other words in Māori has drifted from P-E-P. I guess the phonemes in the particular word lucked out in Māori(?)

    So there’s not only the coincidence of soundalikes. There’s also the coincidence the word in both languages was phonetically stable for 500 years. Is this likely? Are its phonemes somehow ‘core’ or ‘base’ or ‘rooted’ within the sound pattern?

    That wouldn’t explain it changing in some E.Polynesian daughter languages. And none of those made contact with other languages until European times.

    Also: if we expect/anticipate endogenous sound changes in languages (unlike this particular word), doesn’t that increase the likelihood of historical soundalikes, even if the words today sound different?

  161. @AntC: 500 years is not that long, and it depends on what part of the phonological system is affected by sound changes. A hypothetical Latin *cūmara would still be *cumara in Modern Italian, about two millennia later, or a Common Slavic *ku(:)mara would still be *kumara in all major standard Slavic languages, about a millennium later. While I know almost nothing about the historical phonology of Austronesian and nothing at all about the historical phonology of the South American languages involved, it doesn’t seem unlikely to me that such a word could remain mostly unchanged over 500 years (if the languages in the respective regions have a tendency to go totally batshit with sound changes, like French, Albanian, or Old Irish, then all bets are off.)

  162. Hans is quite right; to take another example, a Proto-Semitic *kūmara would be kumər in Algerian Arabic and kūmar in most Middle Eastern Arabic dialects, 5000-odd years later. 500 years is nothing when it comes to phonetic stability. I’m no Quechua expert, but I have looked at pan-Quechua sound correspondences as part of historical linguistics homework (it’s a problem set in Lyle Campbell’s book) as well as at Paul Heggarty’s work. They’re mostly pretty trivial! You can explore a bunch of cognates at Soundcomparisons.com to see just how similar these words have stayed; try the word for “neck”, for instance.

  163. Heh. Mair has a whole post ranting about “a knee-jerk reaction to attribute all distant cultural resemblances to chance coincidence.” Let’s face it, there’s no universally acceptable way to decide these issues; I’m always going to be suspicious of solutions that require too great a leap of faith (about, e.g., premodern trans-Pacific travel), and others like the idea of ancient cross-cultural contact and are suspicious of lexical coincidences. But we manage to coexist!

  164. The full list of comparisons for kumara in Polynesian are here: https://pollex.eva.mpg.de/entry/kumala1/
    My knowledge of historic phonology for the region is minimal, but I think there are some unexpected forms here (e.g. loss of the -m- in Hawaiian, Niue timala etc) which would not necessarily be expected from an old-established word. I’m sure I’ve seen suggestions that this was because it was a later borrowing between languages rather than part of the inherited vocabulary everywhere. (Note: they reconstruct as *kumala for Polynesian as a whole, but if that is meant to contrast with a hypothetical *kumara I think you’d need to be sure that the Tongan and Niuean forms were not borrowings from further East

  165. David Eddyshaw says

    Mair has a whole post ranting

    I note his frankly disingenuous and needlessly offensive response to a perfectly sensible comment there.

  166. I wouldn’t dream of commenting in that thread.

  167. David Eddyshaw says

    500 years is not that long, and it depends on what part of the phonological system is affected by sound changes.

    Kusaal tʋm “send” is cognate with Swahili tuma “send.” The time-depth for proto-Volta-Congo can hardly be less than 4000 years … (even just proto-Bantu itself is said to be well over 2000 years old, though such numbers seem mostly to rely on voodoo lexicostatistics: still, the Bantu languages differ a lot more from each other than the Romance languages, say.)

    On the other hand, Kusaal kɔlig “river” is cognate to Mbelime wuonu “river” (and Mbelime is much more closely related to Kusaal than Swahili is.)

  168. Kusaal kɔlig “river” is cognate to Mbelime wuonu “river”
    Drop the other shoe, would you?

  169. A nice summary of Proto-Polynesian segments and their reflexes.

    kūmara~kumara is also reflected as such in Rapanui and Tuamotuan. The loss of /m/ in Hawaiian and other languages is irregular, but as the POLLEX link indicates, it also occurs elsewhere (examples from Mangaia). *l is reconstructed for the protolanguage by convention: it could have been either /l/ or /r/. The Western Polynesian words are all borrowings, maybe some post-European ones. Sweet potatoes were introduced elsewhere in the Pacific by Europeans.

  170. The plants are usually propagated by stem or root cuttings, says wp. But they also produce seeds. “True seeds are used for breeding only.”

    i’m only an indifferent gardener, but i can, i think, elaborate this properly.

    it’s saying that seeds are not how new kumara plants are generally grown under cultivation – that kumara-growers only grow them from seed when they’re trying a new crossbreed. that implies that kumara probably don’t breed true to strain from seed (like apples, for instance), and possibly-to-probably don’t grow very successfully from seed.

    and that makes me think that any seed-based mechanism of transmission is not the best place to be looking. this doesn’t affect the human/non-human question at all: it just means that the non-human mechanism would be floating chunks of stem or tuber (perhaps as part of a larger clump of flotsam pulled off shore by a storm or washed downriver by a flood) rather than seeds.

    it also might mean that a well-supplied coastal vessel would carry an uncooked tuber or two, either as trade goods (“here’s our best strain, can we have yours?”) or as emergency supplies in case of getting stuck somewhere for an extended period of time.

    but given the odds of drift-carried kumara landing and taking root more than once, it seems like the timeline of its spread among the pacific islands might be revealing, if the genetics can help establish it at any usefully fine scale – especially relative to polynesian settlement wherever it seems to have first turned up.

  171. David Marjanović says

    I note his frankly disingenuous and needlessly offensive response to a perfectly sensible comment there.

    It's factually wrong, too.

    I wouldn’t dream of commenting in that thread.

    I have dared. Let’s see what happens.

    I’ve tried to remind myself that being passive-aggressive seems to be a personality trait some people just have, and need not be intended as an act of card-carrying mustache-twirling cackling villainy even though that’s what it… is.

  172. Stu Clayton says

    Gosh, David, that’s a daring comment there, given how high in the saddle Herr M is riding. I feel like breaking out a beer and pulling up a deck chair, but it would just make me sleepy. I’ve seen it all before anyway.

    I’ve tried to remind myself that being passive-aggressive seems to be a personality trait some people just have

    Trait ? It’s a fully-fledged way of life, driven by what Nietzsche called Ressentiment. So you really have to watch out for the aggressive phases. They are not necessarily just verbal. [see my forthcoming essay Dastards in the Dog Park].

  173. Lars Mathiesen (he/him/his) says

    @AntC, my impression is that the historical developments of the Polynesian languages is quite secure because the number of cognate sets is huge — so if they were saying that loss of -m- was regular in Hawai’ian, they would not just be guessing. But note that both of your examples are marked as “Problematic,” possibly because the changes are not regular. Looking at KUMI.1 ‘ten fathom measure’ in the same list, the correspondences in those languages are ?umi and kumi and not marked problematic.

    On the other hand, Proto-Polynesian kumara has 22 “unproblematic reflexes out of 37 listed languages. Most of them are just kumala. I think that pushes any hypothesis that the form of the word on the Polynesian side has changed a lot over 1000 years out well beyond any reasonable number of standard deviations.

    But on the Cañari side, I agree. One attestation 500 years after the assumed borrowing (and 100 years after a conquest by the Inkas) is compatible with all sorts of protoforms. If there was some loan from a well-reconstructed neighbouring language that was somehow unambiguously Cañari, it could be narrowed down a bit. But even if some miracle allowed us to pinpoint a similar form to Proto-Cañari of 1000AD, there is still the possibility of independent creation — how many different syllables did Proto-Polynesian even have? (And this would have to be compared with the likelihood of even a single successful return voyage, which is not known. I only see claims and counterclaims).

  174. how many different syllables did Proto-Polynesian even have?

    Proto East Polynesian, the language in question, had 11 consonants and 5 vowels (plus 5 long vowels). So at least 55³=166,375 different possible three-syllable words.

  175. January First-of-May says

    If he could have, Columbus would have demanded the monopoly on luck before he set sail. But he didn’t.

    Magellan (and El Cano) surely had more; their expedition is how we have the name “Pacific”, because that particular ocean had seemed unusually quiet to them.

    OTOH apparently they managed to miss basically all the inhabited islands on the way…

  176. Kusaal kɔlig “river” is cognate to Mbelime wuonu “river”
    Drop the other shoe, would you?

    I second Y’s request for more info; I love that kind of unobvious relationship.

  177. (DE, you brought it up recently, using the Mampruli cognate, but without revealing the plot then either.)

  178. David Eddyshaw says

    Fairly sure that I did expound this one in some detail somewhere on LH (and, after all, all issues are treated somewhere on LH) …

    Still, it’s probably easier to rehash the explanation rather than search for it.

    The stem of the Kusaal form kɔlig is actually not far from the proto-Oti-Volta *kpel-; Kusaal inflects this stem in the ga/sɪ class, which is where the -g comes from (the -i- is an epenthetic vowel.) More on the class membership below.

    Labial-velars don’t contrast with velars before rounded root vowels in any current Oti-Volta language, but there is some reason the think that the protolanguage did have a contrast: some of these rounded vowels seem to have been the result of rounding by the preceding consonant, and in such cases eg. Kusaal /k/ may correspond to /k͡p/ elsewhere, e.g. Kusaal kɔbir “bone”, Moba kpabl.

    And in fact, the Moba representative of this word is a handy half-way house to understanding the Mbelime: Moba has kpénû.

    Two things are notable in this: firstly, it has /n/ for proto-Oti-Volta non-initial *l: this change is found regularly throughout all the Gurma languages (which include Moba) and all the Atakora languages except Waama (which include Mbelime.)

    Secondly, the word is in a different noun class: the “long thin things” ŋʊ/ŋɪ class. It’s no mystery why this is different in Kusaal: Western Oti-Volta has lost the ŋʊ/ŋɪ class, with just a couple of stray irregularities remaining as evidence that it once had it too. So the full proto-Oti-Volta form, along with its singular class suffix, was actually *k͡pelu.

    Almost there …

    The *k͡p -> h change is regular in Mbelime: compare Kusaal kpi “die”, Mbelime hii “die.” The rounding of the vowel is of the same nature as in the Kusaal. The only thing remaining to be accounted for is the vowel length.

    Many Oti-Volta roots of the form CVC have CVVC allophones. Kusaal is fairly typical in that it is no longer possible to reduce the allophony pattern to any simple set of rules, presumably because there has just been too much levelling, but Mbelime obligingly does have as simple rule in noun flexion: you always get CVVC when the syllable is open, CVC otherwise. As the suffix is -u, the CVVC allomorph is required, and /uo/ is the long vowel corresponding to short /o/: so, wúónù.

    The stem tones also match regularly: Mbelime wúónù, Moba kpénû simply match, as high tone; Kusaal kɔ̄līg has mid tone, but belongs to a tonal subclass where the mid tone becomes low tone when the word sheds its noun class suffix and appears as a bare stem in composition (which happens all the time in Oti-Volta languages, as that is how nouns are construed with following adjectives and/or demonstratives: kɔ̀lkān, “this river.”) That tone class regularly corresponds to the high tone of Gurma and non-Waama Atakora languages, and comparisons with tones outside Oti-Volta show that the Gurma/Atakora tones are conservative, whereas Kusaal and its closer relatives have innovated.

    So: all forms have arisen by known mechanisms from the proto-Oti-Volta *k͡pélù “river” (a member of the “long thin things” noun class.)

  179. David Eddyshaw says

    CVVC allomorphs, not allophones, sorry. And allomorphy pattern. But you knew that.

    I see I’ve missed a step in the Mbelime form: *hw -> w.
    I have actually handwaved a bit in this: like it’s neighbour (and relatively close relative) Byali, Mbelime normally shows proto-Oti-Volta *k *kp -> h, but there are some inconsistencies. You could arbitrarily deal with these by multiplying the number of proposed consonants in the protolanguage, but in fact I think the inconsistency reflects the fact that this is an Atakora Sprachbund thing rather than an inherited sound change. The unequivocally Western Oti-Volta language in the Atakora, Boulba, shows similar changes, again not carried through with complete regularity.

  180. Beautiful, thanks for taking the trouble!

  181. January First-of-May says

    …I’m missing one thing here: how did h become w? Is the Mbelime form secretly huonu?

    …Oh, I see it now. I guess it was missing when I saw the comment.

  182. David Eddyshaw says

    The “bone” word that I mentioned, corresponding to Kusaal kɔnbir /kɔ̃bɪɾ/ “bone”, Moba kpabl, in Mbelime goes: singular hɔ̰ɔde, plural wɛ̰. (In all three languages, the word has the same tones as “river”, BTW.)

    The disappearing -b- in the Mbelime is a whole ‘nother problem which I haven’t got sorted out yet. There are plenty of parallels for b/w alternations after vowels in Oti-Volta, but I haven’t been able to work out regular rules so far. Judging by the presumed* Bantu cognate (e.g. Swahili mfupa, proto-Bantu *-kupa) there was a *p in there back sometime …

    * The proto-Bantu *u is awkward, as the Oti-Volta forms all suggest a proto-Oti-Volta *a: *k͡pap- or *k͡pab- Still, the resemblance seems too close to be pure chance. The tones correspond properly, too.

  183. Presto Change-o! Lovely.

    *k͡p -> h is odd. Any guesses about intermediate steps? (I know you won’t just say “lenition”.)

  184. but given the odds of drift-carried kumara landing and taking root more than once, it seems like the timeline of its spread among the pacific islands might be revealing, if the genetics can help establish it at any usefully fine scale – especially relative to polynesian settlement wherever it seems to have first turned up.

    Thanks @rozele, it seems the timeline of spread in E.Polynesia was quick. This at a time when there was regular travel throughout — so I’m guessing the languages were still very close. (As others have noted, kumara didn’t get back-transported to W.Polynesia/neither the word — that didn’t happen ’til European spread.)

    The trouble with getting finer detail on the botany today is that Europeans brought other cultivars from elsewhere in the Americas (Caribbean, for example) so now the strains are all hopelessly tangled up.

    the non-human mechanism would be floating chunks of stem or tuber (perhaps as part of a larger clump of flotsam pulled off shore by a storm or washed downriver by a flood) rather than seeds.

    I take the points everybody’s making that the distance to the Marquesas is huge for such flotsam. And how would the tubers root themselves if they just washed up on a beach? But if we’re not positing human agency, there’s a much longer timeline for this to happen: S.Am natives were cultivating Kumara and rafting along the coast from about C6th.

    One of the botany papers (sorry I’ve churned through too much stuff by now) guesstimated from the samples it did test kumara’s first arrival in E.Polynesia would be somewhere around the Marquesas. But it didn’t have a sample from there. This corresponds to the Polynesian myths — if you interpret them generously. But also corresponds to the ocean drift/prevailing winds from Ecuador.

    So kinda this would line up with the hypothesis David M is trying to pursue (on the Log). The trouble is it just wouldn’t work to sail _to_ Ecuador from the Marquesas — precisely because you’d be battling currents and winds and would take much longer than drifting back. So we’d have to hypothesise they set off from somewhere completely elsewhere; then the interaction on S.Am coast; then to Marquesas where no Polynesian had explored.

    ‘Voyages of settlement’ took heaps of stuff/whole families, and only set out _after_ ‘voyages of exploration’ had established there was a habitable island and reliably how to get there. Then if you were off course by so far as the S.Am coast and trying to get home, I don’t see why you’d pause at a so-far uninhabited island to plant these new-fangled kumara.

    I’m not taking anything away from the Polynesians’ powers of navigation/seamanship to say I just don’t see how they’d know how to get back to central E.Polynesia. We’d have to posit all sorts of luck and chance — less chance than a single soundalike.

    Thanks @anhweol, @LarsM. Yes I don’t get why some of those are marked ‘Problematic’. (It’s fair to mark the W.Polynesian soundalikes as ‘Borrowed’.)

    Rarotongan ʔUmi vs Māori Kumi is a surprise: NZ was settled from Rarotonga (say the myths, and the genetics); their languages are still mutually comprehensible/the difference is more of a dialect than a separate language. Rarotongan has /Kuumara/, exactly the same as Māori. Other words starting /k-/ I checked also start /k-/. WP thinks Rarotongan has both /k/ and /ʔ/ (unlike NZ Māori) – allophones? ‘Problematic’ means they’re not sure their data is right? One of the Rarotongan outlying islands has /kūʔara/ OTOH in other places /ʔānau/ for Māori whānau – family. /ʔ/ is some kind of super-allophone?

  185. I have dared. Let’s see what happens.

    Apparently you’ve vanished into oblivion.

  186. David Eddyshaw says

    *k͡p -> h is odd. Any guesses about intermediate steps?

    I presume *k͡p -> hw in the first instance.

    The “bone” words make sense on the basis of the /w/ first being lost before rounded vowels, and then remaining /hw/ -> /w/. (The front vowel in the plural of vowel-stems in this noun class is a systematic thing in Mbelime, where the plural class suffix was originally *ya.)

    Unfortunately, this neat explanation doesn’t account for hii “die.” The imperfective form of proto-Oti-Volta *kpi “die” was probably *kpu, but (a) in Mbelime it’s actually hiimu and (b) of all the verbs in the lexicon, “die” must be about the least likely to have remodelled its perfective on the basis of the imperfective.

    Byali, the other language which usually does *k ->h, actually has yia “die”, wo “kill” (cf Kusaal kpi, respectively.)

  187. Darn it! Both David M’s and @Hat’s comments have vanished already. I’m certainly not going to try.

  188. David Marjanović says

    The trouble with getting finer detail on the botany today is that Europeans brought other cultivars from elsewhere in the Americas (Caribbean, for example) so now the strains are all hopelessly tangled up.

    That’s why the 2013 genetics paper used old herbarium specimens.

    Then if you were off course by so far as the S.Am coast and trying to get home, I don’t see why you’d pause at a so-far uninhabited island to plant these new-fangled kumara.

    Perhaps you needed a longer break of some kind. Shipwrecked?

    I’m not taking anything away from the Polynesians’ powers of navigation/seamanship to say I just don’t see how they’d know how to get back to central E.Polynesia. We’d have to posit all sorts of luck and chance — less chance than a single soundalike.

    A single soundalike plus all sorts of luck and chance to get the sweet potatoes across the ocean without having people carry them.

    Apparently you’ve vanished into oblivion.

    FFS.

  189. RAR ʔumi is ‘ten of cards’. It was probably brought over by Hawaiians (sailors?), Hawaiian having *k>ʔ and the general meaning ‘ten’.

    AntC, there’s also the human genetic signature, discussed in Ioannidis’ et al. recent paper. They infer contact with the Marquesas at about 1150–1200 AD, from the equatorial coast of South America, either by a one-way voyage from the continent or by a round trip from Polynesia.

    Technically an accidental drift of an American boat to the Marquesas is possible, but they’d be even less expected than Polynesian explorers to carry months’ worth of food and water (plus sweet potatoes), but “everything is possible”. I’m OK with that.

    I still don’t know about the chickens, though. I’ve lost track of the back-and-forth arguments.

  190. @Y I still don’t know about the chickens, though. I’ve lost track of the back-and-forth arguments.

    Yes I’ve been staring at this too long now/need to get out and climb up a mountain or something/can’t be bothered to chase it all up perspicuously.

    I think the chickens got rejected. Anyway they’re way down south in Chile/a long way from Ecuador. So either the chickens could fly, or we’re positing multiple Polynesian voyages.

    The human genetic signatures IIRC were from samples (skulls) in Brazil. No explanation for how they’d get there without European transport. And they were dated to right on the borderline of European contact. How secure was that dating?

    What this kumara drama reminds me of is Graham Hancock — a journalist, no sort of historian/anthropologist/linguist. He’s had a go at the Egyptian Pyramids, Gobekli Tepe, the Ceylon ‘Giant’s Causeway’ … He’s a real example of ‘a little knowledge is a dangerous thing’. He learns just enough to ride the tails of some unlikely hypothesis/doesn’t place what he finds in context of a whole discipline/writes a book of romantic imaginings/moves on to some other topic when (if) some academic finally gets called upon to entertain his nonsense semi-seriously.

    So in the kumara case I know that I don’t know a lot. But I have an innate scepticism. An isolated word/not a slew of terms around some technology like horses & chariots to my mind needs an extraordinary level of evidence. We don’t even know whether this S.Am. word belongs to Cañari vs Quechua; if Quechua it wasn’t spoken on the coast at the needed time. And the rendering of the pronunciation was by a Spanish-speaking priest who went so far as to put a Spanish declension on it. (Quechua does put plural declensions, but they’re not `-es`; no info whether Cañari has declensions.) I just don’t trust the priest’s rendering of the sound. And if we don’t have the sound, we don’t have a soundalike.

  191. Whatever it is the so-called Cañaris spoke, the point is that the word kumara was spoken by the coast.

    That the recorder was Spanish is not a big problem. Presumably the kumara neat the coast is similar to that recorded reliably from inland Quechua, and there are no sounds in it that can’t be rendered in Spanish, except possibly an aspirated k which wouldn’t be distinguished from a regular one. The plural suffix is no stranger than that in English banana-s.

    Untrained missionary recordings are quite valuable and informative. They just require more careful handling. These people weren’t making things up just for kicks. That’s why there are conferences and journals on missionary linguistics. There are entire languages only known from European missionary records.

  192. the point is that the word kumara was spoken by the coast.

    No: that’s what we don’t know — what was spoken on the coast **at the time of hypothesised contact**. It was spoken on the coast by 1582. If by “inland Quechua” you mean Inca lands — that is, Inca as at ~1000 CE — it ipso facto wasn’t spoken on the coast as at ~1000 CE. (Well I guess unless the Cañari borrowed it. But since they were rivals, that seems unlikely.)

    Note Quechua as at today has two words referring to two different varieties of sweet-potato, with neither appearing to be super-ordinate. (They nowadays use a Spanish-imported word for sweet-potato-in-general.) So I think we can’t reach for an explanation that one word is Inca-Quechua whereas Cumal is a Cañari ‘regional’ word left over from before Inca conquest.

    That’s why there are conferences and journals on missionary linguistics.

    Yes sure, by the C19th/20th. And missionaries these days get trained in linguistics and ethnology — Everett/Pirahã for example. But (correct me if I’m wrong) in C16th the Spanish were going around destroying native American empires.

    quite valuable and informative

    Doesn’t strike me as reliable enough to build a whole story of contact — in the absence of any other corroborating evidence. Indeed in the absence of any other transfer of artefacts or words.

    There are entire languages only known from European missionary records.

    Well nobody recorded much about Cañari, so we can’t today even reconstruct its sound pattern except by presuming it’s a substrate to Cañar-Quechua creole. The next-door language group Barbacoan is better recorded, but I think was not overrun by Inca. The attempt to reconstruct more Cañari by comparing to Barbacoan didn’t get very far — not even far enough to establish there was a genealogical relationship.

    accidental drift of an American boat to the Marquesas is possible, but they’d be even less expected than Polynesian explorers to carry months’ worth of food and water

    I was taking it the sailors perished in the storm. Only the raft/its cargo drifted.

  193. David Eddyshaw says

    But (correct me if I’m wrong) in C16th the Spanish were going around destroying native American empires.

    Some of the best early work on indigenous American languages was done by Catholic missionaries well before the nineteenth century. A particularly splendid example is Horacio Carochi’s work on Nahuatl

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

    This is recommended even now for serious study of Classical Nahuatl (there’s a nice English translation of it with the Spanish text opposite.)

    Although Carochi was particularly gifted, he was by no means an isolated case.

    On the other hand, these people were hardly untrained linguistically, in the sense that they were as well up in the linguistics of their own era (such as it was) as anybody could well be.

    Not all of the Spanish were destroyers. Many spent a lot of time vigorously opposing the exploitative practices of the destroyers. It’s significant that, calamitous as the European invasions were everywhere, the actual eradication of local peoples was (overall) much more thorough in the Anglophone than Hispanophone colonies.

  194. Even more to the point, there’s Domingo de Santo Tomás, who published in 1560 a Quechua dictionary, Diego González Holguín with his 1607 grammar and 1608 dictionary, and other sources of the same era.

    Anyway, AntC, I get the feeling that you are trying very hard not to look at the linguistic evidence that’s in front of you, and so insist that it doesn’t exist. It gets tiring.

  195. I get the feeling that you are trying very hard not to look at the linguistic evidence

    It’s true that I’m completely convinced by the seafaring evidence, so I’m examining the linguistic evidence (which would contradict the seafaring) particularly closely. Are you suggesting I should relax and ‘let it pass’ that we have no secure attestation of the word in the right place at the needed time?

    IOW we should just let Mair run amok with his woo? (R.Fenwick I’m not so critical of: they readily states they’s not a linguist/ethnologist. They seems to take myths more-or-less verbatim.)

    It’s not as if *[ku:mara] is outside of P-E-P sound patterns that we’d even be looking for a borrowing, if not for this alleged soundalike. It’s no surprise it’s a neologism: it’s a new thing in E.Polynesia.

    If there was language transfer, what further level of contact would be needed to get both the word and the thing and presumably knowledge how to preserve the thing en route and cultivate it upon arrival at dry land. Seemingly not even the existence of remote peoples with a different language survived in myth. (I suppose we could experiment washing up some kumara on a beach to see if/how they survive. Anybody want to fund me to go to the Marquesas for that?)

    Come to that: have we any other examples of a single contact between distant peoples transferring a single word?

    (I’m not trying to browbeat everybody into conceding the point. Yes it’s tiring/I’m only putting effort into it because it’s been blowing hard all week here/I can’t get out sailing. You can just stop. I’ll just stop as soon as the weather comes right.)

  196. . It’s no surprise it’s a neologism: it’s a new thing in E.Polynesia.

    It should be a surprise, actually. When you come across something new, you almost never coin a name for it out of thin air. If you can’t or don’t want to borrow a name, you create one based on words and morphemes you already have (tele-vision, basket-ball, …), or, at a pinch, on onomatopoeia (tuk-tuk). Polynesian is reconstructible enough that the former should be analysable, and onomatopoeia seems completely unmotivated here.

  197. you create one based on words and morphemes you already have

    (Sorry I can only do NZ/Māori vocab/neologisms, but wortbildung is common across Polynesian languages.)

    Ao-tea-roa, Kai-kōura — names of places, indeed not coined out of thin air.

    Piwakawaka a bird endemic to NZ The fantail has 20 or 30 different Māori names.. Most Māori names for NZ endemic birds are adapted from Polynesian names. But of course perhaps the Polynesian names were coinages.

    There’s also plenty of Māori names for endemic NZ plants, including those used as food, which don’t appear in the Polynesian wordlists — and again some that are adapted.

    For *[ku:mara] there’s plenty of candidate words/syllables that could be assembled for coining that.

    There are kumara-associated words (such as for storage pit) that aren’t soundalikes/compounds from kumara. Ok: do soundalikes for hāpoko or kōpiha appear in Cañar-Quechua?

  198. Unetymologisable names for endemic NZ plants is a good counterpoint; that’s the sort of thing I would want to look at more closely for estimating the odds of coincidence here. Bird names tend to be onomatopoeic quite a lot, loosely imitating their cries; plant names are usually harder to associate with sounds.

  199. Did I say (in another place) “Basque seems to be a particularly favourite source language …” (for bogus etymologies/long-distance language contact).

    I bring you ‘kumara’ from Basque. Allative singular of ‘kuma’, a cradle. The picture in wikt looks exactly the sort of thing in which you’d carefully place your kumara for drying before storing in pits over winter.

    Case solved.

  200. David Eddyshaw says

    I think that R.Fenwick actually is a linguist: indeed, I have a copy of their grammar of Ubykh, unless I’ve mistsken their identity (possible, as it”s a bit complicated.) I’m actually a bit surprised at their tolerance for Mair, but then a lot of people are more tolerant than I am.

  201. That’ll be this R.Fenwick on Ubykh.

    Also apparently this R.Fenwick on Kunwinjku dialect of Bininj Kunwok, a non-Pama-Nyungan language of Australia’s Northern Territory — although a few comments on they disclaim detailed knowledge of that topic. They do appear to be based at Uni of Queensland.

    Hmm? Is that the same style of comments as on the kumara thread?

    Anyhoo I’m glad I didn’t make any assumptions about their preferred pronouns — apologies for the awkward (to my ear) morphology.

  202. @AntC: The question is not whether there are neologisms in Maori or Polynesian, but whether there is a good internal etymology for *kumara in Eastern Polynesian that is in accordance with the usual word-forming patterns of the language and makes sense.
    On why only one item and one word – who knows? Maybe there were more, but they weren’t accepted into the culture and didn’t survive; maybe the sweet potato was the only thing that looked attractive and was practical enough to take on the journey back. Considering what you said about currents, the chain would look like this: Polynesians from the Not-Marquesas are thrown off-course and land in South America; they stay some time and learn about sweet potatoes, among other things; they try to get back and land at the Marquesas, where they share their findings with the locals. So there would be three stages of selection – what the seafarers decide to bring back and report, what the locals at the Marquesas would accept, and what of it they would maintain over the centuries. On each step, ideas, items, and words may get lost.

  203. The newly discovered animals kiwi, kākāpō, moa, tuatara, all have transparent etymologies.

  204. we’re talking about an exact phonetic match, with allowance for just one single vocalic epenthesis that Polynesian phonotactics absolutely demand. [R.Fenton, on LLog]

    Can I stretch everybody’s patience a little further with some phonetics probing, because I find R.Fenton’s claim (with all respect) to be just not true. The priest wrote ‘comal(es)’. Why that ‘o’. not ‘u’? What was the sound-value of ‘o’ vs ‘u’ in C16th Spanish?

    Part 1: Soundalikes. There seems to be no Spanish word ‘cumal’, but there is ‘cumel’ — some sort of alcoholic drink; there is ‘comal’ = griddle for tortillas from Nahuatl, also borrowed into English. (So I’m not saying those words were current in Spanish at the time, but the sounds were at least spellable.)

    Then the soundalike pattern we’re looking for is /kOmaL(Ǝ)/, where /O/ is /o/ or /u/; /L/ is a liquid — note E.Polynesian languages have either /r/ or /l/, not both; /(Ǝ)/ is an optional epenthetic lax vowel, schwa or /a/.

    I note that Quechua _doesn’t_ have /o/, does have /u/, /a/ — not that Quechua is the language we need to probe; but again why did the priest use ‘o’? I note some Greenbergian rules that all languages have a velar; if they have any nasals they have /m/; if they have a liquid it’s /r/ or /l/. So I predict before looking that we’ll have heaps of soundalikes for that pattern. For languages with a small phoneme inventory this gets more likely — all those sounds will be available.

    And so it proves. Indian subcontinent name Kumar from Sanskrit; Old Irish cumal = female servant; Irish cumar = ravine; only a small deviation to get VP Kamala; and on and on. And Basque kumara is just icing on the cake — and not at all likely, because Basque has a large inventory of consonants, and consonant clusters, and not necessarily open syllables, and diphthongs (but only the standard-issue 5 vowels, same as E.Polynesian).

    Part 2: possible source languages for a soundalike. Under the drift-on-the-Humboldt hypothesis, we don’t need the kumara to have come from (what is today) Southern Ecuador. The Humboldt comes northwards for the whole length of S.Am. East coast before branching NW anywhere from north Chile up to Ecuador (coincidentally the range of Balsa rafts) — indeed better our plant matter departs from south of Ecuador, else it’s liable to run aground on the Galapagos. (Counter-hypothesis: are there Kumara beached on those?)

    How many native languages were spoken along that coastline circa 1000 CE? What are the chances some/several of those have a soundalike to our pattern? I can’t tell exactly when our kumara story originated. With Heyerdahl? Or at least before Heyerdahl got debunked? But let’s say you go looking with a grab-bag of E.Polynesian words for any sort of match to any W.Coast S.Am. languages: I think highly likely you’ll find a close-enough match. Kumara just was an over-achiever.

    And now despite Heyerdahl being debunked we have ethnologists/philologists who soooo want to believe in long-distance communication in human history, they are deaf to the dangers of soundalikes (that actually aren’t soundalikes); they dismiss all the disciplines of the Comparative method; they diss other specialities (like nautical historians); they’re ready to believe in miracles.

  205. It doesn’t matter that two languages have words which sound alike. If they sound alike (allowing for later known regular sound changes) and have the same meaning—that’s when it’s significant. So what is the chance that Proto East Polynesian and one of 10 or 20 coastal languages have the same trisyllabic word, ☞ both meaning ‘sweet potato’, by random coincidence? About 20 * 55⁻³, or 1 in 10,000.

    You still haven’t addressed Ioannidis et al.’s paper.

  206. The Ioannidis et al papers are mostly behind paywalls. I did skim a couple, but chiefly threw up my hands at “Heyerdahl” woo-woo. The 2020 paper (if you want me to try to take it seriously):

    We find conclusive evidence for prehistoric contact of Polynesian individuals with Native American individuals (around AD 1200) contemporaneous with the settlement of remote Oceania.
    … a Native American group most closely related to the indigenous inhabitants of present-day Colombia.
    … in this analysis the Native American ancestries of the Pacific islanders all fall within, or beside, the Zenu people—an indigenous Colombian population.

    “Sweet potato cultivation in Polynesia as a crop began around 1000 AD in central Polynesia. …with the earliest archaeological evidence being fragments recovered from a single location on Mangaia in the southern Cook Islands, carbon dated between 988 and 1155 AD.” [wp]

    So 1200 is too late. None of these dates are very secure/there might be a little wiggle room. But Ioannidis et al will need a lot of wiggle. They have a 2021 paper I can’t see, perhaps that wiggles. Polynesian expansion into E.Polynesia/Marquesas estimated 1025 – 1120 CE. (We’d need to hypothesise the Americans arrived in the middle of the expansion/100 years earlier than that 1200 date, and whatever they brought to the party got taken back to Rarotonga/Samoa/Tonga and then forward to NZ and to RapaNui — which is not impossible, there was still plenty of two-way contact at the time.)

    But Colombia (or even N.Ecuador, which the paper vaguely entertains) is the wrong place: too far north for anybody to be speaking Cañari, and especially not Quechua/Inca expansion several centuries later. See their Fig. 4.

    Now IANA genome-sequencer, so much of the paper’s ifs and buts went over my head. OTOH they are not Nautical scientists nor ethnologists/linguists, so when they blather on about Cañari (at least they didn’t mix them up with Inca/Quechua, but they’re still several unrelated language groups along the coast away from Zenu) and this:

    For the same reason, these archipelagos [Marquesas] would be the most likely origin for Polynesian individuals discovering the Americas using their characteristic upwind exploration.

    I see magical thinking. Upwind exploration from the Marquesas (which is also up-current: the wind might drop or temporarily turn round; the current not so much) is what the Polynesians just would not attempt, as already made very clear by Finney 1985.

    Columbia/N.Ecuador is also too far north/around the corner to the NE to catch the current to the Marquesas.

    Wikipedia says nothing is known of the Zénu language/unclassified. These native Americans are in the wrong place at the wrong time speaking the wrong language. Why are Ioannidis et al even mentioning kumara and Heyerdahl?

    As at 1970’s (Yen) or 1980’s (Finney), I guess Heyerdahl was still not debunked. Heyerdahl’s hypothesis of Polynesian origins is overwhelmingly rejected by scientists today. — indeed it was not well regarded even at the time. As at 2020’s, academics would add greatly to their credibility by just not mentioning it.

Speak Your Mind

*