Tuesday, 18 June 2013

Pseudoscience: Western Xia Become Navajo? - An Email from Alice Beck Kehoe

I received an email from Alice Beck Kehoe, an archaeologist/anthropologist who holds some ever-so-slightly fringe ideas about pre-Columbian trans-oceanic contact - a bit like Betty Meggers, an anthropologist whose expertise was demonstrable but many of whose ideas were contrary to evidence or parsimony.  She emailed me because of a book I mocked, briefly, in a post on The People of Alor: Ethel Stewart's utterly bizarre The Dene and Na-Dene Indian Migrations 1233 A.D.: Escape from Genghis Khan to America, a book claiming to set out the evidence for a recent migration of Sino-Tibetan-speaking people to North America because of the Mongol invasion of the Western Xia polity in the early thirteenth century.  These migrants then became the Athabaskan-speaking populations of western North America, including the Haida, Navajo, and Apache.  I mocked the book because it's totally unsupported and among the worst of fringe archaeology, claiming something downright silly and in defiance of the facts.

Kehoe recommended reading this article by a guy named Joe Wilson, who apparently now has a PhD based on the ideas originally put forward by Stewart, but (reportedly) ameliorated.  She suggested I email him.  I haven't, but I may do if I have more time.  Like Stewart's book, I found the article almost willfully wrong and obviously incorrect.  Because of facts.

What facts?  Well, first, the fact that the Athabaskan languages are not at all related to Sino-Tibetan languages, and 700-odd years is simply not long enough to obliterate the connection.  Chaucer's English is nearly that old, and it is still legible to readers of modern English, to some extent at least, despite centuries of language contact on all continents, a brief period of massive sound change, and the acquisition and loss of countless vocabulary items.  We've got a ton of examples of Sino-Tibetan languages, including those spoken in the Western Xia region, as well as both close and distant relatives.  They don't look anything like Diné bizaad.  If Athabaskan and Sino-Tibetan are related at all - and I really don't think they are, given that, well, they aren't - then it has to be at such a great time-depth as to make this hypothesis unrealistic.

Second, this migration is not documented anywhere in any source.  No source, anywhere, says that any group of central Asians took to the sea, as Stewart has them do, to flee for new lands in the east or the south or anywhere.  As the archaeology isn't behind this theory either, and there is in fact no archaeological evidence whatsoever to link people from the Western Xia with Athabaskan speakers, this means the theory is lacking in any solid documentation of any kind, given that there is also no linguistic evidence at all.

There were some population movements that resulted from the Mongol invasions, including movements within China, central Asia, and elsewhere, that may not have happened without them.  These may have included the movement of Tai-Kadai speakers into mainland southeast Asia, which was also precipitated by the movement south of Chinese speakers in the Song dynasty (see the story of the Zhuang minority, documented by Jeffrey Barlow in a concise historical article of 1987 - 'The Zhuang minority peoples of the Sino-Vietnamese frontier in the Song period', Journal of southeast Asian history, 18(2): 250-269).  I'm not denying that population movements were caused by the Mongols and by Chinese states.  I just expect those movements to be documented either historically, archaeologically, or linguistically, or all three, as they are in the case of the Tai movements into Assam, Thailand, Myanmar, Laos, Vietnam, and Cambodia in the twelfth century, and as they are decidedly not in the case of Stewart's hypothesis.

Third, Wilson, but not Stewart, relies on an analysis of Navajo sand-painting to make the point.  Frankly, I consider sand-painting much, much less reliable an indicator of population movements than language, as does everyone.  Language is a huge set of independent markers - lexical items, intonation, morphology, and so on - and when two populations speak related languages with overlapping vocabularies and cognates related through plausible sound changes, we can be pretty damn sure that they have some close historical relationship that is demonstrable through the comparative method.  It's about as secure as it gets in prehistory, which is why linguistics is such an important discipline for all prehistorians to be aware of.

If two populations both use sand to make images, then we can say that they are both innovative groups of humans using the resources available to them to express themselves.  Navajo sand-painting is not Buddhist; it isn't like Tibetan sand-painting; and Tibetan sand-painting is not the same as anything known from the Western Xia, as far as I'm aware.  Moreover, Navajo sand-painting isn't found in other non-Navajo Athabaskan-speaking groups, so it's not like it's an overt, obvious pattern found throughout two languages families.  I think it's easier to explain in terms of the fact that Navajo people typically live in a place with lots of sand, much of it beautiful.  Using the phenomenon in the way Wilson does is an attempt to forge a link between these populations based on more or less arbitrary evidence, taking the connection between Athabascan and Sino-Tibetan as justified a priori.  It isn't.

Again, I'm not opposed to some demonstrable examples of pre-Columbian trans-oceanic contact.  But the spurious ones, like Ethel Stewart's claim that refugees from the Mongol conquests fled to the Americas, or Betty Meggers' view that South American ceramics came from Jomon Japan (they didn't), don't deserve to be treated seriously, because I'm not sure they're proposed seriously.  It seems like a lot of these people just want to ruffle some feathers and propose controversial theses because they're controversial, not because they're justifiable.  It's a shame because many of them, including both Kehoe and Meggers, are/were excellent researchers with plenty of expertise.  They just seem to get a hunch, follow it, and expect the evidence to eventually tally up the way their hunch thought it would.  But I don't think that's how science works.


  1. "The linguistic evidence presented at the end of this paper consists of 36 etymologies, that is 36 sets of cognate words that appear to be shared by Yeniseian and Na-Dene, but not (for the most part) by other language families. It is significant that these shared words include basic vocabulary (boil/burn, children, dry, he, hunger, name, night, nit/louse, old, summer, word/speak), body parts (breast, cheek, elbow, foot, guts/stomach, head, shoulder), flora and natural phenomena (birch bark, cedar, clay, fir, lake, river, snow [on ground], snow [falling], stone), fauna (deer, owl, rabbit, skin [animal], squirrel), and cultural artifacts (boat, bow/arrow, dish/plate/basket, rope). It is difficult to imagine that similarities of this nature could exist between language families that do not share a common origin." See this study: http://www.pnas.org/content/95/23/13994.full.pdf

  2. Yes, I'm aware of the proposed connection between Yeniseian and Na-Dene, but that isn't the same as Sino-Tibetan (not even close!), and the connection between Ket and Athabascan is so remote as to be relegated to the ancient past (mid-Holocene?), not 700 years ago. I'm not denying the idea of pre-Columbian trans-oceanic contact, but the idea of Xixia migration to the Americas is undemonstrated and, basically, ridiculous.


You can post anonymously if you really want to, but I would appreciate it if you could provide some means of identifying who you are, if only for the purpose of knowing who has written what.