Since it seems to be Linguistics Week here at Gambler’s House, here’s another post on Jane Hill’s theory that the spread of agriculture into the Southwest was associated with a migration of speakers of Proto-Northern-Uto-Aztecan (PNUA) from somewhere in Mexico. Previously I discussed an article of hers from 2001 in which she tried to show that a set of vocabulary items related to agriculture could be reconstructed all the way back to Proto-Uto-Aztecan (PUA), which, if true, would strongly support Peter Bellwood’s argument that agriculture was introduced to the Southwest by speakers of Uto-Aztecan languages migrating north from central Mexico. I found that article unconvincing. One reason was that, since almost all of the agricultural vocabulary known from Uto-Aztecan languages quite understandably comes from the southern languages of the family, which were spoken by farming groups, reconstructing that vocabulary all the way back to PUA requires the assumption that PNUA is a valid genetic unit combining all the northern languages, because almost all of the agricultural vocabulary known from those languages comes from Hopi, the only one spoken by a people who primarily practiced agriculture in historic times. Hill’s 2001 article, however, doesn’t provide much evidence to show the reality of PNUA, which significantly weakens her argument, as do the many problems with the correspondences she does identify.
The article I’m talking about now, however, is about a related but somewhat different issue. Published in 2008, it seeks to show that a set of agricultural terms from PNUA was borrowed into Proto-Kiowa-Tanoan (PKT), presumed to have been spoken by the indigenous hunter-gatherers who occupied the Southwest before the posited PNUA migration, and that a separate set of vocabulary referring to local wild plants and animals was also borrowed into PNUA from PKT. Here, rather than dealing with the very difficult matter of reconstructing proto-language vocabulary, Hill is dealing with loanword studies, which is generally more fruitful (though still difficult and often frustrating). The theoretical model for how this borrowing would have occurred is straightforward: speakers of PNUA, practicing an agricultural lifestyle somewhere in the Sonoran Desert, migrated above the Mogollon Rim onto the Colorado Plateau, where they found both a different environment and groups of hunter-gatherers who were very familiar with it. Since agriculture is a much riskier and more difficult endeavor in this area, with its shorter growing season and less predictable weather than in the Sonoran Desert, hunting and gathering would likely have become more important for the PNUA farmers, and they would have eagerly sought out knowledge of local resources from the local people, who may also have been intrigued by the potential of the unfamiliar agricultural practices of the newcomers. So, the PNUA speakers introduced the PKT speakers to farming, and in turn the PKT speakers introduced the PNUA speakers to plants and animals important on the Plateau but unknown in the desert. In the process, some words for these things moved between languages as well. Hill notes that this implies both that the PKT speakers, formerly hunter-gatherers, chose to adopt agriculture rather than being pushed to marginal areas by the PNUA speakers and that contact between the two groups was not necessarily always antagonistic. Both of these implications are problematic for Bellwood’s theory of the correlations between language distribution and the spread of agriculture, which holds that hunter-gatherers very rarely adopt agriculture when they come into contact with farming groups expanding out of their homelands with large populations but instead are either assimilated by the farmers or pushed into marginal areas unsuitable for farming. This is somewhat ironic, since Hill actually makes a very good case for these borrowings, which provides considerable support for some version of Bellwood’s general idea that language and agriculture generally spread together.
As in the previous paper, Hill is careful in this one to point out all the potential problems with the etymologies and correspondences she posits here. There are a lot, especially because Kiowa-Tanoan languages are not very well-documented and PKT reconstructions are much more tentative than P(N)UA ones. In this case, however, I find most of the correspondences pretty convincing. With contact linguistics like this, there are some inherent advantages over the sort of “pure” historical linguistics Hill was doing in the earlier paper. The most important is that loanwords are often pretty easy to identify, especially in well-documented language families. If a term is found in one language but not in any others in its family, but it’s very similar to a term with a similar meaning in a nearby but unrelated language, it’s pretty easy to conclude not only that an episode of borrowing occurred but also which direction the borrowing went. This is something of an ideal case, of course, and in practice it’s often not quite as clearcut, but it’s still easier to show that a term was likely loaned into a language or subfamily than that a set of vocabulary can be reconstructed back into a proto-language.
In this case, it’s the borrowings from PKT into PNUA that are most convincing. This is mainly because the PNUA forms are not attested elsewhere in Uto-Aztecan but are quite similar in both form and meaning to what can be reconstructed for PKT (which, again, is not all that reliable). The loans in the other direction are trickier, in part because Kiowa-Tanoan is a small family and comparisons between branches can’t really be done the way they can for Uto-Aztecan, but given the other loans they seem pretty plausible. Among other things, these loans provide pretty strong support for PNUA as a valid grouping, which in turn strengthens the argument of the 2001 paper, although it’s important to note that the issue in the 2008 paper is actually rather different, and it’s easy to imagine a group of farmers speaking PNUA migrating out of Sonora or southern Arizona without concluding that their ancestors necessarily migrating out of central Mexico speaking PUA. PUA could also have been spoken by a group of hunter-gatherers in, say, coastal Sinaloa or Nayarit who adopted agriculture after contact with agricultural groups migrating up from further south, perhaps speaking a language related to Purepecha, much as the PKT speakers later adopted it after contact with PNUA speakers. Nevertheless, the existence of PNUA is important to Hill’s 2001 argument, and the support for it here does strengthen that earlier argument.
The implications of this loanword evidence for archaeology are interesting. It definitely supports R. G. Matson’s argument, based on totally different evidence, that the Western Basketmakers spoke a Uto-Aztecan language and migrated into northeastern Arizona and southeastern Arizona from somewhere further south. In connection with that argument Matson also surmised that the of the Colorado Plateau and that they spoke Keres or a Kiowa-Tanoan language. As Hill notes in this article, Keres is an isolate and it would be difficult to use it in this kind of study. Kiowa-Tanoan, while a small family, does have a sufficient number of languages and enough apparent time-depth to be reconstructed into a form usable for comparisons to PNUA. It is still fiendishly difficult to figure out what language(s) the inhabitants of any ancient site would have spoken, but the integration of linguistic evidence in studies like this has the potential to shed some light on the issue.
To tie this back to Chaco, which seems to have been a pretty important regional center during the Basketmaker III period, the evidence from this article suggests that the Eastern Basketmakers of the Chaco area may have spoken PKT, although they may on the other hand have spoken a language ancestral or related to Keres or Zuni (both isolates). Or perhaps Chaco was inhabited by more than one linguistic group, as many archaeologists have argued for the later period of its more obvious regional dominance. This evidence does suggest that whoever was living at Chaco at this time probably was not speaking a Uto-Aztecan language, although it doesn’t entirely rule it out. There is, after all, no way to tell exactly when this episode of PNUA-PKT contact occurred, although if it involved early contact between farmers migrating in and local hunter-gatherers it would presumably have been rather early in the Basketmaker II period. Importantly, the fact that the loans seem to have gone both ways shows that whatever contact took place involved both groups continuing to exist as social entities of some sort. This is not evidence for assimilation, in other words, but for peaceful contact between agricultural and hunter-gatherer groups involving the exchange of information that enhanced the subsistence options of both parties. The archaeological implications of that are difficult to figure out precisely, but it’s a subject worth thinking carefully about.
Hill, J. (2008). Northern Uto‐Aztecan and Kiowa‐Tanoan: Evidence of Contact between the Proto‐Languages? International Journal of American Linguistics, 74 (2), 155-188 DOI: 10.1086/587703