A University of Copenhagen press release reports on what could be an exciting discovery:
Christophe Helmke and Magnus Pharao Hansen have taken the first steps toward solving a major archaeological mystery surrounding the ancient Mexican city of Teotihuacan. Until now, the language of Teotihuacan has been unknown. […] By analyzing the signs on Teotihuacan’s colorful murals and many other artefacts, they have concluded that the signs constitute an actual writing system, and they believe that this writing records an early form of the Uto-Aztecan language, which a thousand years later developed into the languages Cora, Huichol, and Nahuatl, the language of the Aztecs.
The paper, “The Language of Teotihuacan Writing,” is paywalled, but a free preprint, without the commentaries, is here; there’s Reddit commentary here (“Magnus is an excellent scholar and Nahuatl linguist and I take his ideas seriously”) and here (“it is the first time i see an abstract with a version in nahuatl (beside english and spanish). neat”). Thanks for the great collection of links go to Y, who adds:
The crux is that the rebus principle which has been tried before to interpret the script had assumed a language similar to Classical Nahuatl. Pharao Hansen has been working for a while on strengthening the evidence for a Nahuatl-Corachol subfamily. They argue that the protolanguage yields a better fit for the Teotihuacan rebuses than the Nahuatl of a thousand years later.
I hope it turns out to work as well as the Linear B decipherment!
Thank you for sharing this with us. I’ve already read the press release and the reddit comments. I look forward to reading the free preprint, even if much of it will be over my head.
It’s interesting for me on a personal level to think that there may be a linguistic link between the ancient inhabitants if Teotihuacán, which I have visited, and later Náhuatl-speaking inhabitants of the Valley of Mexico. I once mentioned here how my dad’s home neighborhood is really an old village that got swallowed up by Mexico City. I learned later that area still had Náhuatl speakers in the early 2Oth century so in all likelihood there were still Náhuatl speakers on my dad’s side of the family just a couple of generations ago. I have reasons to believe my paternal grandfather was a speaker, he would have been the right age to have learned it, but if he was he didn’t pass it down to my dad or his siblings, as far as I know. I take this as a sign from above to study Náhuatl.
an early form of the Uto-Aztecan language
Just as Linear B is “an early form of the Indo-European language.”
The paper naturally says no such daft thing; it hypothesises that the writing represents the common ancestor of Cora, Huichol, and Nahuatl. The whole point is that this would support the idea of a Nahuatl-Corachol branch of Uto-Aztecan.
Ah, press releases! Is there anything they can’t distort?
Pharao Hansen and his blog have come up here before.
!!! !!! !!!
!!! !!! !!!
!!! !!! !!!
That is all. I’ll read the preprint later.
From MPH’s blog: an example of a Teotihuacan pun, here, and an exploration of the Nahuatl-Corachol connection, here.
Pharao Hansen’s Ears of Nopal blogpost is a nice anecdotal intro to the idea of a rebus reading of Teotihuacan iconography using an earlier dialect in his proposed Nahua-Corachol clade.
Of course the Ear/cactus etymology may also be part of the formal paper, that, enticed by the blogpost, I now want to read.
Another researcher who writes so well and makes his points so clearly that you want him to be right.
Thanks for the links, Y.
The one about Nahuatl-Corachol doesn’t strike me as making much of an argument that the two constitute a genetic subgroup.* 2017 is a long time ago, though: have you seen anything more recent?
* They’re obviously related, but that’s not the issue. You need common non-trivial innovations that can’t be explained by diffusion to demonstrate that (as MPH certainly knows as well as I do.)
Within Oti-Volta, “live”, “breathe”, “dehusk”, “bury/plant”, “heat” and “sun” are all cognate in Waama and Kusaal (as are a great many other words), and the two languages both lack the merger of non-initial *l and *n found in all the Gurma and Eastern Oti-Volta languages which come between them geographically; but Kusaal and Waama unequivocally don’t form a subgroup within Oti-Volta.
DE, all I can find is a paper (preprint), from last year, “Sapir’s Law and the Role of Accent in the Reconstruction of Proto-Corachol-Nahuan”, and a vocabulary-centered paper, “Familia o vecinos? Investigando la relación entre el proto-náhuatl y el proto-corachol”, in a 2020 Festschrift for Karen Dakin. I haven’t yet read them.
Also, the Teotihuacan paper has detailed supplementary info here.
Thanks!
Y: From MPH’s blog: an example of a Teotihuacan pun, here, and an exploration of the Nahuatl-Corachol connection, here.
I thought we’d discussed both those posts before, but apparently not. The more reason to look forward to a discussion now.
From the (preview) paper:
In a Mesoamerican context we can also see a close equivalence between writing systems and archaeological cultures, with little overlap between these, suggesting that writing systems were each tied to a particular language (or at least language family), in which a given writing system was developed (Helmke and Davletshin 2019; Houston 2004:293-308; Justeson and Mathews 1991).
A tangential point, but when writing is found archaeologically, and it turns out that each archaeological culture has one written language, this seems like archaeological support for the shaky hypothesis that archaeological cultures correspond with linguistic communities.
But not quite. It doesn’t show more than that there’s a dominant language within an archaeological culture. And not even that, it shows that there’s one default language for the specific purpose of writing. And we can’t tell if the invention of writing itself led to the development of a default language, and if cultures without writing might be different.
But how many caveats can you pile upon each other before they too become shaky?
>* They’re obviously related, but that’s not the issue. You need common non-trivial innovations that can’t be explained by diffusion to demonstrate that (as MPH certainly knows as well as I do.)
In a comment in the 2020 preprint (p. 14), MPH seems to believe that he had done that in 2017, arguing against Dakin’s diffusionist reinterpretation of innovations previous authors used to justify the Nahua-Corachol grouping.
I haven’t read 2017 nor all of 2020, and I’m not sure I bring enough perspective to be a good judge once I do. But DE, I’ll be interested to hear whether you read the new paper, and whether it changes your view.
I’m not hostile to MPH’s view (and he knows this material vastly better than I do), and he could well be right, but I’m still not persuaded that he has really proven his point, based on the papers I’ve seen.
The new paper references Pharao Hansen 2020 as the paper showing that Carachol and Nahuatl form a real genetic subgroup: this is one of the papers that Y linked to above.
It seems to do a good job of demonstrating that these two branches are related to one another (no surprise), including in some ways that are not obvious and require some historical reconstruction, but as far as I can see does not really show that the reconstructed features are common innovations: there’s nothing much from other southern Uto-Aztecan languages, which you would need to adduce in order to show that the reconstructed Carachol-Nahuatl features are not shared more widely, and thus inherited in common from a yet earlier protolanguage.
(The supposed phonological common innovations are disputable, as the new paper quite properly points out.)
Now I am by no stretch of the imagination an Americanist, and for all I know the absence of the reconstructed features from other parts of Uto-Aztecan may be perfectly obvious to the relevant experts. But I’d still expect this to be shown rather more explicitly, if so, rather than tacitly assumed.
Retreating rapidly to a family that I actually do know about, I’m somewhat reminded of Coffi Sambiéni’s Le Proto-Oti-Volta-Oriental, a fascinating and detailed attempt to reconstruct proto-Eastern Oti-Volta.
This is an admirable and useful work, but it’s greatly weakened by two things.
Firstly, Waama, one of the four languages used in the reconstruction, is not in fact closely related to the other three within Oti-Volta.* This really only becomes clear when you compare with Oti-Volta languages outside this proposed “Eastern” group.
Secondly, all four languages are part of a Sprachbund, which is characterised (among other things) by some quite extensive shared phonological changes, (for example, devoicing of all *g *g͡b, and merger of *l and *d throughout.) This is confirmed by the lucky accident that there is also an unequivocally Western Oti-Volta language within the linguistic area in question, Nõotre, which participates in these changes: this proves that the changes in question are not, themselves, evidence for any “Eastern Oti-Volta” genetic grouping.
So what Sambiéni has actually reconstructed is not proto-Eastern Oti-Volta (which, in the sense he means, never existed), but proto-Oti-Volta, but with all the data from other branches missing, and with many areal features wrongly ascribed to the protolanguage itself.
* And the remaining three are not all that close to one another, either: the internal relationships within “real” Eastern Oti-Volta are much more distant than within Gurma or Western Oti-Volta, and more on a level with those between Western Oti-Volta and Yom/Nawdm. But I do think this “core” Eastern Oti-Volta is a real genetic group: there are some convincing common innovations. In particular, Nateni and Ditammari definitely form a sub-subgroup of their own, though this is far from obvious: Ditammari has fused noun-class prefixes, having mostly eroded away its inherited suffixes – yet another areal change, but this time, the area doesn’t include Nateni, but instead includes the Gurma language Akaselem and the non-Oti-Volta language Miyobe.
Turning from Uto-Aztecan to Danish onomastics, I was puzzled by the reference to “Pharao Hansen 2020” instead of “Hansen 2020,” because I had assumed “Pharao” was a middle name, i.e. secondary given name, rather than part of his surname. But maybe I was misinformed, and it certainly looks a bit more unusual for the Danish context than either Magnus or Hansen.
I cited it the way the man himself does (after looking in vain for “Hansen 2020” among the references.)
In hindsight, “Pharao” does look more plausible as a rococo spelling of some indigenous Danish surname than as a given name inflicted by an irresponsible parent.
The Social Security Administration tells me that Pharaoh has never been in the top thousand American boys’ names. However, I would be shocked if there weren’t a few guys out there with that name.
Having quickly skimmed the 2020 paper, I had the same impression: good analysis, but it needs to show that these putative common innovations are absent in other UA languages. Which is a shame, because he seems to be generally a thorough and rigorous linguist, and surely showing the absence of a feature in a language is not very much work, relatively.
The great jazzman Pharaoh Sanders (1940-2022) only had to slightly modify his birth name of Ferrell Sanders.* He went to a segregated high school in Arkansas which had been named after the quite-impressively-named Scipio Africanus Jones (1863-1943). Apparently the Danes standardly spell the egyptian-royal-title as “Farao,” but I assume there as elsewhere there’s no accounting for proper names.
*He seems to have started using the “Pharaoh” name professionally early on when he was in the orbit of Sun Ra, who helped to promote Egyptomania more generally in the avant-garde jazz world. You often see his birth name referenced as “Farrell,” which seems at least modestly more likely ex ante, but wikipedia goes with two e’s and I’m gonna assume they’ve got it right.
I see J.W. Brewer just beat me to mentioning Pharoah Sanders (note the spelling). By his telling, even Ferrell was some kind of reference to the Pharaohs in the Bible.
Pharrell Williams’s father and one of his brothers are also named Pharoah.
My point above is that Pharao Hansen is arguing against a line of reasoning from Dakin that pulled Nahuan out of a Corachol-Nahuan grouping proposed by Campbell and Langacker:
From the 2020 preprint:
>The most comprehensive argument for a subgrouping of Corachol and Nahuan so far has been Campbell and Langackers’ proposal of four phonological innovations that are exclusively shared by proto-Corachol (PCC) and proto-Nahuatl (Campbell & Langacker 1978
He may assume readers are familiar with C & L, and that they understand that by countering Dakin’s critique of C & L, and even adding a fifth shared innovation, he is strengthening the case already made by C & L.
I agree with DE that if that’s his line of reasoning, he should probably make it more explicit. But that does seem to be his line of reasoning.
He also wrote that he was making his full annotated cognate set available for analysis in the 2020 paper, but I didn’t find it in the preprint, so I wonder whether that may be in one of the tables left out pending proper publication.
If anyone read the above post in the 2-3 minutes after I posted it, please note that my cut-paste from the PDF initially left out the middle line, making the quote above incomprehensible. It is now restored.
Selecting text with the cursor works very strangely in PDFs at times.
BTW, Ryan, what you call the 2020 preprint is from 2024. The 2020 paper is in Spanish. The supplementary material to the 2024 paper is apparently paywalled like the article (which is unusual and deplored).
Whoops. Thanks for the correction.
Paywalling supplementary material along with the main text hasn’t been unusual in 15 years or more. Some journals keep the very existence of supplementary material from you if you’re outside the paywall.
In 2003 we visited the pyramid at Cholula, near Puebla. In the open-air area next to it the information panels were in Spanish, English and Nahuatl, not too surprisingly. A Japanese couple asked me if the Nahuatl was French. A strange question, but to Japanese eyes Nahuatl, Spanish, English and French must all look equally exotic.
Ryan: My point above is that Pharao Hansen is arguing against a line of reasoning from Dakin that pulled Nahuan out of a Corachol-Nahuan grouping proposed by Campbell and Langacker:
He may assume readers are familiar with C & L, and that they understand that by countering Dakin’s critique of C & L, and even adding a fifth shared innovation, he is strengthening the case already made by C & L.
Also, does it matter for the question at hand if the Corachol-Nahuan grouping is complete? Does it matter if it’s strictly genetic, or areal, or a bit of both? The reconstructed forms would still be valid and can be compared to the rebus readings of the glyphs.
Compare reconstructions of families with only two or three different but clearly related languages, e.g. Baltic.
I think it matters that it’s genetic. Aren’t the interpretations of the glyphs dependent on (or at least our confidence in them strengthened by) the sound laws? If documented similarities are areal (which presumably means younger), then the sound laws aren’t valid, and we can’t be confident in readings of glyphs nearly 2 millennia old.
If the similarities are genetic, it doesn’t much matter whether the family is incomplete. That just affects our understanding of the timing of the breakup of Southern Uto-Aztecan.
Athel, there was also Nahuatl signage when we went to Teotihuacan last winter. Sadly by my way of thinking, they no longer allow access to the pyramids above the base level. Though the Temple of the Feathered Serpent, which had not been opened at the time of my last visit, is more spectacular than the pyramids.
There have been some major advances in archaeology and understanding of the site in the last 20 years.
Admittedly a lot of it probably happened earlier but wasn’t readily available to the public.
Strictly areal would obviously matter. Substitute genetic with a strong areal component.
Edit: Right.
A Japanese couple asked me if the Nahuatl was French. A strange question, but to Japanese eyes Nahuatl, Spanish, English and French must all look equally exotic.
A French-Nahuatl word would be twenty letters but only two syllables.
“Teotihuacan” is actually pronounced [twã].
The truly enlightened eliminate the vowel and simply nasalize the [w].
In Nahuatl-Irish this would also be one syllable, but a completely different one.
Téon /tʃu:n/*. (Written “téodhghn” in the traditional orthography.)
* Add little superscript gammas to taste.
In the original Welsh, it’s Ty De Gwogawn “House of the South of Glory.” The reference is of course to Huitzilopochtli (“South Side of the Hummingbird.”)
For the metaphors involved, compare “the south end of a northbound mule”.
I’m reminded (more or less irrelevantly) of the way the city of Lyon has just one syllable in French, but English speakers usually make it two, stressing a syllable that isn’t there.
Malagasy surnames like Randriamanantena are easy to recgnize when written down, but I have the impression that most of the many syllables are ignored in speech.
At least some of the DJs on the classical station here stress the zeroth syllable of “Mstislav”.
@AC-B:
Re Malagasy, WP says:
Other accounts describe final -a in polysyllables as voiceless.
Final -a was actually added historically after word-final consonants; it’s usually held that this was the result of Bantu influence.
There are apparently dissentients: Alexander Adelaar is distinctly unimpressed with one here:
https://www.jstor.org/stable/23321849
The Adelaar paper mentions a “diminutive” prefix ki- in Malagasy, which he says is borrowed from Bantu, specifically, the Bleek-Meinhof Class 7 prefix; “diminutive” seems more like Class 12, *ka-, though I think Class 7 had indeed absorbed its diminutive function in Sabaki Bantu, e.g. Swahili kitoto “small child”, mtoto “child.”
The Bleek-Meinhof Class 12 prefix is probably cognate with the proto-Oti-Volta *-ka class suffix, likewise often diminutive; it appears in e.g. Kusaal biig “child” (cf Waama bika “child.”)
I love this kind of unexpected connection.
I agree with DE. When I go to large dictionary, I want it to tell me wampum and wapiti are related, and how, with cross-references.
In this regard, the OED Word of the Day for today is kitu kidogo, ‘East African. Money offered or accepted as an inducement or bribe.’ Here is the OED etymology:
That’s it? No morphological breakdown for kitu? I would have thought that the Swahili -tu (older /-tʰu/ from *-ntʊ̀) is etymologically the same *-ntʊ̀ in Bantu and ubuntu, as ‘some one, some thing’. Is this not correct?
The current (2018) etymology of ubuntu in the OED has a cross-reference to Bantu:
The revised (2022) Bantu has no cross-reference to ubuntu, though. But it does have cross-reference to the English slur muntu (which has a cross-reference to the slur munt). If I were organizing the revisions, I would point probably point all the cross-references to Bantu and list the related words there.
By the way, the OED has a nice citation from W. H. I. Bleek, who apparently introduced the the name Bantu for the linguistic group in Zulu Legends (1857):
I would have thought that the Swahili -tu (older /-tʰu/ from *-ntʊ̀) is etymologically the same *-ntʊ̀ in Bantu and ubuntu, as ‘some one, some thing’. Is this not correct?
At least as far as Swahili goes, yes: the human sense arises from the class membership, rather than the stem itself.
I’m not sure how far back that goes in Bantu in general, though. Eton, for example, has mòd “person”, bòd “people”, but that seems to be all for that etymon. And Zulu ubuntu, where the (u)bu- is just an abstract noun-class prefix, not anything specifically human, means “humanity”, not “essence.”
I don’t think the “thing” sense is likely to be primary, despite WP’s implication:
https://en.m.wiktionary.org/wiki/Reconstruction:Proto-Bantu/m%CA%8A%CC%80nt%CA%8A%CC%80
Mukarovsky’s “proto-Atlantic-Congo” *-nintu “person” is historical linguistics done by vibes. It’s presumably based on forcibly deriving the proto-Bantu form from this imaginary creation, along with e.g. Fulfulde neɗɗo “person” (plural himɓe, Because Fulfulde) and the proto-Oti-Volta etymon that underlies e.g. Kusaal nid, plural nidib.
However, the Oti-Volta forms probably really go back to just *nɪ-: cf e.g. Mooré nédà, plural neba. There are other Western Oti-Volta human-class nouns which insert an alveolar consonant after vowel-final stems before the human-class singular suffix -a, like Kusaal wiid(a) “hunter”, plural wiib(a). Kusaal has simply generalised the -d- to the plural in nidib; most other Oti-Volta languages have done the same, though not all.
The combining form for “person” in Kusaal is nin-; there is no trace of a stem-final nasal elsewhere, so this looks like it’s from a reduplicated *nɪnɪ. This is not unparalleled: the Kusaal adjective bil(a), where the stem is monomoraic bi-, with a diminutive class suffix -la, which is otherwise obsolete in Kusaal (but not in Mooré or Farefare), has the plural bibis.
So the POV forms do not support any “proto-Atlantic-Congo” *-nintu “person.” Like pretty much all proposals at that depth, it just evaporates on closer examination.
(Oti-Volta is definitely much more closely related to Bantu than either group is to “Atlantic”, which by common consent nowadays is in any case not itself a unified genetic group; I myself feel that no part of Atlantic has so far been rigorously shown to be related to Volta-Congo at all.)
Fulfulde neɗɗo, that should be, not eɗɗo (Too much last-minute editing.)
Kusaal bil(a) means “little.”
“Atlantic”, which by common consent nowadays is in any case not itself a unified genetic group
The putative cognates in the WP article on Atlantic (after Pozdniakov and Segerer) don’t look too bad: basic lexicon, good superficial similarity, not much unexplained dangling bits.
Fulfulde neɗɗo, that should be, not eɗɗo
Fixed — we don’t want people getting the wrong idea about Fulfulde.
The Adamawa language Samba Leko has nɛ́ “person”, plural nɛ́b; it’s one of the few nouns in the language that still has a distinct plural form.
John Rennison, in his Koromfe grammar, says
Historical linguistics is not his forte, but he’s right about ba, at any rate.
The putative cognates in the WP article on Atlantic (after Pozdniakov and Segerer) don’t look too bad
WAA Wilson in 1989 cited lexicostatistical work suggesting that the branches of Atlantic are as remote from one another and as isolated within Niger-Congo as Ijoid; John Merrill in his 2018 these states that ‟the Northern Atlantic languages are remarkably distinct from each other. Whatever genetic relationships exist between these groups must be extremely distant – perhaps more distant than can be satisfactorily recovered by the tools of comparative linguistics.”
Merrill does proper rigorous comparative work. He’s not keen on Pozdniakov’s work, and his explanations of why not are persuasive, I reckon.
The table of “Atlantic-Congo” cognates at
https://en.wikipedia.org/wiki/Atlantic%E2%80%93Congo_languages
significantly includes nothing from Atlantic.
(The “Gur” line leaves much to be desired; “name” and “blood” are spurious and should be removed: the real forms for “tooth” and “bone” are much more convincing (POV *ɲîn-dɪ́, *kpá̰b-dɪ́), and “eye”, “ear” and “mouth” should all have been given (POV *nîm-fʊ́, *tʊ̀w-dɪ́, *nô-dɪ́.)
If future scholarship does rigorously identify real genetic relatives to Volta-Congo, I wouldn’t be astonished if Dogon actually turned out to be more promising than Atlantic (or Kordofanian, for that matter.).There’s been a lot of sloppy thinking regarding the role in comparative work of noun-class systems (which are absent in Dogon), no doubt because they’re so in-yer-face in Bantu; scholars of “Niger-Congo” have mostly really been trying to create a kind of “Macro-Bantu.”
Thanks to Languagehat for the shout out and to everybody else for the comments. I am happy to answer questions. I agree that the 2020 paper did not adequately show that the shared features were exclusive innovations (though most of them are). But I think in the 2024 IJAL article I do show that.
I recently got a peer review from an anonymous reviewer who gave me strong vibes of being Lyle Campbell (I could be wrong), who told me not to spend too much time arguing for the validity of Corachol-Nahuan in the paper, since that could surely be assumed to be the consensus view (he is generally dismissive of Karen Dakin’s work).
Regarding Danish onomastics:
“Turning from Uto-Aztecan to Danish onomastics, I was puzzled by the reference to “Pharao Hansen 2020” instead of “Hansen 2020,” because I had assumed “Pharao” was a middle name, i.e. secondary given name, rather than part of his surname. But maybe I was misinformed, and it certainly looks a bit more unusual for the Danish context than either Magnus or Hansen.”
In Denmark we generally allow only one last name and any first names inherited from the other parent are placed as “middle names” along with the first name. The rule of thumb is that any patronyms ending in -sen always go last and become the surname. But since most -sen names are very common we tend to use the first surname (which is technically a middle name) for identification purposes. So for example our three Rasmussen primeministers are generally known as “”Poul Nyrup”. “Anders Fogh” and “Lars Løkke”. In the same way I am generally known as Magnus Pharao – Pharao being in fact my paternal surname, since I inherited my patronym from my mother. My daughter’s last name is Pharao, since I gave them my paternal surname in the Spanish style (their mother is Mexican). This means that sometimes people are confused that we don’t share a surname (since they lost Hansen). So a while ago I decided that it makes more sense to use “Pharao Hansen” as my full last name, the way it is done in Spanish speaking countries, which would also give the expected Spanish order of patronym and make me more easily identifiable. I probably will eventually make the change official, but for now I only try to enforce it bibliographically.
As for the history of the name Pharao it can be traced back to a family of craftsmen in Angeln in Northern Germany in the early 16th century. It is hard to know how it might have arisen there.
Thanks very much for those elucidations — I’m especially pleased to know the onomastic details.
Brett says
October 15, 2025 at 4:08 pm:
The Social Security Administration tells me that Pharaoh has never been in the top thousand American boys’ names. However, I would be shocked if there weren’t a few guys out there with that name.
======
Over 300 are listed here:
https://www.findagrave.com/memorial/search?firstname=Pharoah&middlename=&lastname=&birthyear=&birthyearfilter=&deathyear=&deathyearfilter=&location=&locationId=&bio=&linkedToName=&plot=&memorialid=&mcid=&datefilter=&orderby=r&page=1#sr-26909168
Thank you, Magnus.
Is the “consensus” on Nahuatl-Corachol based on unreliable evidence, like lexicostatistics? If so, that consensus means nothing, and rigorous evidence based on shared innovations is absolutely necessary.
The Teotihuacan visual pun matching the reconstruction of PNCH brings to mind Hrozný and the laryngeals in Hittite. That sort of confirmation is vanishingly rare.
Over 300 are listed here
300 Pharoah’s; 137 Pharaoh‘s.
rigorous evidence based on shared innovations is absolutely necessary
Lyle Campbell is of course not renowned for lumpery*, though I suppose that might perversely make him a bit overconfident over such hypotheses about genetic relationships as he has himself espoused. “If even I think these languages are related/form a subgroup …”
And accurate subgrouping is hard. Significantly more so than “merely” showing that two languages are genetically related.
* It also seems to be the case that lumpers in one linguistic family can be splitters in another: Gerrit Dimmendaal and Roger Blench come to mind. But probably irrelevant here, what with Campbell’s primarily Americanist focus.
No, the evidence for the group has been primarily exclusively shared phonological innovations proposed by Campbell and Langacker. The problem is that the original innovations were not that impressive, but the sort that could conceivably arise independently or through contact.
In fact lexicostatistics has been primarily used to cast the grouping into doubt, which is why I have spent time arguing that the lexicostatistics was skewed by unrecognized cognates and missing data. Greenhill et als 2023 paper was the first lexicostatistical study (other than my own counter study) to find support for the grouping based on cognate density. Campbell and Hill were among the coders of cognate in that study so it’s not 100% independent confirmation. But you could argue mine is.
In any case, in the 2024 paper I add another phonological trait that characterizes the grouping, which I think is more complex and therefore weighty, namely the stress pattern that conditions loss/retention of *p. And a bunch of uniquely shared cognates.
Campbell and Langacker write in a footnote that “Ives Goddard (“A Preliminary Survey of the Uto-Aztecan Pronominal System” [manuscript]) has suggested a certain amount of grammatical evidence for this subgrouping (based on pronouns and morphology).” Does that one overlap with your evidence?
Kaufman’s unpublished manuscript on Uto-Aztecan classification (available at AILLA) is unkind to Campbell and Langacker: there’s a section aimed at that paper called “Langacker’s etymological atrocities”, along with general criticism of Langacker’s “Boppesque” methods. Do Kaufman’s arguments invalidate those of C&L about the Nahuatl-Corachol subgrouping, aside from the weaknesses which you point out?
And a bunch of uniquely shared cognates
In Oti-Volta, there are at least fifteen or so cognates shared exclusively between Waama and Outer Oti-Volta (i.e. Western Oti-Volta, Buli/Konni and Yom/Nawdm), including some very basic vocabulary (e.g. “house”, “tongue”, “evening”, “hear”, “swallow”, “kneel”), with none of the rest being particularly obscure items, either.
I was so struck by this that I tried quite hard to find any other evidence supporting a Waama/Outer Oti-Volta subgrouping. Nothing, really.
So presumably they must all be common retentions. There actually is some phonological and morphological evidence that the Oti-Volta branches that lack these etyma do themselves form a subgroup, albeit an internally very diverse one. Geographically, they’re all fairly central, adding some plausibility to their having shared innovations which didn’t spread to the periphery.
(A bit like the satem languages … only with more evidence of common inheritance rather than this being just an areal thing of some kind.)
Talking of areal things, the Western Oti-Volta language Nõotre, which is closely related to Kusaal, shares with its much less closely related Eastern Oti-Volta neighbours not only some systematic sound changes but also some distinctive vocabulary, e.g. hɔ̃si “bite” (Byali hãsi, Kusaal dum), tùbré “spear” (Byali tōbídē, Kusaal kpan.)
However, I don’t have any examples where there would be a clearcut difference in outcomes between cognates and borrowing. (The tone correspondences in “spear” suggest cognacy rather than borrowing, but both Nõotre and Byali have done weird things with their inherited tone systems, so that doesn’t mean a lot.)
Nõotre is the only Western Oti-Volta language that has the original proto-Oti-Volta etymon for “navel”, Nõotre sìifó; cf Mbelime sūdìfɛ̀, but e.g. Mooré yũ̀ugá.
Unfortunately, Byali happens to be the only Eastern Oti-Volta language where the regular development of root-final *d is to /j/ (with frequent fronting of the preceding vowel), exactly as in Western Oti-Volta: Byali sīífə̄. So Nõotre sìifó could also equally well be cognate with, or loaned from, the Byali word.
Y Wrote: “Campbell and Langacker write in a footnote that “Ives Goddard (“A Preliminary Survey of the Uto-Aztecan Pronominal System” [manuscript]) has suggested a certain amount of grammatical evidence for this subgrouping (based on pronouns and morphology).” Does that one overlap with your evidence?
Kaufman’s unpublished manuscript on Uto-Aztecan classification (available at AILLA) is unkind to Campbell and Langacker: there’s a section aimed at that paper called “Langacker’s etymological atrocities”, along with general criticism of Langacker’s “Boppesque” methods. Do Kaufman’s arguments invalidate those of C&L about the Nahuatl-Corachol subgrouping, aside from the weaknesses which you point out?”
I confess I haven’t read Goddard’s manuscript, but there are pronominal morphology that is uniquely shared between Wixárika and Nahuatl – but given that it is mostly not shared with Náayeri (Cora) we can’t really reconstruct tht for PCN. It seems to be the case that Pre-Nahua was in close contact with proto-Wixárika. But the shared changes are really quite astonishing, both pre-Wixárika and pre-Nahua underwent a complete reorganization of the morphosyntactic profile becoming polysynthetic languages with obligatory marking of all arguments. Wixárika falls short of being omnipredicative because it retained the switch reference system, but otherwise very similar to Nahuatl and distinct from Náayeri morphologically.
The main exclusively shared morphological innovation I have noted is the loss of the accusative case system, where accusative turns into locative markers but stop marking syntactic relations. This can be reconstructed for PCN.
Kaufman believed that the traits shared by Corachol-Nahuan were diffused through contact. But he never did any midlevel reconstruction on UA, only at the PUA level and using very simplistic sound correspondences. So no he doesn’t invalidate them, he just contradicts them. And he never seems to have taken a close look at the relation.
Regarding Kaufman’s work on Uto-Aztecan I could write a lot, and it would be very critical. I was raised academically to believe that Kaufman was a genius and everything he said was right and that was the reason we didn’t need to expect him to support his claims with evidence or often even with arguments. As it turns out, he may have been a genius, his Mayan, Mixe-Zoquean and Zapotecan work is very good, but his Uto-Aztecan work is not nearly as good as his own confidence in it would have it. His tirades against Langacker I consider outright shameful, especially because I think it stifled research on Uto-Aztecan consonant gradation and for decades made scholars reluctant to work on Uto-Aztecan historical linguistics because Kaufman had already solved that (and who would want that treatment from the commanding figure in the field…). As it happens, as you can see in my 2024 article, I think Langacker was right and Kaufman wrong – consonant gradation is an important feature in Southern Uto-Aztecan – and if we don’t recognize it we cannot do etymology in Southern Uto-Aztecan and we will only see half the cognates.
One of the problematic consequences of Kaufman’s rejection of Langacker’s proposal of consonant gradation was that he was unable to see that lots of examples of p in Nahuan are inherited since PSUA *p is only lost in unaccented syllables. This in turn caused him to believe that all of the words with p in Nahuan were loans from other Mesoamerican languages, causing him to believe that Nahuan has taken a huge number of loans compared to other languages in the area – which is simply false. By recognizing consonant gradation and finding the conditioning mechanism (or one of them) one recognizes tonnes of new cognates many of which Kaufman claimed to be loans. Kaufman’s major thing of course was contact/diffusion, and he had complete faith in his own ability to recognize loans on sight. But that confidence made him stop short of looking for alternate patterns. Maybe Kaufman ought himself to have been a little more Boppesque?
I find it odd to use “Boppesque” as an insult… I sort-of-recently read a chunk of his “Über das Konjugationssytem”, and while of course it was out of date in plenty of ways, I was overall pretty impressed, given that he was, basically, inventing a new field. I mean, if you just mean “pre-Neo-Grammarian”, sure — but I don’t really see the point of using someone’s name as an actual insult just because they lived a while ago and so predated some important intellectual advances in their field.
Kaufman wrote of Langacker: “Many of RL’s formulations have an uncomfortably Boppesque feeling about them. He derives affixes from unattested roots with great freedom. Among affixes everything that is functionally equivalent and vaguely similar in pronunciation is assumed to be phonologically equivalent, regular sound correspondence not-withstanding”
His critique of Langacker was that his idea of consonant gradation allowed him to great a leeway in proposing cognates, leading to a lack of rigor. Since Langacker expcted less than perfect sound correspondences, explaining away non-matching consonants as results of lenition. This is the basis for much of the list of “Langacker’s etymological atrocites” .
Kaufman on the next page writes that “progress in reconstruction in the framework of the comparative method involves recognition of previously unnoticed regularities and expanding or contracting the scope of previously reocgnized ones”. But Langacker had precisely noticed some previously unnoticed regularities in consonant correspondences, though he was yet unable to account for the factors conditioning them.
Incidentally I agree with Kaufman regarding many of the rejections of Langacker’s cognate sets, which do sometimes include forms that do not seem to be related but only superficially similar.
But the problem is that if Langacker was too quick to accept cognates that did not fully match the recognized correspondences, Kaufman was equally quick to accept his own posited sound correspondences as the final word and conclude that everything that doesn’t match has to be borrowings.
Maybe he was thinking of Bopp’s IE-Austronesian experiment?
As it turns out, he may have been a genius, his Mayan, Mixe-Zoquean and Zapotecan work is very good, but his Uto-Aztecan work is not nearly as good as his own confidence in it would have it.
This is so frequently the problem with geniuses: they do great work in some area, see it universally acclaimed, and conclude that they are universal geniuses and their every thought about everything is automatically correct. (And then they are surprised and infuriated by criticism.)
The point when I realized that this was the case also with Kaufman, was when I read an extremely critical review by a major Celticist scholar of Kaufman’s decipherment of Tartessian as Celtic.
Joseph Greenberg comes to mind: undoubtedly a great Africanist and linguist, with major lasting achievements to his name, who was also responsible for a lot of stuff that was … not so great.
Greenberg does not seem to have been given to furious responses to criticism or jeremiads against dissenters, though.
The odd thing is, Kaufman knew the Comparative Method and its worth, had applied it to other languages, and purported to rigorously apply it to UA too. And he did a lot of descriptive work on Nahuan languages. It’s odd that he’d be off with regard to that one language family.
Wixárika falls short of being omnipredicative because it retained the switch reference system, but otherwise very similar to Nahuatl and distinct from Náayeri morphologically.
Is Corachol certain, then? Or is there a possibility that the primary split is between Náayeri and the other two?
This is so frequently the problem with geniuses: they do great work in some area, see it universally acclaimed, and conclude that they are universal geniuses and their every thought about everything is automatically correct.
Worse yet when people then acclaim their incorrect thoughts, as Magnus described.
On the other hand, all this gives the rest of us the chance to be right when geniuses are wrong.
I wonder how much Kaufman’s manuscript was in circulation.
Belief in your own infallibility is seriously corrosive to the brain. This is also why so many billionnaires are surprisingly stupid.
I gather that the standard convention for depicting wisdom in Buddhist iconography is large ears.
“And he did a lot of descriptive work on Nahuan languages.”
Actually he didn’t. He did field work on one Nahuan variety in the 1970s and never published any of it.
“Is Corachol certain, then? Or is there a possibility that the primary split is between Náayeri and the other two?”
I actually couldn’t entirely exclude the possibility of a Wixárika-Nahua subbranch and Náayeri as a singleton. But there are some shared phonological innovations for coracholan and a lot of exclusively shared vocabulary. And also some grammatical features.
Thanks for the clarifications about Boppism. It is a more fair usage than I’d initially though (sort of a shorthand for the early stage of the field), but it still seems needlessly uncharitable to Bopp. Surely we usually celebrate Grimm (and Rask) for discovering Grimm’s law, rather than rejecting a sound change as “Grimmian” because it has unexplained (Verner-like) exceptions?
I also hadn’t realized this was the same Kaufman from Tartessian! (It’s not an uncommon name, and in particular I have a lot of Kaufman relatives, so I tend to assume it belongs to different people in different contexts. Wrongly, in this case.)
Belief in your own infallibility is seriously corrosive to the brain. This is also why so many billionnaires are surprisingly stupid.
And politicians too. Ⓐ ☮︎
Kaufman was a proponent of Hokan as well. Zhivlov’s recent ongoing work on Hokan speaks favorably of Kaufman’s work, though naturally it amends it. The main problem was that Kaufman published his reconstruction but without the supporting data, which meant it could not be evaluated properly.
The most recent paper is here; it’s a review of the history of the hypothesis, and rather harrowing to read in parts.
I guess peer review hadn’t reached historical linguistics or Americanistics yet?
Zhivlov’s part 2 is just out.
rather harrowing
The bulk of Hokan research for many years was throwing data at the wall, and concluding with the hope that someone in the future will check and see what stuck.
I guess peer review hadn’t reached historical linguistics or Americanistics yet?
That was in a proceedings volume, and took 100+ pages. He should have published it as a monograph, with all the data.
(N.b.: I have just rescued a comment by Magnus from moderation — don’t miss it!)
Actually he didn’t. He did field work on one Nahuan variety in the 1970s and never published any of it.
Thanks. I misread the description he’d written of an extensive language documentation project, which had a lot of languages listed, and assumed he’d worked with all of them.
He led the Mesoamerican Language Documentation Project that aimed to make dictionaries of some 20 languages. None of them were ever published. (Except one which the author Una Canger self-published eventually). Kaufman generally didn’t publish his data for his reconstructions – except for his Mayan etymological dictionary.
Phar[oa]h Sanders … in the orbit of Sun Ra
I should have also mentioned, as within the remit of this blog, that it was embarrassingly recently that I figured out that the name of ESP-Disk’ (the label responsible for introducing Sun Ra to a wider audience) comes from Esperanto. Bernard Stollman’s first release was a sing-along recording.