Category: Linguistics

Over the past week, the New York Times has been running a five-part essay by filmmaker Errol Morris, entitled ‘Whose Father Was He?’. For the most part, this is a fascinating account of the history of a single Civil War photograph of three children found on the dead body of a then-anonymous Union soldier at the Civil War battlefield of Gettysburg. The photo was widely distributed through the press, and thereby became the means by which the soldier, Amos Humiston, was identified.

Part 1
Part 2
Part 3
Part 4
Part 5

For the most part, the essay is a detail-oriented historical and biographical piece focusing on the way in which the print media was involved in shaping life histories relating to the Civil War, and the way in which family histories become entwined in the semi-mythical aspects of military history. You should read it for that reason.

The article takes a turn for the truly bizarre in part four, however, because Amos Humiston’s great-grandson is the Canadian-American archaeologist David Humiston Kelley, who is best-known for his work in the 1970s helping to decipher the Maya script as a phonetic (rather than ideographic/semantic) writing system (Kelley 1976) and more recently for his work in New World calendrics and archaeoastronomy (Kelley and Milone 2005). He is also infamous in archaeological circles for his advocacy of long-range cultural diffusion, from Southeast Asia to Mesoamerica and also from Egypt to Mesoamerica. He is also a genealogist who claims to have traced his own lineage back to the Biblical King David (!!!). In other words, despite his erudition, in parts of his scholarship he is really no different from any number of other hyperdiffusionist pseudoarchaeologists postulating multiple events of long-range cultural contact on the basis of minimal or no evidence. And much of that evidence is linguistic rather than archaeological, bringing this news article well into the scope of this blog.

Morris’s article gives considerable space and attention to Kelley’s work on transoceanic contact. In one sense this is very surprising given that the article is on a very different subject, but since Kelley is one of his major informants about the Amos Humiston story, and since he makes such sensational claims, it’s perhaps not so surprising that this would end up being part of his story. But because this is not an article about transoceanic contact, but rather an interview with Kelley in which the journalistic obligation to be critical of one’s sources applies, Kelley’s theories take on an authority that they clearly lack.

The first sign that something is wrong is in part two of the essay, where Morris writes:

And I’ve got a copy of his most recent book here. It’s called “Exploring Ancient Skies: An Encyclopedic Survey of Archaeoastronomy.” [4] I open this book to any page at random, and it’s virtually unintelligible to me. But I’m sure that it’s the definitive word on the subject.

When someone tells you that they can’t understand something at all, but are convinced of its absolute truth, this should be a warning sign. In part four, Morris conducts an extended interview with Kelley, in which Kelley says:

I’ve found out, recently, that there were Egyptians in Meso-America. I had thought there were connections, but I had thought they were secondhand through an intermediate, perhaps through Phoenicians or Greeks or somebody. But I didn’t think they were directly Egyptian. But I now have massive evidence that they were.

When queried about this evidence, Kelley responds:

Three different calendric types of continuity. That’s one sort. Then I’ve got over 30 deities and mythical place names, starting with Egypt itself. The Aztecs say that they came from Tlapallan, which is the ancient red land. And the Egyptians called their land red land/black land. The Aztecs actually called it Tlillan Tlapallan, which is black land/red land. And they were under the leadership of the inventor of the calendar, who was called Cipactonal. And Cipactli means “crocodile,” and Tona is “day” and is related to the word Tonatiuh, which is “sun god.” And Tona relates to Aton in Egypt. And Cipactli relates to Sebek or Sobek in Egypt. So you’ve got linguistic evidence for a very complex name.

Right now you may be tempted to say ‘Ooh, Aton … Tona!’ And that is exactly how the ‘method’ works: a parallel is presented as obvious, without any further explication deemed as necessary. And if archaeologists and linguists haven’t spotted it, this just confirms that they are either ignorant or part of a dark conspiracy. But the claim falls apart fairly quickly.

The classical Nahuatl language is attested from a set of texts, primarily 16th century, written in the early colonial period just after the arrival of the Spanish in Mexico. Our phonetic reconstruction of Nahuatl is excellent not only because the Spanish and colonial Nahua wrote it down in the Latin alphabet, but because it’s still spoken today. Prior to the 16th century we have basically no direct evidence for the structure of Nahuatl because the Aztec writing system is only minimally phonetic.

The Egyptian languages are a set of interrelated languages spoken and written from at least the fourth millennium BCE through to the Roman period, with one language, Coptic, used primarily as a liturgical language for the past millennium. All phonetic transcriptions of Egyptian are made more complex by the fact that the Egyptian writing system can be highly opaque in its phoneticism, and because in particular, vowel sounds are underspecified. Egyptian seafaring was at its height in the New Kingdom (1570-1070 BCE).

So if there was contact between Egypt and Mesoamerica that left linguistic traces, it was well before the sixteenth century – and ‘realistically’ (and I use that word advisably) was probably in the second millennium BCE, thousands of years before we have an attested Nahuatl language. What did Nahuatl look like at that point? Well, gee, we don’t know. In fact, there probably was nothing even remotely resembling Nahuatl. Now Nahuatl is a Uto-Aztecan language, and we could ask a historical linguist what the word for ‘sun’ looks like in the various languages of the family. Okay, let’s do that, using the work of Karen Dakin, a top linguist interested in Proto-Uto-Aztecan (the source I’ve used is one of the few that won’t require you to have a subscription). We find out (Dakin 1996: 13) that the reconstructed proto-Uto-Aztecan (PUA) word ‘sun, day’ is *ta-pi, and that its modern descendant – however improbable it may seem – in Nahuatl is ilwi-tl (see footnote on p. 13 for an explanation)!

Indeed (from a less scholarly source) in the vast majority of modern descendants, the words begin with /ta/, and almost none of them contain /n/ – only the Aztecan languages. In fact, ‘tona’ meaning ‘sun’ is simply a semantic extension of the Aztecan verb ‘to shine; to shimmer; to radiate heat’ (Hosler 1995: 106). And the Aztecan languages probably only branched off from the other Uto-Aztecan languages in the middle of the first millennium CE (6th-8th centuries) (Luckenbach and Levy 1980). And of course, by the 6th century (not to mention the 16th century), there simply were no Egyptians to make the trip, so to compare 16th century Nahuatl and ancient Middle Egyptian on these grounds is utterly pointless.

But for completeness, let me just point out that another word for ‘disk of the sun’ in Egyptian is ra (Aten is an aspect of the god Ra), the verb ‘to give off light, to shine’ is wbn (phonetically perhaps /uben/), and the sun’s rays are stwt /setut/). When you can pick any of these as a possible source of similarity, and can arbitrarily change them (Aton –> tona), virtually anything can be a parallel. We also have the problem that all we really have is tn in Egyptian writing, which underspecifies vowels. We think that the inital vowel was /a/ but the interior vowel could be almost anything. So we don’t have /ton/; we have /ton/ or /tun/ or /ten/. Now, can we imagine linguistic changes that would turn aton to tona? Sure we can. But are these attested changes? No, definitely not. Even if the chronology were right (which it isn’t), the resemblance is at best a superficial one – and Kelley, who is no stranger to historical linguistics, surely knows this.

Now let’s look at a parallel example from Kelley’s use of iconography, to give a sense of the archaeological side of the argument:

One, rarity of occurrence and two, specificity of unusual arbitrary characteristics. Arbitrary characteristics, particularly ones that are unusual, are good evidence. Things like a lion’s head with pink and white whiskers on a snake’s body. I’ve got the lion’s head in Egypt, and I’ve got jaguar heads in Meso-America, with the pink and white whiskers. I have jaguars with snake bodies, but they aren’t specifically identified with the jaguar with the whiskers. But still, when you put the two together, it makes a reasonable similarity with this Egyptian one. And it’s a very arbitrary similarity.

Here we have an iconographic resemblance, but the argument is essentially of the same structure: ‘Hey, look at these two things; aren’t they similar?’ But Egypt is not a unified entity, but a millennia-old civilization with numerous phases and iconographic styles, and ‘Meso-America’ is incredibly underspecified both geographically and temporally: is it Olmec (ca. 500 BCE), Maya (ca. 500 CE) or Aztec (ca. 1500 CE)? All three groups spoke completely unrelated languages to the other two, and lived in very different parts of Mesoamerica (without denying that there was interregional cultural contact). Is the pink-whiskered jaguar contemporaneous with the snake-jaguar (or, as Kelley admits, even the same jaguar)? Who can say?

Kelley’s argument is that hybrid cat-snakes are an arbitrary combination of forms that are unlikely to occur in multiple regions by chance, because (of course) there is no such thing as an actual hybrid cat-snake. But when Kelley asserts that “Felines, of any sort, do not have snakes’ bodies. And neither do they have red and white whiskers.”, and that this proves diffusion, he is just recycling Fraser’s (1965) argument that the lack of a real percept demonstrates diffusion wherever there are similar hybrid animals. But in fact, as Wittkower (1938-9) showed decades ago, some hybrids (e.g., bird-serpent) are extraordinarily common cross-culturally, so Kelley’s claim that “It isn’t due to the common workings of the human mind” is at best premature.

The thing is, we actually don’t have a very good idea of how the human mind comes up with these things, but there’s pretty substantial evidence that we don’t do so in a way that is arbitrary. The nature of the argument that Kelley is raising (and in fact has been raising for the past several decades) is that these similarities are too numerous and too arbitrary to have occurred by chance. But it’s very easy to assert that, but much harder to demonstrate (what does ‘by chance’ mean? how would we evaluate it? what statistical universe are we talking about anyway?) Human beings, as members of a pattern-seeking species, tend to attribute a lot of meaning to these, but anthropologists and archaeologists have come up with no reliable set of criteria to allow us to distinguish independent inventions from cultural borrowings. We’re pretty sure that if we find a Coke can in Zambia that Coke wasn’t independently developed twice; we’re unsurprised if the bow and arrow is developed in many different places. But in the intermediate ground, we still don’t know too much. The linguists are far better equipped to deal with this sort of thing than anthropologists and archaeologists, and this is a serious problem.

In my (now-in-production-and-hopefully-out-later-this-year-knock-on-wood) book, I outline six criteria used to identify that one numerical notation system is descended from another (rather than being independently developed), roughly in order of importance:

1) Use of two systems at the same point in time
2) Similarity in forms and values of numeral-signs
3) Similarity in structural features
4) Known cultural contact between the regions where the two systems are used
5) Use of the two systems for similar purposes and/or on similar media
6) Geographic proximity

My goal was not to produce a general theory of independent invention – I don’t think that numerical notation systems are borrowed in the same way as iconographic features, for instance. But many of the same principles are going to apply. Kelley has failed to demonstrate contemporaneity, and his linguistic and iconographic similarities are only superficially so. I don’t think we even need to get to the issue of geographical distance (which is obviously immense) between Egypt and Mesoamerica to put this to rest.

I won’t even begin to address the fact that for decades, Kelley has emphasized his belief that the calendric, linguistic, and archaeological evidence demonstrate diffusion from South/Southeast Asia to Mesoamerica (transpacific rather than transatlantic). You can be a diffusionist, sure, but when your data allow you to derive two radically different routes, you really need to pick. And I’ll only mention in passing that even if there were a real, provable analogy, it’s just as likely that it went from Mesoamerica eastward across the Atlantic (a theory the hyperdiffusionists never seem to like for some reason).

What really upsets me is that even though his theories are extreme and contradictory, Kelley’s words are permitted to stand unchallenged in a major news outlet. The parallel that Morris is trying to draw throughout the article is that the reconstruction of the ancient past is analogous to his own task (and Kelley’s) of reconstructing Civil War history out of scraps, or identifying a dead soldier from a single photograph of his children. And sure, there are some similarities. But just as Aton and Tona share a resemblance but no historical connection, the parallel is ultimately a hollow one. I can’t fault him for being interested in Kelley’s work – after all, it is compelling, audacious, and controversial. But the New York Times, very sensibly, does not publish stories arguing that there were numerous Egyptian visits to the New World that left lasting linguistic and material traces. Hyperdiffusionism gets in the back door, because it’s not in the Science section of the paper. It’s bad enough that this stuff gets through into the popular literature, and even past peer review, because of the disjunct between archaeological and linguistic expertise. For it to stand without even a journalistic ‘But this is not a widely accepted theory’ is unforgivable.

Dakin, Karen. 1996. Long vowels and morpheme boundaries in Nahuatl and Uto-Aztecan. Amerindia, vol. 21.
Fraser, Douglas. 1965. Theoretical Issues in the Transpacific Diffusion Controversy. Social Research 32: 452-477.
Hosler, D. 1995. Sound, color and meaning in the metallurgy of Ancient West Mexico. World Archaeology: 100-115.
Kelley, D. H. 1976. Deciphering the Maya script. University of Texas Press.
Kelley, D. H., and E. F. Milone. 2005. Exploring ancient skies: an encyclopedic survey of archaeoastronomy. Springer.
Luckenbach, A. H., and R. S. Levy. 1980. The Implications of Nahua (Aztecan) Lexical Diversity for Mesoamerican Culture-History. American Antiquity: 455-461.
Wittkower, Rudolf. 1938-9. Eagle and Serpent. Journal of the Warburg and Courtauld Institutes 2: 293-325.

I spent much of the past week in Montreal, Quebec, my erstwhile home, with six junior colleagues (thanks again everyone!), conducting field survey for the Stop: Toutes Directions project. This research project involves field survey of around 40 km2 of the city, including part or all of nine distinct boroughs or municipalities, examining linguistic and other aspects of variability in the city’s stop signs. With the completion of last week’s field survey, we have now identified, classified and photographed 3522 stop signs.

Our central research questions relate to the fact that there are three valid stop sign texts in Montreal – STOP, ARRET, and ARRET/STOP. Each variant is distributed differently across the survey area; moreover, each variant displays different patterns of wear/fading that seem to correlate well with age. These three text languages, all abundantly attested in our survey area, coupled with Quebec’s strict language laws and language ideologies prohibiting English in most contexts, make Montreal a unique laboratory for investigating the ways in which bilingualism are reflected in visual and material culture.

A major goal of our most recent survey period was to identify areas where stop signs are stamped with dates of manufacture, thus allowing us to correlate chronological age with the ordinal index we had developed to measure wear. This allows us to treat the linguistic landscape not only as a synchronic snapshot of a particular point in time, but as a diachronic palimpsest of episodes of stop sign erection and replacement. Because there are nine distinct municipalities in our compact survey area, we are able to readily identify differences not only in linguistic practice, but also in attitudes towards replacement and repair of stop signs.

Of further importance is the fact that 27% of the stop signs we have surveyed have vandalism either on the front or back. This means we not only have a corpus of official stop sign texts, but also a corpus of unofficial writings. While stop signs are designed to be read by drivers, the practice of stickering on stop signs is clearly intended for a pedestrian audience. We are then using this data to explore relations between vandalism and all sorts of social data derived from Statistics Canada. We are also able to identify many instances where stop signs are vandalized in ways that reflect dissatisfaction with, or playfulness towards, the existing ‘official’ linguistic landscape.

Conceptually and methodologically, I think the importance of this work is that it shows how methods that are essentially archaeological or epigraphic can be employed to investigate topics in contemporary linguistic anthropology. What’s more, we are confident that our data are producing answers to questions that would be answered very differently – or in some cases couldn’t even be asked – using ethnographic methods. Not that I don’t think ethnography would tell us anything, but it wouldn’t be well-suited to the questions we are asking, and the answers that even experts (i.e. the officials responsible for managing the erection and replacement of stop signs) could give us would be very different (and in many ways more subjective) than what the material culture of signs actually tells us.

At present the Stop: Toutes Directions website linked above contains a large number of excellent but preliminary research reports produced last year. Now that we are (nearly) done our data collection, I and my co-authors are moving into the next stage of the project, including seeking peer-reviewed venues for publishing our work.

Apparently my post ‘Paleolinguistics and archaeolinguistics’ is currently the #2 result for the keyword ‘archaeolinguistics’ (and #4 for ‘paleolinguistics’). Paleolinguistics is a fairly well-known term among linguists): it refers to the extension of historical linguistics deep into prehistory, often using methods that are not accepted by the majority of historical linguists. But while archaeolinguistics hasn’t achieved the same fame (or at least, not Wikipedia noteworthiness!), that’s the topic I want to discuss.

Archaeolinguistics lies at the interdisciplinary intersection between archaeology and linguistics – including but not limited to the interaction of linguistic and archaeological anthropology. I want to distinguish linguistic analyses of prehistory that do not use the archaeological record (paleolinguistics) from those that do. Moreover, much of archaeolinguistics has nothing to do with historical linguistics at all, but rests on different sorts of intersections between the two disciplines.

Archaeolinguistics includes several distinct topics of study:
– The study of the evolution of language and symbolic behavior through the integration of Paleolithic archaeology (lithics, art, notations, etc.) and studies in cognitive linguistics. This has virtually nothing to do with ‘paleolinguistics’ as an extended form of historical linguistics, but it requires a good foundational knowledge of both archaeology and linguistics, and also of hominin evolution.
– The study of prehistory through the comparative use of historical linguistics and archaeology, e.g., to reconstruct proto-language homelands, prehistoric migrations, subsistence patterns, the diffusion of technology, and the like. Where two independent sources of information converge on the same answer, it is more likely to be correct than when one line of evidence alone is used. This is ‘paleolinguistics plus’: the archaeological record is (dis)confirmatory and serves as a check on wild speculation.
– Archaeological decipherment: the decipherment of ancient texts recovered in archaeological contexts. This relies on quantitative analysis of texts and their signs, as well as more interpretive aspects of decipherment that rely on knowledge of social contexts that can mainly be known archaeologically. Maya script decipherment is a classic example of this ongoing process; without the archaeological record, our understanding of the hieroglyphic texts would be substantially hindered.
– The use of written texts to complement the archaeological record of literate societies to discuss topics of interest to linguistic anthropologists: literacy, cognitive categories, language contact, dialectal variation, and so on. In contrast to decipherment, here the script is well-known, and the questions that are being asked are using textual material holistically with archaeological material to talk about linguistic aspects of ancient life.
– The use of the characteristics of written texts to date archaeological material, and vice versa. Paleographic changes in scripts can be highly suggestive if not definitive of the age of texts, and of associated archaeological material. Conversely, archaeological materials that are datable radiometrically can put associated texts in their temporal context.

If this seems big and vague, maybe it is. All that these things have in common is that they require some knowledge of archaeology, and some knowledge of linguistics. But I guess what I’m trying to do here is to make the case that when there are so many areas that require knowledge of both fields, that it is valuable, from a scholarly perspective, to train students who are knowledgeable in both fields, and to publish work that reflects that intersection.

Over the next few months, I’m going to be meandering through a series of posts on some of the specific topics mentioned above, and more generally on methodological, conceptual, and evidentiary similarities between the two fields that make archaeolinguistics ‘hang together’ better than one might think. I don’t know what will become of these thoughts ultimately – maybe even a short book.