onsdag den 9. december 2020

Nahuatl Scholar Interview: R. Joe Campbell

Believe it or not, this was Joe's first selfie.
It's Joe on the left.
(taken in Guadalajara July, 2018)
In this blogpost, I give you an interview with one of the great Nahuatl scholars alive today, R. Joe Campbell. I have had the good fortune to meet Joe a number of times at Nahuatl meetings, and in 2018 when we met at the Uto-Aztecan meetings in Guadalajara Joe and I hung out together for a couple of days after the meetings. I asked if I could do an interview with him for my blog, and he agreed. Then the interview sat there for awhile, so I asked Joe if he wanted to update it with some more details, which he did. And here it is, the first in a series of interviews with great Nahuatl scholars.

Joe is one of the popularizers of the Nahuatl orthography system sometimes known as "Andrews-Campbell-Karttunen" orthography, which was first developed by J. Richard Andrews, subsequently employed in Joe's and Fran Karttunen’s influential "Foundation Course in Nahuatl Grammar", and Frances Karttunen's "Analytical Dictionary". (His first major work, “A Morphological Dictionary of Classical Nahuatl” came out in 1985 and precedes his use of the Andrews orthography.) He is also a pioneer in computerized Nahuatl study, having begun working on Nahuatl computerized lexicography back when computers were the size of a building and used punch cards. He is still working on two major works: an integrated version of all of Alonso de Molina's three dictionaries, analyzed into individual morphemes and with the possibility of running all kinds of searches for combinations of different morphemes and a dictionary combining the vocabulary of both Molina and the Florentine Codex.

Joe has worked with Nahuatl for more than 50 years, and in this interview he tells a little about his career in Nahuatl studies and his different contributions to the field. 

MPH: So Joe, would you mind telling a little bit about how you got into Nahuatl in the first place and when?

RJC: OK, In 1962, I was a graduate student at the University of Illinois studying Spanish linguistics. And I took a couple of courses with Ken Hale, who was a young PhD, fresh out of Indiana University studying with Carl Voegelin, and in the spring of 1962, he said "Joe, some of the anthropology students are going to Mexico this summer and they're going to study Nahuatl. We're going to Tepoztlán and they're going to do fieldwork." And he said "you could go along, and would you be interested in going along and studying the Spanish dialect of Tepoztlán and see what the effects of Nahuatl were on the Spanish?" And I said "Oh, I'd really like to do that." And so, my then-wife and I and two daughters came to Mexico, and they went to spend the summer with her parents in Guadalajara and I went directly to Tepoztlán. And Ken and I went up to San Juan Tlacotenco at the top of the mountain, and while we were there we met a guy that spoke Nahuatl fluently -- one of the youngest people in town that did -- met him, made friends with him. And then later on Ken took me to see a rather elderly man who had fought during the revolution and sat me down to study with him, and as we worked along I found that he was a very nice guy, but he was very impatient. I didn't have a good ear at the time. I think it has gotten a lot better now. But when he said, I asked him how to say "camino" and he said "ohtli", and I said "otli", and he said "no! ohtli!", and I said "sí, otli". And with his reaction, I saw that he and I probably couldn't work together. We turned out to be very good friends all summer, and I found a much
Joe working with a friend from Tepoztlán

younger person to work with, with considerably more patience.

MPH: So what did you do when you came back from Tepoztlán? How did you continue with you Nahuatl studies after your first visit?

RJC:I wrote a paper for Ken, and then with the things I had to do, write the thesis at Illinois and search for a job at Indiana, Nahuatl sort of got put on the backburner. I got my first job at Indiana University and after I had been there well, like five years, one of the young professors came down the hall and he was doing legwork for the chairman, looking for a federal grant. Because the federal government was interested in doing unusual languages, languages that were not likely to be spoken by ...quote ..."the enemy". And he said, if we apply for a grant and say that we can offer Nahuatl, would you be willing to teach it? And I said, "well I am really not prepared to teach it on the basis of six weeks of experience", and also with the intervening years. And so I said "I would be willing to do it if I had an informant, if I had somebody to work with me in the class". And they came back to me and said "We're more than willing to do that; we'll pay for an informant and all the expenses and everything, but ... you would need to go find somebody". And to make a long story short, I was going to be in Mexico all the summer of 1970, and I went to Tepoztlan -- a friend of mine in Tepoztlan, Karen Dakin -- and she paired me up with a friend of hers, the anthropologist Judith Friedlander, and we went to Hueyapan and we found a young lady who was very happy to go to the United States and work for a year, that was Elvira [Hérnandez]. 

MPH: So you really learned Nahuatl when you were teaching it with Doña Elvira Hernández from Hueyapan?
Joe with Elvira Hernández in Hueyapan

RJC: That is exactly right. She and I, we started at the beginning of the school year. And she educated me in class, and then we had three hours a week out of class, and it was just like having an intensive teacher.

MPH: And when did you start getting into studying the colonial version of the language?

RJC: When Elvira came back to Mexico, I taught Nahuatl for two more years by myself and used sources like González Casanova's folktales and a lot of structural material. But during those two years I started realizing that if I really wanted to get more details of Nahuatl I would have to look at printed sources. So I started putting the Molina 1571 Nahuatl-Spanish half dictionary on punched cards. And it ended up that the whole second half of 1571 fit on something like 20,000 cards. That's ten boxes of cards. And then from there on, I worked in San Antonio with the whole card deck, inserting English translations into the boxes. So it swelled up, and so in one year -- from I think it was 75-76 -- I managed to translate all of the Spanish of the second half into English. And while I was doing that I was realizing that I should also enter the record by means of another punched card with the morphology codes typed out on that card.

MPH: So maybe you can explain a little, what is a perforated card and what is a morphology code?

Perforated card used for early digital computing.
RJC: A punched card was a card with eighty columns for punching holes in different combinations to represent letters. And people who used them actually started understanding what the perforations meant: There was a top row and a second row and the top row and second row then correlated with a set down below of one through ten and a combination of top and 1 was an A and top and 2 was a B and so on. 

MPH: Ok, so it was a very simple form of machine readable code? 

RJC: Yes, it weighed a lot more than a floppy disk.

MPH: So basically it would be fair to say that you were probably one of the first people to start using computational technology for working with Nahuatl?

RJC: I was the first one that I knew of, and I don't remember anybody else doing it while I was working on it in the seventies. But here you must remember that when we’re talking about punch cards, ‘computerized’ only means how we stored and re-arranges data. I have seen more than one suggestion that somehow the morphological analyses are done “by computer” and adds. They are NOT. Every morphological analysis is my considered opinion. The computer really has nothing to do with it.

MPH: So, the first work of yours that I became aware of was an article that you had in IJAL in 1976, about the phonology of Hueyapan Nahuatl, which I read for the first time before I went to do field work in Hueyapan myself in 2003. Could you summarize how that article came about and what is the interesting part of the argument?

RJC: I am not sure if I can remember the data very well, but I was working with paradigms that I learned from Elvira and I was trying to come up with a very logical description of the morphology. And it turns out that when you look at the representation of some words it is very straight forward: there is part one, two and three and they sound the same in either present tense or preterit. But then there was some verbs that looked like they had an underlying /w/ that was phonetically a [v] that in the preterit would turn into a velar [ŋ]. And, I thought well it's underlyingly a /w/ its phonetically a [v] and in another form it is an [n], what kind of segment could that possibly be? So because of the influence of an article by Sol Saporta in Spanish where he talked about the problem of palatal l and l and y in Spanish, in words like [mil] "thousand" and [miyón] "million" [this article], I came up with the idea that what that segment was could probably be a /ŋw/ a velar nasal with a labial release. So that would explain why Hueyapan Nahuatl had the present tense form [noga:va] "I stay", and the preterit form  [onoga:ŋ] "I stayed". 

MPH: Then another really important thing you've done has been the foundation course. How did that come about?

RJC: When I went back, well I went from San Antonio to Nebraska and stayed in Nebraska for two years, and then I went back to Indiana. And they had the Nahuatl course waiting for me when I got there and asked me to do that. And so, I didn't have a textbook, and I thought, rather than do a sort of struggling course the way I did in the early 70s, I thought what I ought to do would be to write simply a series of disconnected lessons, starting with the most simple things and doing lots of exercises where people had to do recognition and recall. And I started writing as many lessons as it took to get through the semester. And as I gave the course a couple of times more during the 80s, I wrote more and more lessons. -- Not a lot of explanation, because the explanations were mainly things that I did in class. But I gave them a lot of heavy take-home stuff to work on and come back and build up their memory on things. And as a matter of fact in 1987, I had already shown this rather big stack of exercises to Fran Karttunen who had been a friend of mine for a long long time. And at one point Fran said - that was in 1987 - she was going to have a government funded institute in the summer of ’89, and would I teach the modern dialect and she would do the classical track and could we use a selection from my set of exercises. And if we could, she would supply a denser set of explanations so it would be easier to work on at home; and she would also put in the vowel length which I had not taken the trouble to do. And that's what we did, we used it for a six week National Endowment for the Humanities Summer Institute in Austin Texas. 

MPH: And then the other thing is that when you finished the translation of the 1571 Molina in 1975, that was not at all the last thing you did on the Molinas. Can you tell us a little about what you have been doing with the Molina dictionary since then? 

Frontispiece of Molina's 1571 vocabulario
RJC:The translation of the second part of the 1571 dictionary was just the first step. I followed it up by preparing a morphological analysis of the Nahuatl vocabulary of the Nahuatl-Spanish dictionary and doing the programming to make it possible to sort all Nahuatl entries along with their morphological analyses and English translations under each of their component morphemes.

I happened to show the resulting print-out to professors Lloyd Kasten and John Nitti of the University of Wisconsin during a reception at our house following a lecture that they gave in Bloomington involving the use of computing technology in the study and preparation of old documents. This was probably in 1984. They expressed interest in having the Hispanic Seminary of Medieval Studies publish the morpheme index and gave me explicit instructions to make it possible for me to insert the printing codes into my data so that the final publication came directly from the computer tape that I submitted to them -- by this time I had moved past cards-- since re-setting something so complicated would have been impossible. Somewhere there is a picture of me standing at the Post Office window posing with the 12x12 inch package in my hand. The result was my A Morphological Dictionary of Classical Nahuatl: a morpheme index to the vocabulario en lengua mexicana y castellana of Fray Alonso de Molina (Madison, 1985). 

The morphology I did in the Morphological Dictionary is somewhat more rudimentary than what I am doing now. For instance if you have a word like <tlacatl>, that is a simple morpheme (well actually, it’s two, but people don’t think much about the -tl being separate), but if you have a word like <nitetlacatilia> there are actually five pieces. In the Morphological Dictionary I just coded this as tlacatl-verb-caus for the noun tlaca-tl, the verb-forming morpheme -ti- (become) and the causative -lia. I did not code the ni- (I) -te- (someone) or the other personal prefixes at all. 

Later I made up numerical codes for the prefixes that would set them up in ways that were reasonable instead of spelling them out and I made a number of separate categories for the verb-forming morphemes as well as for the causative and benefactive morphemes and the preterits. Today the morphology for nitetlacatilia is p11-p53-tla:catl-v01a-caus04. I still treat the absolutive suffixes as part of the spelling for the noun morphemes because it makes the nouns more recognizable for most people.
Since I had done the English translation of the 1571 dictionary...The 1571 dictionary is really two dictionaries and I call them Molina 1 and Molina 2 - and I had only worked on Molina 2 - the Nahuatl-Spanish. And so I thought that what I should do would be to put in the 1555 and the first half, the Molina 1, into machine readable form. By then I had graduated from punched cards to putting them in in electronic mode - just keying them in, which struck me as a rather strange thing to do rather than using cards, but I did that. And I by-passed translating the other two dictionaries in favor of spending all my time inserting morpheme codes after each entry in all three Molina dictionaries. My goal here is what I call The Integrated Molina Dictionary -- the gathered Nahuatl entries with their morphological analyses, their Spanish equivalents and, where I have it from M2, the English translations of the Spanish. This is actually very nearly finished, except that I keep setting it aside to work on the Florentine Codex vocabulary.

MPH: So basically, rather than just translating the Spanish translation of the Nahuatl word, what you did was to also analyze the Nahuatl word piece for piece and type it up as part of the entries for the three Molina dictionaries. So about how many entries is that for the entire Molina corpus?

RJC: Well, actually, it wasn’t “in addition” to translating. I only translated the Spanish of Molina2. Ok, I can't remember off the top of my head. But the Molina 2, the Nahuatl-Spanish dictionary has 23,623 words, and the 1555 has something over 15,000 and I don't remember what the Molina 1, the Spanish-Nahuatl, has - but it is somewhere in between those figures. And of course one of the kind of things I did was to write a program to sort Molina 1555 and Molina 1 and 2 together on lexical items, and then I had a cluster that showed what each dictionary said about any given item.

MPH: So how about many entries do you have in your morpheme database at this point?

RJC: So I am a little bit unclear on this, because I also have an extra little ambitious thing that I did, which was to also code items for various kinds of things; say I would code them for an error that Molina did in parsing things, for example when Molina stripped the prefixes of a verb he would sometimes he would cut off the beginning of the verb itself. So I put in a code so that I could gather all of those things. And I was also interested in the fact that there was a certain amount of spelling variation like the dropping of n, y and w, and so I coded for either dropping or retention. And if you simply add all of the codes for those different kinds of notes plus the morphemes, right now, my "morpheme" count is around one million one hundred thousand.

MPH: That's a lot of words. Can you tell us briefly some of the cool things that your program can do with that data? The type of searches you can run and the kind of information you can extract?

RJC: The very first thing I was aiming at was being able to give an alphabetical index of all the words in the three Molinas integrated, and that is pure vanilla dictionary construction. But the next thing I wanted to do was to compile a list of morphemes so that anybody could look up alphabetically in a list of morphemes and find all the words in any one of the three Molinas that had that morpheme. So from the very beginning I was constructing a morpheme index of the whole thing. And then because of some of the advantages I saw in looking at the whole vocabulary from the end of the word, I also wrote a program that could give you a backwards index. So for example everything that ended in -tilia, you'd find every single word that had that. Parallelly and not meaning to get distracted fom Molina, but yes I did get distracted from Molina, I also together with a friend of mine from France, Marc Eisinger, we put the Nahuatl of the Florentine Codex into machine readable form, we each did one book and another, and then he did the entire verification of all the Florentine Codex, just a monstrous job. (Verification is when you re-enter data so that a program can find where one copy differs from the other.) And then I spent three years in the late eighties integrating the Nahuatl and the Dibble and Anderson English translation - it took three years to get them to match sentences. And then I spent a long time adding a regularized spelling line to the Florentine entries so that the computer could bring together instances of a word that was spelled differently in different places. And then, I could correlate words from the Florentine with words from Molina, that's one thing, and then the other thing independently from Molina, since the Florentine was text it was sentences in contexts, it would be interesting for anyone to see how a given word was used, in context, so I wrote a concordance program that will allow you to interrogate the Florentine database for any string that you wanted. For instance... well, there were two different things: one would construct an index based on any alphabetical representation so you could look for "tlacat" or you could extend it and look for "tlacatil", whatever... But the other thing was, and certainly based on favors that people did me in enlightening me - Fred Householder at Indiana had a computer colleague that had been doing some work on texts and he had the idea of hanging codes on words in context… He wasn't working on Nahuatl but I turned it into working on Nahuatl. If you wanted to look at anything that was related to <tlacatl> then you could hang a morpheme under that and you could make an index of every morpheme and then you would get the word in-text, in context. So what I did was to label, I had to fill in for every word that I could, and that means a lot that I couldn't, I made every word in the sentence a key to that sentence and then every translation of those given Nahuatl words was a key to the Nahuatl word which was a key to the sentence. In this way I could retrieve a lot of interesting information.

MPH: So you have been working on this for 43 years. When are you going to be done, and what is the final product going to look like?

RJC: I originally hoped... Well I keep changing my mind about what the goal could possibly be. The original goal was to do Molina. But then I started realizing that it would be a lot more valuable to do a combined study of the Molina and the Florentine. And then I started realizing that it was very unlikely that I would have time in life to do all of it, so I said Ok, I will just do Molina. But while I have been working on Molina I never stopped analyzing sections of the Florentine. As a matter of fact it is not sections of the Florentine, but I look for a given thing in the Florentine, and I go through and find every token of it and mark it for whatever it needs - morphology, phonetic variation and things like that. And if there are errors in the Dibble and Anderson translation, I put in a note on every sentence where I find that going on. And every time I do, I always think, I am in awe of what they did, but we are all humans, we all make mistakes.

MPH: Yes, and we can keep improving on it. Thank you very much Joe, for telling us about your work.

RJC: You're welcome, I enjoyed it.


Here, we end the blog with a list of some of Joe's most significant publications (several coauthored with his wife Mary Clayton who is also a Nahuatl scholar):

  • Campbell, R. J. (1976). Underlying/ŋw/in Hueyapan nahuatl. International Journal of American Linguistics42(1), pp. 46-50.
  • Campbell, R. J. (1985). A Morphological Dictionary of Classical Nahuatl: A Morpheme Index to the Vocabulario en Lengua Mexicana Y Castellana of Fray Alonso de Molina. Hispanic Seminary of Medieval Studies. University of Texas.
  • Campbell, R. J., & Clayton, M. L. (1988). Bernardino de Sahagún’s Contributions to the Lexicon of Classical Nahuatl. The Work of Bernardino de Sahagún: Pioneer Ethnographer of Sixteenth-Century Aztec Mexico, pp. 295-314.
  • Campbell, R. J., & Karttunen, F. E. (1989). Foundation course in Nahuatl grammar (Vol. 1). Institute of Latin American Studies, The University of Texas at Austin.
  • Clayton, Mary L., and R. Joe Campbell. (2002) "Alonso de Molina as Lexicographer." Making Dictionaries: Preserving Indigenous Languages of the Americas. pp. 336-90.