Historians speak in the same hushed, reverent, knowing tones about ‘the archive’ as you lot do about ‘the field’. There must be tales of heroic difficulty and overcoming: the mold to which one is allergic unto death, the bizarre local foods, the grumpy archive director, the grumpy informants. These are the cult mysteries.

Similarly, historians tend to snob any kind of work in translation, despite the obvious fact that we are nothing if not translators of the past. My Italian is near-native and I certainly did find some errors in the main Gramsci translations, including a couple of pretty substantive ones. But overall, I think the translations are pretty good and allow full access to any intelligent reader. There’s also a lot of fretting in Gramsci studies about how the _Prison Notebooks_ were edited and translated. Yet in my view, some of the best work on/with Gramsci has been done based on these imperfect sources.

I also got to know Durkheim’s _Elementary Forms_ very, very well while I was working on my dissertation. I worked with the old translation and Karen Fields’ new one, plus I had a copy of the original French to check in with if I had any doubts. I found both translations to be just fine; the new one in general a bit more readable but with some odd gratuitous changes that lost a little compared to the first one. In no case was there a problem big enough to lose the sense in anything but the most decontextualized reading.

Weber translations are famously worrisome this way. As an example, so much is lost from the German sense of “Herrschaft” by translating it “domination” (just as so much is lost from Marx’s “Aufhebung” by any one-word translation like “overcoming”). So that can be misleading, except that if you read Weber’s typology of legitimate domination even just a little bit carefully, all of those layers of meaning are well-embedded. The same is true of “virtu’” in Machiavelli, a translation-unfriendly piece of polysemia that smoothly finds its meanings in the contexts in which it’s embedded; as do they all, without the rituals of disciplinary purification adding a lick.

Language ability is one of the few methodological issues seen to really count in sociocultural anthropology. Kind of like sieve size or something for archaeologists: “what? you only used 1/4 inch sieves? If only you had used 1/8 inch ones you would have picked up so much more information!”

In the very, very early days people spent quite some effort devising ways to generate data without knowing the language. Linguistic incompetence was built into the method. WHR Rivers, for example, and his lists of questions for quickly elucidating kinship terms during the few hours he had on the beach while touring the Pacific by boat. Published two volumes out of that stuff. (And of course everyone since has delighted in pointing out how wrong it all is).

Lastly there is a power dynamic involved in whether you choose to speak the local language, use an interpreter or speak English in those colonially induced multi-option situations. I had about four language options in my PhD field site – and each one felt different. In one I was the embarrassingly bad foreigner who got everything wrong, in another we were all incompetent, in another I was in control (and in the fourth we all felt pretty happy).

I think of ancient debates about the demarcation of the social facts of interest to the anthropologist from individual psychology. The unmentioned elephant in the room was always the fact that a focus on the public and visible substantially reduced the need for linguistic competence. Hut diagrams, census data, kinship terms, lists of political and ritual offices and politics conceived primarily in terms of inheritance and succession to office: relevant data could be collected by someone with rudimentary language skills and commonsense local knowledge. One largely undiscussed implication of the interpretive turn is that sorting out the twitches from the winks, the wink with a grin that shares a joke from the wink that signals time to act, the fictional wink that may mimic any and all of the above, let alone probing deeply into the possibly idiosyncratic perceptions and feelings of individuals, requires much better language and communication skills than more than a handful of linguistically talented anthropologists are likely to acquire in only a year or two.

I definitely use my language skills to monitor the translation, make sure she doesn’t forget to tell me about things I catch, but I have far more information having used an interpreter than I would have otherwise. I suppose time and peer-review will tell whether that information is reliable, but it just seems bizarre to me to write off the use of interpretors entirely, especially in contexts where those interpretors have a far better fluency in English than I could reasonably hope to acquire in a reasonable time in Twi.

bq. My main language deficits were in catching meaning in difficult auditory circumstances….and in the elaborate language used by matapules in ceremonial occasions.

I think of what I wrote in my chapter on “Traditional Chinese Religions” in Ray Scupin, ed., _Religion and Culture: An Anthropological Focus_,

bq. China is the world’s oldest continuously literate society, and the sheer volume of historical texts is enormous. One source suggests that the 25 imperial histories alone would require 45 million words in English translation. In Chinese the Buddhist Canon is 74 times the length of the Christian Bible, while the Daoist Canon is a library that runs to several thousand pages in its latest edition. In contrast the number of scholars who study these materials is small. In history as well as in archeology, new discoveries continue to appear. Suppressed texts, hidden away sometimes for centuries, surface periodically.

To which I would add that ritual is not the only locus of ceremonial language. In, for example, our work as translators from Japanese to English we have, just within the last year, encountered the specialized vocabularies of audio engineers (algorithms for reproducing Dolbi 5.1 channel sound through earphones), medical biophysicists (ferrite nanobeads for cancer detection and treatment), architecture and urban planning, performance art, dance, early (pre-silver halide film) forms of photography, _gutai_: an art movement that interacted with the French _informel_, lots of marketing and advertising. We can, moreover, no longer count on our Japanese associates for explanations of pop culture. They, like us, are 50+ and frequently as baffled by allusions to the latest trendy coinages as we are. The good news is that using Google to search for them usually pops them up, with enough context to make a decent stab at how they ought to be rendered.

That’s why I went to an island in the Ha’apai archipelago. Not all that isolated; kids studied English in school: they went to high school on the main island in the archipelago or in the capital: a number of the men had worked overseas, in New Zealand. Otherwise, they were subsistence farmers and fishers who sold copra to buy things like needles and flour.

It took me a year and a half (plus some previous language tutoring in the US) to become fairly fluent. I don’t think it’s THAT hard to develop conversational fluency when you’re in a situation where you can’t escape it and you must learn. By the end of my stay in Tonga, I was amused to find that when I interviewed intelligentsia, we were codeswitching. We’d switch back and forth from English to Tongan depending on the topic.

My main language deficits were in catching meaning in difficult auditory circumstances (church conferences in huge echoing churches) and in the elaborate language used by matapules in ceremonial occasions. I would suppose that the latter is the kind of thing John McCreery is discussing: a cultural or literary tradition that takes even native speakers years to master. Not all of them do. Just so, most Tongans were content to let the matapule learn the special counting words for coconuts and pigs, words only used in ceremonial prestations.

As I ask this question, I imagine two quite different scenarios. In one, the anthropologist’s village is isolated. Yes, from a properly distanced world system perspective we may recognize that the lives led there have been influenced by all sorts of history. To the people who live there, life is, however, pretty much the life their parents lived. Life moves in small, largely predictable cycles. The range of local concerns, whether cattle in East Africa or pigs in Melanesia, is limited. In this small, relatively stable community, Basil Bernstein’s hypothesis holds. People speak a restricted code. Grammatically and phonologically speaking, it shares the openness of all natural languages; an infinite number of new things could be said. But the fact of the matter is that they aren’t. The same topics recur. Vocabulary is limited. Most conversation involves short and syntactically simple structures. In this scenario, the anthropologist’s becoming fluent within a year or two is a reasonable expectation.

Kerim’s experience and mine reflects a radically different situation. The people we work with are formally educated, at least through elementary and lower secondary grades. They are citizens of nations that teach and celebrate their roots in global as well as local traditions. They watch TV, they go to movies, they read newspapers and magazines, some even books. These days their access to the Internet may be as good or better than ours. Critically for language learning, their societies are complex, with numerous bodies of specialized knowledge and the jargons that go with them. Meanings are always contested. And, local written traditions that stretch back hundreds and thousands of years and exist in literally millions of printed words provide plenty of fuel for debate. Thus, in Grant McCracken’s wonderful phrase, “meaning flows.” In this situation, “learning the local language” is an enterprise for which a lifetime is not enough. Learning enough to converse even semi-intelligently on a limited range of topics is a serious achievement, even for the native speaker.

These are, of course, ideal types, deliberately polarized models constructed to frame discussion. But where particular research fits between them and the linguistic demands it entails—there’s the rub.