The January issue of Neophilogus includes an essay on the problem of translating Dante’s Divine Comedy into Arabic (abstract here). As you might expect, the challenge of translating a classic of medieval Christian orthodoxy into the language of the Koran is especially great. Setting aside the fact that Dante consigned Mohammed to hell’s eighth circle, the theological and literary differences between Catholic and Islamic civilizations are so extensive that it is impossible to get all of Dante’s Italian subtleties into the Arabic. I doubt that any of my readers are surprised by that news.
Meanwhile, these days I’m reading Tolstoy’s Anna Karenina, a classic novel that I have read a few times before. This time I’m using the (fairly) new translation by Richard Pevear and Larissa Volokhonsky, a husband-and-wife team who have set the literary world agog with their stylish translations of the Russian canon. Of course I can’t help wondering whether I’m enjoying Tolstoy or the translators when I come across a witty passage like, “Each had something demeaning and derisive to say about the unfortunate Mme Maltischev, and the conversation began to crackle merrily, like a blazing bonfire.” [p. 134]
“Began to crackle” really lets the reader perceive the way malicious gossip can enliven talk [the original, Constance Garnett, translation just says “crackled”], while “blazing bonfire” gives a strong sense of delightful destruction [Garnet: “a burning faggot-stack”].
Such thoughts eventually turned me toward my blog and I wondered how long it took for speech to become imperfectly translatable. How long was it before the story of Babel and the world’s confusion of tongues would have made sense to people?
If the first speakers limited themselves to things like —mama, —lion come, and —figs yonder, interpreters could have provided a perfectly literal translation. And it is widely noted that the early speech of children tends to be completely translatable. A two-year-old in Italy speaks like a two-year-old Egyptian; the distinctions that separate Dante from Arabic readers have yet to appear. What happens to destroy that compatibility?
We tend to blame it all on separate histories. One language evolves in one setting, one another. So even if two languages share a common ancestor, time and space will make them incompatible. Commonsensical as that explanation is, it cannot be exactly right. Arithmetic has grown up in a variety of civilizations and been separate from the beginning, and yet Chinese and European mathematics is compatible. Both groups found prime numbers. Why didn’t they also both find gender compatibility?
Prime numbers are real; gender compatibility is an artifact of a language’s phonology, and yet speech starts out as concretely as arithmetic. Mama refers to something that is at least as real as the number one, but along the way something happens. That change is not easy to specify, especially if I say words pilot attention. How can attention be piloted away from the real?
Many people, of course, deny that words pilot attention. They say that words are symbols— arbitrary signs whose meanings emerge from the whole system. Each system is unique and therefore translation can only provide an approximation, rather like trying to fit a tailored suit onto another, stouter or leaner, gentlemen. Some alteration will be required.
That explanation seems plausible when we look around the world today and see how the many ways languages differ, but when you think in terms of origins it makes less sense. We can imagine an evolution toward symbols—begin with “indexes” in which the indicator is directly associated with its meaning, e.g., baboon footprints indicate baboons; then comes “icons” that resemble what they signify, e.g., drawings of baboon footprints; next come symbols, in which the drawing no longer seems like a footprint so that any other symbol would serve just as well if it had been agreed upon instead. But that evolutionary path would still not explain why translation eventually became a matter of trade-offs rather than a simple switch to a different notation. Written numbers, after all, probably did evolve in some sort of icon-to-symbol passage, but the Roman numeral IV and the “Arabian” 4 are perfectly translatable. And children today do not go through the index to icon to symbol route. They begin right off with mama, an arbitrary sign common to many but not all languages.
Of course, mamas are concrete while the number four is an ideal. As Tolstoy said (I translate freely), “All idealizations are alike; each concrete realization is realized in its own way.” That variability poses problems for words. Suppose, for example, one group of H. erectus had a word for baboons, another group had a different word. Translation seems simple and perfect.
But suppose one group used the same word (grok) for baboons and vervet monkeys while the other had two different words for the different species. Translation becomes trickier, especially translating the more general term into the more specific one. You cannot translate grok without looking to see whether it refers to a baboon or a vervet. This problem may have been the first barrier to ready translation: languages do not all divide the concrete world along the same joints. Even so, if this was all the difference between the languages of, say, 1.5 million years ago, translation would have been pretty easy.
This issue of dividing a seamless experience into distinct words becomes still more tricky when we introduce verbs. Like nouns, verbs can make distinctions unknown to other languages. As an English speaker I can say, “Oh, yes, I know London,” and and, “I know how to prove the Pythagorean Theorem.” A French speaker, however, would use one verb (connaître) for knowing London and a different verb (savoir) for knowing a geometric proof. The tricky element here is that, while monkeys are things, knowing is a phenomenon. It is real, but not an object.
I can draw your attention to baboons by simply pointing, but I cannot literally point to your knowing something. I can say, “Joe knows London,” and you might recall seeing Joe enter a tube station and heading toward the appropriate platform without checking a map or asking for help. “Yes, I remember that,” you reply. I’ve said something and you’ve recalled something in response, but it all happens internally. Objects perceived or recalled support the proposition, but do not constitute it.
One of the wonders of language, its ability to make private things public and to draw attention to invisible parts of your awareness. I can ask, “Do you know London?” and you look within yourself and your experience. A dog may know parts of London too, but nobody ever draws the dog’s attention to its own knowledge. Some philosophers see this self-referential activity as a special form of knowledge, but it makes better evolutionary sense to see it as a special form of attention. Dogs know what they attend to, as do we. But we can direct our attention internally much more readily than dogs can. When speech came along it became possible to direct one’s attention inward explore oneself. Naturally people found new things to talk about.
Yet talk of phenomena, personal and public, and splitting experience at different joints still does not explain why translation became so problematic. If that’s all there were too it, translation would seem more demanding but still possible. If one erectus group said know him and know axe making while another said connaître him and savoir axe making, an interpreter could still provide correct translations. I suspect that as long as humans were limited to pidgins (a.k.a. protolanguage), translation remained straightforward, even though the speakers inevitably divided the real world separately. Translation should be just changes in notation.
Indeed today’s pidgins show a phenomenon that is unknown in true languages, a kind of change in notation called relexification, the substitution of one vocabulary for another. When the slave trade began, for example, it was conducted by using a pidgin based on Portuguese words. Then the English took over the business and the pidgin became English-based, the Portuguese words disappeared almost entirely except for a few like savvy (understand). The same relexification occurred in VietNam as a pidgin based on French took on an American vocabulary and the only French traces were words like beaucoup. As long as a language can be relexified, translation seems possible. There are always going to be technical issues—words for things that don’t exist in some other locality; disagreements over exactly how to divide the things and phenomena of this world; maybe even some semi-literary flourishes like puns—but essentially all speakers share a common world. One word is literally as good as another.
My working assumption is that pidgins lasted until some time after Homo erectus. In other words there was a long period during which the Homo lineage used symbols without simultaneously creating those distinct worlds that we now assume separate every symbol-based culture. When full language entered the story, something happened beyond simply getting a richer syntax. True languages still borrowed words from other tongues as needed, but relexification became impossible because speakers hold on desperately to the untranslatable portions of their language. But just what were those untranslatable portions? Where did they come from?
I see I’ve been going on a while and cannot answer in 25 words or fewer, so I’ll save the rest for next week’s post.