by David Bellos
In speech translation, tone, pitch, gestures of the face and hands, and the stamping of feet are thrown to the winds, even though they transmit major clues as to how the speech is to be understood.
In all the many other dimensions and levels of an utterance in speech or writing the criterion of translation is not to be the same but to be like.
A is like B only in respect of C. This is a way of saying that when a thing is compared with some other thing, the act of comparing rests on a third term that is neither A nor B. What makes soupe de poisson like “chowder”? The comparator (called ter-tium comparationis in the old rhetoric) could be “soup” (but that would be a poor kind of likeness), or “seafood” (that’s richer, both are soups made from fish), or the fact that they are both eaten hot, or that they are both available in cans, or that there are cans of both on that shelf. The “likeness” of soupe de poisson and “chowder” is a variable, and its value varies in accordance with the comparator used or implied in any given context of use.
The dimensions of an utterance where likeness is the relevant criterion of translation are of many different kinds. Register, tone, rhythm, style, and wit can only ever be said to be like one another in respect of something external to the text itself. For example, to judge that writing iambic pentameters in English is like Racine’s use of the twelve-syllable line is to base likeness on the social and cultural values of poetic forms in two different environments. In both English and French verse, these are the commonest, most frequently used forms, and thus like each other in that respect. But they are not like each other in any other way. Writing twelve-syllable lines in English to represent French verse, on the other hand, is like the original only in respect of the number 12, but quite unlike it in respect of the underlying rhythms of English, which is a stress-timed language, and of French, which is not.
By choosing which dimensions to connect in a relationship of likeness and the extent to which the likeness is made visible, a translation hierarchizes the interlocking, overlaying features of the original. To that extent at least, translations always provide an interpretation of the source. It’s more obvious in literary texts with relatively few practical constraints, but the same underlying situation holds for all acts of translation between languages.
The nub of the question is this: Given that a translation preserves the information and the general force of the original, in what respect is it possible to say that its manner or style or tone is like those features of its source?
Georges Perec wrote in a wide variety of styles, but a characteristic feature of all his writing is that important information is placed at the very end, making you realize that up to that point you hadn’t understood the main import of the sentence or paragraph—or even the novel. At the level of sentences and paragraphs it is easier to do this in French than in English literary prose, which typically introduces new information in a different manner. Nonetheless, by exploiting the notorious flexibility of English sentence structure and bending it a fair bit, I respected Perec’s “late release” technique as far as I could. By the very fact of doing so I offered an interpretation of Perec’s style, but the likeness of my prose to his is a tightly focused and fragile thing. Because I had to take greater liberties with English than he did with French, my writing is not “like” Perec’s at all in respect of linguistic norms.
No translation is the same as its source, and no translation can be expected to be like its source in more than a few selected ways. Which dimensions are selected depends on the conventions of the receiving culture, the nature of the field involved, or even the whim of the commissioner of the translation. But any utterance is such a multidimensional and many-faceted thing that no translator is ever short of a little elbow room. To put it the other way around, no set of social, practical, linguistic, or generic constraints ever determines completely how a translation is to be done.
If meaning and force are kept the same and if in a limited set of other respects a translation is seen to be like its source, then we have a match. Translators are matchmakers of a particular kind. It’s not as simple as the marriage of content and form. Just as when we match faces and portraits, we rely on multiple dimensions and qualities to judge when a translation has occurred.
Children’s puzzle books exploit and psychologists study our ability to recognize and manipulate the distinct but overlapping relations I’ve called same, like, and match.
Translators use that ability in the specific fields of speech and writing in a foreign tongue. Not all of them are great at their job, and not many have the time and leisure to wait for the best match to come. But when we say that a translation is an acceptable one, what we name is an overall relationship between source and target that is neither identity, nor equivalence, nor analogy—just that complex thing called a good match.
That’s the truth about translation.
THIRTY-TWO
Avatar: A Parable of Translation
On a recent visit to India, where I was trying to learn more about translation, I took an afternoon off to go to the movies and watched a faded copy of what I believe is the most expensive film ever made. To my delight and surprise, Avatar turned out to be a parable about translation, and that’s why I bring it in at the end of this book.
The hero of James Cameron’s science-fiction fantasy is a human transformed by a laboratory technique into another being—nine feet tall, with a prehensile tail and amazing skydiving skills. His task is to penetrate the society of same-looking beasts causing trouble for a galactic mining company, and then to send back to his controllers the information they need to get the local inhabitants out of their way. He is still a human being under his impressive new shape.
But now that he has become a Pandoran in outward appearance, our hero becomes Pandoran in other ways, too. He goes native, so to speak, and becomes loyal to the community that has now accepted him as one of its own. These strange beings are fighting to remain themselves and to pursue the lives they have always had. Our hero makes their right to be different his own.
But respect for difference is clearly intended in the film to be an expression of a human value. So is our hero one of them, or still, at bottom, one of us? Is the mining company the vector of humanity—or are the awkward beasts in its way the true embodiment of our aspirations and souls?
The movie doesn’t quite answer the question at the end. It is the question that translation sets and must also leave open. How can a hugely modified transmogrification of some utterance—incorporating on occasions the verbal equivalent of a nine-foot-long tail—still remain, at some fundamental level, what it was?
Like Cameron’s fantasy, the practice of translation rests on two presuppositions. The first is that we are all different—we speak different tongues and see the world in ways that are deeply influenced by the particular features of the tongue that we speak. The second is that we are all the same—that we can share the same broad and narrow kinds of feelings, information, understandings, and so forth. Without both of these suppositions, translation could not exist.
Nor could anything we would like to call social life.
Translation is another name for the human condition.
Afterbabble: In Lieu of an Epilogue
In most intellectual disciplines, the stories of the Hebrew Bible are no longer used as sources or tools for thought. Translation studies is an exception. Scholars and essayists in this field continue to pay extravagant attention to the account of the origin of linguistic diversity given in the Bible.1 It’s far from obvious that their time is well spent.
The Tower of Babel comes from a story told in Genesis 11. The first verse states that in the beginning “the whole earth was of one language, and of one speech.”
This is not very plausible. Nothing we know or can observe about human linguistic behavior makes it likely that there ever was a single form of speech.
The rest of this section of the Bible, Genesis 11:2–9, offers an account of how the ancestors of
the Jewish people got from their hypothetical state of linguistic unity to the condition of diversity that manifestly characterized the part of the world they lived in some three to four thousand years ago.
The voluminous tradition of Babel commentary weaves religious, philosophical, historical, cultural, archaeological, and philological speculations around the story told in Genesis. Do these verses contain a trace of historical events? Or should we read them rather as a fable designed to account for the way things are, or the way they were long ago? For the purposes of this book, it does not matter whether there really was a ziggurat honoring the Assyrian god Marduk near the place now called Babil (in Iraq), or whether it was visited by Herodotus, or when it fell down. For an understanding of language and translation, it doesn’t matter if or how the Bible story is related to the Sumerian Incantation of Nudimmud. Nor does it make any difference whether we pick from the welter of Babel commentaries those which assume that linguistic diversity is a Dreadful Mess (the vast majority, in fact), those that claim it has a Silver Lining, or those few who argue that it is a Very Good Thing.2
What matters is whether we allow Genesis 11:1 to close our minds to other ways of imagining the origin of human speech. Cynics might say that’s what religious texts are supposed to do. But translation is not a matter of faith. It’s much more interesting than that.
The supposition of an original common form of speech has been taken to mean that intercomprehensibility is the ideal or essential nature of language itself. Such an assumption makes translation a compensatory strategy designed only to cope with a state of affairs that falls short of the ideal. It licenses, indirectly but no less strongly, all the many attempts there have been to devise languages that for some if not all purposes improve upon those that we have.3
This contentious foundation of the Babel story acquired implicit if unintended support in the nineteenth and twentieth centuries from the scholarly work of historical linguists. They sought to group languages into “families” and to reconstruct the hypothetical progenitors of these cousin tongues, as well as the rules by which each had received its inheritance. The discovery of a family likeness among Sanskrit, Greek, Latin, and Old Persian opened a new vista into the past, toward a single source for a whole spectrum of languages spoken between northern India and the Atlantic Ocean.
These exciting advances made it easy to see the historical derivation of modern languages like a cascade trickling down the mountainside of time, branching out into streams and rivers. At the now inaccessible top of the hill, there must have been a single source—Proto-Indo-European, for the great family that joins the languages of northern India to many of those of the West; and, at an even more remote altitude, Nostratic, the supposed ancestor of Indo-European and other language groups of Europe and Asia; and, high above that, “proto-World,” the language of ante-Babel, the original and unitary human tongue.
Some saw the underlying meaning of linguistic change and diversification through spectacles borrowed from Darwinism. For them, the growth in complexity from single-cell life forms to the magnificent machinery of humankind served as a model for understanding the “evolution of language,” from the rough-and-ready speech of hunter-gatherers to the refinements of the Académie Française. Others saw language change as a perpetual fall from the economy and mystery of the ancient tongues to the confusing multiplicity you can hear in the street. But behind these scholarly (and often schoolmasterly) pursuits lay a single barely questioned assumption—that all languages are, at bottom, the same kind of thing, because, at the start, they were the same thing. In fact, there was rather better evidence of the contrary. The Babel story may say that in the beginning all language was one—but what it shows is that, for a single people in the third or second millennium B.C.E., linguistic diversity was a major fact of life.
However, if we accept the proposition that all languages are instances of the same kind of thing, we have to ask: What is it that makes them the same? The most influential answer to this question in the twentieth century has been: a grammar.
The idea that a grammar is the common property of all human languages looks like a hypothesis—something you could test against data, then either abandon or refine. But that’s not the main way in which it has actually been used. Characteristically, the “grammaticality hypothesis” is an axiom, a circular foundation stone. The axiom “explains” why animal and mechanical signaling systems are not languages. Since traffic lights and the barking of dogs seem to have either no discernible rules of combination or no ability to create new combinations, they have no grammar, and because all languages have a grammar in order to count as languages, dog barking and traffic lights are not languages. QED.
In a similarly circular way, the axiom of grammaticality pushes to the edge of language study all those uses of human vocal noises—ums, hums, screams, giggles, mumbles, stammers, exclamations, and interjections, alongside ellipses, nonsense words, gargles, cooing, baby talk, pillow talk, and so forth—that don’t decompose neatly into nouns, verbs, and periods.
Even leaving out the whole range of “ungrammatical” and “nonlinguistic” uses of vocal sounds, the variability and range of the things that the grammars of actual languages regulate make it very hard to see what it can mean to say that a grammar is what all languages have in common. Inevitably, it prompts a second question: What is it that all grammars have in common?
It is hard to find an existing grammatical category that is common to all forms of human speech. Many languages do without determiners such as a and the (Russian and Chinese, for example). Many languages do without gender (in Finnish there is no distinction among “he,” “she,” and “it”). Numerous major and minor languages in the world do not mark number (Chinese, again, has no special form for dual or plural). It’s fairly obvious that you don’t really need adjectives—even in English, you can use “tomato” or “beetroot” if you want to call something red; prefixes allow you to distinguish between big and small versions of the same thing (English minibus, French hypermarché), just as suffixes do in Italian (uomaccio, “big man”), Latin (homunculus , little man) and Russian (, “little lion”). The Argentinean wit Jorge Luis Borges thought up a language without nouns—where verbs and adverbs sufficed for all expressions. “It moons bluely” is all you would need to refer to the presence of a blue-tinted moon in the sky. Aspect is only properly grammati-calized in some languages (Russian, for example), and tense is clearly redundant even in languages that have it. “I go to Paris tomorrow” is perfectly good English, just as Napoléon entre dans Moscou en août 1812 is normal French, with the explicit expressions of time (“tomorrow,” “1812”) making the grammatical marking of time unnecessary. Only some languages have evidentials; vast numbers of them have no prepositions and many others have no agglutinations. The concept of case is virtually absent from English (it subsists in the distinction we still make between he and him, she and her—and that’s about it) and totally alien to Chinese.
And so it goes on. Mood is not part of English grammar (we use separate words, such as may, should, ought, and so forth), but it provides Albanian with an elaborate set of resources for expressing all sorts of affective qualities, including admiration. Vowel harmony is a basic feature of Hungarian: you say a moz-iba if you went to the movies, but az étterembe if you went to the restaurant, because the “o” and “i” sounds of the first require the suffix -ba to match them, and the “é” and “e” sounds of the second call for the suffix to be -be. Nothing like that happens in the vast majority of the world’s languages.
The hunt for what all grammars share—“Universal Grammar”—has been going on for a long while, and has got about as far as the search for the Holy Grail.4 However, at one level, the answer is obvious, because it is definitional: all grammars regulate the ways in which free items may be combined to make an acceptable sentence.
The trouble with that is obvious: “sentence” is a grammatical concept to begin with. Sentencehood is not an observa
ble quality of acts of natural speech. It’s not just in the poetry of Mallarmé that we have difficulty in knowing where to put the period. Just listen to your children! They never finish their sentences properly.
It is true that we can make sentences in any human language. But it is just as true that most of our actual uses of speech do not involve anything that looks much like a grammatical sentence. When we write, of course, we usually try to write in sentences. But not always.
The second major problem with the axiom of grammaticality—with the idea that what makes a language a language is its having a grammar—is that no living language has yet been given a grammar that accounts for absolutely all of the expressions (including sentences) that are uttered by speakers of that language. The “grammar of English”—or any other language—has not yet been completed, and it’s a fair guess that it will always remain a work in progress.
Flaws of this magnitude in aerodynamics or the theory of probability would not have allowed the Wright brothers to get off the ground, or Las Vegas resorts to turn a profit.
The Achilles’ heel of a linguistic theory that places grammar at its core could be put like this: since universal grammar remains elusive and no exhaustive grammar of any single form of speech has yet been devised, every speaking subject on this planet knows something that grammar does not.