by David Bellos
The suspicion that the language of translated works is not quite the same as the language the translations purport to be in has given rise to scholarly work based not on anecdotes and intuition but on the automated analysis of quite large bodies of translated texts in machine-readable form. These techniques allow insights into what is now called the “third code”—the language of translations seen as a dialect that can be distinguished from the regular features of the target language.2 In one such investigation, it’s been found that English novels in French translation have at least one language feature that seems quite at variance with novels originally written in French.
When you want to add emphasis to one part of a French sentence, you take it out of its normal grammatical place and put it right at the start, replacing it in its ordinary location with a pronoun or dummy word. For example, if you want to disagree with what your children ask for as a treat at the fair, you can say—in English—“But I want ice cream,” using the tone of your voice to stress that ice cream is what you want when the kids are clamoring for cotton candy. The regular way to do this in French is to put I in a special form at the head and then to repeat it: Moi, je veux une glace. “Left dislocation,” as this feature of French is most often called, is pretty common in all sorts of circumstances in speech and writing, not just in arguments with kids. In a corpus of extracts from recent prizewinning novels written in French, it occurred 130 times in an expanse of around 45,000 words. But in a parallel corpus extracted from similarly well-seen novels translated around the same time into French, it occurred only 58 times. The difference is quite marked and can’t be explained by any individual translator’s style. A “third code” does seem to exist.3
What’s even more interesting and especially relevant to understanding translation is that the use of left dislocation in the corpus of translations into French is highly concentrated in one kind of context—in dialogue. In the corpus of texts originally written in French, however, more than half of the occurrences crop up in third-person narrative. None of the occurrences of left dislocation in the entire double corpus is grammatically wrong or stylistically inappropriate, but it seems clear that the language norm to which translators of English novels in French adhere (whether they know it or not) is not identical to the language use of novel-writers in French.
The reason for this particular feature of the “third code” in French is not difficult to find. French grammar books and the teaching of French in schools have traditionally categorized left dislocation as typical of oral speech. Translators seem to have internalized that lesson, even though it runs counter to the observable practice of native writers of French. Translators therefore tend to write in a normalized language and are more attentive to what is broadly understood to be the correct or standard form. In fact, anyone who has personal experience of translation work knows this truth. Translation tends toward the center—to whatever linguistic regularities are conceptualized as belonging to the standard language, irrespective of what native speakers typically say. The plight of the English translator edited into “English-minus” is therefore not exceptional in the world of translation. French translators seem to get to the same place even before copy editors go over their work.
The movement of translation toward the standard form of the receiving language can be highlighted by the fate of regional and social dialects. Bournisien, one of the minor characters of Flaubert’s Madame Bovary, speaks with turns of phrase and vocabulary items that are comically typical of the region where the action is set—rural Normandy in the 1830s. English obviously does not have a conventional way of representing the speech of nineteenth-century countryfolk from Normandy. In principle, a translator could make Bournisien speak in English like a Wessex farmer out of Thomas Hardy or a Scottish preacher invented by Walter Scott. But representing one regional dialect of the source by some regional dialect of the target is rarely attempted in translation. 4 Most people currently think it is just silly to make a Bavarian dairy farmer use Texas cowboy slang, or to have a woman on the St. Petersburg tram express herself in Mancunian in order to suggest her geographic and linguistic distance both from the capital and the standard language. The culture of translation as it presently exists in English as well as in French and many other languages eradicates regional variation in the source. It drives written representations of dialectal speech toward the center.
An obvious case of movement toward the center occurs in Charles Baudelaire’s translation of Edgar Allan Poe’s “The Gold Bug.” The African American slave in the story, Jupiter, is represented as speaking in this manner: “Dar! dat’s it!—him never plain of notin—but him berry sick for all dat.” Baudelaire doesn’t try to find a dialect of French to fit, he just says what Jupiter means to say in standard French: Ah! Voilà la question!—il ne se plaint jamais de rien, mais il est tout de même malade.
What else could Baudelaire have done? No resources were available in nineteenth-century French to match an English convention for the representation of African American vernacular.5
Strange to say, the same treatment is not generally accorded to variations in form and style that correspond not to region but to social class. High-flown, pompous, elegant, or regal forms of language in the source are generally represented by forms of corresponding social rank in the target. Real difficulties arise only when the class register is low, and especially when the language of the source represents the speech forms of uneducated folk. This difficulty runs through all kinds of translating, not just literary prose. No consecutive interpreter, for example, would think of adopting lower-class diction to reproduce for the benefit of a visiting foreign dignitary the kind of language spoken to him by a factory hand or collective farm worker: it would surely seem disrespectful and cause a mighty scandal. In written prose, too, translators shy away from giving the uncouth truly uncouth forms of language in the target text. The reason is obvious—grammatical mistakes, malapropisms, and other kinds of “substandard” language must not be seen to be the translator’s fault. It’s actually easier to translate the ravings of a certified lunatic than the intentionally rude and vulgar language of many modern novels. The outright sanitization of bawdy classics carried out in seventeenth-century France (see here) is quite out of fashion—but something of the same sort goes on in almost any translation project.
The “third code” effects that have been revealed in translations (in French, but also in Norwegian, Swedish, and English) and the strong prejudice against regional variation are, even so, mere sidelights on the less easily pinpointed but far more general tendency of all translations to adhere more strongly than any original to a normalized idea of what the target language should be. To put that a different way: translation always takes the register and level of naturally written prose up a notch or two. Some degree of raising is and always has been characteristic of translated texts—simply because translators are instinctively averse to the risk of being taken for less than fully cultivated writers of their target tongue. In important ways, translators are the guardians and, to a surprising degree, the creators of the standard form of the language they use.
EIGHTEEN
No Language Is an Island: The Awkward Issue of L3
The invention of printing, the rise of dictionaries, the spread of literacy, and the establishment of nation-states are probably the main forces that have led us to accept without question that one language is not another, and that the boundaries between, say, English and Yiddish, French and Italian, are real, insuperable, and firmly fixed. The idea that a translation always occurs between an L1 and an L2, between a “source” and a “target,” is only one reflection of this specific culture of language, where different ways of speaking are conceptualized as distinct entities with clear lines between them. But it was not always thus.
On his return to Genoa in 1298 C.E., Marco Polo was flung into jail. He wasn’t put in solitary, and had the additional good fortune to find an old acquaintance inside. Marco Polo told the tale of h
is great adventures on the Old Silk Road to his cellmate, Rustichello da Pisa, who wrote it all down. Marco spoke in what we would call Italian, and Rustichello wrote his words down in French. The “original” Divisament du Monde (The Travels of Marco Polo) was in all probability an improvised translation, and it contains a telltale sign of the way it was composed: the first-person pronoun we sometimes designates Marco and Rustichello, sometimes Rustichello and his readers, sometimes Marco and his companions.1 This kind of person-switching is typical of oral translation and makes it pretty certain that Marco spoke his account in one dialect and that Rustichello wrote it down in another. You can see the same phenomenon of “unstable anchoring” in Claude Lanzmann’s Shoah, a French-language film about the present traces of the extermination of European Jewry in the period of 1941 to 1944. Shoah is quite exceptional among movies because it does not edit out of the final cut the many acts of two-way translation between the French of the interviewer and the Polish, Yiddish, Hebrew, Czech, and German spoken by survivors and informants. (That is partly why it lasts nine hours.) In many sequences, the translator switches between repeating the words of the speaker without changing the orientation of the speech (as in: “I saw the trains being shunted …”) and giving the information provided by the interviewee in indirect report (as in: “He said that he used to see the trains being shunted …”). On occasions, when Lanzmann wants to pick up on an evasive answer and press the witness further, he falls into the same language-situational trap himself and asks the interpreter, not the witness, “What does that really mean?” Instead of transforming such a riposte into a question in Polish for the witness (“What did you really mean by that?”), the Polish interpreter answers Lanzmann directly in French in her own voice, giving him a personal explanation of what the witness had meant to say.2 Such alternations are natural, almost unavoidable departures from the artificial interpreting norm, which overrides the fundamental equation of speaker and voice. In two-way human interaction using a linguistic intermediary who is physically present, it is uncommonly difficult to maintain the fiction of the translator’s nonexistence. Even at the UN, where professionals observe strict rules of noninterference and are put in soundproof glass boxes just to make sure, interpreters still occasionally break off from reproducing the other’s speech (using the same personal pronouns and tenses as the original speaker) and resort to a third-person report when something arises that lies outside the common run of diplomatic speech. Nikita Khrushchev, the Soviet leader in the 1950s and 1960s, was notorious for his impromptu use of impenetrable Russian proverbs and jokes, and his interpreters would often find themselves saying—in the third person—“the general secretary of the CPSU just made a joke.”
Marco Polo and his translator-scribe were using two related but different languages to tell the world about the fantastic diversity of human societies. They were living among a welter of partly intercomprehensible dialects of an originally common tongue, but only one of them was well suited to bringing news of Shangdu to the West—and that was French. In the lands bordering the Mediterranean in the late Middle Ages, French was roughly the equivalent of what English is today;3 the hybrid features of the first manuscript of The Travels might best be likened to the “Globish” written by nonnative speakers the world over nowadays.
But as soon as the manuscript of The Travels got into circulation, other scribes did what any copy editor would do in the modern world—they tidied it up, subjected it to what French publishers call the toilette du manuscrit, and put it into what was considered, respectively, “proper French,” “proper Tuscan,” and, for purposes of wider distribution and retranslation, “proper Latin,” too. By the end of the fourteenth century, there were versions in Czech, Gaelic, German, Tuscan, and Venetian, as well as French, all of them retranslated from the Latin translation, which had itself been done from an early version in the Italian dialect of Venice, based on a source that was either the text of the first known manuscript or something very close to it.4 These progressive emendations of Marco Polo’s narrative mostly suppressed the voice switches of the original translation and turned it into a more situationally consistent narrative. That’s because those later scribes were not translating the traveler’s speech but a story that already was a written text. You could say that something very important was lost; you could also say that The Travels became a classic of exploration literature precisely because, like many modern novels, it was rewritten by professionals. Then as now, the borderline between translating and improving a text—between “helping the reader” and “trashing the source”—is not at all clear-cut.
The borderline between translating and rewriting is in fact no more wiggly than the one between source and target language in the case of many extended texts. Tolstoy’s War and Peace is an oft-quoted example of this. In the Russian original, parts of the novel are in French. This reflects the language practice of its characters—Russian aristocrats of the early nineteenth century used French for much of their social and intellectual lives. Indeed, when challenged by a Freemason to speak of his hopes and desires, Pierre Bezukhov found himself unsure of how to answer, “being unaccustomed to speak of abstract matters in Russian.”5
Translating War and Peace into French is both impossible and easy. Reproduced without alteration in French, the French speech of Russian aristocrats loses all its meaning as a marker of class, and there is no way of indicating by linguistic means alone that a sentence spoken in French is different from the other sentences that are (by force of translation) in French as well. The title page of the French translation may well say Traduit du russe, but that is only partly true. It is “translated” from French as well.
Identical translation problems arise in a vast array of European fiction. The first page of Balzac’s Le Père Goriot contains a sentence in English (“All is true!”) that has an entirely different environment and force when reproduced in an English translation of the text. But what can you do? Translate “All is true!” back into French? Or alter the spelling to “Oll eez troo” to indicate it having been thought by a Frenchman with an atrocious accent? Balzac had no qualms about altering the orthography of French to represent the regional accent of Nucingen, a Jewish banker from Alsace, who also appears in Le Père Goriot. Current conventions don’t allow translators to do that to the diction of narrators—but there’s no strictly logical reason for withholding a lousy accent from Balzac’s narrator, too.
In fact, the more you read in any language, the harder it gets to find an extended text written in that language alone. Two novels I read last year with much pleasure illustrate this point. Michael Chabon’s The Yiddish Policemen’s Union is an entertaining fantasy of a Yiddish-speaking colony in modern-day Alaska. The English dialogue of the characters is understood to represent a translation from Yiddish—or rather, from an imaginary state of Yiddish enriched by fifty years of further existence as a living, growing language on American soil. Chabon’s text is a wonderful hybrid of real and imaginary languages that play with one another—and a translation of it into any other tongue could hardly be considered a translation “from English” alone. Similarly, English, August by Upamanyu Chatterjee mixes Hindi and Bengali with standard literary English to create a language picture of its central character, Agastya, nicknamed August in his English-language boarding school. A keen reader who speaks only English could use it to learn a number of Bengali and Hindi words, just as a reader of the short stories of Junot Díaz can pick up a good amount of Spanish from his hybrid, “Spanglish” texts. But Tolstoy, Balzac, Chabon, Chatterjee, and Díaz don’t switch around between tongues just to provide language lessons. They do so because language alternation (called “code switching” in some kinds of language study) is endemic to all kinds of language use.
I used to have a friend who ran a bank branch in a rural backwater of southwestern France. We always spoke to each other in French, but whenever he came across me in the street or a field, he would begin by saying “Peace and love,” which he p
ronounced pissanlerv. In the same period I knew a Scottish doctor who used to hurry his children along by saying “the tooter the sweeter,” blending tout de suite with something like “the sooner the better.” Both those acquaintances were speaking in a language (respectively, French and English) but they were also speaking another at the same time (respectively, English and French).
Translation is usually thought of as a process involving only L1 and L2, or source and target tongues. But, as we’ve seen, sources typically include smaller or larger amounts of L3, a language that is not either of translation’s traditional twins. When L3 is L2 (as in the case of War and Peace translated into French), it is inevitably rubbed out, but when it is not (in a Swedish translation of Chabon’s novel, for example), it’s not at all obvious how it should be handled. Mind-boggling though they may seem, these problems are not marginal to the way language is commonly used and therefore not irrelevant to translation, either. However convinced we may be that different languages are different things and not to be confused with one another, in practice we never stop muddling them up. The borderline between, say, English and French is more ragged and foggy than grammars and dictionaries would have us believe. “Sayonara, amigo!” may not be an officially English way of saying farewell, but few English speakers have any trouble in knowing what it means.