Is That a Fish in Your Ear?

by David Bellos

  Sixty years of experience have not made it any easier to predict whether an individual can be turned into a conference interpreter or not. Even now, between half and three quarters of all students admitted to interpreter training courses fail to enter the profession.6 At the beginning, in the aftermath of the Second World War, the disastrous history of the twentieth century had produced many thousands of people with outstanding language skills in several of the six official international languages (Spanish, English, French, Chinese, Russian, and Arabic)—children of refugees from the Russian Revolution brought up in Shanghai and educated at the Lycée Français, where they learned English, young refugees from German-occupied France who had spent months or years in Cuba or Mexico awaiting a U.S. visa before going to college in New York, and so on. The first generation of the elite of the translating professions consisted mostly of young people from backgrounds of that kind, who remained in post for thirty years and more. These founding mothers and fathers of the conference-interpreting community have now retired, and it has proved difficult to replace them. The lack of personnel is particularly acute for the two most-needed languages in world affairs today—Arabic and Chinese. Even the Russian- and French-into-English booths are getting harder to fill.

  The structure of conference interpreting at the UN and its agencies and at most other international gatherings that can afford it is not now quite as it was at the Nuremberg Trials. The rules invented for that first experiment were that all interpreters should work only into their “native” language (now called their A language, “A” standing for “active”), and that all interpreting should be done from the “original.” With six UN languages currently in operation, that would require six teams of five translators, or thirty people in all, to service a single meeting. The job is now reckoned to be as stressful as the work of air traffic controllers; the eighty-five-minute slots used at Nuremberg have been replaced with a routine of alternating thirty-minute shifts (the Chinese and Arabic booths change over every twenty minutes) through a normal (short) working day—so that in fact you would need sixty people, not thirty, to service an international meeting if the original rules were still applied. There just aren’t sixty people with those high-level and variegated skills that can be gathered at any one time in any one place in the world, not even in New York City. The following schema allows the illusion of seamless language transfer to be achieved with a team of just fourteen members:

  In the French booth: two interpreters, one listening in Spanish and English, the other listening in Russian and English, and giving out in French

  In the English booth: two interpreters, one listening in French and Russian, the other listening in Spanish and French, and giving out in English

  In the Spanish booth: two interpreters, both listening in English and French, and giving out in Spanish

  In the Russian booth: two interpreters, both listening in either Spanish or French as well as English, and giving out in Russian

  In the Chinese booth: three interpreters working shifts, taking in English and Chinese and giving out in Chinese and English

  In the Arabic booth: three interpreters working shifts, taking in French or English and Arabic and giving out in Arabic and English or French

  In other words, Chinese gets into Spanish, French, and Russian by relay from the English channel, and Arabic gets into Spanish and Russian by relay either from English or, most often, from French; Spanish and Russian get into Chinese by relay from the English channel, and into Arabic by relay from French. If the Russian interpreter in the English booth has gone to the bathroom, then the Russian channel also gets into English by relay from the French booth; similarly, if the Spanish interpreter in the French booth has a nosebleed, Spanish gets into French by relay from English.

  Relay, or double translation, is in principle a bad idea, as the possibility of error is increased, as is the time lag between the delegate’s speech and the output in listeners’ headphones. Also, the fact that Chinese and Arabic interpreters work both into their A language and from it into English is not a good idea—working both ways at once more than doubles the mental stress involved. But the devices of relay (double translation) and retour (one interpreter working in two directions) are godsends for the UN officials whose task is to ensure the smooth running of the meetings. Without relay and retour the whole system would be vastly more expensive—and it’s not exactly cheap as it is.

  In the European Union, further refinements are used to ensure that meetings of a body with twenty-four official languages can be coped with. Full symmetrical interpreting under Nuremberg rules—that’s to say, each translation direction being supplied by a single dedicated interpreter—would require a team of 552 interpreters, exceeding by far the number of delegates taking part in any meeting, and that’s clearly not feasible. The system works like this:

  When all participants in a meeting understand at least one of the EU’s working languages (English, French, German, and Italian)—and this is nearly always the case—then an asymmetrical language regime is used. “Asymmetry” means that participants may speak in any of the official languages (as long as they let the interpreting service know which one ahead of time), but may listen in only one of the four working languages. Such a meeting would be said to have a “24:4” language regime. If each translation direction were served by a dedicated individual, that would require up to eighty interpreters per session, which is still far too many.

  The number is further reduced by interpreters with two A languages who can work into both, a device called cheval, but also, most crucially, by retour—interpreters who work into their B language as well. The greatest economy of all is of course made by relay. When the Lithuanian delegate speaks, an interpreter with Lithuanian B provides a simultaneous German translation, which the German–English, German–French, and German–Italian interpreters use for their versions in the working languages (and in a 24:4 regime, no further language versions are required). In this example, the hub or pivot language is German; for other languages at the same imaginary meeting, the hub may be English, French, or Italian, bringing the total number of actual bodies needed to service a meeting under 24:4 to a maximum of twenty-eight, and quite a lot fewer if (for example) the Portuguese–French interpreter also does Spanish when French is the hub language, or the Swedish–German interpreter also does Danish when German is the pivot. Because all EU interpreters must have two B languages, the use of asymmetric regimes together with cheval, retour, and relay suffice to provide just about affordable simultaneous interpreting in Brussels and Luxembourg, and at the European Parliament in Strasbourg.7

  At the UN, the system is often invisible to users. Interpreters are placed at the rear or the side of the assembly hall behind soundproofed and tinted glass screens. You can attend a dozen meetings without even realizing the interpreters are physically present—so it’s only natural they should get taken for granted. What’s more insidious than the occlusion of the interpreting magic, however, is the impression that anything you say can be simultaneously heard in all other tongues. Conference interpreting, glamorous though it is, buries the real difficulties—and the real interest—of language transfer beneath sophisticated, almost circuslike tricks of the language trade. It makes people think that it’s only a matter of time before we can all have a device to stick in our ear—the “Babel fish” of The Hitchhiker’s Guide to the Galaxy—to provide us with instant communication with all the peoples on earth.

  Unlike most translators in written mode and a high proportion of consecutive interpreters, conference interpreters are rarely specialists in any particular field and come closest to being pure language professionals. Few domain-specific organizations are sufficiently large to justify having salaried interpreters on their books: only sixty-seven organizations in the world employ members of AIIC (the interpreters’ professional body) as full-time staff, and only four (the UN in Geneva and New York, and two of the International Criminal Tribunals in The Hague) employ more tha
n ten. As a result, most of the three thousand members of AIIC (and a roughly equal number of nonmembers) work freelance and travel from conference to conference, dealing with all sorts and kinds of topics. Fast-talking yet good listeners, interpreters must be both alert and relaxed, able to tolerate unspeakably boring harangues but also quick to pick up the gist when something entirely new comes on the agenda. They belong to a rare breed.

  They might become even rarer, because there are several threats to the survival of the species. First, the precipitous decline in the teaching of foreign languages in the English-speaking world in the last fifty years means that there are ever fewer entrants to the profession with English A. If you prevented boys from having bicycles, then the Tour de France would become a celebration of geriatric fitness in a decade or two, and then stop. If you don’t teach native English speakers two languages out of Spanish, Russian, Chinese, Arabic, and French intensively to high levels while they are young, you will not have candidates for interpreter training within ten or fifteen years. There are many English–Spanish bilinguals, of course, but very few of them have another UN language to the requisite degree of fluency. If the requirement were lowered from two to one foreign language for English A, then the system could be run on relay and retour, and staffing problems would be less acute. However, because ten applicants to a translators’ school produce no more than five entrants, and because barely one third of those graduating will be found good enough to enter the profession, large investment in language education throughout the English-speaking world is urgently needed. Without it, the next cohort of our politicians and diplomats, businessmen and consultants, human rights campaigners, international lawyers, and policy wonks may well be reduced to stuffing fish in both ears.

  A second threat to maintaining current language practice in international organizations is that some states may become unwilling to finance simultaneous interpretation into languages that are ceasing to be global vehicular tongues—but the replacement of Russian (for example) may prove politically impossible for many decades yet, and nobody has a clear idea of what might replace French.

  But the bigger threat looming on the horizon is something that’s going on right now in research labs in New Jersey and elsewhere. Using the technology of speech recognition that allows a widely available word processor to generate text from speech, alongside the speech synthesis systems that power today’s automated answering machines, the FAHQT target that current U.S. science policy encourages could well become FAHQST—fully automated, high-quality speech translation. Experimental systems not very far from commercial release already produce running English text from Spanish speech. I may not live to see or hear it, but many of you probably will: automated interpreting for the secondary orality of predictable international diplomatic prose, for tourist inquiries at hotel reception desks, and maybe for other uses as well.

  You will then enter the era of tertiary orality. It will be another world.

  TWENTY-FIVE

  Match Me If You Can: Translating Humor

  A relatively uncontentious way of saying what translation does is this: it provides for some community an acceptable match for an utterance made in a foreign tongue. This doesn’t go very far, but as it applies equally well to conference interpreting, comic strips, legal contracts, and novels, it’s a reasonable place to start.

  What it leaves open are three huge questions:

  1. What makes a match acceptable?

  2. Which of the infinite catalog of qualities that any utterance has are those that a translation may or must make match?

  3. What do we mean by “match,” anyway?

  Those are the questions that translation studies has always sought to answer, sometimes under heavy academic disguise. “Translation quality evaluation criteria,” for example, is a label for answers to question 1. But whatever way you ask these three questions, the answers are not easy to provide.

  All sorts of criteria may be involved in judgments made by different people at different times about the acceptability of a match—theoretical criteria, or practical, social, or cultural ones, and no doubt, on occasions, purely arbitrary ones, too (such as the translator is a famous prizewinner and must have got it right). Trying to rank these criteria or to distribute them to classes of situations where they might apply seems too complicated by half. It is perhaps more fruitful to work in from the outside edge and to begin by looking at places where matches are commonly believed to be extremely difficult to find.

  One area flagged by nearly all translation commentary as being match-poor is utterances that raise a laugh or a smile. Here’s an old Soviet joke about Stalin:

  Stalin and Roosevelt had an argument about whose bodyguards were more loyal and ordered them to jump out of the window on the fifteenth floor. Roosevelt’s bodyguard flatly refused to jump, saying, “I’m thinking about the future of my family.” Stalin’s bodyguard, however, jumped out of the window and fell to his death. Roosevelt was taken aback.

  “Tell me, why did your man do that?” he asked.

  Stalin lit his pipe and replied: “He was thinking about the future of his family, too.”1

  Well, that’s a translation (from Russian), and even in Russian it’s a translation already, because exactly the same joke has been told over the centuries about other brutal potentates, starting with Peter the Great. We can safely assume that this joke form can be preserved together with its point in any human language under two conditions that are only incidentally linguistic ones: the target language must possess an expression for “thinking about your family” that can apply to two slightly different projects (to provide support for your spouse and children, and to protect them from persecution); and that the listener understands or can guess that evil potentates punish disobedient underlings by persecuting their relatives. These two conditions may not be met in all cultures and languages in the world, but they are surely widely available. The “untranslatability of humor” hasn’t survived the very first dig of the spade.

  Provided the two general conditions given above can be met, the jump-for-Stalin joke can be rejiggered to fit a wide variety of other historical and geographical locales in the same language or any other, and still be the same joke. There are very many transportable, rewritable joke patterns of that kind—including those politically incorrect ethnic disparagements of near neighbors that you hear in structurally identical form when the French talk about Belgians, Swedes about Finns, the English about the Irish, and so on.

  Translating these kinds of circulating jokes means matching the pattern made by the interplay of presupposition and meaning that constitutes the point, and then rewriting all the rest to suit. An ability to recognize the match is not rare, and may be almost universal. But the ability to find a good match is one that only some people have. However, we don’t have to go far to find humorous uses of language that work in a slightly different way.

  A Brooklyn baker became deeply irritated by a little old lady who kept standing in line to ask for a dozen bagels on a Tuesday morning despite his having put a big sign in his window to say that bagels were not available on Tuesday mornings. When she got to the head of the line for the fifth time in a row, the baker decided not to shout and scream but to get the message through this way instead.

  “Lady, tell me, do you know how to spell cat—as in catechism?”

  “Sure I do. That’s C-A-T.”

  “Good,” the baker replies. “Now tell me, how do you spell dog—as in dogmatic?”

  “Why, that’s D-O-G.”

  “Excellent! So how do you spell fuck, as in bagels?”

  “But there ain’t no fuck in bagels!” the little old lady exclaims.

  “That’s precisely what I’ve been trying to tell you all morning!”

  There are different ways of saying what the point of this—admittedly paltry—joke is. It makes a character speak out loud a truth she had been unable to internalize. There’s no reason to suppose that matches cannot be found in any langu
age to make fun of some person in the same way. The overall point is made by playing on a difference between written and oral language: structurally similar plays can probably be found and constructed in any language that has an imperfectly phonetic writing system. But once we get down to the implementation of these two features, hunting for matches becomes much more difficult. The assimilation of the present participle of a taboo word to the stem of that word plus the preposition in is possible only because in English the distinguishing mark of the first—the final consonant, g—is habitually dropped in colloquial speech. That’s a low-level, local feature of a particular language, and it turns on the slight mismatch between its spoken and written forms. A structural match in any other language would most likely have to turn on a phonetically and grammatically different feature that may or may not allow the same point—making someone stupid say what they don’t want to understand by diverting their attention from the issue through an intentionally deceptive spelling game.