The educational gyroscope
he inertia that besets the translation business is almost wholly due to persistence of the algorithmic approach. So what is the explanation for this persistence? After all, there is nothing particularly original or revolutionary in my explanations proving why an algorithmic approach is misguided; this knowledge has been part of mainstream linguistics for the best part of half a century! So why do the immense majority of translators still practise an algorithmic approach? I can think of three main reasons.
Belying the frequent charge that schools are staffed by overidealistic revolutionaries overeager to experiment suspiciously novel teaching methods on our unsuspecting overvulnerable children, the fact is that language teaching in schools has changed very little in a hundred years, despite overwhelmingly conclusive evidence that the traditional "grammar-translation" method is grossly ineffective. And when I say "overwhelming," I mean it would be difficult to find even a single serious work that justified the continued use of this method. This is not even a controversial issue: the whole body of linguistic and pedagogical thought over the last five decades concurs on the inefficacy of the grammar-translation method of language teaching. There is simply no serious dissent at all.
And yet, the French and German exercise books of my young nieces look identical to mine, which looked identical to my father's, which looked identical to his father's. Like my grandfather, like my father and like myself, my nieces will leave school after six or seven years of foreign-language instruction without any kind of practical fluency in a foreign language. They will be able to recite verb conjugations and other grammatical rules, but they will be unable to apply them in real-time for the purposes of an actual conversation.
Oh, I was forgetting... Yes, they will be able to perform a rather curious circus trick, a neat little number that consists in aligning words in one language alongside other words in another language, in accordance with algorithms so basic they would occupy a tiny pinpoint at the very tip of our grammar triangle. Their teachers call this "translation." Excepting the tiny minority with the good fortune to have language teachers in tune with current scientific thinking, all schoolchildren get six or seven years' sustained training in bad translation practice. They get regular hands-on practice in how to do it wrong! To be fair to teachers, I admit that it is not easy to teach modern languages properly to large classes, especially when they suffer from discipline problems. But there is still no excuse for misuse of translation in the language classroom. So why do teachers fall back on this technique, which is proscribed in every serious work on language teaching written in the last fifty years? There are three reasons. One: "that's the way I was taught, and if it was good enough for me it's good enough for them." Similar gyroscopic reasoning purportedly explains why battered children become child batterers. But it fails to explain why teachers today go to work in cars whereas their grandfathers sat atop mules. And it fails to explain why people prefer modern medicine to the sawbone's gruesome toolkit. Two: it's perceived as a quick and easy exercise. Elementary algorithmic transposition is seen as a convenient way to check the pupil's understanding of grammatical constructions. So instead of getting the pupil to develop agility in his foreign-language grammar generator, the teacher instructs him to apply low-level conversion algorithms. Instead of getting the pupil to observe a situation and describe it spontaneously in the foreign language (however falteringly, at first), the teacher will ask him to "translate" an English-language sentence describing the situation.
The wrong way:
The right way:
But not only do schoolkids get regular hands-on practice in how to do it wrong: they are actively discouraged from doing it right. Since the purpose of the transposition exercise is to check understanding of grammatical points, there is a strong disincentive to translate meaning. In an exercise whose stated purpose is to check knowledge of the past tense of "avoir", how will the pupil behave when confronted with the sentence "nous n'avions pas assez d'argent"? He will always opt for "we didn't have enough money" and never for "we couldn't afford it." Now I know that no half-way responsible teacher would knowingly set an exercise liable to mislead the student in this way. And I realize that only a monster of a teacher would mark "we couldn't afford it" wrong. But the fact remains that current language-teaching practice strongly and Pavlonially programmes pupils into a diametrically wrong view of translation. Very few, even among those who go on to study translation and become translators, will ever manage to correct this view. Those who go on to study modern languages get more of the same: bad translation is a requisite throughout the university curriculum, right up to the very highest level of academic achievement in France (the agrégation competitive examination, which offers access to a career in teaching at university level). This brings us full-cycle and puts extra spin on the educational gyroscope, to keep our bearings constantly fixed in the diametrically wrong direction.
The third reason why teachers misuse translation in the language classroom is because it offers a semblance of objectivity. As a fervent convert from the sciences to the arts, I have never understood why so many teachers and practitioners in the subjective realm feel so inferior to their counterparts in the objective realm. But the inferiority complex is very real, very widespread and very acute. I left science because I was attracted by the prospect of fuzzy borders, and was astonished to discover that many of my colleagues in the subjective realm actually feel naked and embarrassed without the hard borders I had fled from. To seek the opaque clothing their inferiority complex demands of them, many will resort to blatant intellectual dishonesty, and for all its fragility and elusiveness, the most popular figleaf of all is that of objectivity. (The other big one is standardization, as we shall be seeing.)
Steven Jay Gould used the term "physics envy" (by analogy with the "penis envy" of Freudian psychology) to chide his fellow life-scientists for their apparent unease at not being able to work within the physicist's reference frame, whose perceived rigidity was seen as conducive to intellectual credibility. The expression applies perfectly to translators and translation theorists.
The problem stems from the desire of the educational establishment to impose some kind of intellectual parity across all disciplines, in subjective and objective realms. It is felt that a three-year degree course in modern languages should be "worth" a three-year degree course in physics, though students of both subjects realize this is absurd. Physics students suffer from French envy at university because they have to work much harder; they'd get kicked out if they tried to keep up with the partying pace of their pals in the French department. But they get their own back after graduation, when they're wooed with offers of well-paid jobs while their pals struggle for years from one bum, boring, temporary office job to the next. Nobody seems willing to face up to the fact that you do not need to go to university to learn French: two years living in France will be more than sufficient, especially if you were lucky enough to get a head start by hitting on the five percent of enlightened language teachers at school. Modern-language departments therefore find themselves in the uncomfortable position of having to find something to keep the students busy for three years. The bulk of the time is spent on literary criticism, the main result of which is to ensure that most students never dream of opening a book in any language ever again after graduating. And, again, the professors' physics envy is assuaged by compulsory practice of pseudo-objective mechanical transposition between the two languages, which renders the students unemployable for useful translation work.
So aren't things any better at translation school? Well, no, yes, no and no:
- No, first of all, if we judge by the results: whereas you could reasonably trust a graduate from medical school to diagnose and cure a simple ailment, you most certainly cannot trust a translation-school graduate to translate properly. This I know from extensive personal experience over a twenty-year career in translation: I see none but a negative correlation between formal training and actual competence in translation.
- Yes, because translation schools do at least give recognition to serious linguistics; few translation school teachers would quibble with the analysis set out in this article so far, for example. Translation schools also do valuable work in translation theory, but because much of this work seems highly abstract, it is overtly derided by many practising translators. While I do understand their perplexity at the contrived obtuseness of some learned papers, I rarely join in the derision. I see translation theory as taking us a good couple of metres down into the comparative grammar triangle, and though this is still the tip of the iceberg, it is valuable because all knowledge is valuable. We might be daunted by the size of the task, but that shouldn't stop us from making a start. Then there are parallels between translation theory and work on universal grammar. Universal grammar proves much more elusive than was originally expected, but we learn from its very elusiveness.
Two good recent books on translation are Mona Baker's In Other Words and Mildred Larson's Meaning-Based Translation. But though both claim to be translation textbooks, I see them more as fascinating studies in comparative linguistics. They are full of stuff like this:
"If theme is whatever occurs in initial position we would have to acknowledge that some languages prefer to thematize participants (expressed as subjects in SVO and SOV languages) on a regular basis while other languages prefer to thematize processes (expressed as verbs in VSO languages)."
"The sequential and simultaneous relations [...] relate events which are of equal prominence; that is, in an addition relation. There are also units related to one another which are not of equal prominence. One of the units supports the other. There is a support-HEAD relation between the two units."
"[...] some English texts make little or no use of conjunctions. There are often pragmatic reasons for the preference of certain types of conjunction and the frequency with which conjunctions are used in general. [...] some genres are generally 'more conjunctive' than others and [...] each genre has its own preferences for certain types of conjunction. Religion and fiction use more conjunctions than science and journalism."
Mona Baker (citing work by Smith and Frawley)
Technical studies like these are admirable feats of observation and analysis. But they are not, and hardly claim to be, "how to" books. To take Mona Baker's observation that "religion and fiction use more conjunctions than science and journalism," for example, we cannot imagine any translator telling himself to "lay off the conjunctions because this is a scientific paper and not a sermon"! A scientific writer (or a translator writing in a "scientific" register) will simply find himself using the features of that register. If that register happens (as Smith and Frawley claim) to be relatively light in conjunctions compared to a religious register, the writer will not be consciously aware of this fact as he is writing. Again, we are very much among the fuzzy rules. So the real lesson a translation student learns from this observation is simply that there is never any reason to assume that a conjunction in one language should generate a conjunction in another. Books like Baker's explain the non-parallel, non-linear, non-algorithmic nature of translation so meticulously, so convincingly and so relentlessly that the onlooker is left dumbfounded by the inability of translation school graduates to put the principles into practice. This leads me to my second "no."
- No, because translation schools too are frequent victims of physics envy. In her introduction to In Other Words, Mona Baker displays unmistakable symptoms of the malady herself:
"Like doctors and engineers, [translators] have to prove to themselves as well as others that they are in control of what they do; that they do not just translate well because they have a 'flair' for translation, but rather because, like other professionals, they have made a conscious effort to understand various aspects of their work."
This kind of reasoning, born of the translator's perpetual yearning for intellectual parity with professions like medicine and engineering, leads to educational syllabuses modelled on those of medical and engineering schools. But nothing could be more misguided.
A doctor facing an unusual pathology can seek information in medical literature then apply this information to solve his problem. An engineer designing a machine can run tests and take measurements to ensure it will run properly, and once the machine is running he can even run further tests and take further measurements in order to gauge the accuracy of his design assumptions. Both are consciously applying learned knowledge, and it would be inconceivable that an engineer or doctor, having learned and assimilated all the relevant principles, would then be unable to apply them. Sure, there is always a gap between theory and practice, but an educational syllabus in engineering or medicine can realistically include practical work to cover a reasonably representative sample of typical situations.
The situation is very different in translation. It is perfectly conceivable for a student to read and fully understand the principles set out in books like In Other Words, yet still be unable to apply them. It is even perfectly conceivable for a linguist to write books on these principles yet be unable to apply them: I have several badly-written books on linguistics, and, surprisingly enough, linguists actually have a rather bad reputation for writing style. I pride myself on understanding these principles, even to the point of attempting to explain them to others. Yet I know that in certain situations I am unable to apply them properly; some translations are just too tough for me! If an engineer is able to identify and explain an engineering phenomenon, this, for him, is the end of the problem. But for the translator it is not even the beginning. Many translators would even say that identification and explanation are not even relevant to their task. I don't agree with this outlook, especially when it condones the pursuit of ignorance, but I certainly do understand it.
The problem, again, lies in the fuzziness of the rules, which are not rules at all but merely approximate descriptions of observed behaviour. An engineer can look up the physical properties of the materials he will be using but a translator cannot look up the properties of his linguistic raw materials, first because nobody has ever come anywhere remotely near to cataloguing these properties completely, second because any such catalogue would be so huge and complex as to be unworkable, and third because language just does not work like that. A software engineer can read any number of practical "how to" books on application design, but there is no such thing as a "how to" book on translation, again, because language just does not work like that.
So how does language work? The best explanation I have read comes in Pinker's The Language Instinct: highly recommended (despite what I consider an excessively Darwinian slant). One key point is that most utterances are novel, individual, creative and largely unrepeatable acts. Utterances are not formed by consciously assembling parts listed in a dictionary according to rules listed in a grammarbook. The parts are assembled, and they are assembled coherently, but the assembly process is masked from us, just like the biomechanics of walking is masked from us. The act of walking involves many interrelated muscle, nerve and bone processes, but we do not consciously trigger our nerve paths, actuate our muscles or shift our bones. We just walk. In the same way, we form ideas in response to the situations we experience, and to convey these ideas, we just talk. Yes, writing is more deliberate than speaking, but it is governed by exactly the same principles, in the same way that an athletics discipline is more deliberate than the mere act of walking, but governed nevertheless by identical biomechanical principles.
Knowledge of biomechanics does not make a champion athlete, and knowledge of grammar or linguistics does not make a good writer or translator. And I am setting aside the fact that our understanding of linguistics lags far behind our understanding of biomechanics: we do know a great deal about bone and muscle, but very little about the brain circuitry responsible for language. Language, then, is not a learned skill like differential calculus but an innate skill like walking. And writing is a slightly more elaborate form of language much like athletics is a slightly more elaborate form of walking.
To be really effective, a translation school would have to offer something more akin to coaching in athletics than formal lecture-theatre instruction in differential calculus. Personal experience corroborates my theoretical musings to confirm that if there is any way at all to teach translation, that way can only be one-to-one coaching, along the lines of an apprenticeship. If engineering really is 99% perspiration and 1% inspiration, translation is just the opposite. I hate to disagree with Ms Baker, but 'flair' is just about everything to the translator. And while flair can be nurtured, it cannot be taught.
- No, finally, because translation schools will often attempt to cure physics envy by inoculating a virus that is in fact more dangerous. To counter the sometimes legitimate accusations of irrelevance, inefficacy and inapplicability, they will resort to almost anything to show their willingness to keep in touch with the world outside and address the real-life needs of the booming worldwide translating industry. I will be examining this more fully in Part 3, but for now will just mention two points. The first concerns the fad for filling out the teaching staff with real-life translators with some kind of business prominence. The problem here is that because translation is the ideal salesman's productsomething people need but are unable to assessbusiness prominence often means good salesmanship alone. The result is that translation students are being taught by some excruciatingly bad translators, with neither flair nor theoretical understandingfor whom Mona Baker is about as intelligible as Albert Einstein. The second concerns the fad for terminology courses, which are about as useful to the professional translator as a stylophone to the concert pianist. Terminology courses are the climax in physics envy because they stand safely in the objective realm, protected against any risk of interference from the messy business of actually writing. Of course, they are very easy to teach. And lower-ability students love them because they offer the prospect of neat, precise, numbered "right answers." By accepting terminology as a useful field of study, the translation school tacitly denies that the translator needs familiarity with the subject matter. If a translator is sufficiently familiar with his subject matter, he will have very little need of heavy-duty terminological assistance. Heavy-duty terminology aids only become useful if the translator has no idea what he is writing about. And under those circumstances he should not be translating that kind of material in the first place.
As project manager with a large French translation agency I would grope around for analogies to give new recruits a glimpse of what I meant by good translation practice. Often, I would ask them to consciously distance themselves from the source text like a visitor to an art gallery will stand back to view a painting. I'd tell them that if ever they found themselves struggling with a particular non-technical word in French (like "animation"), it meant they weren't standing far enough back. If ever they found themselves reaching for the dictionary for ideas on how to translate a word they had in fact understood, it meant they were letting the brushstroke whirls conceal the irises.
My other favourite, particularly apt for the agency's aerospace projects, was escape velocity. To translate well you have to overcome the gravitational pull of the source language. This means your own text needs its own momentum. If your own text is to have its own momentum, you obviously have to know your subject matter, which brings us back to the requirement on understanding discussed in Part 1. But that's not all: you have to actually appropriate the text. You have to get into it. Pinker postulates that children learn languages spectacularly more easily than adults because language-learning brain circuitry is redeployed to other functions after a certain age. An alternative theory, supported by applied linguists like Stephen Krashen, stresses the importance of affective factors; since language plays such an important part in personal and group identity, adults are psychologically ill-prepared for going through the language-learning process. Of course, there is nothing incompatible between these theories, since if an affective factor is real, it too must be implemented in brain circuitry. I mention this because I think affective factors are important in translation. If a translator is to get into a text, he must be an actor. And quite aside from subject-matter familiarity, I doubt that any actor is equally at home in any role. As a life-long physical and intellectual nomad, I believe I am blessed with the sense of universal curiosity that is the translator's second most important quality (after writing ability). I thrive on variety, and this enables me to enthusiastically accept a wide range of acting roles. But everyone has his limits. Different translators will quake at the prospect of impersonating scientists, engineers, poets, analysts, literary figures or advertising copy writers. None of these roles intimidates me unduly, and I welcome technically tough projects as an opportunity for learning. No, what throws me as a translator is the piece of writing that says nothing. The infatuation with communication is a disease of modern society. Public relations departments seem beset by the conceit that captive customers will have nothing better to do than read the repetitive verbiage coaxed out of would-be journalists cluttering up the payrolls of a hundred sad little PR agencies. Personnel departments imagine that the company magazine will enjoy pride of place on the worker's bedside table, guarding against the risk of painful withdrawal from corporate culture over the weekend. In this kind of publication, nobody ever has anything interesting to write about, and the sterile, thankless task of writing about nothing will usually go a junior office employee who can't yet be trusted to perform useful work on his own. The translator thus finds himself in the position of the naughty schoolboy punished with a three-hundred-word essay on the importance of being earnest, which would actually be fine were it not for the fact that the junior office employee got in there first. On this kind of project, I can experience great difficulty getting up to escape velocity. On occasions I have managed to persuade the customer to let me write the thing from scratch, without going through the junior office employee at all. But when I am unable to do this, I observe a curious mechanism at work. As a rather boisterous schoolboy I got a lot of practice writing three-hundred-word essays on the importance of being earnest, and partly because I suspected the essays would never be read, I would colour them with parodies and the occasional unpleasant insinuation about the teachers and school life in general. And I'd do this gently and subtly enough to escape the teacher's wrath if ever the thing was read. Much later, as a language teacher, I imagined training up a total novice to full competence in how not to speak business English. Day after day, month after month, I'd get him to practise proper English grammar, but using a slightly modified register. My student, preferably a senior government official or company president, would attend his first important international meeting fully confident in his English language skills. But instead of saying "I think this matter calls for careful reappraisal," he would come out with "this is a heap of crap and you fucking well know it, asshole." In an impeccable RP accent, of course. There is a mischievous streak in me.
Sometimes, on a particularly numbing, particularly vacuous project, I will read through my work and berate myself. "No native English-speaking writer could possibly have written that. How could you ever attribute that to an authentic junior office employee who can't be trusted to do useful work in Melton Mowbray? Come on, Paul! You know about the read understand write principle. You know about sawbone mentality and saddle-shyness. You were the one that invented them, man! You know perfectly well that with the two muscular extensions at the tip of its trunk, the elephant can 'remove a thorn, pick up a pin or a dime, uncork a bottle, slide the bolt off a cage door and hide it on a ledge, or grip a cup so firmly, without breaking it, that only another elephant can pull it away' [Pinker]. So what is this crap?" And then one day it hit me. Perhaps this wasn't ordinary crap at all. Didn't I recognize the mischievous essence of those subversive schoolboy essays? Wasn't there just a hint of the spiked handshake soundtrack: "How do you do? Can't believe it: you're even fatter than in the photos"? I mean, if I really was down there unentrusted with useful work in a seedy little office in Melton Mowbray, wouldn't I be tempted by the parody option? You bet! But more than that, wouldn't the company actually benefit from this option? Isn't parody better than nothing? I mean, if you've got something to sayeven if it's only figures, statistics and stuffthen predictable, styleguide-compliant language with buzzwords might just about work. The Economist gets by just fine! But when you've got nothing to say...? I'm only half joking.
Better than the Van Gogh analogy, and better than the escape velocity analogy, is the recalcitrant horse analogy. Novice riders can find it difficult to get the horse to do what they want. And there are easy horses and difficult horses, just like there are easy and difficult translations. One way to tackle the problem is to get angry. Often, the horse will respond immediately by doing what you asked, as if to say "why didn't you say so in the first place?" But good riders hardly ever get angry. They don't need to. As soon as they even approach the horse, their energy level and confidence is sufficient to ensure that the horse will follow them trustingly through any difficulty, as it would a natural leader in the herd. I haven't got to that stage yet, but when riding a recalcitrant horse, I will replace the anger by a conscious increase in energy level, almost as if I were preparing to run a race. It works much better! And so it is with translation. Sometimes it really can help to get angry with the text. Sometimes, this will be sufficient to raise your energy level and kick your grammar generator into effective independent action. But I have a nagging suspicion that anger might distort things, producing all sorts of interference. Might that not be where the mischief and parody come from? Much better to work on the confidence and energy directly! One thing is certain: a hesitant, lethargic translator will not be a good translator.
The other parallel between translating and riding is that novice adult riders will try to rationalize what they are doing, as if they were learning to operate a machine, an airplane, perhaps. The habit of an algorithmic approach to learning is so extraordinarily difficult to shed that I would be tempted to explain it in terms of Pinker's brain-circuit redeployment rather than Krashen's affective factors. Even though I am acutely aware of the problem (algorithm within algorithm?), my riding still suffers seriously from it. Riding instructors use a number of tricks to fool students into riding with their bodies rather than their brains. Night-riding, for example, introduces sensory deprivation to develop the rider's sensitivity to how the horse responds to changes in body position. Then there are games, which divert the rider's conscious attention onto something else, thereby developing more natural and more effective interaction with the horse. The neatest trick I have seen so far is in the beginners' jumping class. The instructor gets the students to jump a low fence as best they can, encouraging them to stay as relaxed as possible. When they have jumped the fence enough times not to be frightened by it, the instructor places a second fence just after it. What happens is that the rider does not have enough time after the first fence to start rationalizing about how he's going to jump the second. So the first fence gets jumped as clumsily as usual, owing to brain interfering with body, but the second fence gets jumped smoothly, owing to the brain being switched off. The smooth second jump gives the student a precious initial experience of what a smooth jump should feel like, and the spectacular difference between the two jumps proves the importance of riding with the body rather than the brain.
Now I'm not suggesting that translation students should switch off their brains (though I would say they should be encouraged to use them differently). My point is simply that if riding instructors are imaginative enough to invent such brilliantly simple and effective techniques to focus the right kind of attention on the right tasks, why should translation teachers not do likewise? Some do, I know. One trick, for example, is to have the students read the source text once then translate it from memory. But just like a rider learns more from the horse than from the instructor, so a translator learns more from translating than from studying translation. In my opinion, effective translator training requires an overwhelming emphasis on writing skills, and writing skills cannot be taught in the lecture theatre. The only satisfactory way to train translators is through some kind of apprenticeship arrangement.
Many, myself included, consider language to be the most important human specificity, the defining feature of humankind, as it were. In The Language Instinct, Pinker likens the human's language capability to the elephant's trunk, which any child's drawing will tell us is the defining feature of this most endearing animal. Though one feature is decidedly more fleshy and tangible than the other, both are fabulously sophisticated pieces of equipment. Pinker takes no fewer than twenty-two lines to simply list some of the most extraordinary functions of the elephant's trunk. (It is, apparently, "lined with chemoreceptors that enable the elephant to smell a python hidden in the grass or food a mile away.")
Now the elephant owes its superiority over the human being to a total ignorance of economics. But let us imagine a two-fold cultural disaster among elephantkind. Not only do they get economics, but the ensuing rush of market forces brings compelling reasons why it is no longer cool to "pull up clumps of grass and tap them against their knees to knock off the dirt, to shake coconuts out of palm trees, and to powder their bodies with dust." The market has dropped right out of walking on deep riverbeds using the trunk as a snorkel. What matters most for the modern elephant is wings! Clearly, those overdesigned trunks will have to be reengineered, and there's no way we can wait for evolution to fix it.
In Part 2, I have attempted to set out the reasons behind the academic inertia that has held back progress in the translation industry. But however irritating it might be, academic inertia is far less dangerous than the backlash against it, caused by mounting impatience at academia's failure to keep pace with booming market demand for business, rather than literary, translation. Whatever its failings, academia is founded on reason, and is eventually and inevitably amenable to reason. From academia we can at least expect a slow but sure convergence upon truth. The backlash movement, on the other hand, is not in the least interested in truth. It is motivated solely by economic expediency, and instead of convergence upon truth we can only expect from it a divergence into chaos. Ludicrously, grotesquely, the business-oriented backlash movement within the translation industry seems intent on rapidly reengineering the human equivalent of the elephant's trunk. This will be the subject of Part 3.
In His great wisdom, God the supreme engineer configured the universe to ensure that any species unfortunate enough to open the Pandora's box of economics would also find therein the hope of science. The balance between the two is up to us.