• Front Page
   
  
• January ‘01 Issue
   
  
• October ‘00 Issue
   
  
• July ‘00 Issue
   
  
• April ‘00 Issue
   
  
• January ‘00 Issue
   
  
• October ‘99 Issue
   
  
• July ‘99 Issue
   
  
• April ‘99 Issue
   
  
• January ‘99 Issue
   
  
• October ‘98 Issue
   
  
• July ‘98 Issue
   
  
• April ‘98 Issue
   
  
• January ‘98 Issue
   
  
• October ‘97 Issue
   
  
• July ‘97 Issue
   
  

Sounding the Language-Elephant’s Trumpet

(a guide for intelligent buyers of translation services)

Part 1

  used to be an engineer, and part of my work involved testing the materials and components that my company would buy for making its products. I worked with standards and specifications, procedures and protocols. One day my boss asked me to produce a report in Spanish for one of our contracts. Unthinkingly, I opened the phone book and called the translation agency whose name most appealed to me. That was it! I was unaware that product quality across the translation market ranged from something like last Christmas's plasticine up to space-grade alloy. And though I prided myself on an enthusiastic interest in foreign literature (somewhat exceptional among my engineer colleagues), I had not the slightest idea of what translation involved. With the modern emphasis on procedures and processes, a young engineer in the same situation today might well attempt to apply a rigorous selection procedure when buying in translation services. But curiously, the result will be identically hazardous: one type of ignorance is being displaced by another, and the net result is no change at all. The first type of ignorance arises out of the inertia of tradition and slow-moving academic practice. The second, a result of chaotic growth, is often deliberate and much more pernicious. In the second part of this article I shall be putting both under the microscope in an attempt to contribute to the development of a cure.

I realize that my article may well end up being read more by translators and translation students than by users of translation services, but I insist that translation users and translation buyers are my real ultimate audience, since the change I believe to be so necessary in the translation industry can only be effected under the pressure of an upward shift in user expectations. At some point in the future I hope to build better bridges capable of reaching this intended audience more effectively. For the time being, I should simply reassure non-linguist readers that the technical content of this article is no more difficult than the briefest introductory course on, say, metallurgy or electronic components for the staff of a company's purchasing department. That being said, one of my main tenets is that you cannot buy translation as if you were buying metal strip or integrated circuits. One of the things you will be learning from this article is precisely why this is so.

A fascination with language eventually led me away from engineering, and though my mind quivered eagerly with excitement at the prospect of working outside the neatly charted territories of applied science, I soon discovered that there was a price to pay for this thrill. My first surprise was that whereas knowledge had been everything in my previous world of science and engineering, it seemed to be generally mistrusted—despised, almost—in my new world of translation. My second surprise was to find that this mistrust was if not justified, then certainly understandable, since what passed for knowledge was usually little more than hearsay and vogue, myth and superstition. Much sales talk of dubious veracity filled the vacuum of genuine understanding: the translation industry is very much the salesman's paradise, because he is selling a product that the customer is, almost by definition, largely unable to understand. So in many ways, my career transition was like a leap back several centuries in time. My sparkling laboratory had become an alchemist's cave, and the inquisition was omnipresent. Exhilarating, well yes. But disconcerting too.

Twenty years' close working familiarity with the translation industry has done nothing to lessen my initial impression that it is in a mess. A terrible mess. My immediate circle of family, friends and acquaintances includes several engineers, a blacksmith, several teachers, a vet, a metallurgist, two biochemists, a naturalist, a builder, a doctor, a riding instructor, a plumber and an aircraft mechanic. Though several express occasional discontent with their employment situations, none goes as far as to condemn his profession as a whole for messiness. And all consider it a little odd that I should so condemn mine. I am seen to criticize the translation business a little like I might criticize the culinary skills of my former compatriots, with the mild and jovial treachery of a willing convert to a superior world. But whereas English cooking has allegedly improved over the last two decades, I see not the slightest sign of improvement in the translation business. On the contrary, I think it is getting worse!

One of the engineer friends did come close to understanding my exasperation after an unfortunate experience with an incompetent translator. I fixed the problem for him, and he put the incident down to bad luck, laughing when I told him that bad luck in purchasing translation services can come in very long runs. If you phone five plumbers selected at random from the telephone directory, you will find at least four who are capable of repairing your leaking pipe. And at least four out of five metallurgy graduates will correctly differentiate between mild and stainless steels. But as a young project manager with a large French translation agency, I found that things were very different in the translation business. For a huge aerospace project, I tested dozens and dozens of candidates, from many different backgrounds. Alarmingly, I found that only a tiny percentage of candidates actually knew how to translate, and that competence in translation seemed to correlate with nothing at all: graduates from the big translation schools in Paris and Geneva translated no better than anyone else, and expensive freelances translated no better than cheap ones. Whether you select a translator from the phone book, the membership directory of a translators' association or the graduation rolls of a leading translation school, your only certainty is that he or she will speak two or more languages. Beyond that, you are in the dark: totally!

I would like to think of this article as a candle.

In this first part I shall be explaining some very basic linguistic concepts, and in the second part I shall be examining in detail some of the specific misunderstandings that combine to perpetuate and exacerbate a climate of quite astonishing confusion.

The basics

So why is the translation business in such a mess? There are many reasons, most of which arise out of fundamental misunderstandings as to the nature and working of language, and the remainder out of deliberate attempts to exploit these misunderstandings for reasons of commercial interest. Here, the astute potential buyer of translation services will step in to eagerly point out that if the mess is caused by a defective understanding of how language works, all he needs to do is hire a linguist. Alas, I will have to disappoint him: as I shall be explaining later, there is no necessary connection between theoretical understanding of linguistics and practical competence in translation, and the reason for this is the same as the reason why no amount of Shakespearean studies will enable a student to write anything remotely like Shakespeare.

I've already hinted at the first misunderstanding, which is that knowledge of two languages is sufficient for translation between them. But in fact this misunderstanding is by far the least dangerous. Indeed, if interpreted correctly, the statement actually crystallizes nothing less than the whole truth of the matter: knowledge of two languages will be sufficient for translation between them, provided we are very careful with the terms "knowledge" (which we must break down into receptive knowledge of the source language and productive knowledge of the target language) and "language" (which we must break down into language, dialect and register).

Receptive knowledge

Receptive knowledge means understanding what the author has written. This sounds obvious, but very many translators regularly translate texts they do not understand. They understand the language and think this is sufficient: it isn't. If the translator doesn't understand the actual ideas that are being expressed, he stands little chance of being able to convey them to his readers. Quite aside from problems of terminology (which are often overstated, for commercial reasons), translators who do not understand the content of a text are forced to resort to simple mechanical conversions between syntaxes and lexicons, and this is a guarantee of bad translation, as we shall be seeing in detail later. Incredible as it might seem, the requirement on understanding is very widely ignored, even in the highest places and at the highest salary levels. I once worked as a translator with a technical agency of the United Nations, and was astonished to find that the staff (mostly Oxford classics graduates) were not only totally ignorant of the technical aspects of the agency's work, but actually appeared to take pride in this ignorance: "We are translators, and it's not our business to know about wires and circuits and things." I remember translating "la tension sur les bornes de la résistance" as "the voltage across the resistor," an expression that is so standard and so commonplace in electrical engineering that even a novice would be bound to pick it up from the most casual skimming through the relevant literature. But for the head of the English translation department at this UN organization, it was "far too risky." I felt like a junior astronomer being rebuked by his learned society for daring to suggest that the earth revolved about the sun. This was fifteen years ago, and friends in Geneva tell me things have changed. But a brief visit to the websites of the big international organizations tells me they have not.

Then in linguistics, we speak of denotational meaning, which refers to the face-value of words, and connotational meaning, which lies deeper. This is not just a fanciful invention of linguists: in everyday language, we speak of reading between the lines, though few but the trained linguist will consciously realize just how much of the meaning is on the surface and how much below. In fact, linguists identify over a dozen different types of meaning, but the denotation/connotation split is the most obvious and important. For an illustration of connotational meaning, consider the expression "quivered with excitement" that I slipped in earlier. This is a rather striking metaphor, but its connotations go beyond the metaphorical effect itself. In this article, the expression stands out starkly against its surrounding material, and a translator would have to accurately identify why the author might have wished to produce this effect. A good translator, with a thorough knowledge of English, will first have to realize that the expression is not only a metaphor but a cliché that might have been drawn from a teenage boy-meets-girl magazine. He will then have to determine what such a cliché is doing in a rather dry text on basic linguistics. It might, for example, belong to the author's personal idiolect (some people really do speak in clichés), but in that case there should be similar instances scattered throughout the text. It might have been an attempt at humour. It might just be fortuitous (the first thing that came into the writer's mind). It might have been inserted to guard against reader somnolence. It might have been inserted with a view to serving as an example later in the text. And there might well be a mixture of reasons. The translation (provided it is a good translation) will differ depending on the reason. This analysis sounds very complex, especially when you realize that hundreds of similar analyses will be required in even the shortest text. But it is essential to understand that no translator will attempt to perform this kind of analysis consciously, except, perhaps, when writing about the complexities of translation. The analysis is performed instinctively, in real-time, using our usual language-processing mechanism. I will be making much of this in the second part of the article.

I can hear the buyer of technical translations for industry protesting that his texts are all face-value only, for precision, for unambiguity, for repeatability. No poetic iceberg stuff for him, mate! And yet I will have to disabuse him too: however unpoetical the machine manual, the submerged, stylistic, connotational portion can determine whether the technician follows the proper procedure or decides to brave it alone, with risk to life, limb and plant maintenance budget. And however dry and dusty the contract specifications, the submerged portion can determine whether a company is perceived as a trustworthy partner or a shoddy nightflyer.

Productive knowledge

Productive knowledge means being able to write. This is the number-one, crucial, overriding, essential, critical factor. It cannot be overemphasized. If you can't write, you can't translate. It is impossible. There is no workaround. And the formula can be expanded at will:

If you can't write well, you can't translate well.

If you can't write technical documentation, you can't translate technical documentation.

If you can't write advertising copy, you can't translate advertising copy.

Why is it the critical factor? Because whereas it is fairly easy to acquire full receptive knowledge, it is extremely difficult to acquire full productive knowledge. But my conscientious translation buyer over there has been studying a little linguistics, so he knows that, excepting serious brain damage, everyone does acquire full receptive and productive language capabilities as a child, and that this acquisition appears to be automatic and effortless. Right! But receptive capability can refer to either listening (to spoken language) or reading (written language); and productive capability can refer to either speaking or writing. What every child acquires effortlessly is a capability to speak and to understand speech. Writing came much later than speech. It is the product of fast human cultural development whereas speech is the product of slow biological evolution. And though writing is obviously based on speech, it does have a number of very special characteristics. The most obvious is permanence, but the most important is the loss of reciprocity.

I am fond of saying that language offers humankind the power of "unlimited reciprocal elucidation," and need little encouragement to diverge into colourful speculation on how this capacity must take most of the credit for the development of human civilization. Fashion-followers would doubtless prefer "interactive" to my "reciprocal," by which I simply mean that when one person says something, the other can expand upon it, ask for explanations, latch onto to a particular point, turn the conversation in a different direction, or do any number of things that will determine the subsequent course of the verbal exchange. By "unlimited," I mean that this language-assisted exploration of the world has literally infinite scope: there is no boundary on the conversational web, and this, I feel, is the most important implication of Chomsky's famous observation that there is an infinite number of possible utterances. If you doubt the existence or the importance of the reciprocity characteristic, I suggest you try a little experiment: the next conversation you have, make a point of staying silent. Very soon, the other person will stop speaking, and probably ask if you are feeling allright. The effect will be even more noticeable over the phone, since you will also be filtering out non-verbal cues. You can resume normal conversational service by just interjecting the occasional "uh," "right" or "really"; even if you don't actually say anything, you are providing your part of the interactivity without which speech cannot really exist. (In fact, you are saying something: you're saying "I understand, please go ahead," which ties in with what we saw about reading between the lines.) So what about speeches, which are by definition spoken but obviously lack reciprocity? Well, I think we must say that speeches have more in common with writing than with speech, and, indeed, that writing originated as a way of lending permanence to a very special form of speaking: special because it required considerable preparation and memorization, whereas normal speech is necessarily improvized, by virtue of the fact that you never know what the other person's going to say until he's said it.

If you listen carefully to anyone speaking, you'll realize that speech is packed full of reciprocity cues, inviting the other person to contribute, or at least confirm that he's following. The key characteristic of writing as opposed to speech is that it must get by without any such confirmations. This makes a number of heavy demands on the writer, and one of those demands can be even heavier on the translator than on the native writer.

• The writer must know his readers well enough to know what they are likely to follow and what they are not. This is usually fairly easy, because the writer shares a common basic psychology and a common cultural background with his readers. However, it gets more difficult if the writer does not share the same cultural background. Whenever we cross into another cultural community, we find a change in language. Sometimes, the language is completely different, as when we cross the Pyrenees, but sometimes the change is more subtle, as when we walk out of the laboratory into the sales department. As I shall be explaining later, it is crucially important to understand that the difference between these two transitions is one of degree only. Both transitions (into Spain and into the sales department) are of precisely the same nature. Translators are by definition writing for an audience across a cultural divide. But in addition, they are often asked to write for audiences whose specialized knowledge puts them in a cultural community that is slightly removed from the mainstream population and speaks a slightly different language: it is this that places a heavier demand on the translator. The difficulty is often overstated for commercial reasons, but it is nonetheless real.
• Since there is no reciprocity, the writer must elucidate alone. Because the reader will not be able to interrupt the flow to ask questions explicitly, the writer must anticipate them and answer them implicitly. Again, few writers (even among the very best) will consciously realize they are doing this. Again, conscious awareness of linguistic mechanisms has little to do with actual practical writing skill, and this apparent incongruity is the source of much industry confusion, as we shall be seeing later.
• Whereas a spoken exchange will meander naturally in response to each successive utterance of each participant in the conversation, the writer must chart his course carefully. This does not mean that he must plan his document rigidly beforehand; indeed, I am convinced that the best writers leave themselves the greatest organizational flexibility. What it does mean is that the writer must lead the reader carefully through each of the stopping points in the plot. Again, this has a lot to do with anticipation. And yet again, it works almost entirely at a subconscious level.

There's just one little loose end I have to tie up. I started this section with the comment that reading was easy but writing difficult. Everybody will realize the truth of this without my having to give detailed explanations as to why it should be. But I think is it important to emphasize that reading, unlike writing, is a wholly internal process that is wholly hidden from scrutiny. The only possible way to evaluate a piece of writing is by reading it, yet reader response itself eludes accurate analysis. And, of course, there can be little useful response to a text written in a language that the reader does not understand. This is why the many misunderstandings that blight the translation industry have been able to prosper unchecked.

Language, dialect and register

My claim was that knowledge of two languages would be sufficient for translation between them, provided we were careful with the terms "knowledge" and "language." I've dealt with the two aspects of "knowledge," so let's now take a brief look at "language." We can consider language as an immensely complex open-ended coding system that enables humans to engage in unlimited reciprocal elucidation. It appears specific to humans, because whereas other animals can send and receive quite complex signals to communicate information on aspects of their environment, none appears able to limitlessly pursue elucidation on the content of these signals.

Working from macro to micro scale, we can start by saying that all the inhabitants of the world have a language capability, and that this capability can take any instantiation: a child will speak Chinese if brought up in Beijing but Spanish if brought up in Madrid. These different instantiations are usually known simply as "languages," and most of them are mutually unintelligible. Examining each broad language community (often delimited by national frontiers), we in fact find that it is made up of many smaller communities speaking different but mutually intelligible languages, which are referred to as "dialects." In common parlance, the historically dominant dialect is usually granted "language" status, while the others remain known as mere dialects. And often, the speakers of a humble regional dialect will be regarded with less esteem than the speakers of the national-language dialect, a phenomenon that owes more to primitive tribal instinct than to logic, sense or decency. For a very long time, linguists have known that there is no qualitative difference between a dialect and a language: a very old joke has it that a language is a dialect with an army and navy. Then if we move in closer still, we observe that there are communities within communities, dialects within dialects. At the strongest possible magnification, we find that dialects are composed of idiolects, an idiolect being the dialect of a single person. (If you think this sounds fanciful, consider it as just another way of saying that no two people speak in exactly the same way: an idiolect is like a fingerprint.) We also observe that people sometimes appear to switch between dialects, but here we must distinguish between two cases:

• A person who was brought up in Birmingham but works in London might speak with a Black Country accent when visiting family but suppress the features of this accent when at work, possibly because of the low prestige traditionally attached to regional accents. Such a person is bilingual in a very similar way to a person who speaks French and Spanish. The only difference is that French and Spanish are mutually unintelligible, whereas Black Country and standard English are (mostly) mutually intelligible.
• Salespeople and engineers might speak different dialects while they are in different departments at work, but the same dialect when they are together at the pub. Linguists use the term "register" rather than "dialect" when referring to the way the same individual will speak differently in different situations within the same speech community. The principle is very similar, and the border between dialect and register is very fuzzy in places. The main difference is that a speaker will shift registers freely and intentionally within his dialect, which is set fairly firmly, chiefly by geographical factors, though social and other factors may also be involved. However, if a register is used consistently and intensively among a community, it can diverge and coalesce into a fresh dialect in its own right. This is the origin of the dialects (also known as jargons) used by professional communities like engineers and lawyers.

As I have repeatedly hinted, there is no difference between a dialect and a language. So while everybody knows that a translator translates from one language to another, we could say more precisely that he translates from one dialect to another. If this is not generally realized, one reason is possibly that the term "dialect" is rarely used with reference to written material. But whether or not it is generally realized, it has some monumental implications for the translator:

• A translation buyer will obviously specify the target language explicitly, but it is usually up to the translator to determine the dialect of the target language. This is often self-evident, but the translator might need to check exactly who he his writing for: it makes a difference whether he is writing for a machine operator or the company president. Things get more complicated when the source document itself represents communication across dialect barriers within the source language, especially since dialect breakdowns can differ widely from one language to the next. Imagine, for example, an engineer writing for the sales department, a scientist writing for the general public, or a defendant from a minority community addressing the judge. Here, the difference between register and dialect comes into play, and we will recall that this difference hinges on intentionality. Specifically, the translator must determine the extent to which the author is merely attempting to conform to the usual dialect of his source or target speech communities and the extent to which he is intentionally pitching his writing style (i.e. register and combinations of registers) with respect to the dialect of his audience. This sounds complicated until we realize that in the immense majority of business translation situations the author is simply writing to be understood. (Incidentally, it is precisely this balance of intentionality that differentiates literary from business translation.) Very occasionally, the text might not be sufficient in itself to reveal the author's precise pitching intentions. This can happen if the author is particularly unskilled, but it can also arise simply because of the differences in dialect breakdown within different languages. In this kind of situation, the translator will need to ask the author how the text should be pitched. If the author is not available, the safest bet, again, is to assume he wishes to be fully understood by his audience. Yes, it sounds so obvious, and yet...

• The target dialect has its own lexical and syntactical rules. These are usually subsumed by the grammar of the standard language variety, but not always. Some technical dialects, for example, can considerably stretch the rules of the standard language, as can the attention-grabbing dialects used in advertising copy. And the relationship between dialect and standard variety may differ considerably from source to target language. For example, technical English can differ more from standard English than technical French from standard French. This leads many English speakers to believe that French technical writing is inherently flowery or that French is generally ill-suited to the expression of technical concepts. I have heard these assertions made gravely by established translators, but both are the purest nonsense. It is simply that French and English are different: very obviously different in syntax and lexicon, and more subtly different in the permissible distribution of registers. Another glaring example of this is the fact that written French diverges much more from spoken French than written from spoken English.
• The only valid way to judge a translation will be by reference to the target dialect. This means that a translated software manual must read like a native software manual and a translated report like a native report. The translator's job is to understand what is meant (more precisely, what is intended) in the source dialect, then wholly convey this meaning in the target dialect. If he is unfamiliar with the target dialect, then this disqualifies him from doing the job just as surely as if he were unfamiliar with the target language in general. His commission is to produce a document that belongs fully and properly in the target dialect, regardless of where the source dialect is situated within the dialect bundle that forms the source language as a whole. And we can take this reasoning a step further, noting that whereas all spoken dialects are equal, we cannot say the same of written dialects. There is no such thing as a defective spoken dialect, but there is such a thing as a defective written dialect (or written idiolect, to be more precise, owing to the lack of reciprocity). This is simply because, as we have seen, speaking is easy, and mastered effortlessly by every normal human, whereas writing is rather difficult, and mastered by few. Translators are very frequently called upon to translate badly written texts. This is a fact of life. Unfortunately, many incompetent translators will attempt to blame poor output on poor input, but there is not the slightest linguistic justification for this, since the translator is plainly being asked to translate into a coherent rather than a defective dialect. Think of a translator offering translations into English from Spanish, French or Italian: the output language is the same regardless of the input language. This structure is exactly equivalent to a translator offering translations into clear English from clear French, muddy French or downright illiterate French: the output dialect is the same regardless of the input dialect. As far as I know, no translators overtly offer translations into muddy dialects of English or any other language. Yet this is what translation buyers actually get, around 80% of the time!

• The term "best practice" has been violated into near insignificance, much like its sister "quality," but if there can be any residual meaning to it in a translation context today, it is that a translated document must be indistinguishable from a first-class example of a native document. The translation must be capable of standing on its own merits as a good piece of writing. No allowance can be made for the fact that it is a translation: the translation process must be absolutely transparent. Decades of massively bad translation practice, with massive official backing, have greatly reduced readers' expectations of translated documents. This must stop: readers must realize that if a translated text reads badly, it can only be because the translator has not done his job properly. If a source document reads badly, this is very probably because the author's written idiolect is defective; there's not the slightest stigma to this, because the author is rarely a professional writer, but an engineer, a financial analyst, a designer, whatever. But the translator is by definition a professional writer, since to translate at all he needs full productive knowledge of the target dialect. His written idiolect is therefore not defective, and the target dialect as a whole is certainly not defective, otherwise it could not have developed as a viable channel of communication in the first place. So there is no legitimate reason why a translated document should read badly, excepting negligibly rare instances in which an author intends it to read badly.