Is That a Fish in Your Ear?

TEN The Myth of Literal Translation

With bilingual dictionaries to get them started and Roget’s Thesaurus to help them polish their work to a nice finish, translators ought not to find it too hard to tell us what the words on the page really mean. In practice, however, it’s the words on the page that hang like a dark veil over what a piece of written language means. Words taken one by one obscure the force and meaning of a text, which is why a word-for-word translation is almost never a good job. This isn’t a new insight: arguments against literal translation go back almost as far as written translation itself.[51]

After immersing himself for several years in the history of translation, George Steiner discovered that it consisted very largely of repeated arguments over this same point. “Over some two thousand years of argument and precept,” he wrote with perceptible frustration, “the beliefs and disagreements voiced about the nature of translation have been almost the same.”[52]

When Don Quixote’s favorite bedtime book, Amadis de Gaula, appeared in French, for example, the translator gave his patron two reasons for not having stuck to the literal meanings of the Spanish words:

These twin justifications for “free” translation—literal translation just isn’t appropriate for the target audience and isn’t suited to the original, either—were familiar themes in the sixteenth century, as they had been for many centuries already. In fact, few commentators on translation have ever come out in favor of a literal or word-for-word style. Literal translation is precisely what translators in the broad Western tradition don’t do. But if literal translation is not a widespread practice, why do so many translators feel a need to shoot it down—often with overwhelming force? Octavio Paz, the Mexican poet and man of letters, stated the standard view in more recent times: No digo que la traducción literal sea imposible, sino que no es una traducción: “I’m not saying a literal translation is impossible, only that it’s not a translation.”[54]

How back far does it go? There are references to the issue in the writings of Cicero (106–43 B.C.E.) and Horace (65–8B.C.E.), but a long sentence written by Saint Jerome, the first translator of the Bible into Latin and subsequently the patron saint of translators, can be taken as the first full formulation of the lopsided dispute between “literal” and “free.” In 346 C.E., when he was near the end of his labors, Jerome wrote a letter to his friend Pammachius to counter the criticisms that had been made of the translations he had done so far. Jerome said this about how he had gone about his task:

A provisional translation would give the following sense: “Thus I not only confess but of my own free voice proclaim that apart from translations of sacred scriptures from the Greek, where even the order of the words is a mysterium, I express not the word for the word but the sense for the sense.”

Jerome’s expression verbum e verbo, “the word … for the word,” can be considered synonymous with “literal” translation, and his sensum exprimere de sensu, “to express the sense for the sense,” corresponds to the idea of “free” translation. He proclaims that he doesn’t do “literal” except when translating “sacred scriptures from the Greek.” That seems clear until you realize that the exception clause drives a cart and horses through the main claim, because what Jerome did throughout his long life was to translate sacred scripture, more than half of which he translated from Greek.

Jerome also says he abandons sense-for-sense translation not just when translating scripture from Greek but specifically ubi et verborum ordo mysterium est, in those places “where even the order of the words is a mysterium.” As the meaning of the word mysterium is uncertain, there’s no final agreement as to what Jerome was really talking about. At the root of Western arguments about how best to translate lies a mystery word that nobody is quite sure how to translate.

In late Latin written by Christians, mysterium most often means a holy sacrament. Jerome’s sentence therefore seems to recommend sticking to the exact order of the words of the Greek New Testament because its word order is sacred. Louis Kelly understands Jerome to be saying:

This reading supports the view that Jerome is not really defending “sense for sense” translation, as he first seems to be doing, but “word for word.” But why would Jerome treat Greek word order as sacrosanct and not do the same for the scriptures he translated from Hebrew and Aramaic? The “Greek exception” doesn’t make a lot of sense if holiness is the dominant reason for mimicking the word order of the source.

However, Jerome may have meant something else by mysterium. He may have wanted to explain his approach to an issue that confronts every translator at some point: what to do with expressions that you don’t understand. It’s a real problem for all translators, because every utterance ever made in speech or writing has something blank or fuzzy or uncertain about it.

In ordinary speaking, listening, and reading, we cope with the gaps in various ways. An impenetrable phrase may be treated as a transmission error—a mispronunciation, a typo, a scribal glitch. We have no trouble replacing it with what we instantly guess to be the true form, and in spoken interaction we do this automatically, without noticing the corrections that we bring to what we hear. When reading, we use the context to prompt a meaning that fits. Where the context isn’t good enough to allow this, we just skip it. We skip-read all the time! Nobody knows the meanings of all the French words in Les Misérables, but that’s never stopped anyone from enjoying Hugo’s novel. However, translators are not granted the right to skip. That’s a serious constraint. It hardly arises in most kinds of language use; it’s one of the few things that sets a problem for translation that is almost unique to it.

Jerome was working with many different sources, but his main text for the Old Testament was the Greek Septuagint, translated from now lost Hebrew sources several centuries earlier. According to legend, it had been commissioned around 236 B.C.E. by Ptolemy II, the Greek-speaking ruler of Egypt, for his new library in Alexandria. He had sent men to Judea to round up learned Jews who understood the source text, then wined and dined them and set them up at Paphos (on the island of Cyprus) to get down to work. There were seventy (or seventy-two) participants in this foundational translation workshop, which is why the text they produced is called the Septuagint—a way of writing (not translating) the Greek word meaning “seventy.”

The Seventy wrote not in the language of Homer and Sophocles but in koiné, the popular spoken language of the Hellenistic cultures dotted around the Middle East. They also wrote it in a peculiar way, perhaps because koiné was their vehicular language and not completely native to them. So it would hardly be surprising if some words, phrases, and sentences in it baffled Saint Jerome seven centuries later. One telltale sign of the Seventy’s difficulty with Greek is the way they handled Hebrew words referring to Jewish religious mysteries. For example, they represented the Hebrew as Xερoυβíµ, which is not a translation, but just the same word sounded out in a different alphabet. Jerome followed style—he wrote out approximately the same sounds in Latin script, making cherubim. English Bible translators have done the same, giving us a Hebrew masculine plural form (-im) for a concept that has stumped all translators since the third century B.C.E. In addition, the transfer of letters through three scripts and four languages has altered the sound of the word almost beyond recognition, from “kheruvím” to “cherubim.”

This way of dealing with an untranslatable by not translating it while making it pronounceable (sound translation, homophonic translation: see here) could be considered the primary, original meaning of the term literal translation. It represents a foreign word by putting in place of the letters of which it is made the corresponding letters of the script of the target language. But we do not call that literal translation nowadays—we call it transliteration. And it probably wasn’t what Jerome had in mind in the famous passage from his letter to Pammachius.

What, then, did Jerome mean by mysterium? Here’s an alternative translation of the mystery passage by a canon of Canterbury Cathedral:

To put it in a slacker style, “I translate word for word only where the original—even its word order—is completely impenetrable to me.” That is, of course, what translators have always done. For the most part, they transmit the sense; where the sense is obscure, the best they can do—because unlike ordinary readers they are not allowed to skip—is to offer a representation of the separate words of the original. This may even explain the style of the translation of the extract quoted here. Maybe Derrida’s translator, far from trying to sound foreign, was simply baffled.

What, then, is a literal translation? Not a substitution of letters, since we call that transliteration. A one-for-one substitution of the separated written words? Maybe. When confronted with a decidedly loose French translation of “The Jumping Frog of Calaveras County,” Mark Twain decided to back-translate his story into English using a single-word substitution device intended as the opposite of his French translator’s overuse of rephrasing.

This schoolboy prank mocks French, French grammar, the school teaching of French, and so forth. But the main thing it demonstrates is Octavio Paz’s point: “literal translation” is not impossible, but it is not a translation. You can only understand the target text if you can do a reverse substitution of the words of the source and read the French through its representation in English. In other words, to make any sense of “The Frog Jumping” you have to know French, whereas the whole purpose of translation of any kind is to make the source available to those readers of the target who do not know the source language. A translation that makes no sense without recourse to the original is not a translation. This axiom incidentally explains why the meaning of cherub will forever remain a speculation.

The term literal also hides other mysteries. It is used to refer not only to a translation style that barely exists but to say something about the way an expression is supposed to be understood.

The distinction between the literal and figurative meanings of words has been at the heart of Western education for more than two millennia. The literal meaning of an expression is supposed to be its meaning prior to any act of interpretation, its natural, given, standard, shared, neutral, plain meaning.

However, when we say, “It was literally raining cats and dogs last night,” we mean the adverb literally in a figurative sense. Studies of large corpora of recorded speech have shown that the majority of the uses of literal and literally in English are figurative; similar results would no doubt be extracted from written texts in all European languages.[57] This is a curious irony, because expressions that mean one thing and its opposite were a thorn in the flesh of precisely those Greek thinkers who invented the distinction between literal and figurative in the first place. But language is like putty. The figurative use of literal is one among a thousand cases of expressions meaning this and its opposite, depending on what you use them to mean.

Literal is an adjective formed from the noun littera, meaning “letter” in Latin. A letter in this sense is a written sign that belongs to a set of signs, some subsets of which can be used to communicate meanings. Speech communicates meaning, writing communicates meaning—but letters on their own do not have any meaning. That’s what a letter is—a sign that is meaningless except when used as part of a string. The expression “literal meaning,” taken literally, is a contradiction in terms, an oxymoron, and a nonsense.

What we probably meant in the distant past when we asserted that something was “literally true” in order to emphasize that it was really true, true to a higher degree than just being true, was that it was among those rare things that were worthy of being “put into letters,” of being written down. All the uses of literal with respect to meaning and translation implicitly value writtenness more highly than oral speech. They are now among the surviving linguistic traces of the fantastic change in social and cultural hierarchies that the invention of writing brought about. They carry the shadow of the early stages of literacy in the Mediterranean basin between the third and first millennia B.C.E., when alphabetic scripts first arose together with the texts that through translation and retranslation have shaped and fed Western civilization ever since. This is presumably why the same words and the same terms still persist in debates about how best to translate.

Yet even in the modern era we do not always know quite what we mean when we claim that something is literally true, and even less when we call a translation a literal one.

Toward the end of the nineteenth century, a Franco-Egyptian mountebank with a medical degree and a talent for social climbing and free composition in French published a new version of The Arabian Nights. It was a commercial and cultural success, feeding a wave of interest in the Sexy Orient among the elite, and it impressed many writers of the day, including Marcel Proust. The translator, Joseph-Charles Mardrus, knew Arabic, and he used some Arabic texts as the basis of his rewriting of the collection of ancient Eastern tales, which he titled, in a daring Arabism in French, Les Mille Nuits et Une Nuit, with a subtitle as clear as can be: Traduction littérale et complète du texte arabe, “The Thousand Nights and One Night: A Complete and Literal Translation of the Arabic Text.”[58]

The subtitle is less a description than an assertion of status. Calling the work “complete” is obviously intended to give it a higher value than previous versions—but why should “literal” have seemed to Mardrus an effective way of enhancing the status of his work?

It wasn’t a slip: Mardrus’s preface emphasizes and magnifies the meaning of his subtitle:

Mardrus was not a conventional translation theorist, and scholars of Middle Eastern languages claimed that he was not a translator, either. A professor of Arabic at the Sorbonne demonstrated that there were no textual sources for many passages and stories in Mardrus’s entertaining and readable compilation. But Mardrus was a personage on the Parisian cultural scene and would not suffer such slings and arrows without returning fire. Friends came to his defense: André Gide argued that despite the demonstrations of Professor Gaudefroy-Demombynes, Mardrus’s work was “more authentic than the original.”[60] The translator’s own riposte built on Gide’s extraordinary claim. Academic critics learned Arabic in the classroom, not from living in the Middle East.

Mardrus’s translation was therefore the “literal” version of an essentially oral source. His written word in French stands for the spoken word of Arabic culture. If academic critics insist on having a textual source for the authentic Arabian Nights, which he wrote, well, no problem: “One day, in order to please M. Demombynes, I want to settle once and for all the Arabic text of The Arabian Nights by translating my French translation into Arabic.”

What stands out from this literary squabble is that the idea of what a literal translation consists of is culturally conditioned to a high degree. Mardrus wanted to say that his work was authentic, that it gave the true voice of the Arabic culture that he rightly or wrongly regarded as his special native privilege to possess. His solution to the argument—to manufacture a source to give textual scholars the evidence they demanded—may appear quite nutty, but it is not illogical from Mardrus’s point of view.

What all other Western commentators mean by “literal translation,” on the other hand, is unrelated to authenticity, truthfulness, or plainness of expression. It really refers only to the written form of words, and even more particularly to the representation of words in an alphabetic script. When that technology for the preservation of thought was still relatively new, and for those many centuries when it was not widely shared and was used for a restricted range of needs and pursuits (law, religion, philosophy, mathematics, astronomy, and, occasionally, the entertainment of the elite), it made sense to attach high prestige to the writtenness of written texts.

But in a world of near-universal literacy, that’s to say, for the last two or three generations, where alphabetic script is used for entirely ordinary tasks (to label packaged food; to advertise underwear; and to write blogs, horror comics, and pulp fiction), the fact that something is worthy of being written down in letters gives it no added value at all. “Literal” isn’t “Word Magic” anymore, it’s just a hangover from the past. The terms of debate about translation and meaning need to be updated, and the long-lasting scrap between literal and free should now be laid to rest.

However, there is one important area where the transposition of meanings at the level of individual words is a valuable, inescapable tool: in school and, more particularly, in foreign-language lessons.

There are many different ways of teaching languages. The Ottomans rounded up youngsters in conquered lands and brought them back as slaves to be trained as dil oğlan, or “language boys,” in Istanbul. Modern direct methods are gentler but rely on the same understanding of how languages are best learned—through total immersion in a bain linguistique, a kind of baptism of the brain.

Throughout the period of learned Latin in Western Europe, immersion was not an option. There was no environment in which everybody spoke Latin as a native tongue, and so the language had to be taught by teachers, in classrooms, through writing. Reprising Roman methods in the teaching of Greek, the European language-teaching tradition was heavily skewed toward the use of translation as the means of imparting written skills in the foreign tongue, and also as a means of assessing students’ progress toward that aim. The teaching of modern European languages in schools and universities got off the ground toward the end of the nineteenth century and borrowed its methods from the translation-based traditions in the teaching of Latin and Greek. It is generally reckoned to have been a disaster. However, if the aim of learning Latin (or French, or German) is to be able to read texts in that language fluently and also perhaps to be able to compose and thus to correspond with other users of Latin or French or German (whose native languages may be quite varied), then translation and composition skills are quite appropriate educational aims.

Translation-based language teaching is no longer in fashion, but its ghost still inhabits a number of misconceptions about what translation is or should be.

Teaching a foreign language when an actual linguistic environment is not available and in the absence of technologies that allow a linguistic environment to be simulated (television, radio, film, sound recording, and the Web) was obliged to rely on writing—on slates or on chalkboards, in exercise books or in print. With only those tools available it’s not obvious how to explain that the expression у меня большой дом is to be understood as “I have a big house” unless you also explain that it can be broken down into “At me big house” and use this item-by-item representation of the foreign in English disguise to introduce basic grammatical features—for instance, the fact that Russian doesn’t have a definite or indefinite article; that adjectives agree in number, gender, and case with the nouns they qualify; that there is no place for the verb to be in a Russian expression of this kind; and that possession may be expressed by a preposition before a personal pronoun, which has to be put into the appropriate grammatical case. Indeed, the grammar explanation I’ve just given is almost meaningless until you have seen it in action in a written expression and been told what each written item stands for.

Some people call this “literal translation,” but it would be better to adopt a distinct term for the parallel, item-by-item explication of an expression in a foreign language for the purpose of teaching how the foreign language works. “Wording” is invaluable, and I don’t think even the most direct of direct methods can do without it at some point. In fact, language learners taught by other methods always reinvent wording for themselves when grappling with a sentence just beyond the level of competence they have reached.

“Wording” gives you a first approach to the shape and order of the language you are learning. It helps not so much to translate as to produce acceptable expressions in the foreign tongue. To translate into the foreign language, you learn first of all to put the source into foreign dress. You learn that “My father has a big car” must first be translated into “At father big car” before you can even start to slot in the Russian expressions that will add up to the sentence with the stated meaning.

Wording is neither a language nor a translation, just an uncommonly helpful intermediate stage in learning how to read and write in a foreign tongue. School translation into L2 also gives the instructor a means of checking whether students have grasped and remembered the shape and order of the language. It’s not a test of an abstract grammar point but of grammar in a context of use. That’s how I learned languages at school. Given good teachers and keen students, it works.

But often it does not. Worse still, it often leaves ex–school students who failed the test pieces with a horror of doing translations, and sometimes a lingering resentment of those who can.

Since the expansion of education in the nineteenth century down to the present day, facing-page printed translations of standard works in foreign languages have helped countless students improve their grasp of grammar and vocabulary and allowed them also to read foreign works at greater speed and thus to understand them more completely. Some facing-page translation series, particularly of Latin and Greek, use techniques very close to wording and are often called “cribs.” Others aim at a more fluent target text, but the constraint of fitting paragraph to paragraph, if not quite line to line, limits the reorganization of material and rephrasing normally found in a literary translation. Penguin Parallel Texts, in the U.K., and the series currently published by Folio in France are of great use to foreign-language learners of Italian, Spanish, Russian, and so forth, and also to people like me who were taught a language long ago and are glad to have some help when revisiting the key texts of youth.

Wording translation and facing-page translation (which almost always uses matching sentence length) are not “bad” ways to translate. They are language operations with specific finalities, serving communicative and educational purposes proper to them and to nothing else. Translation is not just one thing; how best to do it depends on what you are doing it for.

However, wording is not what people mean when they call something a literal translation. The so-called literal translation of У меня большой дом is not “At me big house” but “I have a big house.” That’s to say, all that is actually meant by calling something a literal translation is a version that preserves meaning in grammatical forms appropriate to the language of the translation. Octavio Paz was right to say that there is no such thing as a literal translation! It’s just a translation—a plain, ordinary, actual translation of the source. The left-side player in the long and frustrating game of squash between “literal” and “free” doesn’t really exist. It’s just the shadow of another, more ancient world. But shadows can be quite frightening even when you know they don’t exist.