The task of the translator, to borrow the title of what is probably the twentieth century’s single most influential commentary about the goal of translation, is to create a text that improves upon the original. In all fairness to Walter Benjamin, this is not what he says in “The Task of the Translator.” Benjamin proposed that a good translation puts the same kind of pressure on the target language that the original puts on the source language, and so “to some degree all great texts contain their potential translation between the lines.” To claim that a translator aims to improve the original text flies in the face not only of Benjamin’s idealism but also of conventional wisdom, which holds that translation is impossible from the outset. As John Ciardi once said, translation is “the art of failure.” That the quote is frequently misattributed to Umberto Eco seems to back the point.

Yet this clichéd wisdom has little bearing on reality. In his marvelous book Is That a Fish in Your Ear?: Translation and the Meaning of Everything, David Bellos demonstrates many of the ways that translation is not only possible but ubiquitous, so thoroughly woven into the fabric of our daily lives—from classrooms to international financial markets, from instruction manuals to poems—that if translation were somehow to become impossible, the world would descend into the zombie apocalypse faster than you can say “je ne sais quoi.” The European Union, for example, has twenty-four official languages; every legal document within the EU has to be translated into all of them, and every official translation is legally the original. There is clearly a tension between the varieties of “translation” happening all around us—every moment of every day, truly one of the fundamental activities that hold our world together—and the persistent recycling of platitudes about how this activity, so basic and ubiquitous, is impossible. If the platitudes are recalled more often than translation’s pervasiveness, it is only because translators are usually invisible, their work mysterious.

One reason translation is superior to the original is the access it grants. Without translation, Stieg Larsson would have had no appreciable presence on the beaches of the world these last few years, and Proust’s À la recherche du temps perdu would only have been a mesmerizing epic novel to those who could read French. For the rest of us, it could only be a doorstop. We tend to assume that all the best literature in a given language finds its way into English, and that—making a leap that sounds more sensible than plausible—if it’s worth reading, it’s probably already available in English. But this is simply not true. What gets translated and published in English in any given year is such a tiny fraction of literature available in other languages that we Anglophones can never hope to read all the worthwhile works of literature in other languages. Anyone who knows a foreign literature well would have little trouble naming titles, including major works by major writers in that language, that are unavailable in ours. The odds are strong that you will never be able to read what might have been your favorite book.

That a translation is superior because you can read it without knowing its source language seems obvious, yet it is a fact easily overlooked. Without that access, most literature might as well not exist—and within the range of our own experience, it doesn’t. The tasks of the translator for the most part are to increase the availability of information and to stage that information’s effect in the new language. A funny sentence about cats in Japanese cannot just be about cats in English: it also has to be funny, because a joke that doesn’t make you laugh is not really a joke. By the same token, the notorious difficulty of translating humor might be chalked up to the sad fact that so many brilliant translators just aren’t that funny. Whether on the stage or on the page, the difference between someone who comes up with a joke and someone who makes people laugh is entirely in the telling.

The same is true of everything that might be termed “style.” There are many examples of a work’s stylistic features being amplified or augmented by the translator. I once attended a reading where Joseph Brodsky invited his friend Derek Walcott to read his translation of Brodsky’s poem “Letters From the Ming Dynasty,” because Walcott’s English had more effectively drawn out sonorities that Brodsky had wanted to produce in Russian. (For the record, both versions are superb.) And the popular Japanese author Haruki Murakami says quite openly that the style of any of his books in English depends upon which of his three main translators has taken it on. If Edgar Allan Poe sounds more like Charles Baudelaire in the latter’s renditions of Poe’s writing, it can only be to Poe’s benefit.

How do translators go about their work? A number liken translation to the theater—Jake Donaghue, the hero of Iris Murdoch’s Under the Net, describes it as “opening one’s mouth and hearing someone else’s voice emerge,” and this seems apt. Translation can be likened to forensics, that is, the competitive rhetoric and oratory practiced in our more traditional high schools and colleges. It, too, is competitive, at least insofar as it produces multiple versions of the same text, which is itself an invitation to compare. The assumption that an original is always superior to a translation is just such a comparison, if one that usually has no basis in actual experience.

Even in stagecraft, though, one has a choice of approaches. Some translators, like some actors, project the source material through their own indelible style and personality, so that even in a bravura performance we never lose sight of the person on stage, as it were. We might call this the Jack Nicholson School of translation. Charles Simic, an excellent poet and prolific translator, is a member. When he calls translation “an actor’s medium,” he is less interested in transforming himself or imitating someone else than in making himself believe that he’s the one writing the poem in the first place. Translation becomes a way of taking over the material. While the rare bird with access to both versions may be disappointed—Brodsky objected to what happened to the Russian poet Osip Mandelshtam at the hands of W.S. Merwin, “from whom more should have been expected than a translation of Mandelstam into Merwin”—Merwin’s Mandelstam becomes a performance all its own. Just because you enjoy reading Shakespeare doesn’t mean you can’t appreciate Derek Jacobi as Lear.

Some actors, however, prefer to disappear into the role, and here, too, there is a clear analogy to translation, which can become its own vanishing act. A translator may spend hours laboring over a word choice or the placement of a comma, pondering whether an exclamation point means the same in German (where it appears frequently) that it does in English, all to create a text that sounds as if it were written by someone else. This kind of translation is similar to Method acting. Even “off-set,” long after the work of translation is finished and the translator has moved on to another project, another role, he or she will forever remain in character in the printed book, hidden behind the author’s name. In the early 1960s, Jiri Levý, a Czech theorist whose work has attracted increasing attention in recent years, developed a practical approach to literary translation based in part on Method acting. He called his approach “Illusionist,” by which he meant that the translator would do everything he could to make the reader forget that he was reading a translation. He likened it to a totally immersive experience of the theater.

Yet, as we often find when we try to “translate” theory into practice, principles can be compromised by reality. How, after all, are we supposed to get into character when that “character” is another human being whom we know only through certain movements of mind, through the traces he or she has left in language—and a non-native language at that, since there is a strong bias in Anglo-American literary culture that translators be native speakers of English?

Jorge Luis Borges parodied this predicament brilliantly in his story “Pierre Menard, Author of the Quixote,” which just about every student of literary translation encounters early and refers to frequently thereafter. The title character, a French translator devoted to creating a new version of Cervantes’s masterpiece, decides that the only way he can truly come to inhabit the text is to relive the life of its author, so that his production of the text will mimic Cervantes’s. “Initially,” Borges writes, “Menard’s method was to be relatively simple: Learn Spanish, return to Catholicism, fight against the Moor or Turk, forget the history of Europe from 1602 to 1918—be Miguel de Cervantes.” These kinds of mimetic absurdities are common in Borges, for whom the artist’s ambition to fill his work with more and more detail creates not merely an imitation of life, but a carbon copy of it, a “map of the Empire whose size was that of the Empire,” as he once wrote. But Menard’s ambition comes crashing against the reality of his time and place. He is a Frenchman of the twentieth century, not a Spaniard of the seventeenth. No matter what he does to re-enact the life of Cervantes, the world where this bizarre performance occurs will have long since moved on.

Still, behind Borges’s very serious silliness is the idea that an effective approach to translation is one that reconstructs the composition of the original text. The notion’s germ is already contained in the truism that one has to be a poet to translate poetry, which nevertheless can’t account for why some of our best translators of poetry—John Felstiner from German, for example, or Allen Mandelbaum from Latin—are poets primarily when they are translating. They create their greatest works as poets when they are collaborating with the poets (living or dead) whose work they translate. But the idea that translation benefits from some technical expertise in the material being translated still holds, not least because the translator would then presumably have a better grasp of the meaning of the original.

This principle of specialization underwrites much of the translation that affects us directly on a daily basis. Legal or medical or diplomatic translators, to name just three fields where the absence of translators would be felt most immediately, may not themselves be lawyers, doctors or diplomats, but they know the jargon specific to those professions, as well as how members of those professions communicate with one another, in the source language as well as in English. This is especially evident in technical fields, where most professional translators happen to be employed. If the person translating the manual to your new computer has never brought a computer online, you’re likely to know. Ikea, the Swedish housewares firm that has elevated production efficiency to a core part of its corporate ethos, has almost entirely eliminated the need to retranslate its manuals by rendering most assembly instructions as illustrations. But as anyone who has assembled Ikea furniture is aware, even this form of translation is not without ambiguity, confusion and smashed particleboard.

With literary translation, this degree of specialization is often developed through the painstaking, case-by-case process of becoming an expert on an individual text. This is why several writers have referred to translation as, above all else, an act of interpretation, albeit one in which the translator’s act of reading is recorded and reproduced, word by word and thought by thought, for a new audience. This involves not only having a strong command of the history and literary traditions of the source culture, which most translators acquire as part of their language training, but also being at least as proficient in the history and literary traditions of the target culture. Not only does one have to catch idiomatic expressions and sly allusions; corresponding idioms and references must leap to mind in English.

Here, another form of forensics comes into play, the kind that is most familiar to us from television crime procedurals. Think of it this way: the translator breaks down the original text, then reverse-engineers it—in effect figuring out how it happened by relying on evidence at the scene (the book), questioning witnesses (the author, when available, or the author’s editors and acquaintances) and consulting the various, sometimes conflicting postmortems produced by scholars. The translation then becomes the reconstruction of an event.

If you really want to get into the author’s head, short of reliving his or her life, you find yourself reading everything he or she was reading while thinking about and writing the book you are translating. I sometimes think of this as reading over the author’s shoulder. It became a particularly big part of my life in 2010, when I was working on two books that are thick with quotations: Krzysztof Michalski’s The Flame of Eternity: An Interpretation of Nietzsche’s Thought (Princeton; paper $22.95) and Marek Bienczyk’s Transparency (Dalkey Archive; paper $14.95), a book-length essay. Bienczyk is a novelist and a specialist in Romanticism at the Polish Academy of Sciences in Warsaw, as well as a translator of fiction and philosophy from French. His book, a discursive, novelistic meditation on the theme of transparency, quotes liberally—nearly two hundred instances over the course of the book—from Polish Romantic literature and French philosophy, not to mention just about every major European text on transparency itself. Michalski, who died of cancer about a year after his book on Nietzsche appeared in English, was a philosopher based in Vienna. His book contains over 500 references, many of them to Nietzsche, as well as to ancient and medieval theology.

The number of quotations in each book has stuck in my head because it fell to me to track them down. After all, if we were going to be quoting Nietzsche and Roland Barthes, I wanted to use translations by those people who, for me as a reader, have best captured what the originals should sound like in English (Walter Kaufmann and Richard Howard, respectively). Where the text being quoted had never before appeared in English translation—as I’ve mentioned, you might be surprised at how much work by major authors remains invisible to us—I wanted to find the original source and render it in a style consistent with the translations that are available.

Bienczyk’s and Michalski’s books were not very helpful in this endeavor. Both had been released by Znak, one of Poland’s most venerable literary publishers, and, despite their rarefied subject matter, both were presented as literary prose for the general reader, without any notes. In fact, both authors were somewhat resistant to my intention to include citations in my translation, preferring instead that I simply work from their prose, despite the possibility that there would be long passages consisting of translations of translations of translations. I insisted on weighing their sources with my own ear.

When the stars were properly aligned, this meant reading a quotation in Polish and finding its corresponding passage in the English translations I was using. In practice, it was rarely that easy. Michalski and Bienczyk frequently quoted from memory, and there were more than a few instances where their memories had produced paraphrases instead of quotes. Sometimes they were working from a translated text that had made troublesome modifications to the original. Very often they would quote an author without any indication of their source; having read the quotation in Polish, I would back-translate it into several versions of what the English, French or German might be. Once I had tracked down the source, I would either find the best available translation of the passage or produce one myself.

As in television detective work, one relies on a combination of old-fashioned sleuthing, technological aids and dumb luck. Witnesses help, but they can be unreliable; never intending to fact-check or source their literary prose, both authors had forgotten precisely where several quotations had come from. My own familiarity with some of the sources helped—literary translation is almost always served by elective affinities between author and translator, which in this case meant that we had already been reading a lot of the same stuff—but I still had to become conversant with authors and texts that had never meant much to me before. Digital searches, which cover millions of texts in a fraction of a second, help a lot. A typical day’s work brings all these resources into play and still ends with a note to one’s tomorrow-morning self about where to pick up the thread. After all, most CSI-type crime dramas mark their entr’acte with a scene of frustration, that moment when the investigators hit a wall (literally or figuratively, depending on the quality of the direction and/or set design), so that after the commercial break (or, for the translator, a good night’s sleep), we can backtrack and re-evaluate. That was my daily life for about a year.

What could possibly motivate someone to perform this kind of labor? Not money. The per-hour wage for literary translation in the United States usually works out to cents. The appeal, what sustains the translator from one jaw-clenching moment to the next, is the purely poetic activity of matching a word or phrase, of reorchestrating the tonal shifts from one language to another. But in the end, all you really want to do is to climb inside this text in order to understand how it works, what it was that sparked the enthusiasm bordering on obsession that is the sine qua non of labor so generous in its demands and poor in its pay. Sometimes it’s only by reconstructing the machine from the inside out, switching it on, and watching it work that you can move on to something else. If you can outfit it with a new bibliography or index along the way, more’s the better.