How do we assert our qualifications and competency as translators? Certainly, for most of us, familiarity with the source language is essential; or, to put it the other way around, ignorance of that source language would be a disqualification. Translating from translations is mostly what one doesn't do, or doesn't want to admit to doing. Yet the history of translating around the world is crowded with translations made from other translations. The reasons given for relay or chain translation, which is a big part of the hurried modern literary history of East Asia among other places, are usually practical. But the main fact I want to foreground about relay translation is that it's not often spoken about: though frequent and expedient, it seems a derivative and unauthentic way to translate, and translations of this kind are often treated as mere harbingers of serious, genuine translations.Footnote ¹

Thus translations of translations are apt to pass unnoticed. To counteract this special type of “invisibility,”Footnote ² I propose that we be on the lookout for translatedness, as a feature of both provenance and style. When the omitted or ignored stages in composition are acknowledged, familiar documents may become strange and unwieldy—in a good way. Here's a thought experiment. Let's say someone wants to dub or subtitle the movie West Side Story into another language. A truly careful translator will not be satisfied with rendering the approximate meaning, but will want to find ways to express other dimensions of the text: not only rhyme, melody, and rhythm but also the different idiolects of the characters (the more and the less recently arrived Puerto Ricans, the Jets and the Sharks, the cops, the teachers, and so on). Stephen Sondheim put a lot of thought into his writing, and the translator should honor that, no? But there's more. If the translator is translating the recent Spielberg film, based on the 1961 Robert Wise and Jerome Robbins film, should the translation acknowledge the fact of the remake being a remake, now that our ethics of translation calls on us to notice such things? Most audiences in English-speaking countries will recognize, if they didn't know it already before they got to the theater, that the story closely follows that of Romeo and Juliet. So, in a broad sense of the word “translation”—an extension that is already common critical practice—we could call the Spielberg film a translation into current cinematic codes of the 1961 film, which was a translation from the stage play, which was a transfer into the New York slums of Shakespeare's play set in Verona, which was itself a translation or adaptation onto the stage of a story Shakespeare had read in Arthur Brooke's dawdling English couplets—Brooke's version being as well a translation from an Italian tale, the predecessors of which recede into unknowability. The “originals” submitted for translation are translations themselves, if you care to look into their background. And the question for the translator is this: When you translate version N, should your translation include echoes or reminiscences of version N-1? Sondheim's libretto often picks up words from Shakespeare's playscript, and a translator sensitive to this sort of thing would want to try to capture it, so far as possible, in the language of arrival, but how?

Then again, someone might protest that the important thing is to reflect faithfully the impression that the observer of version N has of that version, not including all the previous versions or the poetic effects that may rely on the relation of N to the versions before N. In that spirit, we might say, the translation is a synchronic thing, a snapshot of one link in the chain, one moment in the history of the star-crossed lovers’ tale. Translations written for performance can often economize on allusion, since a word spoken on the stage or screen is quickly supplanted by another. So a practical translator may say that my quibbles about the “originals” of the “original” are misplaced. The synchronic task of the everyday translator defines the “original” with respect to an everyday audience that may not care about too many layers of diachronic development and difference. It is no more improper than what we do every day, when looking through the window at trees and people outdoors: we don't usually focus our gaze on the window, it is the thing through which we see, just as our cornea is, and to pay attention to it would mean losing sight of the usual objects of our vision. Translation, as we learn from Karen Emmerich, makes the originals, or, in the imaginary case above, determines what will count as a relevant “original” for present purposes.

When we look at, not through, the cornea or the window glass, it's usually because something has gone wrong: a cataract is clouding our vision, dust or frost has coated the windowpane; and our aim in focusing on the medium is to fix it. The “glassy essence” of translation requires its disappearance as an object of attention. But what about a case of translation the mission of which could not be fulfilled without noticing and trying to reproduce the translatedness of the translation taken as the nonce original for translating? An example is ready to hand. I do not think anyone can understand American poetry of the twentieth century and later without considering Ezra Pound's Cathay, a booklet of fourteen poems translated from Chinese and published in 1915. As Pound acknowledged on the title page, it was a case of relay translation: the poems were “for the most part from the Chinese of Rihaku [the name of Li Bai (李白) in Japanese pronunciation], from the notes of the late Ernest Fenollosa, and the decipherings of the Professors Mori and Ariga.” The roll of names calls for a backstory. Fifteen or twenty years before Cathay's publication, Ernest Fenollosa had sat down for many sessions with two Japanese colleagues, one a scholar of law and international relations, the other a cultivated practitioner of kanshi (漢詩), Chinese-character verse. Working from anthologies of Chinese poetry ranging from the Book of Songs (ca. 800–500 BCE) to the Tang dynasty (618–907 CE), Fenollosa's informants glossed the poems word by word in English, annotations that Fenollosa took down and paraphrased in a series of notebooks. Pound then mined these notebooks for phrases and lines that he recomposed according to his own sense of what Chinese poetry had to offer the American English-language poet of 1915. So there is our series of relays: from the Chinese texts to an unregarded but presumably active stratum of Japanese thought in the minds of Fenollosa's interlocutors, to their utterance in English, to Fenollosa's recasting of that utterance in the notebooks, to Pound's reformulation of those notebooks in a new style of English free verse. But few readers of Cathay have paid much attention to the translational relays going on within the windowpane (to vary the previous simile): one proof is the fact that the notebooks had to wait until very recently to receive a comprehensive editing and interpretation under the hand of Timothy Billings. None of that mattered for the first readers of Cathay. Or perhaps it mattered for five or ten among the thousands who read it in the first ten years of its existence as a sample and manifesto of modernist, Imagist verse writing. The five or ten who knew enough to see what Pound had done to the Chinese poems, the mistakes or shortcuts he had committed, might have said something about it to one another, but that didn't enter the conversation very prominently. For the vast majority of readers, Cathay mattered as an articulation of a new way of using language. It had taken its lead from the Chinese poets, but most of all it laid out its example and said, as all ambitious and innovative works do, “Go thou and do likewise.” It broke with many of the conventions that enabled readers to recognize something as a poem.

Where Georgian verse was continuous, regular, expressive, polished, Cathay was pleased to be rough, flat, broken, asymmetrical (Saussy). A vast, pointless European war was devouring the young men, and the best the British national verse standard could produce was this:

Cathay permitted the soldiers of ancient China to fall out of step with such pentametric marching and say:

Setting Cathay against Brooke's marching meter doesn't just highlight Georgians against modernists. Even contrasting Cathay with Pound's prior practice, or with “Prufrock,” shows what a profound case of arrhythmia is affecting Pound's line. Many reasons could be advanced for Pound's defiance of metronome meter here, but one reason for the use of awkward, asymmetrical prose must have been to remind us that what we are reading is a translation, the awkwardness being a kind of guarantee of fidelity. (Fidelity is just role-playing in Pound's Cathay; as is well known, Pound couldn't read a line of Chinese and got interested in learning the language only in prison and in the asylum.) Not only in metrics, but in syntax and word choice too, Cathay avoids poetic patterning: clauses are chained together in an additive rather than subordinating way, with little but commas to join them. The reader doesn't know what's coming next. I get from this the feeling of a first-draft translation, when one is just taking inventory of the foreign text without a clear sense of what it all adds up to, no anticipation or completion yet. It's just this, and this, and this, and this.

In the depiction of states of mind, too, there's a certain flatness, bluntness, or externality. You could put this down to Oriental restraint, if you wanted to go the cultural-essentialism route and make the perception derive from properties of the distant original rather than from the mediator; or you could see this as an effect of translation, which is often hard put to capture fine shades of semantic difference and implication for which the language of arrival offers few affordances.

I am reminded—as a symmetrical inverse—of Leo Spitzer's analysis of “pseudo-objective causality” in certain Parisian novels of night life.Footnote ³ It's clear that the Chinese poems as Pound understood them through Fenollosa's glosses showed the way that Pound thought modern poetry should go, away from the indefinite, inward suggestion and toward the luminous and evident image. In all these ways, at any rate, the fact that Cathay consists of translations is inseparable from its intended effects on the reader and on literary history.

Now imagine translating Cathay from English into yet another language. Would not the retranslator of Cathay have to attempt to reflect these very qualities of translatedness that define its poetic innovations in the literary world of 1915? A translation that did not represent in some way the hesitancy, the off-key qualities of Pound's Cathay would miss the mark, by my reckoning—though I do not in the least mean to imply that I know what those qualities would mean in the new situation into which a translation would introduce them. A hasty survey of existing translations of Cathay into French and modern Greek shows the translators compensating for what must have seemed to them deficiencies in the text: supplying, in place of Pound's repetitions, blurred semantics, and additive syntax, such indicators of stylistic competence as clausal subordination, lexical precision, and word variation (Alferi; Lorentzatos; Mendrakos).Footnote ⁴ But then again, the responsibility may not lie entirely with the translators, for French and Greek demand grammatical markers that English and Chinese often do without.

To get more precisely at what I am saying from an absurd counterexample, one might imagine a lazy translator, commissioned to render Cathay into French or Farsi, who discovers after a little reading that the poems are versions of famous Chinese poems. Would it not make sense simply to copy into the manuscript the existing translations found in, say, Paul Demiéville's Anthologie de la poésie chinoise (Anthology of Chinese Poetry) or the Farsi equivalent?Footnote ⁵ But that would not do at all. It would amount to obliterating the specificity of Cathay as a translation performed by a specific person or persons at a specific moment for particular reasons. But we enact exactly that obliteration every time we describe Cathay as a “translation”—which it both is and is not. To put it differently, the very idea of retranslating Cathay into another language shows that a translation is not a substitute for the original, insofar as two translations of the same work are not substitutes for one another. Yet the conventional ideology of translation tells us that a translation must be a substitute for the original, that such is the condition of its being a good translation. To the usual way of thinking, translation is already a substitute for something, so to back-translate a translation, all you have to do is substitute the original for the substitute. But no. A translation doesn't necessarily fit precisely back into the mold of its original. That misfit, that lack of correspondence, must be what Lawrence Venuti describes as a “scandal” (Scandals). Well, let the scandal go on!

In the same spirit of encouraging scandal, Christopher Bush has proposed that someone might translate Hamlet into French as the Hamlet of Wilhelm, not William, Shakespeare—the Shakespeare, that is, who matters so much for German literary history, specifically as a totem of kinds of writing that defy the French neoclassical canon (7).Footnote ⁶ That Hamlet would not be identical to the other Hamlets that may be found circulating in translation. It would demand a different engagement from the reader, who would need to be ready to spar with multiple frames of reference and their different angles of refraction. Translating translations makes comparison obtrude on our vision.

A third and last example. It is generally known that the tales and essays of Edgar Allan Poe were translated by the French poet Charles Baudelaire. It is often said that Baudelaire improved them in the process. But English-speaking readers have not had much of an opportunity to judge the results for themselves. I once proposed to translate back into English a representative selection from Poe's “Extraordinary Tales,” “New Extraordinary Tales,” and “Tales of the Grotesque and Serious” (groupings made by Baudelaire or his publisher, by now familiar to French readers), without checking the translations against Poe's English. The purpose was to reveal experimentally something that one often hears said loosely or metaphorically: that a translator “invents” or “discovers” the writer being translated. (T. S. Eliot: “Ezra Pound is the inventor of Chinese poetry for our time” [14].) Could I then introduce English-language readers to a new author: Baudelaire's Poe? Sadly, the agency to which I directed my grant application rejected my proposal, deeming it frivolous. But since I hope this essay will be read by the sort of people who might appreciate such frivolity, let me cry in my beer for a moment.

The versions of Poe, published between 1848 and 1864, were a bigger success, commercially speaking, during Baudelaire's lifetime than his own poetry, and probably did more to accredit him in the French literary world than Fleurs du mal (despite the éclat achieved by the censoring and trial of the latter). They have remained the only translation of Poe's tales that is read in France. Through Baudelaire, Poe has taken up permanent residence in the imagination of psychoanalysts, surrealists, situationists, the authors of graphic novels, and many other artists and thinkers in French-speaking territories. Dozens of paintings by René Magritte take their titles from Poe stories translated by Baudelaire; Salvador Dalí returns to Poe again and again; André Breton's “Second manifeste du surréalisme” (“Second Surrealist Manifesto”) fishes for outrage: “Crachons, en passant, sur Edgar Poe” (“Let's spit, en passant, on Edgar Poe”).Footnote ⁷ Marie Bonaparte's reading of “La lettre volée” in 1933 preceded by some thirty years Lacan's famous “Seminar on the ‘Purloined Letter’”; both relied on Baudelaire's translation, abusively replaced in English versions by a restoration of Poe's English. From France, Baudelaire's Poe spread to Spain and Russia, for many years obviating the need for direct translations from English. Like the King James Bible or Antoine Galland's Les mille et une nuits (The Thousand and One Nights), Baudelaire's Poe is a translation that arguably surpasses the influence of the original. In those parts of the world and to that degree, it can accurately be said that Edgar Allan Poe is an authorial identity created by Charles Baudelaire.

It is not absurd, stubborn, or Franco-chauvinistic to prefer, as I do, Baudelaire's Poe to Poe's Poe. Baudelaire is a precise, economical, vivid stylist, more consistent in his effects and narrative voice than Poe, whose writing I often find lurching, awkward, and spotty. Whether consciously or unconsciously, Baudelaire improved Poe, editing him for the purposes of presenting him to French audiences as an author of a specific kind, an author of understatement, irony, and abrupt (but purposeful) changes of tone. Models to avoid—despite resemblances—would have included Eugène Sue, Victor Hugo, and the Alexandre Dumas of Les crimes célèbres (Celebrated Crimes). It is by not coinciding with any of those models that Baudelaire's Poe gains a particular dignity. Among the features of Poe's tales that corresponded to or stimulated parallels in Baudelaire's own writing are the emergence of gothic or irrational motifs on a background of contemporary urban life, an appreciation for common sense as the source of errors in perception and evaluation that must be corrected, and a narrator quick to hypothesize general laws on the basis of anecdote. Now that Baudelaire has been translated carefully and sensitively into English for over a hundred years (early Englishings of Baudelaire are, to my ear, too melodramatic and Poe-like, not sufficiently attentive to his classical dryness), the conditions are ripe for offering a translation of Baudelaire's Poe that will be recognizably Baudelairean, though in another language.

By not consulting Poe's original I mean to make it clear to myself and my potential readers that the object of the translation is the text that Baudelaire made out of Poe's materials; it is not in the least the aim to return to an “original” state. My purpose will be to render Baudelaire's Poe faithfully, accurately, and idiomatically, as if I were conveying an elegant work of nineteenth-century French narrative prose. I set this aim for myself on the principle that the process of literary translation (unlike that of diplomatic or scientific translation) is always to some extent adaptation. Baudelaire transformed Poe in certain ways, while nonetheless fulfilling the main requirements of a good (faithful, reliable, attractive) translation. In retranslating, I must retain the differences that Baudelaire introduced. Instead of closing a circle, my retranslation would chart an ellipse—a flyover of the United States by the satellite launched from Paris in 1848. Academic readers would find in it a case study of the early worldwide circulation of American literature, which is not simply an American undertaking; they would also find there material for thinking theoretically, historically, and comparatively about translation. And surely such a book would further an understanding of the sensibility that European and South American writers and artists have found themselves responding to in Poe—not so much our familiar Poe as Baudelaire's and Stéphane Mallarmé's Poe, the international Poe bearing a French forwarding stamp.

We learn from translating translations how much literary history ramifies. An intervention in language or poetics made at a certain place and moment through translation impels changes that affect readers and writers downstream from it, while leaving other branches of the causal tree untouched—unless we should try, scandalously and anachronistically, to reintroduce elements of one causal stream into another causal stream, complicating history and definitely alienating the advocates of linearity. That's one thing translation already does: bringing twelfth-century Provençal poetry, let's say, into the conversation of twentieth-century England, or making the Sufi poets address the concerns of contemporary Norwegians. But translating translations as translations does that even more efficiently, if it's permissible to offer efficiency and complexity in the same package. I say let's try.