Translation as a daily, difficult practice often looks like a story of repeated failures, but we should be clear about the sort of failure we have in mind. First, it cannot be absolute or final: many words, images and modes are persuasively translated all the time. Second, even when force or connotation is lost in the passage from one language to another, something may be gained, a glimpse of new meaning, or a resonance in the very gap between idioms. And third, our standard of success cannot be the perfect transmission of all aspects of a text or speech. This would not be a translation at all but an unimaginable replica, rather like the map of an empire in a Borges story which turns out to be as big as the empire itself. Traduttore tradittore, the Italian proverb eagerly says. But the translator is not a traitor, only the orchestrator of a second, different form of life. Or not necessarily a traitor. Treason is possible, even frequent, but not all departures from the original are treasonous.

At the other end of the scale of our received ideas translation appears as a large, loose figure of speech, a synecdoche for reading itself, or for intertextuality, or even for literature. ‘In the end all literature is translation’, writes Novalis, cited by Efraín Kristal. In the end, maybe, but meanwhile there are still distinctions to be made, and although Borges adduces the idea of translation within a single language, thereby diffusing the meaning of the term considerably, he still separates what he calls ‘direct writing’ from ‘imitation’. Translation, he says, helps us to understand the ‘modest mystery’ of literature because it tackles ‘a visible text’, and different versions of that text ‘are a partial and precious documentation of the changes the text suffers’ [translated by Eliot Weinberger, in Jorge Luis Borges, Selected Non-Fictions, New York, 1999, p. 69]. Borges literally writes of the ‘vicissitudes’ the text suffers, which suggests that translation is a form of trouble as well as a second life. Borges would not, I think, resist the idea that literature itself is a form of trouble; but he would not say it was only that.

Translation is the ‘invisible work’ of Kristal’s title, and he seeks to situate Borges’s writing between its specific and metaphorical applications, between the daily practice and the figure of speech—translation is the ‘process whereby a writer remodels one sequence of words into another’. ‘I see translation’, Kristal writes, ‘as more central to Borges’s literature than the celebrated labyrinths, mirrors, tigers, and encyclopedias that abound in his literary world’. An appealing argument, but the subject remains slippery, and this admirable book develops a curious and instructive stutter in its opening pages, defining its ‘purpose’ in four different ways. ‘The purpose of this book is to demonstrate that translation . . . is central to Borges’s reflections on writing and to his contribution to literature’; ‘The purpose of this book is to make visible Borges’s creativity as translator’; ‘The main purpose of this book is to offer an account of the role translation played in Borges’s creative process’; ‘My purpose is to underscore the significance of translation in Borges’s œuvre’.

All of these purposes are accomplished to varying degrees, and Kristal’s hesitation is understandable, since he has at least three distinct avatars of Borges on his hands: the Borges who thought and wrote about translation; the Borges who translated; and—most interesting and most elusive—the Borges who folded ideas about translation into almost all of his fiction. Kristal devotes separate sections to each of these three figures, and argues, in an afterword, that the complexities of translation should displace those of philosophy in the critical account of the energy driving his work—we must understand Borges’s deep games with philosophy but not mistake him for a philosopher; nor, on the other hand, do we have to accept all his ironies as ultimately sceptical. The role of translation offers Kristal a bridge between the two positions: Borges thinks in versions to avoid any commitment either to first causes or their absence.

The first named purpose (to discuss the centrality of translation in Borges’s work) is the clearest and most fully achieved; the last (regarding its significance to his œuvre) looks like a rephrasing of the first, but invites us to wonder where the œuvre begins and ends. The second purpose is also accomplished, though the result is more disappointing: Borges translated important and interesting writers—Woolf, Faulkner, Emerson, Carlyle, Poe, Stevenson, Kafka, Gide, among many others—into Spanish, and his first published work, printed in a Buenos Aires newspaper in 1910, was a rendering of Oscar Wilde’s ‘The Happy Prince’, written, according to his mother, when Borges was nine.

He ‘felt proud’, according to Kristal, ‘that his first publication was a translation’, and liked to introduce himself as a translator. But this seems to have been a mask rather than a profession; and the actual run of Borges’s work in that line appears to be tuned to the perceived needs of Argentinian culture rather than to any writerly programme. He is credited, among other things, with bringing German Expressionism ‘to the attention of the Spanish-speaking world’; he also translated from English, French, Italian, Anglo-Saxon and even Old Norse (the Prose Edda), which he learned as his eyesight faded in the 1950s. Not that ignorance of a language barred him from rendering it in Spanish: he became fascinated by Chinese literature, and translated fragments from collations of English and German versions. Some of these works were composed in the dark basement of the Municipal Library, where he escaped the ‘solid unhappiness’ of his ‘menial and dismal existence’ as a First Assistant Librarian, a job he took on shortly before his father died to lighten their dependence on his mother’s fortune.

Borges’s creativity as a translator does come across in Kristal’s account, though it seems rather slight on the whole, effectively limited to a few switches and interpolations. Kristal makes heavy weather claiming importance for these moves, and he himself finally backs down. ‘Borges’s most radical engagements with an original’, he says, ‘did not take place in his translations, even in the most daring of them’; and in another context he tells us why:

One can translate within the same language, and one can copy from one language to another. Borges would call a text a ‘copy’ if the most pertinent observations to be made about it could also be made of the original. In contrast, a ‘version’ is a text with relevant differences with respect to either the original or another translation of the same work . . . A ‘re-creation’ . . . omits many details to conserve the emphasis of the work, and it may add interpolations.

Although Kristal wants us to see Borges as a re-creator, he was in most of his translations a copyist. His fiction is another matter, and if we do not quite see his ‘creative process’ at work—Kristal’s third purpose—we do see what that unlovely phrase is getting at: the role of translation not as a figure for literature but as a model of complex human interaction.

At the end of the story ‘Tlön, Uqbar, Orbis Tertius’, for example, what looked like an entertaining fantasy turns into historical desolation. The world has been invaded by a ruthless system of unreality—Borges lists Nazism, Anti-Semitism and Dialectical Materialism as its equivalent—and the narrator dedicates himself to what he calls ‘an indecisive Quevedian translation’ of Sir Thomas Browne’s Urn Burial, which he does not intend to publish. The ‘indecision’ of the narrator’s translation is a direct answer to the emphatic certainty of the ruling system; and the unpublished text is an instance of the invisible work that remains when the visible world has gone wrong. Kristal cites Christopher Maurer, who identified a strain of Quevedo (the 17th century Spanish satirist) in Borges’s late verse as the source of his ‘stoic calm’ at the prospect of oblivion.

But there is more. Along with his friend Bioy Casares (who makes an appearance as a character in the story), Borges himself translated a section of Urn Burial, which excised references to Christianity and shifted ‘a meditation on the stoic values rewarded in the afterlife’ to a ‘quiet protest in the face of obliteration’. He also made an interesting interpolation about the burial of facts in silence.

Many more facts have been buried in silence than registered, and the most copious volumes are but epitomes of what has taken place. The chronicle of time began with the night, and obscurity still serves it; some facts never see the light; many have been declared; many more have been devoured by the obscurity and the caverns of oblivion.

‘There is no better commentary’ on ‘Tlön, Uqbar, Orbis Tertius’, Kristal says; and there seems to be more than stoicism at work. By filtering the voice of a fictional Browne through a pastiche of Quevedo, the narrator evokes a use of literature which is not merely mimetic or playful, or submissive to a world rapidly actualizing the Nietzschean saying that there are no facts, only interpretations. Although Nietzsche, as it happens, did not quite say that. He said he would say it. ‘Against positivism, which insists on the phenomenon that there are only facts, I would say: no, facts are just what there are not, there are only interpretations’. Of course, the world of Tlön is rigorously idealist; in it, there are no facts, only phenomena. But the two conditions look alike, flourish in the same hedgerow. The task of translation grows between them, where what is lost or obscured, may also be preserved and retrieved; where fact and phenomenon continually translate into each other and remain at play.

Kristal quotes Borges’s discreetly unfaithful version of a sentence by Kipling—not quite a copy and not yet a re-creation. Kipling writes ‘I had just discovered the entire principle upon which our half-memory falsely called imagination is based’. Only a half-memory, and ‘imagination’ is the wrong name for it. Borges writes, ‘In that instant I had discovered the principles of the imperfect memory which is called imagination’. The imagination is a form of memory, imperfect because it is human, but loyal to what has happened, however elusive that may be. In this idiom, to imagine the past is not to invent it but to find it, and the past itself begins to resemble a frequently translated text.

Thinking of Borges’s views on translation, as distinct from his uses of it in fiction, Kristal suggests that ‘it would be possible to construe an approach, even a doctrine, on the basis of his general observations’. More than an approach, I think, and less than a doctrine. The difficulty with Borges’s pronouncements, on translation or on anything else, is their infallible mischief. This is also their charm, of course; and more than their charm, part of their meaning. But then paraphrase becomes clumsy, because so much depends on what is not being said. The great virtue of Kristal’s book is that he never loses hope or track, and he makes the consistency of Borges’s views on translation, as well as their shifting complexity, very clear. But even he misses a few swerves, and following Kristal’s hints, it may be worth looking more closely at Borges’s quiet irony, not to pin it down but to trace its movements.

The various translations of a single text, Borges says, are ‘different perspectives on a mobile fact’, a lottery of what is left out and what is underlined (‘un largo sorteo de omisiones y de enfasis’). He goes on to make his most frequently quoted remark on translation—he himself printed it, Kristal tells us, on at least three separate occasions. ‘To presuppose that every recombination of elements is necessarily inferior to its original is to presuppose that draft nine is necessarily inferior to draft H—for there can only be drafts. The concept of the “definitive text” corresponds only to religion or exhaustion’ [translated by Eliot Weinberger, in Selected Non-Fictions, p. 69, slightly modified]. Borges then writes of the ‘superstition of the inferiority of translations—coined in the well-known Italian adage’, but he does not mean translations cannot be bad, only that they do not have to be. By the same token originals do not have to be good, and one of Borges’s funniest jokes in this domain is his suggestion that an original can be unfaithful to its translation. Even so, Borges is proposing a mischievous model of reading rather than sheer relativity or recklessness. ‘There can only be drafts’ means we should, as readers, treat all texts as works in progress—assuming there is both work and progress. The ‘definitive text’, similarly, is a function of our needs rather than the work itself. The mobile fact will keep moving until we stop it by dogma, or become too tired to follow its movements. But it is a fact, not a fantasy or the free-form invention of the reader.

The above quotations come from the splendid essay ‘The Homeric Versions’, but there is more delicate and ironic detail in ‘The Translators of The Thousand and One Nights’, which Kristal calls ‘Borges’s most sustained and boldest essay on translation’. Borges starts out by invoking the concept of ‘translating against’, as if translation were always (or often) an attempt to erase the anxiety of influence induced by past translations. ‘We must understand this hostile dynasty’, Borges says. ‘Lane translated against Galland; Burton against Lane’ [translated by Esther Allen, Selected Non-Fictions, p. 92]. Each seeks to ‘annihilate’ his predecessor. We might think the translated text gets a little lost in the process, and Borges does think so. He also thinks that losing a famous text in translation may be a way of finding it has entered another literature. The virtue of these battling translations is that they ‘can only be conceived of in the wake of a literature. Whatever their blemishes or merits, these characteristic works presuppose a rich (prior) process’. But the initial text does not disappear, and Borges has no hesitation in speaking of the ‘falsification’ of Burton and others.

He is most subtle and amusing, though, on the work of J. C. Mardrus—called Madrus throughout Kristal’s book—who published a 1906 translation of the Arabian Nights, which drew Borges’s suspicion by its very subtitle: ‘Traduction littérale et complète du texte arabe’. Borges, it turns out, admires this work but chiefly for its ‘happy and creative infidelity’. Or does he admire it? Mardrus ‘translates luxuriantly’, Borges says; ‘strives to complete the work’; makes ‘personal contributions’—where the word for ‘contributions’ is ‘obsequios’, gifts or hand-outs. ‘In general’, Borges says, ‘Mardrus does not translate the book’s words but its scenes: a freedom denied to translators, but tolerated in illustrators, who are allowed to add these kinds of details’. ‘Mardrus never ceases to wonder at the poverty of “oriental colour” of The Thousand and One Nights’, and ‘with a stamina worthy of Cecil B. de Mille, he heaps on the viziers, the kisses, the palm trees and the moons’. Borges declares that of course he is not attacking Mardrus for these divagations; or rather, to be more precise, he says that he would be sorry—‘not for Mardrus, for myself’—if we were to read in his remarks a ‘propósito policial’, nicely translated by Esther Allen [Selected Non-Fictions, p. 106] as ‘constabulary intent’. His aim, Borges says, is not to demolish but to document the widespread admiration for Mardrus’s version. But one can be suspicious without being a policeman, and it is hard not to see here, beneath the acknowledgement of the charms of Orientalism, an attachment to the stricter idea of the less adorned East.

What Borges is offering, I take it, and what makes his argument so hard to fix or summarize, is an oblique but devastating attack on the ideas both of creativity and of fidelity. Creativity is fine, he implies, if you do not care about the difference between the author of Exodus and Cecil B. de Mille. Fidelity is good if you do not mind being a yawn-inspiring pedant. In all translations, case by case, there will be chances for creative choice, and fidelity does not have to be dull, but the simple embrace of either of these abstractions will put a stop to the only activity that matters, the proliferating and potentially endless talk among texts. When Borges, writing of Mardrus, uses the stealthy phrase ‘feigns to translate’, ‘finge traducir’, he unmistakably implies the reverse, a translation that would be the real thing. But the real thing would be multifarious, and nobody said it was easy.