Back Sides of the Tapestry

Cervantes fancifully imagines his reception in China in the dedication of part 2 of Don Quixote (1615). Still smarting from the unauthorized sequel published by the pseudonymous Alonso Fernández de Avellaneda a year earlier, the author boasts of having received breathless overtures from around the globe to revive the “authentic” knight errant:

Cervantes leaves readers to wonder how, and in what language, his literary self and the Chinese messenger conduct their fleeting encounter. Since Cervantes almost certainly knew no Chinese, how, then, could he understand the content of the emperor's missive, which he makes a point to note is written “en lengua chinesca”? Does the imperial emissary speak Spanish and therefore translate it (and, if so, how did he learn it, particularly before the founding of the school)? Or are there unmentioned intermediaries who tender services in translation and interpreting? How much faith should one place in their efforts to translate the language of a nation invoked most commonly, as it is here, as a metonym for distance and foreignness?

Such questions become more pointed when posed in the light of a metafictional scene toward the end of the novel, when Don Quixote visits a printing shop in Barcelona that, much to his chagrin, is publishing Avellaneda's unsanctioned spin-off. While surveying the operations of the press, out of curiosity he interviews a translator who is rendering an Italian book into Spanish. The obliging worker satisfies his queries about how to translate various words, after which Don Quixote shares his own opinion on translation: “me parece que el traducir de una lengua en otra, como no sea de las reinas de las lenguas, griega y latina, es como quien mira los tapices flamencos por el revés, que aunque se veen las figuras, son llenas de hilos que las escurecen y no se veen con la lisura y tez de la haz” (“it seems to me that translating from one language to another, unless it is from the queens of all languages, Greek and Latin, is like someone looking at Flemish tapestries from the back side, for although the figures can be seen, it is full of threads that obscure them, and they cannot be seen with the smoothness and color of the front side”; 1249). Read at face value, so to speak, the visual metaphor casts translations as subordinate to works written in their original language, always inferior in their ability to capture the richness of the source text. Scholars like Michel Moner have adduced this denotative reading to portray a Cervantes who shared the disdain for translators that was apparently permeating the Spanish ether (520). While it is true that Don Quixote goes on to compare translation with copying something from one paper to another, a mechanical process requiring neither “ingenio ni elocución” (“ingenuity nor eloquence”), such interpretations fail to acknowledge the plurivocal reality of early modernity, which could be subject to what Coldiron has termed the “Babel effect,” even as individual writers and translators flouted its prescriptive deference to auctoritas and its demotion of translation to “a necessary but fallen, secondary, stopgap measure” (“Beyond Babel” 313; see also Newman and Tylus 2–3).

For even as Don Quixote recites these commonplaces and, in his test of the translator's Italian vocabulary, reifies translation as a series of word-for-word correspondences, numerous other details of the scene betray an esteem for translation as the métier of an autonomous, accomplished intellect. The knight-cum-pundit laments the likelihood that the translator's “loables trabajos” (“laudable deeds”) have gone unrecognized. The depiction of him as “un hombre de muy buen talle y parecer y de alguna gravedad” (“a man of very good bearing and appearance and some seriousness”) deviates meaningfully from that of a Bartlebyan copyist, particularly as the narrator refers to him as an “autor.” The equivocal nature of this term in the early modern era, patent even in its most authoritative dictionaries (Covarrubias Horozco 1483), is echoed in Don Quixote's exaltation of Cervantes's contemporaries Cristóbal Suárez de Figueroa and Juan de Jáuregui as translators who “felizmente ponen en duda cuál es la tradución o cuál el original” (“happily put in doubt which is the translation and which the original” [1248–49]). And, despite his allegation that translation admits of no ingenio—an “untranslatable” term whose semantic field encompasses intelligence, wit, acumen, and creativity (“Ingenium” 485–86)—in his exchange with the translator Don Quixote uses the word a total of four times, subtly restoring to his interlocutor's labor the transformative power of the imagination possessed by the ingenioso hidalgo himself.Footnote ⁷

Some six centuries before Cervantes's novel was published, the Chinese Buddhist monk and scholar Zan Ning compared 翻譯 (fanyi; “translation”) “to turning over a piece of brocade—on both sides the patterns are the same, only they face in opposite directions” (qtd. in Cheung 199). Though Zan Ning and Cervantes each deploy textile metaphors that appeal to the act of upturning, as Martha P. Y. Cheung observes, the former refrains from vitiating translation, validating it instead as distinct from yet commensurate with the obverse side (187).Footnote ⁸ If we tug at its threads a bit more, however, even Cervantes's simile of the tapestry can be seen to belie its own insistence on representing translation as the degenerated imitation of an original. By studying the handicraft of Renaissance tapestries, Kathryn Vomero Santos calls attention to the fact that they were actually woven from the back, allowing her to argue insightfully that the side with all the knots and loose fibers thus becomes “the site of their creation rather than a symbol of secondary imperfection” (345). In addition to this material referent, she draws on the preface of Leonard Digges's 1622 translation of a Spanish romance, in which Digges offers an alternative reading of the tapestry metaphor popularized by Cervantes: “as the faire out-side could ill be seene, without helpe of the knots within; no more can the fame of a wel-deserving Author be far spred, without the Labor of a translator” (qtd. in Santos 344). Vindicating “the knottie wrong-side” of the tapestry as a prerequisite for the existence of the face typically viewed by spectators, Digges's commentary analogizes translation as indispensable for the global reach of a given author, thereby defying abiding hierarchies of textual creation and anticipating in embryonic form the uneven processes of canonization that Pascale Casanova coins “littérisation,” or how translation consecrates the stature of peripheral literatures in the “world republic of letters” (136). For Santos, attending to “the knots within” foregrounds the craft of translators—as Cervantes does in the Barcelona printing shop—while inviting us to “read backwards toward a history of translation” (352). In the spirit of what Santos calls Digges's “early revisionist reception” (344) of the tapestry metaphor and Cervantes's own gestures toward the generative virtues of translation, I want to take up this invitation to develop a historically informed reading of translation that keeps to the fore its capacity for creative, and sometimes radical, transformation. If for Don Quixote the hidden, disheveled back of a tapestry figures translation in opposition to its front-facing original, then here I seek to uncover translation's own symbolic back side, turning Cervantes's metaphorics of inversion over again to render visible the Gordian knots of back-translation's unseen history.

Backstories of Back-Translation

The children's game of telephone, known in Commonwealth English by the racially freighted name “Chinese whispers,” derives its entertainment value from the distortions that a short phrase undergoes as it is whispered from person to person around a circle, coming back finally to its creator, who then announces the initial message and its often comically nonsensical end result. To achieve the desired effect, the game requires the cumulative deterioration of information as it travels along a circuit of relays and is acted upon by mishearing (or facetious) participants and phenomena like sibilance and ambient noise. A digital analogue of telephone, and by far the most visible display of back-translation today, is the use of Google Translate to generate absurdly garbled versions of instructions, recipes, movie dialogue, and, especially, pop songs. As dozens of catchy videos on YouTube illustrate (Reese), running lyrics through the automated translation service serially across multiple languages returns a parodic shell of the original (and a warning, if unheeded, to students of foreign languages about the perilous temptations of machine translation).Footnote ⁹

Though seemingly ready-made for the Internet age, experiments with back-translation increasingly pop up on the printed page as well. The project Twin Spin “radically” translates a compilation of Shakespearean sonnets into German before shuttling them, laden with technologically infused neologisms, portmanteaus, and anachronisms, into English once more (Shakespeare). The inventive anthology Multiples, edited by Adam Thirlwell, compounds this premise with a dozen short stories by the likes of Franz Kafka, Søren Kierkegaard, and László Krasznahorkai, all retranslated four to six times through such languages as Urdu, Hebrew, Dutch, Swedish, Icelandic, and Arabic. What the book's inside cover teases as the “complicating factor,” akin to aural fallibility in the game of telephone, is that the translators’ uneven expertise in the source languages deals the stories remarkably dissimilar fates—some emerge from the trial relatively intact, and others undergo marvelously Frankensteinish transformations (Thirlwell). Writers like Sawako Nakayasu have engaged with back-translation in other formally diverse, provocative ways, and their work, while at home at the fringes of the avant-garde or what Adriana X. Jacobs labels “extreme translation,” is beginning to earn its critical due.Footnote ¹⁰

Yet if we work backward from these experimental avatars of “born-translated literature,” we can unearth in literary history ample, albeit scattered, evidence of back-translation as a no less stimulating practice. Murasaki Shikibu's early-eleventh-century Tale of Genji received in the 1920s and 1930s its first full treatment in English by Arthur Waley, whose rendition, known as much for its revisionary license as for its stylistic refinement, has been a fountainhead for indirect translations ever since. Its towering stature, together with the fact that Murasaki's text necessitates modernization for all but specialists, led to petitions that Waley's translation furnish the undercarriage for Genji's transport back into Japanese, a venture realized in 2008 with a translation touted as Ueirī-ban Genji monogatari (Waley's Genji [Samata]). As Michael Emmerich has elucidated, the unorthodox drive to back-translate a keystone of Japanese cultural heritage was impelled by broader, translocal forces of canon formation, the aspiration to stake Genji's claim, now largely undisputed, as a venerated classic of world literature (315–42).

The most notorious example of back-translation is probably Mark Twain's “The Jumping Frog: In English, Then in French, and Then Clawed Back into a Civilized Language Once More by Patient, Unremunerated Toil,” first published in 1875, a decade after the short story that had granted him his first brush with fame. The impetus to translate his own tale back into English was motivated less by his nose for satirical deadpan—of which his intractably stilted, word-for-word translation has some measure—than by a nascent animosity toward the French, and especially his zeal to avenge a withering review that accompanied the translation of the folksy “Jumping Frog” in the urbane Parisian monthly Revue des deux mondes (Bentzon). Twain's complaint that the “unlucky experiment” had failed to capture his trademark humor is disingenuous, however (3). Though it declines to reproduce his cacographical vernacular, the French translation neither betrays its source nor originates the syntactic calques, solecisms, macaronics, and loanwords that beleaguer his absurdly literal transcription back into English. Never has an author aired grievances against a translator with such painstaking verve, but a few—from Erika Mann to Haruki Murakami and Lao She to Vladimir Nabokov—have had their writing, for sundry yet typically more benign motives, translated back into its native language, whether by their own or another's hand.

Before modernity, and particularly in the plurilingual age of Cervantes, the conditions favoring back-translation were in many ways even riper, above all because of two avenues of preserving Latin and Greek in the midst of the blossoming European vernaculars. The first, a systematic program known as double translation to train students in a foreign tongue, consisted of translating a text into another language and then, after a period of time, translating that draft back into the original language without consulting the source. Though Roger Ascham is often credited with developing its praxis, it had already been described at some length by the Spanish humanist Juan Luis Vives in 1531: “as soon as they have learnt syntax, let the pupils translate from the mother-tongue into Latin, and then back again into the mother-tongue” (113–14). Ascham outlines in The Scholemaster (1570) a similar process but inverts the order of Latin and the native vernacular (26). In 1612 John Brinsley the Elder, who wrote several language-instruction manuals based on translation, likened double translation, in a beguiling analogy, to a round-trip voyage: “for there is the same way from Cambridge to London, which was from London to Cambridge” (104–05). The method was, according to Ascham, successful in cultivating a reputed mastery of classical languages for his student, the future Queen Elizabeth, who “onely by this double translating of Demosthenes and Isocrates dailie without missing everie forenone, for the space of a yeare or two, hath atteyned to soch a readie utterance of the latin … comparable with her Majestie” (96).

To understand the second principal vector for early back-translations, we must go back further still, to the translatio studii et imperii—the transfer of cultural knowledge from one region of imperial dominance to another—of premodern Europe and the countless indirect translations it yielded. Before the advent of printing, when a source text was inaccessible—because lost, geographically distant, or suspended in a language with few capable readers or lexicographical tools—translators naturally resorted to antecedents written in other languages, a compromise that on occasion could engender unwitting back-translations. Thanks to the Renaissance vogue of Neo-Latin, eager translators sometimes duplicated the labor of their forerunners, unknowingly remitting texts to the language of a preexisting version, as was the case with the Roman des sept sages, translated from French to Latin and back again (Cañizares Ferriz 66), and Alain Chartier's De vita curiali, which rounded the same lap but in reverse (Fery-Hue 11). These reciprocal movements between a Romance language and Latin undo the standard axial grid that pegs translations from one vernacular to another as horizontal and those hailing from a classical language of prestige as vertical (see Stierle), attesting further to back-translation's indifference toward the orthodoxies of dialectical convention.

Yet such phenomena were by no means confined to Europe and the Near East (such as the Greco-Arabic translational hot spot of Baghdad), as indirect translation was both a veritably global tool and a tool of incipient globalization. The intercultural conduit of the Silk Road facilitated East-West commerce not just in material goods but also in ideas, importing Buddhism from the Sanskrit of India to China by way of stopovers in various intermediary languages of Central Asia, and whose return trip could sometimes give rise to a particular class of back-translation (Nattier). Colonialism also laid the conditions for a dominant pivot language that connected peripheral languages with those of other European polities—Spanish in the Philippines, French in Vietnam, Dutch in Indonesia (St. André, “Lessons” 83–84)—which quickened the use of indirect translation and has likely spawned an untold number of back-translations awaiting their own discovery.

Biblical history supplies like examples, beginning with the Septuagint, which made the Hebrew Bible legible to speakers of Old Greek and itself formed a principal node in a vertiginous network of transmission, interpolation, and recension of often fragmented ancient manuscripts. These complexities eventually generated predictable controversies over the concept of religious truth in translation, pitting hebraica veritas, graeca veritas, and latina veritas against one another. While the corpus of Latin translations known as the Vetus Latina were based on the Greek Septuagint, Jerome turned to the original Hebrew in his fourth-century composition of the Vulgate, the authoritative Latin translation promulgated at Trent centuries later as the official Latin Bible of the Catholic Church. Revering the Septuagint as the legitimate word of God, despite its being an interface for indirect translation, Augustine chided Jerome for his recourse to the Hebrew Tanakh, a choice the early translator defended by fretting that scripture would become tainted or watered down “in tertium vas transfusa” (“by the transfer into a third vase”; qtd. in “To Translate” 1145). Such vessels would multiply over the course of the Renaissance and Reformation, when translations into various European vernaculars, boosted by Gutenberg's technological innovation, grew into a dendriform genealogy of textual lineages. Polyglot Bibles allowed readers to trace these wayward circuits of transmission and their ensuing phraseological variations and, in a sense, to visualize the process of back-translation as an assurance of scriptural fidelity. Nevertheless, the Novum instrumentum omne (1516), Erasmus's diglot New Testament whose later editions informed the Luther, Geneva, and King James Bibles, proffers a telling caveat. In a rush to finish the work but missing part of the manuscript of Revelation, he made the irreverent yet expedient decision to back-translate the last six verses into Greek from his own Latin.

Yet well after presses across Europe were daily churning out printed folios by the thousand, indirect translations would continue unabated as eminently practical portals of access for translators and their readers. In the burgeoning market for foreign texts in early modern England, for instance, many came from intermediary languages, even as their title pages frequently failed to acknowledge a work's author, translator, or lingual genealogy. So pervasive was such translingual crossing, splicing, and recombination that one can easily see how a manuscript gone astray could beget unsuspecting back-translations, as could misattributions, whether accidental or voluntary. Along these lines, back-translation has been harnessed as a pragmatic device for comparatively testing the prospect of a preexistent source, source translation, or transmission pattern, or as a reconstructive vehicle for rehabilitating a work in the language of a “missing manuscript,” so romanced by literary historians as to become a trope in its own right. To be sure, the protracted, nonlinear history of back-translation that I have summarized here is punctuated by salient differences—its uses ranging from the ludic and parodic to the pedagogical, technical, historiographical, and ideological; its discovery, concerted or incidental; its genesis, by design or by chance. What I wish to emphasize is that the multifunctional applications of back-translation equip it with an unusual capacity at once to expose and to conceal, to cover one's tracks even as it plants a guidepost for retracing them once more. Not unlike the vexed duality of a Renaissance tapestry, back-translation fuzzies the picture it weaves with the very threads that wind back to its origins.

Perhaps it was this textual mutability and incertitude that inspired the humanist Girolamo Bargagli to describe in the mid–sixteenth century a parlor game whose premise bears a striking resemblance to the modern game of telephone (15). Of course, such historical ambiguities as covert translation, pseudotranslation, and “found” manuscripts themselves underpin the fictional world of chivalric romance and the fodder for Cervantes's parody: after the story ends abruptly in part 1, chapter 8, the “author” embarks on a desperate search for the missing continuation, which he finds among some papers in a Toledan market and hires a young “morisco aljamiado” (bilingual Moor) to translate from Arabic (Cervantes, Don Quijote 1: 118). The story we read from then on is thus the translation of the “true history” of Don Quixote, written by the Arab pseudo-historian Cide Hamete Benengeli and, owing to other cryptic, metaleptic references throughout the novel, further mediated by an array of secondary and tertiary authors, narrators, versions, and chronicles. And—in the editions of the novel in languages other than Cervantes's Spanish—translators, which brings us back to Lin Shu.

From Lin to Cervantes and Back

Lin cut a rather quixotic figure. In a career that straddled the turbulent period of Chinese history between the late Qing dynasty and the early republic, he championed time-honored literary values just when many of his compatriots were seeking to throw off the shackles of an imperial, Confucianist past. While the revolutionary zeitgeist of the New Culture and May Fourth movements ushered in the widespread adoption of the modern vernacular in literary works, Lin defended the value of 古文 (guwen; “classical Chinese”) in his prolific translations of Western literature, which in turn slaked a growing bourgeoisie's thirst for cultural products from overseas. His methods, too, were anachronistic, relying on oral interpreters to decipher the foreign languages in which he was unread. Yet, as Michael Gibbs Hill explains, despite wielding these purportedly “broken tools” (25), Lin and his team pioneered new techniques of tandem translation (對譯; duiyi) that eschewed both the exploitative legacy of the practice in missionary and colonial contexts and the main alternative that had prevailed up to that point, the indirect translation of Western texts through Japanese. The political thinker Kang Youwei summarized the utility of the latter procedure with a canny agricultural metaphor: “the West will serve as the ox, Japan will be the peasant, and we [China] will sit back and eat [the harvest]. With little expense, all of the most important books will be available to us” (qtd. in Hill 35). In the face of a waning body of translators with the expertise needed to work from Japanese, Lin and company's collaborative model proved at least equally fertile by showing that the nonnative seeds of Western literature could germinate and thrive in the ancient soil of guwen (Hill 48). And, as the translators drew upon literary commentary (評點; pingdian) and sequel writing (續書; xushu), traditions that “were always enmeshed within a dialectic of transmitting and innovating, between reproducing and remaking a text both in spirit and at the level of the word,” those seeds yielded curious new hybrids (Hill 35–36).

Transplanted back into the Spanish of its origin, Cervantes's Don Quixote still drives the narrative, but Lin's Moxia Zhuan is a manifest presence, and, though more recessive, so are the phylogenetic traits it inherited from earlier strains. Transliterations and mistranslated homographs, homophones, or polysemes become conspicuous markers of the text's early pedigree, shibboleths or lexical bread crumbs for tracing the route back to the English of Motteux and Jervas and, in some cases, all the way to Cervantes's Spanish. In the author's partially autobiographical story “The Captive's Tale,” interpolated into Don Quixote, translation proves vital for the protagonist's ability to navigate the interreligious pressures of the Algerian bagnios and mount an escape to freedom with the aid of his future bride, Zoraida. While narrating an exceptionally tense moment, the captive remarks that Zoraida understood Sabir, the historical Mediterranean pidgin that originated the term lingua franca, already lexicalized in English when Motteux composed his translation. However, Lin's transliteration of it as lingua fulanka (387n569) would have struck most of his readers, not without sufficient irony, as incomprehensible. Another idiosyncrasy emerges when Zoraida lavishes on the captive a coin worth “dos coronas en la moneda española” (“two crowns in Spanish currency”; 378, 378n558), even though the corona (“crown”) was never Spanish legal tender. The incongruity stems from the challenge of translating the phrase “diez reales de los nuestros” (“ten reales of ours”) uttered by the Spanish captive to explain to his intradiegetic listeners (and Cervantes's domestic readership) a foreign monetary unit. Motteux confronts this crux with the hypernaturalizing “two crowns of our money,” setting for Lin the trap of an equivocal first-person possessive pronoun coupled with decidedly British coinage. Typographic liabilities also become more discernible as they snowball down the slope of subsequent versions and Relinque scrupulously translates them back into Spanish (164, 370), while other felicitous faux pas crop up from Lin and Chen's own process of oral dictation—namely, its susceptibility to transposing Chinese homophones like jian, which could mean either 劍 (“double-edged sword”) or 鑑 (“ancient bronze mirror”; Relinque 85, 85n98)—and recall once more the blithe upshots of the game of telephone.

The most prominent lexical slippage befalls the priest from Don Quixote's village, who makes his way into English as the cognate curate but, starting in chapter 5, transforms into a 醫生 (yisheng; “physician”) in Chinese (translated by Relinque as médico [“doctor”]), likely because of the graphic proximity of curate to cure (which happens to be homonymic in the Spanish cura [Relinque 36–37, 79n84]). Rocinante undergoes a more subtle metamorphosis from a “rocín flaco” (“nag”) in Cervantes to a 駿馬 (junma; “steed”) in Lin to a “caballo veloz” (“fast horse”) in Relinque, a permutation she ascribes to Motteux's and Jervas's use of “lean,” an adjective that can signify scrawny and malnourished as much as fit and toned (Relinque 51, 51n4). Such nuances produce ripple effects detectable not solely across the various translations but, naturally, in other aspects of the narrative as well. With a racing stallion instead of a work horse, Don Quixote becomes, as early as the second sentence of the novel, transfigured. Receiving from Lin the cognomen “Quisada” and thus recalling Leo Spitzer's classic study of the “polyonomasia” or “linguistic perspectivism” (41) already baked into Cervantes's text, the protagonist likewise moves nearer to an object of veneration than of derision, exchanging unhinged derangement for a more fully developed sentimental ethos. Suitably, his relationship with Sancho Panza veers toward one of 師弟 (shi di; “teacher and disciple”), at least in part because of the challenge of translating escudero (“squire”) into a cultural tradition to which knight errantry was historically alien. Though it could be tempting to impute Lin's choice to discount the many Chinese words for “servant” wholly to a Confucianist worldview that revered teaching, filial piety, and respect for one's elders, Relinque suggests that this, too, may proceed from the fact that in English master, an honorific applied to Don Quixote, can indicate a relationship of either tutelage (maestro) or hierarchy (amo [Relinque 35–36n29, 65n57, 69n65]). Wanton domestication is sometimes but the spectral guise of erstwhile ambiguity.

Yet scarcely had the sinuous, inky strokes of Lin's classical Chinese characters dried than his fellow translator Zhou Zuoren excoriated Moxia Zhuan for what he deemed grievous oversights and unconscionable liberties (47–49; see also Relinque 26, 33, 147n209). His was not the only scathing review that objected, among other things, to the reduction or outright elimination of certain plot details, poems, proper names, and even Cervantes's own prologue. Modern critics have questioned more generally whether Lin and his collaborators’ work merits the status of translation proper, qualifying it instead as “rewriting” (Chan 16) or marking it by throwing scare quotes around the word translation (St. André, “Lessons” 76). Though I wish neither to neutralize the text's productive capability to muddy the distinctions between translation and terms like adaptation and re-creation—boundaries that have forever been porous, dynamic, and culturally contingent—nor to flatten the philosophical and methodological diversity of textual creation in China (Chan; Cheung; Shei and Gao; St. André and Peng), I take Moxia Zhuan definitively as a translation. Unlike, say, the Restoration playwright Thomas D'Urfey's Comical History of Don Quixote (1694) or Dale Wasserman's musical Man of La Mancha (1965)—to name but two intersemiotic adaptations that have been translated “back” into Spanish—Moxia Zhuan does not advertise itself as anything other than a translation, and thus it aligns with the primary criterion of other inclusive yet historically responsive definitions of the practice (Hosington et al.). This approach has the added benefit of attending to debates on translation that in late Qing and early Republican China were at once more nuanced—for instance, the stylistic gradations between liberal “free translation (自由譯 ziyouyi)” and more literal “straightforward translation style (直譯 zhiyi)” (Dai 112)—and inseparable from wider social currents. Zhou and others’ early critique of Lin's work is symptomatic of the reality that translations of Western literature had become an ideological touchstone, and the preservation of foreignness therein, an improbable yardstick for their translators’ yen for domestic reform.

Amid the tumult of this polarization, Lin, born into poverty and deprived of formal training, confessed in other writings that his translations were “plagued by errors that pass unnoticed” (qtd. in Relinque 26; see also Chan 118n41), even as the uncertain provenance of Moxia Zhuan's polysemic slippages might be seen to mitigate its own answerability. Yet if to accept Lin's self-attributed “errors” as such would be to reinscribe the very deficit-based norms of translation that I am suggesting his text prompts us to recalibrate, then to disavow them altogether would be to dispossess the translator of the generative authority upon which such a project just as vitally rests. With these twin pitfalls in mind, and taking a cue from Cervantes's and Lin's own penchant for creative renaming, I would propose that its slippage errors make Lin's text errant: straying from the well-worn path of fidelity while, like Don Quixote, roving adventurously, traveling freely without a fixed course or destination (evoking the term's own etymological entanglements with the Latin errāre and iterāre).Footnote ¹¹ Errant translation, then, skips the guardrails of orthodoxy for the unexpected encounters born of traversing a vaster, more capacious semantic field. Of course, as Cervantes's knight errant substantiates in archetypal fashion, adventure always risks misadventure, or what in a translative idiom Berman would call défaillances, alternately translated as “failing” (Lane, “Literary Back-Translations” 308), “faltering,” “deficiency,” “defaults,” “shortcomings,” and, in Lane's recent attempt to redeem the practice of retranslation, the less disciplinizing “struggle” (309). In the conventional interpretation of Cervantes's tapestry, these might figure as its most tangled and unseemly strands. But such défaillances are precisely what endorse Moxia Zhuan's value for back-translation (and, one presumes, what make Relinque's own text commercially marketable, albeit to the relatively small, specialized audience of a young Spanish boutique press). To affirm the peculiar jouissance of the knots tied by its errancy is not to fetishize the text but to unsettle the punctilio of normative prescriptions for “successful” translation.

Lest we brook a double standard, we should remember that Cervantes, too, was criticized for the mistakes that stole their way into the princeps edition of Don Quixote, most infamously the disappearance (and subsequent, mystifying reappearance) of Sancho's donkey. This has been the albatross of many a translator—emend an erratum, you betray the author; leave it intact, you risk its being ascribed to yourself; footnote it, and you magnify it, again betraying the author—but Lin is decisive in remedying extant lapses in Cervantes's text, even when they have “passed unnoticed” by his English translators. Rather than cast judgments about their propriety, I take Lin's corrective interventions as part of a larger interplay of mutual emendation along the back-translational loop from Cervantes to Relinque, thereby upending traditional hierarchies of authorship while underscoring the translator's creative agency. Nowhere is this agency more notable than in several digressive glosses that Lin intersperses throughout the body of the text and that Relinque reproduces parenthetically in Spanish, ranging from strictly explanatory statements or innocuous musings—“Este pensamiento es particularmente extraño” (“This thought is particularly strange”; 197)—to comparisons with Chinese customs, folklore, political history, or current events. For example, amid the emotive scene when Agi Morato is reproaching his apostate daughter, Zoraida, for abandoning him for European Christianity, Lin interjects into the text an observation about how much the characters’ plight resembles an ancient legend immortalized in a poem by Yuan Mei (Relinque 392, 392n574). In the middle of a metafictional discussion among Quisada, the magistrate (transformed from the ecclesiastical canon in Cervantes's novel), and the priest-cum-physician, in which the latter demands more robust governmental control over the content of books and plays, Lin similarly injects an allusion to nascent bids for literary censorship in early-twentieth-century China (Relinque 440; 440n630). Another topical reference surfaces in a conversation between Quisada and Sancho, in which the knight reassures his squire-disciple that the promise of a conquered territory for him to govern will yet materialize, but that he should exercise patience because worldly change cannot happen overnight. Here Lin intervenes again, writing, “Tomemos el ejemplo de la revolución, tan imponente, tan grandiosa, y según yo lo veo, es algo que siempre ha sucedido” (“Let us take the example of the revolution, so imposing, so grandiose, and, as I see it, it is something that has always taken place”; Relinque 91). This is probably the most patent act of editorializing by the Chinese translator, who otherwise tends to avoid broadcasting his subject position or projecting overt political ideologies onto the text, even as a multitude of subtler cases betrays the influence of his own worldview and cultural milieu.

What interests me, in any event, is less the content of such utterances than their meta- or extranarrative status as an expression, once again, of the translator's agency, exerted reciprocally along both directions of the translational feedback loop. Most remarkable about Lin's varied interventions is the degree to which they embody those of Cervantes's fictional aljamiado translator and those of his pseudo-historian Cide Hamete Benengeli, not to mention more rigorously historical conditions of interlingual transaction and textual production. The prodigious medieval Toledo School of Translators relied on oral interpreters of Arabic to articulate in Romance what would be set down by scribes versed in Latin,Footnote ¹² and trujamanes—a term derived from Arabic to denote a dragoman or intermediary whose vocation could also encompass that of translator—plied their trade in Iberia up until Cervantes's day (Roser Nebot; Tolosa Igualada 82–85). Moriscos, like the one encountered in Toledo's Alcaná marketplace early in Don Quixote, frequently served as trujamanes, and this character's labor in translating Cide Hamete's rediscovered manuscript, like Lin and his team's own collaborative methods, would likely have been carried out in an oral register (Villar Lecumberri 838).

As the novel progresses, however, Cide Hamete becomes more conspicuous, and the translator increasingly asserts his own authority over the text. Early in part 2, the translator judges Sancho's diction to be implausibly sophisticated and therefore concludes that the fifth chapter must be apocryphal. In order to fulfill his duty, he translates the chapter anyway but within a few more pages starts to redact and adapt the text more freely (723). When Don Quixote visits the home of Don Diego de Miranda in chapter 18, readers are told that the “author” included a thick description of the domestic interior, minutiae the translator decides to omit “porque no venían bien con el propósito principal de la historia” (“because they did not go along well with the main purpose of the history”; 842). In a satirical treatment of religious oaths in chapter 27, we find him glossing a flippant vow by the Muslim Cide Hamete (“Juro como católico cristiano” [“I swear as a Christian Catholic”; 934]), whose intention, the young translator explains, was merely to denote a commitment to the truth of Don Quixote's exploits as if his historian were a devout Christian. And by chapter 44 he is exercising full creative license over the text, editing at will, even if spurred by fatigue and a nagging resentment of the subject matter:

This passage has long confounded critics, beginning with the first ever annotated edition of Don Quixote, edited by Diego Clemencín, who spurned the passage as an “algaravia que no se entiende” (“incomprehensible hullabaloo”) that itself should have been discarded (368–69). Yet, without claiming to resolve the enigma of its apparent logical fallacies (how can the “original” record the actions of its translator?) and equivocation over who precisely is doing what—which are nonetheless in keeping with the authorial binds of other parts of the novel—these lines make more sense when read in the light of the built-in ambiguities of the early modern autor and alongside preceding examples that likewise blur the distinction between author and translator, source text and translation. As with the metaphor of the tapestry, it behooves us to recall that without the translation—the knotty back side—the story would have remained, at least from the ninth chapter on, unread and as good as unwoven. With a self-sufficient translator who is unsatisfied to linger in the background and who, by way of edits, glosses, and parings, makes incessant incursions into the front-facing side, Cervantes showcases translation as a chiefly sovereign site of creation as well as contestation.

Doubling Back: Translators and Their Doppelgängers

It is a fittingly Cervantine irony not only that Lin never published a translation of part 2 of the novel, in which the work of translators gains even greater visibility, but also that he excised Cide Hamete and the Moorish translator entirely from Moxia Zhuan. It is almost as though he felt no need to retain them because of a consciousness that he and Chen Jialin would somehow occupy their place, a prescience that their own creative translation, with all its first-person interventions, would either suffice to fulfill, in those characters’ absence, the story's metanarrative thrust or, were they to be preserved, exceed it. One could even argue that such liberties make Moxia Zhuan innately more faithful to the spirit of Cervantes's Don Quixote. I say this not to perpetuate increasingly outmoded, absolute models of normative translation criticism but to register that they were already being consciously defied and superseded long before our present moment, even as back-translation's circularity would stymie the thrust of such progressivist teleologies. Accordingly, and contrary to what some might desire or tacitly assume about the academic or commercial animus behind the Spanish back-translation, it bears emphasizing that Relinque's work is no more derivative than that of her predecessors. To resolve the laborious difficulty of converting the classical Chinese, a language meant strictly for reading, into the predominantly oral and dialogic register of the narrative in Spanish, she opted for what she classifies as an “estilo literario neutro” (“neutral literary style”) infused with “algunos elementos … arcaizantes” (“some archaizing elements”; 40). Unlike the bulk of back-translational practices in the premodern era, Relinque's process of what Karen Emmerich might call “translingual editing” (2) involved actively consulting Cervantes's source text, which aided in reconstructing Lin's onomastic omissions and parsing pronominal forms of address held in abeyance by the classical Chinese (41). Whether or not we agree with these more discrete choices, they disclose the mark not just of a glossator but of an interpreter and stylist as well. Hence, one can detect on a performative plane of Relinque's text the interventionist traces of Lin, Jervas, Motteux, and the aljamiado translator even as its journey back into the Spanish of Cervantes threatens to efface, or at least to dilute, their own distinctiveness.

This brings us back around to what Coldiron characterizes as “Schleiermacher's Janus-like dilemma” (“Visibility Now” 198), the dichotomy of domestication and foreignization with which this essay began. True to this binary formulation, Motteux's translation was faulted for its affected use of cockney slang, a domestic style if ever there was one (Stavans 184–86); Jervas accused Motteux of “loose paraphrase” “foreign to the design of [Cervantes]” (“Translator's Preface” iv); while Lin's “translation practice” has been branded by Venuti as “thorougly domesticating” (Scandals 179). Given his imperial nostalgia and fealty to classical Chinese, Lin thus serves for Venuti as an ideal foil to Zhou and his brother Lu Xun, who promoted their own translations of Western literature as explicitly foreignizing,Footnote ¹³ of a piece with their sympathies for the modernizing impulse of the republic. Leaving aside the fact that such narratives sometimes dovetail all too handily with the political ideologies of figures for whom the historical record is seldom transparent or disinterested,Footnote ¹⁴ it is true that Lin's translation marshals discursive strategies that Venuti would deem “domesticating,” including its abundant use of 成語 (cheng yu; “traditional idioms”), the decorous suppression of certain mores, and the assimilation of European cultural and religious references to Confucianism, Taoism, or Buddhism. Importantly, Venuti concedes that “domesticating strategies … can still result in a powerful hybridity that initiates unanticipated changes” and that Lin's translations of Charles Dickens, Alexandre Dumas fils, and Rider Haggard are “at once domestic and foreign, Chinese and Western” (Scandals 182; see also Dai). Yet it is not at all clear how Venuti would account for a text like Moxia Zhuan, much less Historia del caballero encantado, engendered of intermediary translations of Cervantes's already hybridized, pseudotranslated novel. For how is one to winnow the domesticating discourse the story acquired when it was Englished from that which newly manifested when it was Sinicized? With Relinque's text and the cumulative translational chain of which it forms the latest link, who is to say where one nativizing act ends and its alienizing corollary begins? As a domesticating or foreignizing motif travels around this chain, does it accumulate, get mutually reinforced, or self-correct? Even if we can track, piecemeal, the evolution of “domestic” or “foreign” terms as they move and morph across different translations, a practice for which back-translation historically has proven useful, a holistic analysis of the sort Venuti himself has long clamored for remains more elusive (Translator's Invisibility 19). Unlike retranslation—the much more thoroughly studied concept by which two or more translations are produced from the same source text, thus easily inviting comparison among them—back-translation makes us contend with preexisting interference that obfuscates, in Thomas O. Beebee's words, the “X-ray of the otherwise invisible process of interpretive drift” that standard, one-way translations afford (123). To borrow an aural analogy, the signal arrives home to its source language always mediated by the noise of previous translations; the signal-to-noise ratio increases in relation to the number of intermediary translations in between, and it does so under perpetual threat of degradation and entropy. Yet, as with the game of telephone, back-translation's susceptibility to loss is also its gain.

In this sense, Relinque's Historia represents a conceptual inversion of Jorge Luis Borges's “Pierre Menard, autor del Quijote,” the short story that Steiner anointed as “the most acute, most concentrated commentary anyone has offered on the business of translation” (73). While Menard's compulsive fantasy of writing another Don Quixote verbally identical to Cervantes's text draws its novelty wholly from the contextual vicissitudes wrought by chronological and cultural distance, Relinque's Historia is animated by the collective transmutations of its forebears. This means, however, that each text accrues value by virtue of displacements across time and space, of the transportation into a foreign milieu that, according to the story's sly narrator, makes Menard's version “casi infinitamente más rico” (“almost infinitely richer”; Borges 449), enabling both Borges's and Relinque's texts to incite, from opposite flanks, provocations about origins, rewriting, and the paradoxes of translating monolingually. But the most remarkable parallel, because couched in the terms of a characteristically inscrutable, apparently marginal, and therefore assuredly meaningful detail, is Borges's oblique reference to back-translation. In a footnote to the inventory of Menard's body of work, the bereaved narrator reports that he has failed to locate “una versión literal de la versión literal que hizo Quevedo de la Introduction à la vie dévote de San Francisco de Sales” (“a literal translation [into French] of Quevedo's literal translation of Saint Francis de Sales's Introduction à la vie dévote”), ascribing its absence in Menard's library to “una broma de nuestro amigo, mal escuchada” (“one of our friend's jokes that we must have misinterpreted”; 446). The ambivalence of Menard's jest hinges on whether his hypothetical back-translation, despite being “literal,” would diverge from Introduction à la vie dévote, or whether it would match Sales's text to the letter and thereupon replicate the outcome of Menard's grander, quixotic enterprise. Either way, Borges, whose own translations notoriously enfold “interpretive recreation” (Kristal 51) and “creative infidelity” (Basile 23n2) and who, like Lin Shu, readily translated translations of works written in languages of which he had no knowledge, appears to ironize as much the pieties of translational literalism as the palimpsestic follies of interlingual exchange. Pushed further, these maneuvers might even urge us to fancy, in true Borgesian fashion, the fate of literary works that would circumnavigate infinitely recurrent concatenations, hypertextual Möbius strips by which a back-translational loop, after amassing myriad languages, is closed only to commence the process anew, weaving with its coiling filaments an immense tapestry to be hung from the walls of a Babelian library. Like its extravagant eponymous character, “Pierre Menard” underscores, in microcosm, the whimsical futility yet exceptional critical paybacks of back-translation.

Without discounting these ludic assets, if we take back-translation seriously as a literary-historical practice, our disciplines too have a good deal to gain. The obstacles that back-translation throws up to pinpointing the origin of a given translational style or strategy trouble the very concept of originality, thus betraying a dormant capacity to intervene in world literary concerns around alterity, authorship, appropriation, transculturation, (un)translatability, and adjacent fields of translation criticism and postcolonial translation studies, to name just a few. The errant, deterritorializing impulses of the round-trip, multidestination journey of back-translation—one of literary émigrés repatriated to find that they, their homeland, and their home language are not quite the same as before—likewise press us to renegotiate the binaries of fidelity and betrayal, center and periphery, vertical and horizontal, inter- and intralingual, translation and adaptation. As a recursive, uroboric loop, back-translation moves us back but also forward, affording glimpses of new translational horizons. This is not to erect either an aporia by which the asymmetries of these polar dynamics become hopelessly indeterminate or a utopia in which they are rendered inert but, rather, to suggest that the past and futures of back-translation may open a productively different vantage from which to unravel them. Back-translation, with its radical potential to reveal and obscure, invites us to peer, anamorphically and defiantly, at the tapestry's back side.