TRANSLATION AND THE TRIALS OF THE FOREIGN
Translated by Lawrence Venuti
THE GENERAL THEME of my essay will be translation as the trial of the foreign (comme epreuve de l'etranger). "Trial of the foreign" is the expression that Heidegger uses to define one pole of poetic experience in Hölderlin (Die Erfahrung des Fremden). Now, in the poet, this trial is essentially enacted by translation, by his version of Sophocles, which is in fact the last "work" Hölderlin published before descending into madness. In its own time, this translation was considered a prime manifestation of his madness. Yet today we view it as one of the great moments of western translation: not only because it gives us rare access to the Greek tragic Word, but because while giving us access to this Word, it reveals the veiled essence of every translation.
Translation is the "trial of the foreign." But in a double sense. In the first place, it establishes a relationship between the Self-Same (Propre) and the Foreign by aiming to open up the foreign work to us in its utter foreignness. Hölderlin reveals the strangeness of the Greek tragic Word, whereas most "classic" translations tend to attenuate or cancel it. In the second place, translation is a trial for the Foreign as well, since the foreign work is uprooted from its own language-ground (sol-de-langue). And this trial, often an exile, can also exhibit the most singular power of the translating act: to reveal the foreign work's most original kernel, its most deeply buried, most self-same, but equally the most "distant" from itself. Hölderlin discerns in Sophocles' work - in its language - two opposed principles: on the one hand, the immediate violence of the tragic Word, what he calls the "fire of heaven," and on the other, "holy sobriety," i.e., the rationality that comes to contain and mask this violence. For Hölderlin, translating first and foremost means liberating the violence repressed in the work through a series of intensifications in the translating language - in other words, accentuating its strangeness. Paradoxically, this accentuation is only a way of giving us access to it. Alain addressed the topic of translation in of his remarks on literature:
I have this idea that one can always translate a poet — English, Latin, or Greek — exactly word for word, without adding anything, preserving the very order of the words, until at last you find the meter, even the rhymes. I have rarely pushed the experiment that far; it takes time, I mean, a few months, plus uncommon patience. The first draft resembles a mosaic of barbarisms; the bits are badly joined; they are cemented together, but not in harmony. A forcefulness, a flash, a certain violence remains, no doubt more than necessary. It's more English than the English text, more Greek than the Greek, more Latin than the Latin [. . .]
(Alain 1934: 56-7)
Thanks to such translation, the language of the original shakes with all its liberated might the translating language. In an article devoted to Pierre Klossowski's translation of the Aeneid, Michel Foucault distinguishes between two methods of translation:
It is quite necessary to admit that two kinds of translations exist; they do not have the same function or the same nature. In one, something (meaning, aesthetic value) must remain identical, and it is given passage into another language; these translations are good when they go "from like to same" [. . .] And then there are translations that hurl one language against another (...) taking the original text for a projectile and treating the translating language like a target. Their task is not to lead a meaning back to itself or anywhere else; but to use the translated language to derail the translating language.
(Foucault 1969: 30)
Doesn't this distinction simply correspond to the great split that divides the entire field of translation, separating so-called "literary" translations (in the broad sense) from "non-literary" translations (technical, scientific, advertising, etc.)? Whereas the latter perform only a semantic transfer and deal with texts that entertain a relation of exteriority or instrumentality to their language, the former are concerned with works, that is to say texts so bound to their language that the translating act inevitably becomes a manipulation of signifiers, where two languages enter into various forms of collision and somehow couple. This is undeniable, but not taken seriously. A superficial glance at the history of translation suffices to show that, in the literary domain, everything transpires as if the second type of translation came to usurp and conceal the first type. As if it were suddenly driven to the margins of exception and heresy. As if translation, far from being the trials of the Foreign, were rather its negation, its acclimation, its "naturalization." As if its most individual essence were radically repressed. Hence, the necessity for reflection on the properly ethical aim of the translating act (receiving the Foreign as Foreign). Hence, the necessity for an analysis that shows how (and why) th' aim has, from time immemorial (although not always), been skewed, pervert') and assimilated to something other than itself, such as the play of hypertextual transformations.
I propose to examine briefly the system of textual deformation that operates in every translation and prevents it from being a "trial of the foreign." I shall call this examination the analytic of translation. Analytic in two senses of the term-detailed analysis of the deforming system, and therefore an analysis in the Cartesian sense, but also in the psychoanalytic sense, insofar as the system is largely unconscious, present as a series of tendencies or forces that cause translation to deviate from its essential aim. The analytic of translation is consequently designed to discover these forces and to show where in the text they are practiced - somewhat as Bachelard, with his "psychoanalysis" of the scientific spirit, wanted to show how the materialist imagination confused and derailed the objective aim of the natural sciences.
Before presenting the detailed examination of the deforming forces, I shall make several remarks. First, the analysis proposed here is provisional: it is formulated on the basis of my experience as a translator (primarily of Latin American literature into French). To be systematic, it requires the input of translators from other domains (other languages and works), as well as linguists, "poeticians" and . . . psychoanalysts, since the deforming forces constitute so many censures and resistances.
This negative analytic should be extended by a positive counterpart, an analysis of operations which have always limited the deformation, although in an intuitive and unsystematic way. These operations constitute a sort of counter-system destined to neutralize, or attenuate, the negative tendencies. The negative and positive analytics will in turn enable a critique of translations that is neither simply descriptive nor simply normative.
The negative analytic is primarily concerned with ethnocentric, annexation-ist translations and hypertextual translations (pastiche, imitation, adaptation, free rewriting), where the play of deforming forces is freely exercised. Every translator is inescapably exposed to this play of forces, even if he (or she) is animated by another aim. More: these unconscious forces form part of the translator's being, determining the desire to translate. It is illusory to think that the translator can be freed merely by becoming aware of them. The translator's practice must submit to analysis if the unconscious is to be neutralized. It is by yielding to the "controls" (in the psychoanalytic sense) that translators can hope-to free themselves from the system of deformation that burdens their practice. This system is the internalized expression of a two-millennium-old tradition, as well as the ethnocentric structure of every culture, every language; it is less a crude system than a "cultivated language." Only languages that are "cultivated" translate, but they are also the ones that put up the strongest resistance to the ruckus of translation. They censor. You see what a psychoanalytic approach to language and linguistic systems can contribute to a "translatology." This approach t also be the work of analysts themselves, since they experience translation as n essential dimension of psychoanalysis.
A final point: the focus below will be the deforming tendencies that intervene ■ the domain of literary prose - the novel and the essay.
Literary prose collects, reassembles, and intermingles the polylingual space of community. It mobilizes and activates the totality of "languages" that coexist in language. This can be seen in Balzac, Proust, Joyce, Faulkner, Augusto Antonio Roa Bastos, Joao Guimaraes Rosa, Carlo Emilio Gadda, etc. Hence, from a formal point of view, the language-based cosmos that is prose, especially the novel is characterized by a certain shapelessness, which results from the enormous brew of languages and linguistic systems that operate in the work. This is also
characteristic of canonical works, la grande prose.
Traditionally, this shapelessness has been described negatively, that is, within the horizon of poetry. Herman Broch, for example, remarks of the novel that "in contrast to poetry, it is not a producer, but a consumer of style. [. . .] It applies itself with much less intensity to the duty of looking like a work of art. Balzac is of greater weight than Flaubert, the formless Thomas Wolfe more than the artistic Thornton Wilder. The novel does not submit, like proper poetry, to the criteria of art" (Broch 1966: 68).
In effect, the masterworks of prose are characterized by a kind of "bad writing," a certain "lack of control" in their texture. This can be seen in Rabelais, Cervantes, Montaigne, Saint-Simon, Sterne, Jean Paul Richter, Balzac, Zola, Tolstoy, Dostoevsky.
The lack of control derives from the enormous linguistic mass that the prose writer must squeeze into the work - at the risk of making it formally explode. The more totalizing the writer's aim, the more obvious the loss of control, whether in the proliferation, the swelling of the text, or in works where the most scrupulous attention is paid to form, as in Joyce, Broch, or Proust. Prose, in its multiplicity and rhythmic flow, can never be entirely mastered. And this "bad writing" is rich. This is the consequence of its poly-lingualism. Don Quixote, for example, gathers into itself the plurality of Spanish "languages" during its epoch, from popular proverbial speech (Sancho) to the conventions of chivalric and pastoral romances. Here the languages are intertwined and mutually ironized.
The Babelian proliferation of languages in novels pose specific difficulties for translation. If one of the principal problems of poetic translation is to respect the polysemy of the poem (cf. Shakespeare's Sonnets), then the principal problem of translating the novel is to respect its shapeless polylogic and avoid an arbitrary homogenization.
Insofar as the novel is considered a lower form of literature than poetry, the deformations of translation are more accepted in prose, when they do not pass unperceived. For they operate on points that do not immediately reveal themselves. It is easy to detect how a poem by Hölderlin has been massacred. It isn't so easy to see what was done to a novel by Kafka or Faulkner, especially if the translation seems "good." The deforming system functions here in complete tranquillity. This is why it is urgent to elaborate an analytic for the translation of novels.
This analytic sets out to locate several deforming tendencies. They form systematic whole. I shall mention twelve here. There may be more; some combine with or derive from others; some are well known. And some may appear relevant only to French "classicizing" translation. But in fact they bear on all translating at least in the western tradition. They can be found just as often in English translators as in Spanish or German, although certain tendencies may be more accentuated in one linguistic-cultural space than in others. Here are the twelve tendencies in question:
ennoblement and popularization
the destruction of rhythms
the destruction of underlying networks of signification
the destruction of linguistic patternings
the destruction of vernacular networks or their exoticization
the destruction of expressions and idioms
the effacement of the superimposition of languages
This bears primarily on the syntactical structures of the original, starting with that most meaningful and changeable element in a prose text: punctuation. Rationalization recomposes sentences and the sequence of sentences, rearranging them according to a certain idea of discursive order. Wherever the sentence structure is relatively free (i.e., wherever it doesn't answer to a specific idea of order), it risks a rationalizing contraction. This is visible, for instance, in the fundamental hostility with which the French greet repetition, the proliferation of relative-clauses and participles, long sentences or sentences without verbs - all elements essential to prose.
Thus, Marc Chapiro, the French translator of the Brothers Karamazov, writes:
The original heaviness of Dostoevsky's style poses an almost insoluble problem to the translator. It was impossible to reproduce the bushy undergrowth of his sentences, despite the richness of their content.
(cited by Meschonnic 1973: 317)
This signifies, quite openly, that the cause of rationalization has been adopted. As we have seen, the essence of prose includes a "bushy undergrowth." Moreover, every formal excess curdles novelistic prose, whose "imperfection" is a condition of its existence. The signifying shapelessness indicates that prose plunges to the depths, the strata, the polylogism of language. Rationalization destroys all that.
]It annihilates another element of prose: its drive toward concreteness. Rationalization means abstraction. Prose is centered on the concrete and even tends to render concrete the numerous abstract elements bobbing in its flood (Proust, Montaigne). Rationalization makes the original pass from concrete to abstract, not only by reordering the sentence structure, but — for example - by translating verbs into substantives, by choosing the more general of two substantives, etc. Yves Bonnefoy revealed this process with Shakespeare's work.
This rationalization/abstraction is all the more pernicious in that it is not total. It doesn't mean to be. It is content to reverse the relations which prevail in the original between formal and informal, ordered and disorderly, abstract and concrete. This conversion is typical of ethnocentric translation: it causes the work to undergo a change of sign, of status - and seemingly without changing form and meaning.
To sum up: rationalization deforms the original by reversing its basic tendency.
This is a corollary of rationalization which particularly concerns the level of "clarity" perceptible in words and their meanings. Where the original has no problem moving in the indefinite, our literary language tends to impose the definite. When the Argentine novelist Roberto Arlt writes: "y los excesos eran dcsplazados por desmedimientos de esperanza" (the excesses were displaced by the excessiveness of hope; Arlt 1981: 37), French does not tolerate a literal rendering because everywhere, in this passage from Los Siete Locos, excess is still in question. French asks: an excess of what?
The same goes for Dostoevsky. Chapiro writes: "To render the suggestions of a Russian sentence, it is often necessary to complete it" (cited by Meschonnic 1973: 317-18).
Clarification seems to be an obvious principle to many translators and authors. Thus, the American poet Galway Kinnell writes: "The translation should be a little clearer than the original" (cited by Gresset 1983: 519).
Of course, clarification is inherent in translation, to the extent that every translation comprises some degree of explicitation. But that can signify two very different things:
(1) the explicitation can be the manifestation of something that is not apparent, but concealed or repressed, in the original. Translation, by virtue of its own movement, puts into play this element. Heidegger alludes to the point for philosophy: "In translation, the work of thinking is transposed into the spirit of another language and so undergoes an inevitable transformation. But this transformation can be fecund, because it shines a new light on the fundamental position of the question" (Heidegger 1968: 10).
The power of illumination, of manifestation, (1) as I indicated apropos Hölderlin, is the supreme power of translation. But in a negative sense, (2) explicitation aims to render "clear" what does not wish to be clear in the original.
The movement from polysemy to monosemy is a mode of clarification. Paraphrastic or explicative translation is another. And that leads us to the third tendency.
Every translation tends to be longer than the original. George Steiner said that translation is "inflationist." This is the consequence, in part, of the two previous tendencies. Rationalizing and clarifying require expansion, an unfolding of what, in the original, is "folded." Now, from the viewpoint of the text, this expansion can be qualified as "empty." It can coexist quite well with diverse quantitative forms of impoverishment. I mean that the addition adds nothing, that it augments only the gross mass of text, without augmenting its way of speaking or signifying. The addition is no more than babble designed to muffle the work's own voice. Explicitations may render the text more "clear," but they actually obscure its own mode of clarity. The expansion is, moreover, a stretching, a slackening, which impairs the rhythmic flow of the work. It is often called "overtranslation," a typical case of which is Armel Guerne's translation of Moby Dick (1954). Expanded, the majestic, oceanic novel becomes bloated and uselessly titanic. In this case, expansion aggravates the initial shapelessness of the work, causing it to change from a shapeless plenitude to a shapeless void or hollow". In German, the Fragments of Novalis possess a very special brevity, a brevity that contains an infinity of meanings and somehow renders them "long," but vertically, like wells. Translated by the same Guerne (1973), they are lengthened immoderately and simultaneously flattened. Expansion flattens, horizontalizing what is essentially deep and vertical in Novalis.
This marks the culminating point of "classic" translation. In poetry, it is "poetization." In prose, it is rather a "rhetorization." Alain alludes to this process (with English poetry):
If a translator attempts a poem by Shelley into French, he will first spread it out, following the practice of our poets who are mostly a bit too oratorical. Setting up the rules of public declamation as his standard, he will insert their thats and whichs, syntactical barriers that weigh upon and prevent — if I can put it this way — the substantial words from biting each other. I don't disdain this art of articulation. . . . But in the end it isn't the English art of speaking, so clenched and compact, brilliant, precise and strongly enigmatic.
(Alain 1934: 56)
Rhetorization consists in producing "elegant" sentences, while utilizing the source text, so to speak, as raw material. Thus the ennoblement is only a rewriting, a "stylistic exercise" based on - and at the expense of - the original. This procedure is active in the literary field, but also in the human sciences, where jt produces texts that are "readable," "brilliant," rid of their original clumsiness and complexity so as to enhance the "meaning." This type of rewriting thinks itself justified in recovering the rhetorical elements inherent in all prose — but in order to banalize them and assign them a predominant place. These elements — in Rousseau, Balzac, Hugo, Melville, Proust, etc. — restore a certain "orality," and this orality effectively possesses its own norms of nobility — those of "good speaking," which may be popular or "cultivated." But good speaking in the original has nothing to do with the "rhetorical elegance" extolled by the rewriting that ennobles. In fact, the latter simultaneously annihilates both oral rhetoric and formless poly logic (see above).
The logical opposite of ennoblement — or its counterpart — occurs in passages judged too "popular": blind recourse to a pseudo-slang which popularizes the original, or to a "spoken" language which reflects only a confusion between oral and spoken. The degenerate coarseness of pseudo-slang betrays rural fluency as well as the strict code of urban dialects.
This refers to the replacement of terms, expressions and figures in the original with terms, expressions and figures that lack their sonorous richness or, correspondingly, their signifying or "iconic" richness. A term is iconic when, in relation to its referent, it "creates an image," enabling a perception of resemblance. Spitzer alludes to this iconicity: "A word that denotes facetiousness, or the play of words, easily behaves in a whimsical manner — just as in every language worldwide, the terms that denote the butterfly change in a kaleidoscopic manner" (Spitzer 1970: 51).
This does not mean that the word "butterfly" objectively resembles "a butterfly," but that in its sonorous, physical substance, in its density as a word, we feel that it possesses something of the butterfly's butterfly existence. Prose and poetry produce, in their own peculiar ways, what can be called surfaces of iconicity.
When translating the Peruvian chuchumeca with pute (whore), the meaning can certainly be rendered, but none of the word's phonetic-signifying truth. The same goes for every term that is commonly qualified with savoureux (spicy), diu (robust), vif (vivid), colore (colorful), etc., epithets that all refer to the iconic physicality of the sign. And when this practice of replacement, which is most often unconscious, is applied to an entire work, to the whole of its iconic surface, it decisively effaces a good portion of its signifying process and mode of expression -what makes a work speak to us.
This refers to a lexical loss. Every work in prose presents a certain proliferation of signifiers and signifying chains. Great novelistic prose is "abundant." These signifiers can be described as unfixed, especially as a signified may have a multiplicity of signifiers. For the signified visage (face) Arlt employs semblante, rostro and cara without justifying a particular choice in a particular sentence. The essential thing is that visage is marked as an important reality in his work by the use of three signifiers. The translation that does not respect this multiplicity renders the "visage" of an unrecognizable work. There is a loss, then, since the translation contains fewer signifiers than the original. The translation that attends to the lexical texture of the work, to its mode of lexicality — enlarges it. This loss perfectly coexists with an increase of the gross quantity or mass of the text with expansion. For expansion consists in adding articles and relatives (le, la, les, qui, quc), explicative and decorative signifiers that have nothing to do with the lexical texture of the original. The translating results in a text that is at once poorer and longer. Moreover, the expansion often works to mask the quantitative loss.
The destruction of rhythms
I shall pass rapidly over this aspect, however fundamental it may be. The novel is not less rhythmic than poetry. It even comprises a multiplicity of rhythms. Since the entire bulk of the novel is thus in movement, it is fortunately difficult for translation to destroy this rhythmic movement. This explains why even a great but badly translated novel continues to transport us. Poetry and theater are more fragile. Yet the deforming translation can considerably affect the rhythm - for example, through an arbitrary revision of the punctuation. Michel Cresset (1983) shows how a translation of Faulkner destroys his distinctive rhythm: where the original included only four marks of punctuation, the translation uses twenty-two, eighteen of which are commas!
The destruction of underlying networks of signification
The literary work contains a hidden dimension, an "underlying" text, where certain signifiers correspond and link up, forming all sorts of networks beneath the "surface" of the text itself — the manifest text, presented for reading. It is this subtext that carries the network of word-obsessions. These underlying chains constitute one aspect of the rhythm and signifying process of the text. After long intervals certain words may recur, certain kinds of substantives that constitute a particular network, whether through their resemblance or their aim, their "aspect." In Arlt you find words that witness the presence of an obsession, an intimacy, a particular perception, although distributed rather far from each other — sometimes in different chapters — and without a context that justifies or calls for their use. Hence, the following series of augmentatives:
which establishes a network:
This simple network shows that the signifiers in themselves have no particular value, that what makes sense is their linkage, which in fact signals a most important dimension of the work. Now, all of these signifers are augmentatives, appropriately enough, as Arlt's novel Los Siete Locos contains a certain dimension of augmentation: gates, wings, cages, entrances, giants, alleys acquire the inordinate size they have in nocturnal dreams. If such networks are not transmitted, a signifying process in the text is destroyed.
The misreading of these networks corresponds to the treatment given to groupings of major signifiers in a work, such as those that organize its mode of expression. To sketch out a visual domain, for example, an author might employ certain verbs, adjectives and substantives, and not others. V.A. Goldsmidt studies the words that Freud did not use or avoided where they might be expected. Needless to say, translators have often inserted them.
The destruction of linguistic patternings
The systematic nature of the text goes beyond the level of signifiers, metaphors, etc.; it extends to the type of sentences, the sentence constructions employed. Such patternings may include the use of time or the recourse to a certain kind of subordination (Gresset cites Faulkner's "because"). Spitzer studies the patterning system in Racine and Proust, although he still calls it "style". Rationalization, clarification, expansion, etc. destroy the systematic nature of the text by introducing elements that are excluded by its essential system. Hence, a curious consequence: when the translated text is more "homogeneous" than the original (possessing more "style" in the ordinary sense), it is equally more incoherent and, in a certain way, more heterogeneous, more inconsistent. It is a patchwork of the different kinds of writing employed by the translator (like combining ennoblement with popularization where the original cultivates an oralitv). This applies as well to the position of the translator, who basically resorts to every reading possible in translating the original. Thus, a translation always risks appearing homogeneous and incoherent at the same time, as Meschonnic has shown with the translation of Paul Celan. A carefully conducted textual analysis of an original and its translation demonstrates that the writing-of-the-translation, the-discourse-of-the-translation is asystematic, like the work of a neophyte which is rejected by readers at publishing houses from the very first page. Except that, in the case of translation, this asystematic nature is not apparent and in fact is concealed by what still remains of the linguistic patternings in the original. Readers, however, perceive this incosistency in the translated text, since they rarely bestow their trust on it and do n see it as the or a "true" text. Barring any prejudices, the readers are right: it is not "true" text; it lacks the distinguishing features of a text, starting with it systematic nature. Homogenization can no more conceal asystematicity than expansion can conceal quantitative impoverishment.
The destruction of vernacular networks or their exoticization
This domain is essential because all great prose is rooted in the vernacular language. "If French doesn't work," wrote Montaigne, "Gascon will!" (cited by Mounin 19SS: 38).
In the first place, the polylogic aim of prose inevitably includes a plurality of vernacular elements.
In the second place, the tendency toward concreteness in prose necessarily includes these elements, because the vernacular language is by its very nature more physical, more iconic than "cultivated" language. The Picard "bibloteux" is more expressive than the French "livresque" (bookish). The Old French "sorcelage" is richer than "sorcellerie" (sorcery), the Antillais "derespecter" more expressive than "manquer de respect" (to lack respect).
In the third place, prose often aims explicitly to recapture the orality of vernacular. In the twentieth century, this is the case with a good part - with the good part - of such literatures as Latin American, Italian, Russian, and North American.
The effacement of vernaculars is thus a very serious injury to the textuality of prose works. It may be a question of effacing diminutives in Spanish, Portuguese, German or Russian; or it may involve replacing verbs by nominal constructions verbs of action by verbs with substantives (the Peruvian "alagunarse," s'enlaguner, becomes the flat-footed "se transformer en lagune," "to be transformed into a lagoon"). Vernacular signifiers may be transposed, like "porteno," which becomes "inhabitant of Buenos Aires."
The traditional method of preserving vernaculars is to exoticize them. Exoticization can take two forms. First, a typographical procedure (italics) is used to isolate what does not exist in the original. Then, more insidiously, it is "added" to be "more authentic," emphasizing the vernacular according to a certain stereotype of it (as in the popular woodcut illustrations published by Epinal). Such are Mardrus's over-Arabizing translations of the Thousand and One Nights and the Song of Songs.
Exoticization may join up again with popularization by striving to render a foreign vernacular with a local one, using Parisian slang to translate the lunfardo of Buenos Aires, the Normandy dialect to translate the language of the Andes or Abruzzese. Unfortunately, a vernacular clings tightly to its soil and completely resists any direct translating into another vernacular. Translation can occur only between "cultivated" languages. An exoticization that turns the foreign from abroad into the foreign at home winds up merely ridiculing the original.
The destruction of expressions and idioms
Prose abounds in images, expressions, figures, proverbs, etc. which derive in part the vernacular. Most convey a meaning or experience that readily finds a parallel image, expression, figure, or proverb in other languages.
Here are two idioms from Conrad's novel Typhoon:
He did not care a tinker's curse
Damme, if this ship isn't worse than Bedlam!
Compare these two idioms with Gide's amazingly literal version:
II s'en fichait comme du juron d'un etameur
(He didn't give a tinker's curse)
Que diable m'emporte si l’on ne se croirait pas a Bedlam!
(The Devil take me if I didn't think I was in Bedlam!)
(cited by Meerschen 1982: 80)
The first can easily be rendered into comparable French idioms, like "il s'en fichait comme de l’an quarante, comme d'une guigne, etc.," and the second invites the replacement of "Bedlam," which is incomprehensible to the French reader, by "Charenton" (Bedlam being a famous English insane asylum). Now it is evident that even if the meaning is identical, replacing an idiom by its "equivalent" is an ethnocentrism. Repeated on a large scale (this is always the case with a novel), the practice will result in the absurdity whereby the characters in Typhoon express themselves with a network of French images. The points I signal here with one or two examples must always be multiplied by five or six thousand. To play with "equivalence" is to attack the discourse of the foreign work. Of course, a proverb may have its equivalents in other languages, but . . . these equivalents do not translate it. To translate is not to search for equivalences. The desire to replace ignores, furthermore, the existence in us of a proverb consciousness which immediately detects, in a new proverb, the brother of an authentic one: the world of our proverbs is thus augmented and enriched (Larbaud 1946).
The effacement of the superimposition of languages
The superimposition of languages in a novel involves the relation between dialect and a common language, a koine, or the coexistence, in the heart of a text, of two or more koine. The first case is illustrated by the novels of Gadda and Giinter Grass, by Valle-Inclan's Tirano Banderas, where his Spanish from Spain is decked out with diverse Latin American Spanishes, by the work of Guimaraes Rosa, where classic Portuguese interpenetrates with the dialects of the Brazilian interior. The second case is illustrated by Jose Maria Arguedas and Roa Bastos, where Spanish is modified profoundly (syntactically) by two other languages from oral cultures: Quechua and Guarani. And there is finally - the limit case - Joyce's Finnegans Wake and its sixteen agglutinated languages.
In these two cases, the superimposition of languages is threatened by translation. The relation of tension and integration that exists in the original between the vernacular language and the koine, between the underlying language and the surface language, etc. tends to be effaced. How to preserve the Guarani-Spanish tension in Roa Bastos? Or the relation between Spanish from Spain and the Latin American Spanishes in Tirano Banderas? The French translator of this work has not confronted the problem; the French text is completely homogeneous. The same goes for the translation of Mario DC Andrade's Macumaima, where the deep vernacular roots of the work are suppressed (which does not happen in the Spanish version of this Brazilian text).
This is the central problem posed by translating novels - a problem that demands maximum reflection from the translator. Every novelistic work is characterized by linguistic superimpositions, even if they include sociolects, idiolects, etc. The novel, said Bakhtin, assembles a heterology or diversity of discursive types, a heteroglossia or diversity of languages, and a heterophony or diversity of voices (Bakhtin 1982: 89). Thomas Mann's novel The Magic Mountain offers a fascinating example of heteroglossia, which the translator, Maurice Betz, was able to preserve: the dialogues between the "heroes," Hans Castorp and Madame Chauchat. In the original, both communicate in French, and the fascinating thing is that the young German's French is not the same as the young Russian woman's. In the translation, these two varieties of French are in turn framed by the translator's French. Maurice Betz let Thomas Mann's German resonate in his translation to such an extent that the three kinds of French can be distinguished, and each possesses its specific foreignness. This is the sort of success - not quite impossible, certainly difficult — to which every translator of a novel ought to aspire.
The analytic of translation broadly sketched here must be carefully distinguished from the study of "norms" — literary, social, cultural, etc. — which partly govern the translating act in every society. These "norms," which vary historically, never specifically concern translation; they apply, in fact, to any writing practice whatsoever. The analytic, in contrast, focuses on the universals of deformation inherent in translating as such. It is obvious that in specific periods and cultures these universals overlap with the system of norms that govern writing: think only of the neoclassical period and its "belles infideles." Yet this coincidence is fleeting. In the twentieth century, we no longer submit to neoclassical norms, but the universals of deformation are not any less in force. They even enter into conflict with the new norms governing writing and translation.
At the same time, however, the deforming tendencies analyzed above are not ahistorical. They are rather historical in an original sense. They refer back to the figure of translation based on Greek thought in the West or more precisely, Platonism. The "figure of translation" is understood here as the form in which translation is deployed and appears to itself, before any explicit theory. From its very beginnings, western translation has been an embellishing restitution of meaning, based on the typically Platonic separation between spirit and letter, sense and word, content and form, the sensible and the non-sensible. When it is affirmed today that translation (including non-literary translation) must produce a "clear" and "elegant" text (even if the original does not possess these qualities), the affirmation assumes the Platonic figure of translating, even if unconsciously. All the tendencies noted in the analytic lead to the same result: the production of a text that is more "clear," more "elegant," more "fluent," more "pure" than the original. They are the destruction of the letter in favor of meaning.
Nevertheless, this Platonic figure of translation is not something "false" that can be criticized theoretically or ideologically. For it sets up as an absolute only one essential possibility of translating, which is precisely the restitution of meaning. All translation is, and must be, the restitution of meaning.
The problem is knowing whether this is the unique and ultimate task of translation or whether its task is something else again. The analytic of translation, insofar as the analysis of properly deforming tendencies bears on the translator, does in fact presuppose another figure of translating, which must necessarily be called literal translation. Here "literal" means: attached to the letter (of works). Labor on the letter in translation is more originary than restitution of meaning. It is through this labor that translation, on the one hand, restores the particular signifying process of works (which is more than their meaning) and, on the other hand, transforms the translating language. Translation stimulated the fashioning and refashioning of the great western languages only because it labored on the letter and profoundly modified the translating language. As simple restitution of meaning, translation could never have played this formative role.
Consequently, the essential aim of the analytic of translation is to highlight this other essence of translating, which, although never recognized, endowed it with historical effectiveness in every domain where it was practiced.