QUANTITY SPEAKS

 
Quantities replace the translator

 
Material economies know no such thing as non-quantity: all commodities, including all texts, are of certain sizes and at relative distances from each other. In discourse, however, as in monetary systems, there exists a basic point of reference that is rigorously non-quantitative and indifferent to the relative distance of all other terms. This point, the position we have called the I-here-now, is the non-dimensional zero from which discursive distances are measured (cf. Bühler 1934, 102 ff).

To the extent that all human languages would appear to have elements that function as personal pronouns (Benveniste 1966, 261), all speakers can identify with this singular point of departure. Equivalence-producing subjectivity, however, which in principle has no personal pronoun and thus perhaps no human tongue, is unable to express the non-quantitative point from which its utterances depart; it has no I-here-now. However, as we have seen, translational subjectivity can still be expressed through the third-person quantities of the relation TT:Y. That is, translating translators can express their subjectivity from the space between two stretches of text, between two quantities.

The translator's peculiar position between quantities may be approached through T. S. Eliot's classical description of poetic creation:

"The poet has something germinating within him for which he must find the words; but he cannot know the words he wants until he has found the words; he cannot identify this embryo until it has been transformed into the right words in the right order. When you have found the words, the 'thing' for which the words had to be found has disappeared, replaced by a poem." (1953, 17)

This searching for the right words in the right order is no doubt an important part of translating. Yet even when the translator might believe the right words to have been found, the "thing" these words are meant to replace remains sitting there, as a material Y text which, at least for the translator who wants to remain a translator, simply should not disappear. Of course, if it did disappear, the translator could become an author and speak from a new I-here-now, but he or she could no longer regard the result as a translation.

The obstinate presence of the Y text thus makes it difficult to maintain any strong notion of equivalence while translating. This is one of the reasons why individual translators tend not to believe too enthusiastically in equivalence and may indeed privately shun it as an unnecessary falsehood. The basis of comparison is all too available; the translator knows how many alternative TTs have been suppressed or could prove superior in the future. In sum, translators as individuals might well be in a position to claim that their actual activity has nothing to do with equivalence at all. But is this position of importance to anyone except individual translators?

To the extent that translated texts are destined to move away from the individual translator and towards a receiver for whom that translator is in principle anonymous, the collective interest of the translating profession should have more weight than transitory doubts. It is thus fair to regard ideal equivalence as a guiding principle of translation in general, regardless of private dissent. And since translations are ultimately condemned to create some kind of equivalence in the space of reception, the attainment of equivalence may adequately describe the outward aim of the professional translating subjectivity itself, independently of whether or not physically individual translators actually accept that such equivalence has been attained. Storytellers need not believe in their fictions in order to establish fictionality.

This perhaps pedantic distinction between general and individual perspectives helps us to locate some of the surprising ways in which equivalence is related to questions of textual quantity. Unlike the receiver of TT, the translating subject does not perceive equivalence as an immediate relation between two quantities. Equivalence-while-translating, if there is such a thing, must instead be a kind of triangulation between these quantities and a suppressed I-here-now, a point of non-quantity. For the translator, the basic but unexpressed operation is not "Y translates as TT", but "I am translating Y as TT". However, since this truth cannot be expressed as an operator or verb function within TT, the only way it can be recuperated is through the comparison of quantities. Relations between quantities thus replace and represent the professionally anonymous translator.
 

Quantity is of practical and theoretical importance

The importance of quantity is often unappreciated or even entirely ignored in translation theory. Structuralists and their manifold heirs effectively relegated questions of substance to a kind of retrograde positivism, as if the nineteenth century had said all there was to say about material quantities. But despite the numerous theories that continue to be based on the myth of relations without support, many aspects of translational practice happily still depend on how much text is to be produced and how its proportions are to be distributed in the space and time of reception.

Quantity in translation concerns the distribution of textual rhythms, not only in regular verse but also the rhythms that beat most deeply within cultural identity, within the most poetic and sacred dimensions of belonging (Meschonnic 1986). Quantity is also the problematic underlying the more disturbing rhythm of super-printers in the National Security Agency, churning out more than 22,000 lines of recorded and processed conversation per minute, far more than can be humanly translated (Laurent 1986). Or the sexual rhythms mimicked in erotic language, exhausting synonyms and metaphors in virtuoso efforts to last the distance and perhaps to write future generations. Quantity is moreover the editorial problem of fitting texts into pages, flowing around brochure pictures, to the exact bottom of this or that column. It is the key to successful interpreting, which is also editorial in that texts must be fitted into limited temporal space. Or again, it is the main criterion in film dubbing and sub-titling, where linguistic values should ideally correspond to the shape and duration of moving mouths or the available space on the screen. In all these areas - the poetic, the sacred, the sociological, the erotic, the commercial, the printed, the spoken and the photographed... what is left? - , criteria of textual quantity can have priority over questions of strictly qualitative representation.

This does not mean that one should believe everything that quantities have to say. When the Y text "Amérique du Sud" is dubbed as TT "Mexico" because the lip movements are similar (Nida 1964, 178), the kind of quantitative equivalence thus established blatantly misrepresents geographical difference. But theorists of ideal equivalence have been relatively unconcerned by such problems, probably because the very function of ideal equivalence is to hide whatever it is that quantity might have to say beyond the equation TT:Y=1.

Ideal translational equivalence does not deny the existence of two quantities - indeed, it depends on their presumed existence - , but it does suggest that, since Y and TT should have the same inherent qualities, there should be no difference between their dimensions. Translating is thus perhaps the only kind of work with respect to which the quantitative difference between the input Y and the output TT is mostly believed to be non-significant. And yet there usually is a difference.

If no account is taken of these differences between quantities, translatability becomes an abstractly facile concern. When Katz, for instance, attempts to base translatability on the naturalist semantic hypothesis that "each proposition can be expressed by some sentence in any natural language" (principle of effability), he must first declare that length is not a semantic consideration (1978, 205). As difficult as translating may be, Katz assures us that it can always be done, given world enough and time. But world and time are fairly important pragmatic considerations. As Keenen (1978) argues in his reply to Katz, a sentence of a hundred or so words might well be a very precise semantic translation of a five-word proposition, but it will by no means be universally accepted as a worthwhile translation. Effective translation is not necessarily restricted to the absolute equivalence relation TT:Y=1, but it must depend on some kind of reasonable relationship between the quantities of TT and Y.
 

Equivalence is absolute, relative, contradictory or not at all

In what follows, I propose to take the simplest quantitative relations (=, [does not equal], [approximates], >, <) and see how far they can reveal categories of ideal and less-than-ideal equivalence, thus hopefully making translations talk about what it is to be a translation. The working hypothesis is that, thanks to the fundamental nature of the operations and the mostly overlooked importance of quantity, translation itself will manifest an internal organisation that has escaped approaches based on more complex or less pertinent terms.

Certain initial distinctions can be based on the material presence or absence of Y and TT texts in the space of reception. Four cases are to be considered here: transliteration (Y without a visibly distinct TT), double presentation (Y with TT), single presentation (TT without Y) and multiple presentation (TT1, TT2...TTn, with or without Y). These are then four quite different reception situations which can be defined in terms of the fundamental duality of presence and absence, independently of any awareness of transfer or translational subjectivity.

Equivalence would seem to behave quite differently in each of these situations, basically because the quantities involved replace the translator in different ways:

a) Transliteration: Obviously - perhaps too obviously - , if Y is presented without any visibly distinct translational covering, the translator is conspicuous by a certain refusal to work and there is no quantitative difference between Y as a transferred text and TT as a translated text. The result is absolute equivalence (TT=Y, thus TT:Y=1).

b) Double presentation: If TT is presented alongside the transferred text Y and the two are visibly non-identical, then the space in which the translator has worked is at least overt in the sense that it is there for all to see and minor differences cannot be hidden. The optimal general relation is in this case what might be termed overt or strong relative equivalence (TT [approximates] Y1).

c) Single presentation: When TT is presented in the absence of Y, fewer questions are likely to be asked, the illusory possibilities of equivalence are much greater and receptive awareness of a translating subjectivity is more likely to be eclipsed. This is the ideal situation for an amorphous form of weak relative equivalence, since it is only from the position of reception that quantity TT can be projected onto a virtual Y (thus, for the translational second person or implied receiver, TT:Y [approximates] 1).

d) Multiple presentation: An important modification of the above modes is when a TT1 exists in the presence of TT2...TTn. That is, there are at least two different translations of the same text. In such cases it is usually the more recent TT which is considered the better or more equivalent version (principle of translational progress), although values of pseudo-originality might cause this relation to be reversed. Either way, the situation involves contradictory equivalence in reception (for the translational second person or implied receiver, TT1:Y [approximates] 1 and TT2:Y [approximates] 1; but TT1 [does not equal] TT2).
 

There would appear to be little more to say on the level of simple presence or absence. But are the actual quantitative differences between Y and TT then without any significant thresholds?

A significant threshold is breached in court:

"In a New Jersey homicide trial, the prosecutor asks whether the testimony of a witness is lengthier than the translation. 'Yes,' responds the Polish interpreter, 'but everything else was not important." (Alain Sanders, "Libertad and Justicia for All: A shortage of interpreters is leaving the courts speechless", Time 29 May, 1989).

Even when Y is entirely opaque for the actual TT receiver, its physical appearance can continue to support equivalence on the minimal condition that the quantities of the two texts are more or less the same (TT:Y [approximates] 1). Although the limits of this "more or less" are necessarily approximate, they remain of practical importance. Newmark (1981, 130) refers to a Danish pedagogue who, perhaps like perplexed court prosecutors, assumes that a translation is likely to be good as long as it has about the same length as the original. That is, if TT is significantly shorter or longer than Y, suspicion is aroused, the translation will probably be checked and the translator's name might even be noted. Translational subjectivity will thus become manifest, for better or for worse.

The "more or less" of relative equivalence is thus not necessarily blind. Most editors are quite aware that English translations are normally shorter than their German originals, since there are several principles of (assumed or real) compensation written into the structural features of the tongues themselves (German illocutionary particles are compensated for by more expressive word order in English; German compounds are unworried by length, whereas English is relatively unworried about the creation of neologisms to abbreviate compounds). For the purposes of the pedagogue or editor who has to accept or check translations, such principles operate as trivial normalising conditions, in themselves of no significance for the value of the work.

However, beyond these normalising conditions, are there situations where TT can legitimately be considerably shorter than Y (principle of authorised reduction)? Can other TTs be significantly longer than Y (principle of authorised expansion) and still remain translational? If so, how might such thresholds be determined?

To the extent that these limits are operative beyond the virtually obligatory level of linguistic compensation, they must be assumed to be determined by certain choices - including the choice to transliterate - made by translators under certain intercultural conditions.

The challenge such choices present for ideal equivalence can be approached in terms of several types of thresholds: the points at which absolute equivalence (TT:Y=1) could give way to relative equivalence (TT:Y [approximates] 1), and the points where relative equivalence is itself broken by outright reduction (deletion) or unacceptable expansion (addition).

I propose to investigate these thresholds on the basis of the four fundamental reception situations outlined above.
 

 
The proper name is sometimes improper

The strategy of transliteration - transfer without visible translation - potentially leads to the exact quantitative relation TT=Y and thus to strict or absolute equivalence TT:Y=1. That is, the transliterated or transcribed text is exactly as long as the transferred text. But is it still possible to talk about equivalence in such cases? Can Y really be read as a TT (TT:Y=1), or is TT in such cases no more than a zero-value without pertinence for translation (TT:Y=0)? Indeed, given that the relationship is often perceived as one of identity, is it possible to talk about values or a properly translational work process at all?

As can be gleaned from comments by Borges and Derrida , most cases of transliteration are at base problems of proper names or lead to textual results that can be analysed as if they were proper names. Why, for instance, should a "shekel" be left as such in the Bible or a "gentleman" be "in English in the text"? Although the referents concerned can no doubt adequately be described in visibly translated discourse, the foreignness of the foreign term is recognised as distinguishing the distanced concept, like a proper name distinguishes the relative independence of a new-born baby. Yet this clearly does not imply that no proper names can or should be transformed. Transliteration may determine absolute equivalence, but it is not automatically ideal translation. Let us indulge in a few examples to demonstrate the point.

Ezra Pound was very aware that "there is no surety in shifting personal names from one idiom to another", basically because names are already translations of very particular perspectives:

"Sannazaro, author of De Partu Virginis, and also of the epigram ending hanc et sugere, translated himself as Sanctus Nazarenus; I am myself known as Signore Sterlina to James Joyce's children, while the phonetic translation of my name into the Japanese tongue is so indecorous that I am seriously advised not to use it, lest it do me harm in Nippon. (Rendered back ad verbum into our maternal speech it gives for its meaning, 'This picture of a phallus costs ten yen.')." (1918, reprinted 1954, 259)

Some proposed cases of absolute equivalence are in fact unwitting transformations. The eighteenth-century island "Otaheite", for instance, meant "This is Tahiti". According to Garcilaso Inca de la Vega - itself a name to be considered - , "Peru" comes from "Pelú", the name of the first Indian to be interrogated by Spanish conquistadores, who mistook the question "What is the name of this land?" for "What is your name?". In the case of Tahiti, a certain authorship has since been restored to the native islanders. In the case of Peru, the translation has perhaps become an improper proper name. In both cases the supposedly absolute equivalents were in fact authoritative impositions, or rather, impositions of mistaken authority on the part of translators.

Other uncovered names require openly partisan selection akin to questions of diplomatic recognition: "Taiwan" or "China", "Cambodia" or "Kampuchea", "South Africa" or "Azania" (cf. Newmark 1981, 72; 1988, 216). The same holds for questions of disputed discovery ("Desnos's disease" = "maladie de Grancher", noted in Newmark 1988, 199). In the case of names of people, distance and partiality can be regulated by the simple selection of family or given names, the bestowal or deletion of titles, or strategies of morphological adaptation. The Biblical "John" (the Baptist) is necessarily brought closer by the ecclesiastically authorised series "Jean", "Juan", "Joan", "Johann", etc., thus allowing an individual author like Oscar Wilde (in his play Salomé) to create exotic distance by calling him "Iokanaan", destined to be read as a back-translation of the translated series. Thomas More translated himself as "Morus" and may be further translated as a knight or a saint, but he was none of these for the first Spanish translation of Utopia, where he appeared as "tomás moro", a lower-case Moor. One cannot avoid choosing who has the right to produce the name, and the name of the father does not always have priority.

The quantitative identity associated with transliterated names should thus be treated with considerable suspicion. It should not be regarded as an automatic index of untranslatability (the starting point for Mounin 1964), nor as a sign of universal comprehension (for Lotman and Uspenki 1973, proper names are untranslatable because they do not have to be translated). Exact quantitative equality should instead be analysed a special kind of translation, to be understood in terms of the absolute equivalence established between a transferred text Y and a transliterated text TT. As is revealed by cases of isomorphism like "i vitelli dei romani sono belli" , which can be read in either Latin or Italian, the combination of transfer and transliteration does indeed concern translation whenever cotexts or contexts are radically changed. Or more specifically, an apparently untranslated term in an otherwise translated text must be received as a term which has at least been processed translationally. Absolute equivalence is thus established between a manifest text and a virtual text, both of which look exactly the same and occupy exactly the same position.

Curiously enough, the combination of transfer and transliteration also concerns the location of a translational subjectivity. Each time quantities are exactly equal and in a situation of possible translation, each time Y is presented without visible TT covering, this presentation indicates the location of a potential translator who, although deprived of an I-here-now, has adopted a position by refusing to intervene. Thus, almost paradoxically, absolute equivalence points towards a translational subjectivity.

Levy (1965, 81) distinguishes transliteration (Mr Ford as Mr Ford) from substitution (Mr Newman as Herr Neumann) and loan translation (Everyman as Jedermann). For out present purposes, the simple existence of such alternative strategies serves to indicate that absolute equivalence - Levy's transliteration - is the result of a choice and thus a product of properly translational work. It also indicates that the choice is made by someone and thus depends on a more primary distinction between the proper names of author and translator, the two positions from which, as we have seen in the case of Everest-Chomolungma, the name can be selected.

Quantity replaces the translator. But exact quantitative equality - symbolised in the uncovered proper name - indicates the place of the translator with respect to an author, on one side of the mountain or the other.

However, the main problem with absolute equivalence is that it is rarely acceptable equivalence. As Aristophanes' poets discovered with their scales, semantic content also has its specific weight, and in most cases the extreme geographical imbalance between the referents of "Amérique du Sud" and "Mexico" is more important than any phonetic approximation. Absolute quantitative equivalence normally has to be sacrificed in favour of some kind of semantic isomorphism.

Other kinds of equivalence are necessary.
 

Relative equivalence presents asymmetry

Double presentation, the showing of both Y and TT as distinct texts, can be understood as immediate minimal representation of the distance Y­ST.

To the extent that a receiver may or may not be able to understand the transferred but untranslated Y text, there would appear to be two obvious subordinate possibilities for double presentation: Y may be opaque or transparent. Having hurriedly assumed the validity of these possibilities, a theorist like Mossop would then have to find out if and when the Y text was actually read. In practice, however, this is a false dichotomy, since the macrostructural second person implied by the mode of presentation is at once able to understand something of Y (if not, there would be no reason for its reproduction) but not all of it (if understanding were total, the only possible reasons for the translation would be pedagogical, and wastefully pedagogical at that). In principle then, double presentation implies a reception situation in which the transferred text is only relatively opaque or relatively transparent.

Why should this mode be particularly well suited to relative equivalence? Quite simply because the fact of double presentation raises significant doubts about the possibility of absolute equivalence. Obviously, if equivalence were certain, there would be no reason for reproducing the transferred text Y; and if it were entirely impossible, there would be no way of producing a translated text TT. Double presentation thus necessarily questions the principle of absolute equivalence as a guarantee of translatability, and does so even before exact quantities are calculated or actual texts are read.

However, if relative equivalence is thus signaled by double presentation, it is not signaled in the same way for both the texts involved. There is a principle of asymmetry by virtue of which the embodiment of value will be unequal, no matter which of the texts is the longer.

This basic principle can be expressed as follows: If two texts are presented as being in a relationship of relative equivalence, the implied receiver will necessarily consider the more opaque text to have been transferred across the distance Y­ST and will see that text as embodying more value than does the more transparent text, assumed to be the TT.

It should be stressed that this asymmetry is purely a function of the exchange situation at the base of the translational relationship; it does not imply any detailed evaluation. For example, when Benjamin writes

"Auch im Bereich der Übersetzung gilt: en adch hn o logoV [it's supposed to be in Greek] im Anfang war das Wort." (1923, 59)

no physical reader is obliged to prefer the uncovered Greek Y ["in the beginning was the Logos/Word"] to the German TT ["in the beginning was the Word"], but the exchange situation itself indicates a preference. After all, if the German were just as good as the Greek, why should the Greek have been transcribed? Logically, the Greek text must be worth more than the German because, in addition to a slight quantitative difference, its presence in a foreign form embodies the considerable labour expended to bring it across centuries and cultures.

The above principle of asymmetry seems to hold in all cases of approximate quantitative equality. It would moreover appear to be commutative in the sense that it is indifferent to the order of presentation (Y,TT; or TT,Y). However, it should be stressed that this basic principle of presentation always depends on the maintenance of some kind of reasonable relationship between the quantities of Y and TT, such that the receiver can accept that TT:Y1, without undue distractions. Several minor thresholds should be considered here.

What happens in cases where TT is clearly shorter than Y but does not quite break equivalence? Consider, for example, the possible intralingual translation "Orthohydroxybenzoic acid (salicylic acid)". Does not the longer Y text sound more scientifically exact and thus of greater scientific value? Is it not more opaque and thus in some way superior to the shorter, more vulgar version? But does this possible reception mean that the principle of asymmetry is then valid in all cases of significant quantitative discrepancy (Y>TT, but TT:Y [approximates] 1)?

There must be some doubt as to whether such a translation actually accords greater value to greater length or is merely employing normalising conventions to correlate two distanced levels of discourse. Since this latter hypothesis would suggest that the phenomenon depends on the dimensions of the transfer situation involved, we should consider at least a few of the possible variants.

Now, if this hypothetical transfer is over a short distance and the receiver is thus close to the values of scientific discourse - for example, in a teaching situation - , the potential analytical value of the longer and more opaque term is likely to be appreciated and the principle of asymmetry may indeed hold. But a mid-distance interlingual version of the same example - "acide ortho-hydrobenzoïque (salicylic acid)" - has quite a different effect, since it overtly refuses criteria of quantitative equivalence and is able to suggest that the French term is perhaps pompous or unnecessarily long. The visibly shorter TT does not break equivalence, but its effect is potentially ironic.

A further example of translational irony would be sub-titled opera arias, where short TTs like "Don't go!" typically correspond to a great deal of Y-text singing. In a mid-distance or even long-distance situation, ungenerous minds tend to find the result quite comic, although relative equivalence might yet escape unscathed. It simply depends on how near or far the receiver is from the values of opera.

Thus, although the asymmetrical distribution of value in such cases certainly concerns relative quantities, it also requires a certain codification of the transfer situation.

Since the categories of this codification will be taken up in the next chapter, it should suffice for the moment to present as hypotheses the general principles drawn from the above examples.

Here then are two rather peculiar phenomena which occur within the realm of double presentation. First, when the quantitative relation is approximately equal such that TT:Y1, the attributed value of Y tends to be greater than that of TT. However, when the quantitative relation is Y>TT (but TT:Y[approximates]1) and the transfer is mid-distance or long-distance, there may be an ironic inversion of this evaluation such that TT is apparently worth more than Y.

Black irony from a socialist paradise:

Ferri-Pisani's appropriately obscure text Antipodes: L'Australie, Paradis Socialiste (1934) begins with the following sentence:

"Le Commonwealth (ou République) d'Australie date du 1er janvier 1901."

One supposes that, for the implied French reader, the Y text "Commonwealth" is more opaque and perhaps of more (exotic?) value than the TT "République", although the underlying "res publica" could be rendered quite adequately as "common wealth" and relative equivalence might remained unchallenged. But this particular equivalence was very relative. In 1975, the fact that Australia called itself a Commonwealth and not a Republic meant that its Governor-General (the representative of Queen Elizabeth II) could dismiss a freely elected government. Some Australians now believe that "République" is worth much more than "Commonwealth".

Note that the distance Y­ST below the line (France 1934 - Australia 1975) has significantly altered values above the line; mid-distance transfer has altered the way the translation can be read.

But do the above principles apply when TT is visibly much longer than Y? Surely it is here, more than in Y>TT, that the principle of asymmetry would seem to extend beyond the limits of defendable equivalence?
 

Relative equivalence tends to paraphrase ("La Movida" moves)

The range of possibilities for Y<TT can be demonstrated through the following presentation of the one Y text ("La Movida") within TTs which vary from a superficial evocation of untranslatability to extended paraphrase:

(1) "La Movida c'est intraduisible. La Movida, c'est la société espagnole tout entière qui choisit d'avancer." (Elisabeth Schemla, Le Nouvel Observateur, Paris, 5-11 July 1985)

(2) "[...] the phenomenon known as 'La Movida' ('The Happening')." (Newsweek, European Edition, 5 August 1985)

(3) "[...] 'La Movida', was wörtlich übersetzt 'Die Bewegung' heissen würde. 'Aber mit Vorwärtsgehen hat das nichts zu tun' sagt selbstkritisch der Maler El Hortelano. 'Es ist mehr wie ein Schluckauf, der einen auf der Stelle schüttelt'." (Charlotte Seeling, Geo, Hamburg, November 1985)

The transferred text "La Movida" is related to a series of longer possible equivalents: TT1 "intraduisible" [untranslatable], TT2 "choix d'avancer" [choice to advance], TT3 "(The Happening)", TT4 "Die Bewegung" [The Movement], TT5 "ein Schluckauf" [a hiccup], and many more, since the above passages are in fact fragments from three fairly long articles. But what becomes of these strictly translational relations (TT:Y) when, as here, different amounts of text are used to suggest equivalence on one level or another?

The German commentary is explicitly longer than any approximately absolute equivalent, since it proposes and discounts TT4 "Die Bewegung". Yet the French commentary, in declaring Y to be untranslatable and then immediately translating it, merely imitates the German procedure in an implicit way, arriving at similar doubt about the equivalence of TT2 "choix d'avancer". Much the same procedure underlies the Newsweek version, where the presentation of Y and the bracketing of TT3 "(The Happening)" could be read as "there is no equivalent for this term, but it means something like...". In short, the English-language approach involves essentially the same strategy as the German and French rejections of simple attempts at equivalence. The differences between these examples lie not in their procedures as such but in the quantities of interceding text used to convey doubt about ideal equivalence.

What consequences do these mediating quantities have for the implied reader's perception of "La Movida"? It would seem that the longer TT texts accord Y far more value as an original form than does the immediately substitutive TT3 "(The Happening)". The relative quantitative equivalence of the Newsweek translation - the shortest TT - in fact suggests that the Spanish term has no originality beyond its function as a touch of local colour: "La Movida" is made to look like a much belated Spanish version of a term belonging to the American 60s; it has become a kind of bad translation in itself, ironically reversing the principle of asymmetric equivalence. Happily, a more recent Newsweek article redresses the injury, using textual expansion to put "La Movida" back onto the authorship side of the mountain, thus revealing that there was no linguistic reason for referring to "Happenings" in the first place:

(4) "[...] in the years after 1975, Spain celebrated La Movida, a sense of explosive artistic, cultural and political excitement." (Newsweek, European Edition, "Spain's Weary Generation", 30 October, 1989)

In 1985, as Spain was deciding to enter NATO, Americans were not particularly interested in translating anything more than a superficial Spanish culture based on American precedents. In 1989, however, during national elections following Spain's presidency of the EEC, the same magazine saw fit to accord Spanish culture a recent history of its own, with corresponding TT quantities.

These examples suggest a correlative of the general principle of asymmetry: the more the TT quantity exceeds Y, even beyond TTY, the greater the value accorded to Y. Unlike the general principle, however, here it does matter how much more TT there is than Y. Here the logic of asymmetry goes beyond the leeway of relative equivalence.

Why this should be so can perhaps be explained in terms of the washing powder commercial in which a housewife is offered two packets of product TT in exchange for one packet of product Y, and of course stays with the latter because it must be the genuine article. If the logic of marketing were pushed a little, it is even possible that she would not accept three or even four packets of the inferior product TT: they would be too difficult to carry home. Only product Y puts ideal quality into an ideal size, and the more TT is offered in exchange, the greater the implied value of Y.

Attentive television watchers will have noticed that the above illustration of translational exchange just managed to reverse domestic marketing principles: double presentation has just sold the relatively unknown washing powder, without there being any question of taking it home and trying it out first. Only in cases of translational irony will TT win out as the tried and true home-grown product.
 

Why translational paraphrase tends to stop at sentence level

The final question to be asked with respect to double presentation concerns the upper limits of textual expansion. The "Movida" example suggests that paraphrase can be analysed as a legitimate - if sometimes unhappy - mode of translation. But are there limits beyond which TT expansion is no longer translational?

It can be observed that in order to continue beyond the rejected TT "Die Bewegung", the German rendition of "La Movida" has to incorporate a subjective voice. It cites the painter El Hortelano, presumably coming from somewhere within the content of the Spanish term. The extended French text (continuing the fragment reproduced above) does the same, citing a Spanish banker and then unnamed Spanish sources to make "La Movida" equivalent to an acceptably French concept: "...the latest news is that 'La Movida' has taken another name, awaiting the next: postmodernisme" (my translation from the French). In both cases, paraphrase of the one term beyond the limits of the sentence requires location of an authoritative subject external to the translating subjectivity. That is, continued paraphrase of a Spanish term requires citation of a Spanish source (a painter, a banker, or "the latest news"). If no such external subject were indicated, the second sentence - the second verb form, as a rule of thumb - would have to use an operator like "in fact means" or "sometimes means" or "could mean", implying a modulated subjective authority incompatible with the translator's suppressed pronoun. This means that, since the translator's falsely objective "Y translates as TT" cannot simply accumulate paraphrases and seriously propose them all as equivalents, someone, some authoritative subjectivity, some author, has to say "I say ST means TT". A translator who fails to choose a definitive TT the first time round, with the first sentence or verb form, fails definitively to establish equivalence and must turn to or invent the verb of an author.

Thus, since translators are not supposed to be authors - since "I am translating" is necessarily false - paraphrase extended beyond the sentence tends to break with the situation of translational exchange, becoming a discourse of addition or non-equivalence. As a legitimately translational procedure, paraphrase is common but necessarily short-lived.
 

Single presentation hides at least one quantity

Single presentation (TT without Y) makes the significance of relative quantities more difficult to grasp, quite simply because one of the quantities is absent. Indeed, if single presentation is strictly defined as a mode of translation which denies the implied receiver all material access to Y, it is by no means evident how the quantity of TT can be of any significance at all. Translators can and often do omit large stretches of text or add numerous private fantasies, and no receiver need be any the wiser. Moreover, since the definition of single presentation allows for pseudotranslations, one must also talk about cases in which the Y text is singularly unavailable for quantitative comparison because materially non-existent. In such cases, how can quantity have anything at all to say?

Cases of eloquent absence are rare but revealing. Consider, for instance, Jerónimo Fernández's knightly epic Don Belianís de Grecia (1547, 1579) which, presented as a pseudotranslation from the Greek sage Fristón, ends with the pseudotranslator asking anyone finding the lost original to continue the tale. Don Quixote, who sincerely lamented the hundred-and-one serious wounds received by Belianís in the first two books of the epic, nevertheless much appreciated Fernández's astute way of thus "ending the work with the promise of a never-ending adventure" (Cervantes 1605, 72). An absent quantity can always promise more.

However, there must be doubt as to whether such cases of explicit reference to an absent untranslated quantity allow the presented text to function as a translation. Surely the never-ending ending is more paratextual than properly translational? But then again, is not this question in itself indicative of a threshold?

It must be agreed that when a translation is read strictly as a translation, the equivalence operator should ride over any half-suspected discrepancies and nobody should know or care about what might have been left behind. The social conventions determining the limits of relative equivalence remain unexpressed for as long as they are respected, or, more exactly, for as long as the resulting TT can be believed to be equivalent to the absent Y text. If no doubts are raised, no one need be any the wiser. But when significant doubts are raised, a threshold has been reached.

This means that single presentation can only really be investigated through cases in which social conventions have not been respected or the TT itself indicates certain quantitative discrepancies. Happily, Jerónimo Fernández was not alone in presenting the traces of such differences.

 
Simple signs indicate expansion and addition, abbreviation and deletion

There are several basic paratextual signs that allow quantity to gain at least minimal significance in cases of single presentation. Translators' notes or added terms in square brackets mean that the TT is presumably bigger in these places; a row of dots across the page or three of them between square brackets indicate that the TT is quantitatively smaller than the Y text and then there are peculiar possibilities like the bracketed question-mark "(?)" used by Mossop (1983, 248) to indicate that an omitted Y passage made no sense anyway (although this latter example would seem to induce non-equivalence to the extent that it presents a questioning subjectivity). Like the never-ending ending, these signs allow translated texts to be read in terms of their quantitative relation to absent transferred texts, thus providing an obvious way of approaching the thresholds of single presentation.

Why should such signs exist? It is not always because TT has to be longer or shorter for the purposes of just fitting into a page. These signs also respond to the non-quantitative criteria involved in strategies of explanation and simplification, in movements between the implicit and the explicit, or in dealing with problems of excessive semantic density. Formalisation of these phenomena thus requires that Y texts be attributed a minimal notion of content.

Let us assume that the transferred text Y and the translated text TT, which have fixed sizes, each has a relatively stable content or semantic material when in the place of the translator. We shall symbolise these quantities of semantic material as "sY" and "sTT". Even before questions are raised about the exact nature of this content - and many questions do need to be raised - , it is possible to analyse TT:Y in terms of four relationships between textual quantity and semantic material:

- Expansion: Approximately the same semantic material is expressed by a greater textual quantity (TT>Y; sTTsY).

- Abbreviation: Approximately the same semantic material is expressed by a smaller textual quantity (TT<Y; sTTsY).

- Addition: New semantic material is added to Y, with a corresponding increase in textual quantity (TT>Y; sTT>sY).

- Deletion: Semantic material is removed from Y, with a corresponding decrease in textual quantity (TT<Y; sTT<sY).

In theory, the first two relations lie on the properly translational side of the relative-equivalence threshold; the latter two involve non-equivalence and possibly pseudotranslation or non-translation. But can these abstract relationships really be analysed as separate strategies? And how, in cases of single presentation, should one go about determining the thresholds between expansion and addition, abbreviation and deletion?

A careful approach is needed, starting from the most visible manifestations of these relations.

Caveat: Nida unhelpfully calls expansion "addition" (1964, 230-231):

"Although we may describe the above techniques as involving 'additions', it is important to recognise that there has been no actual adding to the semantic content of the message, for these additions consist essentially in making explicit what is implicit in the source-language text. Simply changing some element in the message from implicit to explicit status does not add to the content; it simply changes the manner in which the information is communicated."

Similarly, Nida calls abbreviation "subtraction", even though it "does not substantially lessen the information carried by the communication" (1964, 233).

As much as I would like to respect established terminology, what words does Nida give me to describe processes of addition and deletion which do affect semantic content?
 

Notes are expansion by another name

The translator's note is of interest here as a mediation between double presentation and single presentation. A typical note like "In English in the text", although manifesting considerable aesthetic indolence and usually of absolutely no practical consequence, does at least serve to indicate that a particular TT fragment can be read as a transliterated text for which no translation has been deemed necessary. The note enables the text to be read as both a Y and a TT, as simultaneously single and double presentation. As such, the note-form may act as a general introduction to single presentation.

Translators' notes usually present or explain transferred fragments, offering a partial double presentation that nevertheless allows the receiver to opt for single-presentation strategies. Parts of the Y text are there if the receiver is interested, but their distancing from the corresponding TT indicates that no one is obliged to find them. Notes can be at the bottom of the page, the end of the chapter, the end of the book, in a supplementary glossary or even in specialised dictionaries. In principle, the more the notes are physically distanced from the associated text, the more reception is likely to be in terms of single presentation.

Translators' notes have deservedly become unfashionable. First-year university students of translation tend to strew them all over the place, explaining every half-shaded detail, insulting the implied receiver's intelligence with massive overtranslation, and perhaps reflecting initial frustration at suppression of the translator's discursive position. Thankfully, students tend eventually to accept that receivers are to be presumed intelligent until proven otherwise, that not all details are pertinent to adequate TT comprehension, and that there are several hundred more cunning ways of directing the receiver's attention. But why do they still think I am joking when I describe footnotes as confessions of defeat? Why do students' jaws drop just a little when it is explained that certain footnoted information can equally well be inserted into the text? It seems that first-year students share a common misbelief in equivalence as quantitative equality (as if all equivalence were absolute). The insertion of in-text explanations is thus felt to be cheating, even when the semantic material to be explained is demonstrably implicit in the Y text. It takes some time before notes are appreciated as simply less elegant ways of expanding textual quantity.

Nevertheless, as much as I prefer my students to avoid notes and to insert explanations into their translations, I must recognise that there are no hard and fast rules about such details and that, pedagogically, my personal preferences should be balanced against solid arguments in favour of notes.

One could argue, pedagogically, that notes offer two potential advantages. On the one hand, they help keep the translated text deceptively clean, readable and quantitatively equivalent. On the other, they enable the translator to approach maximum fidelity to semantic content, since the notes themselves are in principle relatively free of quantitative constraints. It is in this spirit of partitioned fidelity that Nabokov called for "translations with copious footnotes, reaching up like skyscrapers to the top of this or that page [...]" (1955, 512).

But these two advantages set up and manipulate a very misleading dichotomy, since the spatial partition they depend on is incompatible with the temporal progression of textual reception. For example, Milosz has reportedly had to retreat to a footnote to explain that the terms "government" and "people", which have apparently become pejorative in Polish, were nevertheless positive when used in Lincoln's "government by the people and for the people". If the note is not read, then misunderstanding is likely to result. And as soon as the note is read, it is bound to increase the semantic density of the translated text and thus annul the initial reason for the separation of note and text. This makes notes difficult to justify. In practical terms, the only real advantage of notes is that they might not be read. Which is indeed a fairly dubious advantage in cases where they were deemed necessary in the first place.

Arguments for and against notes could probably go on indefinitely. But the simple point I want to make here is that translators' notes and in-text expansion both have much the same effect on textual density. That is, notes are just a special case of expansion. Demonstration of this only requires a few examples:

Footnotes to Paul in tongues:

A series of footnotes gloss the Biblical passage about "speaking in tongues" (1 Corinthians 14) as "speaking in tongues, in other languages" (New International Version, 1979). The notes significantly expand the TT. That is, for relative equivalence, 1 unit of Y 2 units of TT; or TT:Y2. The translators have in fact used notes to double the meaning of the text, interpreting Paul's words as a sociological theory of translation. However, the gloss does not annul the more common belief that Paul - in the lost original, in another translation, in the next translation, or in the TT without notes - was talking about a divinely inspired ability to speak in all languages at once. The expansion thus invests the absent original with two parallel readings, not only producing increased textual complexity in the space of TT, but also projecting greater semantic density on ST.

The textual expansion of Australian drovers:

The term "drover", a tired commonplace of Australian literature, can be rendered in Spanish as "pastor" (shepherd), "vaquero" (cowboy), "ganadero" (cattle-owner/ grazier) or, as in the Spanish version of the classic film The Sundowners, "conductor de ganado" (literally, cattle-driver). And yet it is at once more and less than all of these: the term applies to work with either sheep or cows, has nothing at all to do with westerns, excludes ownership, includes movement from place to place, and predates "driving". But the economies and social structures of Hispanic culture would seem never to have produced drovers and thus have no exactly corresponding term. In Spanish translations of Australian short stories (ed. Pym 1992), two of which are entitled "The Drover's Wife", the rather unhappy solution to this problem was to insert the phrase "uno de estos ganaderos trashumantes" ("one of those itinerant sheep/cattlemen") with the first appearance of the term, thereafter using the simple word "ganadero" and trusting that contexts and the reading imagination would retain the necessary disambiguation. To do any more would have been to obstruct the narrative workings of the texts as literature, or to suggest that their value was more sociological than aesthetic.

This simple example might illustrate why textual expansion is often a necessary part of translation. But there is also the correlative: the initial expansion was used to invest the term "ganadero" with a complex semantic value that it did not have previously. The term was made to say more in less textual space; the Spanish language was slightly altered; the effect of the strategy was to increase semantic density. The expansion thus had the same effect as a footnote.

 
Abbreviation and deletion can be difficult to justify

Abbreviation and deletion are generally more difficult to justify than expansion and addition. A trivial reason for this lies in the fact that quantity is commonly the basis for determining the actual price of translated texts. Whatever the value of their work, freelance translators tend to be paid so much per line or page. In such circumstances, any translator who wilfully seeks to improve a text by reducing the initial quantity is simply going to lose money. The economy of this part of the profession habitually fails to recognise the labour of the file.

There are other reasons as well. Just as no one likes to be excluded from a secret, the more one feels excluded from the implied second person of a text, the more value one tends to attribute to that text. One of the rarely repeated secrets of simultaneous interpreting is that, no matter how impenetrable the incoming Y text, no matter how lost you are, you should try not to leave prolonged silences in the outgoing TT: as long as the receivers are getting a fairly continuous non-contradictory text, they will not be too worried about what they are missing. If, on the other hand, the receivers are confronted by substantial gaps in the interpreting, they will begin to doubt the value of the translation as a whole and will inevitably feel frustrated at having missed out on precisely the part that would have most interested them. As much as the interpreter's silence might have been in the interests of not conveying any false content, the silence itself is likely to do more damage to the status of the TT than would a few reasonably informed guesses. For similar reasons, when in real trouble, interpreters should blame all gaps on technical faults. The social logic of secrets makes deletion very difficult to justify by any other means. Overt deletion is likely to promote distrust.

The same is true of written dots, be they across the page, between brackets or editing a quotation. Surgery scars create doubts about the health of the surviving body - something must be missing.

That is, since the citizens of democracies believe they have a right to know everything, it is usually better for undemocratic translators to delete unseen, or not at all. But does this mean that there is no ideologically valid justification for abbreviation and deletion?

As reduction of waste, abbreviation is obviously not as difficult to justify as is outright deletion. In a world of deadlines, complex bureaucracies and technological text production, translators are increasingly called upon to write summaries and reports, compressing source materials in accordance with the needs and interests of highly specific receivers. At the same time, this inevitable restriction of the TT public means that abbreviation often resembles simple deletion in that it provokes potentially unhealthy phenomena of second-person exclusion.

Sigla and acronyms, to take the most obvious cases, are a cultural bane. Living in Paris but excluded from the inner mysteries of French culture, I must have written "Cedex" on several hundred letters without knowing that it was an acronym I spent about three years catching an "RER" train without knowing what the letters stood for. Then there are the virtually uncountable euronyms - 524 of them are listed in a 1990 EEC glossary - which separate the European Community from most of its citizens:

"The European in the street is not thrilled to be told that this week two ICGs start - one that could turn the ERM into an EMU, another that might bring EPC, and conceivably even WEU, within the EC." (The Economist, 15 December, 1990).

Zipf's law may well allow frequency of use to explain abbreviation, but frequently restricted use tends to justify no more than esoteric secrets, of dubious social virtue. In translation, which presupposes transfer away from the centres of maximum frequency, abbreviation must inevitably seek its legitimacy in terms of a second person who is suitably informed. However, as a personal plea against a world of opaque officialdom and stratified communication, I would ask that full and translated versions be given at least once in each TT, usually upon first mention of each siglum or acronym.

Thus, although abbreviation has a certain economic justification, it should not always be considered socially desirable.

As for deletion, the arguments are a little more complicated.
 

Authoritative subjectivity allows addition and deletion

It is by no means easy to distinguish between expansion and addition, abbreviation and deletion. According to the above formalised definitions, expansion and abbreviation should involve semantic material which is itself neither significantly expanded nor abbreviated. This means that addition must be of material which is in no way implicit in the Y text, and deletion must be of material which is in no way implicit in the TT text.

But it is very difficult to say exactly what is or is not implicit in a text. Cunning readers can pull rabbits out of the most unlikely hats, and if a rabbit is considered worth adding to a text, it is presumably also worth reading into that text. Rulings for and against implicitness ultimately depend on authoritatively limited interpretation, and thus on an authority with a specific position in time and space.

Some additions avoid this problem to the extent that they could not have been written by the author of the ST and thus cannot be considered expansions of the TT. Such material could concern events that took place after the author's death, references to posterior works, critical variants, indeed all the textual items in which translators cease to translate and become editors or critics, thus escaping criteria of equivalence. Since such additions necessarily emanate from a subjectivity posterior to the author of the ST, they cannot be attributed the anonymity of a translator. Precisely because they assume a non-translational subjectivity, they are additions and not expansion. That is, additions may have been written by translators, but the translators were not translating at the time.

In this way, close analysis of implied first persons should be able to distinguish between expansion and addition.

But how could such an analysis be applied to cases of deletion? Since dots and brackets tend not to be eloquent about their subjective sources, the only way of knowing who has deleted is to look for an authoritative defence of the deletion.

As we have seen, democratic principles make it extremely difficult to justify the omission of content. To find authoritative support for deletion, it is perhaps easiest to turn to aristocratic politics. Peter the Great wrote the following introductory notes to a Russian translation of the Georgica curiosa :

"Since the Germans are given to filling their books with all kinds of stories for the sole purpose of making large books, none of this was worth translating except the essential points and a brief introduction. But so that this introduction may be more than idle glitter, and for the readers' greater illumination and edification, I note that I have corrected this treatise on agriculture (effacing all that serves no purpose) and present it so that all books will be translated without useless stories, which only waste the readers' time and make them bored." (cited by Lotman 1979, 54)

Whatever the ethics of such justifications, in this case we can be fairly sure that the TT is shorter than the Y text because of deletion and not mere abbreviation. But we can only make this distinction on the basis of a paratextual authoritative "I", Peter the Great, who is in no way to be confused with the actual translator. The treatise has been "corrected" through deletion; what remains can thus be assumed to be both equivalent and authorised.

Although the subjectivity authorising deletion is necessarily extra-translational, it can yet be very close to the translator. The following "I" is from the middle of Wace's Roman de Brut, which may be regarded as a translation of Geoffrey of Monmouth's Historia Regum Britanniæ (Durling 1989, 9) :
 

"Then Merlin spoke the prophecies

Which you have, I believe, heard,

of the kings who were to come

and who should govern this land.

I don't want to translate this book

since I myself don't know how to interpret it.

I wouldn't want to say anything

unless it were as I myself would say. "

 

Now, how is it that when oral translators leave obvious gaps, the effect of deletion is not to authorise their text but to caste doubt on what has been translated? What is the difference between deletion in the hands of Wace and deletion by a conference interpreter? The answer is to be found in the status of the subjectivity involved. The simultaneous interpreter remains anonymous and thus open to either radical trust or radical doubt, whereas some translators - minimally visible in that their products are written - have been able to become authoritative in their own right, mixing trust and doubt to the extent that they assume an individuated identity, taking some of the responsibility that might otherwise have been left to authorities like Peter the Great and their censors, challenging the limits of equivalence.

French translators of the seventeenth and eighteenth centuries often adopted similar authorisation, basing their selection criteria on audience acceptability. Letourneur's well known introduction to his 1769 translations of Young give the aim of translation as:

"...to appropriate all that is good in our neighbours, leaving behind all that is bad, which we need neither read nor know about." (cited by West 1932, 330; D'hulst 1990, 115)

Here the "I" of the authority, censor or individual translator is explicitly based on the "we" formed by the professional translator and his public. That is, although the translator in this case acts individually - the rest of Letourneur's introduction is mostly in the first person - , he does so in the name of the entire receiving culture, assumed to be superior to the ST culture. It is interesting to note that the translator in this case actually claims to delete very little ("...just one or two fragments by a Protestant disclaiming the Pope...") but in fact cuts out everything considered half objectionable on aesthetic or theological grounds and relegates it to his "Notes". The function of the introductory paratext is thus partly to instruct the reader not to read the notes, or at least - more prudently - not to consider the translator in any way responsible for any blasphemies that might be concealed there. Was everything bad really "left behind"? If, as we have remarked, translators' notes are a form of expansion, their combination with deletion may become a way of dividing a translational subjectivity troubled by problems of censorship and public authorisation.

In principle, authorisation should work indifferently for both deletion and addition, since if one can decide what is not worth translating, one can also decide what would have been worth translating if it had been written. Letourneur did in fact admit to several minor stylistic additions, although his presumed mandate did not extend to the limits claimed by Abraham Cowley in another well known introduction, written in 1656:

"I have in these two odes of Pindar taken, left out, and added what I please; nor made it so much my aim to let the reader know precisely what he spoke, but what was his way and manner of speaking." (cited by Nida 1964, 17)

The strategy of deletion is in this case inseparable from addition, since the theory concerns the liberty to betray both quantity and content in favour of a personalist conception of style. Cowley's repeated first person is apparently sufficient justification for these radical changes, since his actual content is in fact another first person, an individual "way and manner of speaking". In theory, no external authority should be necessary for this intersubjective one-to-one communion. But Cowley did not really get away with it. His individuated subjectivity failed to attract sufficient authorisation to escape Dryden's 1680 distinction between his "imitations" and a properly translational discourse based on modes of relative equivalence (analysed by Dryden as "metaphrase" and "paraphrase"). One of the operational criteria of addition and deletion must thus be the authority and ability to maintain the translational status of that which is neither added nor deleted. When this status is radically questioned, the subject responsible for the text risks losing the authority necesary to establish relations of equivalence. The result is sometimes less than translational.
 

Expansion and addition can run into political trouble

It is possible that fully authorised deletions only take place in situations where the TT culture considers itself superior to the ST culture. This historical hypothesis no doubt deserves more investigation than can be offered here.

But what is more worrying is that the inverse hypothesis seems to be unfounded: It appears less than probable that fully authorised additions only take place when the TT culture perceives itself as inferior to the ST culture. The main problem with this second hypothesis is that, once authority is granted, deletions and additions tend to happen together, in the same translation, signed by the same translator. Cowley, for instance, claimed both modes of liberty. So it is difficult to make any generalisation about the relation between the cultures concerned.

A more interesting historical hypothesis might be based on the status of expansions and additions within the receiving culture. If authorised deletion can be interpreted as a form of censorship, the authorisation of addition can become a question of considerable political import. Indeed, if the thin boundary between expansion and addition partly depends on the authoritative status of the translating subjectivity, only those with effective social power can decide, in each situation, what is to be regarded as (justified translational) expansion and what is to be rejected as (unjustified untranslational) addition.

History offers at least three significant cases of institutional conflict between notions of expansion and addition:

a) Fray Luis de León:

"He who translates has to be true and faithful [...] so that those who read the translation can understand the entire range of meanings they would have if they were reading the original, and they may thus select the meaning which they find the best." (1580, 12)

In declaring his total fidelity, Fray Luis ostensibly effaced his personal responsibility for erotic elements in his translation, the Cantar de los cantares (Song of Songs), displacing responsibility to the mind and imagination of the receiver, as have many latter-day defenders of pornography. Unfortunately, the Inquisition disagreed with this theory, declared the translator at once responsible and irresponsible, and imprisoned Fray Luis in 1572 for this and other concerns with fidelity. The text was later published.

b) Etienne Dolet's Axiochus :

"[...] death has no hold over you, for you are not yet at the point of dying; and once you have died, death will still be powerless against you, since you will no longer be anything at all." (cited by Lloyd-Jones 1989, 369; italics mine)

The addition of the last three words, suggesting (correctly?) that Plato did not believe in the immortality of the soul, earned Etienne a trial and an execution in 1546. But perhaps he was not really afraid of death.

c) Luther explained by Nida (1964, 29):

"Luther translated: '...dass der Mensch gerecht werde ohne des Gesetzes Werke, allein durch den Glauben' (Romans 3:28). This word allein, making the last phrase mean "through faith alone", quite understandably provoked the ire of Luther's enemies, who insisted that he was adding to the Scriptures. Luther, however, contended that the added word was fully justified by the theological significance of the passage as well as the grammatical structure, even if it were not to be found in the original."
 

In each of these cases, the difference between expansion and addition is a question of conflict between the translator and a rather important institutional authority. In the last analysis, only authority determines what is or is not justified expansion.

Now, to return to our problematic hypothesis, can expansion be interpreted as a sign of felt inferiority? Perhaps there should be no simple reading in this sense, since the limits of expansion, which require authorisation, thus also concern a potential questioning of authorities, which surely often lie within receiving cultures and have little to do with any ST. And yet, considered carefully, even the more spectacular and complex cases - Luis de León, Dolet, Luther - are oppositional precisely to the extent that they place authority more with an idealised ST - Plato as text, the Bible as God's word - than with the institutions of the receiving culture. A certain critical cultural inferiority is thus implied, and may remain a hypothesis of interest.
 

Some translations become originals

What I have termed contradictory equivalence is the result of situations in which it is difficult to consider a TT1 as a relative equivalent of Y because its function is partly blocked by TT2, TT3... TTn, which may be either physically present or well known by the receiver. We have already come across some examples of this phenomenon. When the Y text "La Movida" is translated as TT1 "Die Bewegung" [The Movement] and then as TT2 "ein Schluckauf" [a hiccup], the translational series splits equivalence between the two possibilities, informing the receiver that neither TT1 nor TT2 is entirely satisfactory. Much the same creation of doubt is achieved when Y content is split through translators' notes, which might also be analysed from the perspective of contradictory equivalence.

When contradictory equivalence becomes a prolonged mode of reception, it must be regarded as a serious challenge to the translational status of the text concerned. But it is more often a transitory situation, a momentary doubt put to rest in the place of reception itself.

In theory, the receiver can resolve contradictory equivalence by regarding either TT1 or TT2 as the "better" or "more ideally equivalent" translation. This is done according to two suitably contradictory principles, one of them based on progress, the other on pseudo-originality.

The principle of translational progress says that the most recent TT tends to be regarded as the most ideally equivalent. In the German "Movida" example, the order in which the TTs are presented means that maximum value is accrued in the last-mentioned version, "Schluckauf". The linearity of the translational series is such that each TT which is proposed and then rejected passes its positive value on to the next TT, creating a sort of accumulative interest released in the last term of the series. In principle, the more terms in the series, the greater the value of the final TT. But this logic is obviously subject to major situational constraints, since although a TT like "Schluckauf" can function as a good equivalent of "Movida" within this particular translational series, it will not necessarily do so independently of the series. The value invested in the term can only be released within the sequence of German TTs.

Translational progress also has its historical projection, since any new version of a previously translated text implicitly presents itself as an improvement. After all, if a text like Wollschläger's German Ulysses (1975) were not somehow "more equivalent" than Goyert's previous German version, why should it have been undertaken and published? According to this common-sense principle, versions of the most translated texts should be coming progressively closer to ideal equivalence. But such sense is perhaps not quite so common, since progressive approximation to the ideal TT:Y 1 (here ideal equivalence of form and semantic material) would then be inversely proportional to the concomitant increase in the distance crossed by transfer (Y­ST). This in fact means that the further one is from the original production of a text, the more one's translation can be equivalent to that text. Which is at least an interesting notion of progress.

Clearly, principles of translational progress do not always correspond to the real way of the world. The main reason for this is that previous TTs rarely disappear as automatically or as gracefully as they should. They remain in cultural canons or receivers' minds, becoming pertinent points of departure in themselves, representing transfer not in terms of Y­ST but as TT2­TT1. That is, some translations complicate matters by becoming pseudo-originals.

Germans are extremely lucky to have a very good second translation of Ulysses. The French badly need a second translation but are unlikely to get one, since the first French version, which is very different from Joyce's own text, nevertheless received the author's imprimatur, benefiting at the same time from Larbaud's largely undeserved prestige and even influencing the final form of the original. It has thus attained pseudo-original status, so far blocking future development of the translational series. But such blocking is not necessarily eternal: Goyert's now surpassed German version also received Joyce's authorisation; luckily, its status as an equivalent was publicly contested in the press.

The principle of translational progress is thus to some extent contradicted by a right of first possession, such that the equivalent which is historically first produced will be accorded special consideration as a candidate for authorisation. For example, the Spanish translator of Adorno's essay on Der Untergang des Abendlandes could not agree with the initial Spanish translation of Spengler's title, but could not avoid recognising its historical authority as a precedent:

"The term Untergang is rendered herein as ocaso [sunset, ending] in every context in which what matters is the conceptualisation of Spengler's thesis. Only in citations of Spengler's book have we used decadencia [decadence], which was the term preferred by the translators of Der Untergang des Abendlandes." (Sacristán 1984, 25)

The Spanish reader of the essay is thus condemned to alternation between a TT1 recognised as authoritative and a TT2 presented as progressive, although coexistence of the two principles is in this case in keeping with the function of the essay as a progressive critique of Spengler's authority.

Perhaps the most common cases of contradictory equivalence are those arising from modern translations of the Bible, particularly in English, where the King James Version is very much a pseudo-original carried around in a collective cultural memory. When the Good News Bible states "Then God commanded, 'let there be light' - and light appeared", few English speakers could resist misreading it as a bad translation of "and there was light". Commenting on this example, a recent article states:

"If the original version of that line, which Coleridge considered 'the perfect poem', haunts Christians, it is because Good News has murdered it. Clearly, whatever else may be sacred, poetry is not." ("Language Debased: Modern English Abusage", The Economist, 22 December 1990).

Both Coleridge and The Economist appear to refer to the King James Version as a sacred source text (an "original version"), not as a translation. Contradictory equivalence may thus lead to an imaginary reversal of ST and TT positions. As Buber noted, the passage of time makes all writing seem a palimpsest of a more original text, and this is true "not just of translations but also of originals". For the modern reader, argued Buber, "even the Bible in Hebrew reads like a translation, and a bad translation at that" (cited by Kloepfer 1967, 83).

To the extent that they can block or even reverse translational series, pseudo-originals negate the principle of progress. But what do they do as representations of transfer, figuring the distance Y­ST? If the principle of progress means that one can come closer to ideal equivalence as one goes further away from the moment of original production, what might a pseudo-original have to say about such distance?

Littré's L'Enfer mis en vieux langage françois (1879) or Pézard's Pléiade translation of Dante into archaised French (1965) may be read as overt and partly ironic attempts to write pseudo-originals, since they both use archaic language. But what they do as representations of distance has not universally been appreciated. Meschonnic (1986, 78) criticises Pézard by arguing that Dante was not archaic for his contemporaries and so should not be archaic for the modern reader. In what kind of French, asks Meschonnic, would the translator have rendered Homer? (Littré had in fact already answered: Homer should be translated into thirteenth-century French. ) However, if Pézard's TT cannot be equivalent to the fourteenth-century Dante (ST), it can yet establish a level of equivalence based on transfer to the twentieth-century Dante (Y). But how does this relate to contradictory equivalence?

If an archaising translation is truly archaising, its manifest TT1 should refer to an implicit TT2, the text that the translator and the receiver would otherwise have exchanged as a non-archaic equivalent. When, to take an almost random example, Pézard gives the TT1:

 

"Ce seigneur des très hautes chansons,

qui sur tous les autres vole à guise d'aigle" (1965, 904)

a contemporary French reader will recognise the contextual import of the last phrase but will inevitably mentally translate it into a modern French TT2 like "à la façon d'un aigle" or "comme un aigle", which in fact has greater quantitative equivalence to the Italian Y text:

 

"Quel Signor dell'altissimo Canto,

Che sovra gli altri, com'Aquila vola" (Inferno IV, 95-96)

Prosodic justifications apart, it would clearly be wrong to argue that archaic translation in this case approaches equivalence with the Y text itself, which is in fact quite readable in terms of twentieth-century Italian. Equivalence is with neither ST nor Y. Pézard's archaism instead provokes awareness of the distance TT2­TT1, potentially running parallel to the distance Y­ST and thus proposing equivalence not to a text, but to transfer, to the temporal distance of Dante.

It matters little that there was no French language in Homeric times or that the King James Version is chronologically posterior to the Hebrew Bible. The essential value of these archaising or archaised pseudo-originals is not that they are condemned to misrepresent archeological orders, but that they take the distance Y­ST from the level of transfer and convert it into a value on the level of translation. That is, they use contradictory equivalence to represent the fact of transfer.
 

Some translations last as monuments

Benjamin claimed that translations themselves are the most untransferable texts, since their peculiar intertextuality has meaning adhere to them with "all too extreme fleetingness" (1923, 61). Translations are supposed to be temporary; they do not last as long as originals; they are disposable, like paperback books and ballpoint pens. Contradictory equivalence nevertheless allows some translations to outlive their translators, becoming hardcover fountain-penned pseudo-originals in themselves, overcoming the fragile mortality of most common forms of equivalence.

Does this pseudo-originality necessarily imply a strong subjectivity? Is it a serious threat to the translational status of the texts concerned? The above examples would seem to suggest quite the reverse. Split or diffuse equivalence tends to obfuscate the translator's discursive position; many of the great pseudo-originals were written by several hands, without a unified subjectivity; and archaising translation by definition requires a certain suppression of the I-here-now, avoiding the cultural traits of the translator's historical locus. Indeed, the more such translated texts are discursively distanced from the place of their production, from the potential I-here-now of the translator, the better they may age in step with their projected originals. Translational pseudo-originality should thus not be confused with the direct volition of an auctorial subjectivity. Translators do not become authors. The pseudo-original instead survives on its own, traversing time as a monument to self-suppressed subjectivity.

Eliot explained poetic creation in terms of the disappearance of the "thing" words are meant to describe. But he also explained creativity in quite translational terms, referring to anterior texts which refuse to disappear: "the existing monuments form an ideal order among themselves". Moreover, for Eliot, the right to enter this ideal order of texts required "a continual self-sacrifice, a continual extinction of the personality" (1919, 294-96). Perhaps the ideal extinction of individual personality is translation.

So much for what quantity would appear to say.

Next chapter