Transfer and translation work against belonging
ONE OF the recurrent shortcomings in general approaches to translation is that we are all constantly prepared to say how translation works but we rarely stop to ask what it might be working against. If, according to a not uncommon Manichean vision, translation were all goodness and enlightenment, what would then be the badness and ignorance it should oppose? And even if translation were found to be less than ideal, where should we look to find the values it lacks? In short, what is the opposite of translation?
I have so far avoided this problem by associating minimal non-translation with transliteration (the text is transferred but not visibly translated). But this is a facile evasion of the issue since, as we have seen, the decision not to translate is still a translational decision; it puts no illogical spanner into the works of translation as a process; its minor elitism is entirely describable in terms of the basic principles of translational quantities. Transliteration might be a degree of apparent non-translation, but it is certainly not anti-translational.
A far more formidable opposite lies not within translation as such, but in the negation of transfer, in denial of the necessary precondition for any translational situation. Maximum anti-translationality - a word not to be repeated - is surely to be found somewhere in non-transfer, in the state of texts whose movements are so restricted that they almost never become candidates for properly translational treatment. Translators are most successfully thwarted by the material they cannot get their hands on; translation can fail because some aspects of texts are difficult to transfer to a Y situation and sometimes impossible to convey to the situation of translational reception. The fundamental task of translators must thus be to work against the bonds restricting the movements of texts; they must attempt to attain some kind of maximum transcultural mobility towards a specific receiver. In short, translation works against constraints on transfer.
Why should there be constraints on the transfer of texts? Quite simply because - reading "transfer" in its juridic sense - certain texts are felt to belong to certain people or to certain situations, making the movement of those texts a question of bestowing some kind of ownership or right to full textual meaningfulness. Seen in this light, constraints on transfer can be thought of as bonds of belonging.
Here I am not particularly concerned with the mechanical reproduction of texts or with the way a strictly legalistic sense of belonging might formalised in terms of copyright laws and international publishers' agreements. I am more interested in the way texts themselves resist reproduction and transfer, undergoing a transformation of values when moved from their apparently rightful place. I am concerned with the way texts themselves are felt to belong, whatever the influence of external decrees.
In English, this belonging can be thought of with or without a subjective compliment; that is, in terms of specific direct ownership or in terms of a situational adequacy analysable as collective ownership. Texts can belong to someone, or they might simply be felt to belong in a certain social place such that they do not belong when they turn up elsewhere. Eliot's search for "the right words in the right order" can be seen as an idealisation of situational belonging; the experience of correcting students' translations is sometimes counterproof of the same belonging, since "the wrong words in the wrong order" often conform to every teachable rule but are still felt to be out of place. We know that texts belong because sometimes they simply do not belong.
How does belonging actually affect the transfer and translation
of texts? Any substantial response would have to consider and account for
various modes of belonging itself. In what follows, I shall briefly look
at the discursive function of texts as they enter actions (forming explicit
and implicit performatives); I shall touch upon several notions of social
embedding through contiguity, secrecy and forgetting; and I shall openly
recognise the powerful binding role of tongues. If it can be shown that
belonging in general is anti-translational, these various specific modes
of belonging should enable us to say something of interest about what kinds
of constraints translation has to negotiate. Awareness of these constraints
should then in turn provide some insight into the eminently pedagogical
problem of why certain aspects of translating are more difficult than others.
There are no solo performances
Explicit performatives are no longer performative once translated. An utterance like "I declare the meeting open" may well be translated as "Je déclare la réunion ouverte" or "Se abre la sesión", but only the first, non-translational version - whatever its tongue - can actually open the meeting. This is a simple example of a text that belongs in one place and not in another, denying full meaningfulness in the translational situation. We could say that translations of performatives are necessarily constative, or, to adopt the terminology of Christiane Nord (1988), the instrumental status of the non-translated text becomes documental when translated. In a sense, the instrumental text - the performing performative, the text as action - is owned by the people who originally open the meeting, and cannot be owned in the same way by those who receive only a report of it.
This is not a trivial example. In their hard discursive core, all contracts, treaties and official declarations are extended performatives. That is why they often require a clear distinction between original and translational versions, ensuring that problems of interpretation bear only on the non-translational text. Such translations are explicit auxiliaries to their originals and cannot become effective performatives.
In terms of translatability, the instrumentality or effective performance of performatives obviously has nothing to do with language choices. It does not matter which language is used to open the meeting, since only the first language to be used will be the one to belong. What resists transfer is not language, but the text's situational ability to become part of an action. We are concerned only with this ability to resist transfer.
Now, although performatives do not concern tongues they do very much concern discourse. Austin noted that properly performative verbs are marked by an asymmetric relation between the first-person indicative active and other persons (1955, 63) and that, in English, this peculiarly performative first person cannot be used with present continuous forms (47). That is, "I declare the meeting open" is performative, but "You declare the meeting open" and "I am declaring the meeting open" are not. The grammatical singularity of explicit performatives is thus intimately associated with the I-here-now, partly explaining their peculiar power to belong to their situation of original production. Moreover, since performatives are linked to a discursive feature that is by definition unavailable to the translator - the translator has no I-here-now - , a translated performative can only be expected to lose instrumentality.
Of course, this is not a sufficient explanation of why performatives belong in certain situations and not in others. The place within which performatives are instrumental also depends on selection and transfer to an appropriate implied receiver or second person. Despite the necessarily unique opening of meetings, examples like Pope John Paul II's Christmas blessings manage to remain performative in numerous languages because each blessing is directed to a different "you" invited to participate in the performance. Criteria of transferability stand or fall on the basis of this relation to the second person. And yet relatively little has been made of the implied receiver of performatives.
When Austin transforms the implicit performative "Go!" into the explicit performative "I order you to go!" (32), the restitution concerns the second person just as much as the first. Similarly, when a universalistic level of performance is projected onto an implicit operator like "I say..." (the performative operator implicit in all utterances), it would be equally legitimate to rewrite the operator as "I tell you that...", incorporating the second person who, after all, has no reason to be excluded. This means that the discursive person most directly involved in the workings of performatives is not really the "I" of the I-here-now, but an inclusive "we" ("I" plus "you"). The most general implicit operator should thus probably be explicated not as "I say..." but as something like the mathematical "let", rewritten as "let us ('I' plus 'you') suppose that...". Alone, the first-person singular can make as many propositions in as many languages as it likes, ordering and organising the world to suite its own desires. But a particular proposition can only become a non-hypothetical performative when an implicit "we" incorporates a second person.
There are several quite obvious reasons why the second person is necessary if performatives are to be instrumentally performative. It should not be forgotten that someone has to receive and understand the text; someone must be prepared to accept the speaker's social power and right to perform. The second person or implied receiver creates space for this someone - a physical receiver - to understand and accord authority to the first person. Such a second person, able to make a performative perform, can be described as potentially "participative".
There are several parallel theories that should be noted here.
In terms of Arendt's categories, a performative structure may well be a happy "fabrication", but it cannot become an "action" until at least two persons are involved (1958, 188ff).
Or again, according to a similar terminology, if things are to be done with words and the words are to be received and understood, it follows that, as BakhtinVoloshinov argued, "every act of understanding is dialogic" (1930, 167).
Participation and dialogue are the conditions of having a text belong in a certain situation as a performative. Yet democratic idealism should not be considered universal. Some degree of non-reciprocal authority is usually also operative, since the first person can only perform if the second person recognises the right to perform. That is, if the first person is to do things with words, the second person must initially consent to listen to things within words, and there is no rational guarantee that this consent will be reciprocal. As Austin recognised, the belonging of performatives is based on an asymmetric "we".
One final reminder before we leave successful performatives:
Perhaps obviously, the existence of discursive positions does not mean
that there are always real people available to take up those positions.
As a text, Finnegans Wake projects an implied receiver able to concretise
maximum polysemy, but it might fairly be supposed that no flesh-and-blood
reader except perhaps Joyce has been able fully to occupy that position.
A distinction must thus be maintained between the asymmetric "we" which
discursively presupposes or invites participation in textual performance,
and the social participants - actual senders and receivers - to whom the
text as real performance can be attributed.
Distance can break performance
Since the discursive second person is defined simply as a potentially participative "non-I", there is in principle no distance between the two positions concerned. If the performance is to be fully performative, both "I" and "you" must be mutually present.
Unfortunately, beyond the world of discursive positions, there is always time and space between the entities filling the "I" and the "you". This extra-textual distance interests me because it might be able to determine thresholds beyond which certain texts cease to be properly instrumental. The isolated question "What is the time?" - a key example for Bakhtin's dialogic principle (1930, 164ff.) - is not only discursively unique in terms of the "now" of each performance, but also implies the presence of a receiver located very close to the I-here-now. That is, unavoidable factors like the distance involved in the passage to a real receiver, the processing of the message, the formulation of a reply and the sending and reception of the reply should, according to discursive logic, not involve any time at all, since the "now" of the question sent should correspond to the "now" of the reply received. And yet it is obvious that the entire double transfer process must take time and the two "now"s cannot be the same. The discursive logic is at odds with the fact of material distance.
For how long can discursive logic win out? For how long can participants pretend that there is no distance between them, that the "now" of the question is referred to by the "now" of the reply? Within a few seconds perhaps, if the speaker's requirements do not include extreme exactitude. But a minute would create an anomalous and slightly absurd situation, and it is difficult to imagine a situation in which the isolated question could be written. The utterance "What is the time?" can only be transferred away from its "I-here-now" within certain pragmatic limits; it has a certain elasticity, a capacity to stretch out and remain valid as meaningful communication. But beyond a certain point, this elasticity will be broken and communication will be defeated by distance. The utterance will reveal its limited transferability.
It is possible that the relative transferability of a text can be analysed simply by isolating and assessing all those elements that refer to or imply an I-here-now calling for participation by a second person.
Hegel discovers transfer (1807, 151):
"To the question, 'What is the Now?' we reply, for example, the Now is night-time. To test the truth of this certainty of sense, a simple experiment is all we need: write that truth down. A truth cannot lose anything by being written down, and just as little by our preserving and keeping it. If we look at the truth we have written down, look at it now, at this noon-time, we shall have to say that it has turned stale and become out of date."
But exactly who is this "we", beyond the supposedly naïve consciousness?
And why is the "now" of night-time marginally more transferable than the
"now" of "What is the time?"
Transferability has second-person thresholds
The analysis of language as action has its limits. After all, not all texts require the same kind of immediate response as "What is the time?". If I write and send a letter in search of a reply, the second person should presumably be located beyond the range of my voice but not beyond the range of my expected lifetime or the value of the stamps put on the envelope. In this case, it would make no sense to ask for the time. However, if I should write a novel like Robbe-Grillet's Les Gommes, where time receives a cotextual literary structuration, I would in principle be content and even gratified to think that I might reach a reader beyond my life span, beyond calculated transmission costs, beyond all earthly possibility of participative reciprocity, and it would indeed be meaningful to have one of my characters ask "What is the time?".
Participative second persons can be sought on at least three levels, associated with three different degrees of transfer:
- Shared transfer: For some texts, the implied receiver calls for an actual receiver who is situationally present so that a reply can reach the sender and discursively share the same I-here-now. That is, the reply should ideally be in the same present tense ("The time is...", or "Do you mean that...?"). Transfer beyond this restricted and usually oral situation will leave the text decidedly out of place. (Example: "What is the time?")
- Referential transfer: Where the physical receiver called for by the implied receiver need not share the same I-here-now but should be able to refer to at least one of these coordinates, transfer may well extend beyond short-distance situational constraints. But if some reply is required - if the text is not indifferent to the location of the second person - , factors of age, space and time impose practical constraints on transferability. (Example: my monthly letters to my mother)
- Indefinite transfer: Some texts need neither
reply nor reference to an I-here-now they can in principle be transferred
indefinitely, within the constraints of world and time. (Example: Les
The thresholds between these three modes are obviously imposed on a practical continuum that goes from short-distance to mid-distance to long-distance transfer, regulating the degrees to which increased distance may break initial elasticity.
It might prove possible to relate these thresholds to
particular text-types, perhaps along the lines of the categories Reiss
(1971) adopts from Bühler, originally based on the three linguistic
persons. But such correlations are more apparent than real, since transfer-based
analysis concerns the position and role of the second person within an
inclusive "we"; it is in principle indifferent to the relative dominance
of first or third persons or to relative focus on form, content or effect.
In fact, the problem of text-types could turn out to be something of a
red herring in this context. Although it is true that the limited elasticity
of certain texts restricts their ability to remain properly instrumental
after various degrees of transfer, there is no fatal determinism at stake:
a text can be transferred beyond its initial elasticity and assume a new
performative status; or the same discourse can be rewritten, with the same
pronominal structure but with more elaborate coordinates, in order to heighten
its transferability. That is, texts can be transformed in order to pass
over one threshold or another, and the modes of rewriting need not affect
text-type models based on criteria other than those of transferability.
This is one reason why I do not expect much help from text linguistics
on this point.
Textual worlds increase transferability
The possibility of breaking bonds of belonging and yet maintaining instrumental status clearly has something to do with referentiality and the existence of codes shared by separated sending and receiving positions. But it would be inadequate to describe transferability as a quality of the referents or the codes themselves. Its conditions are not strictly semantic; its investigation does not present problems as to whether or not we really understand the intentions that originally gave rise to the utterances. Transferability has much more to do with how well the text itself can function as a semantic world.
Let us accept that the question "What is the time?" is relatively untransferable because its referent - the time - is highly specific to the "now" of its I-here-now and because its assumed code - mechanical clock-time - is not as universally available as the sun and the moon. Now, accepting these restrictions, is it possible to imagine modes of textual presentation in which the same question could become eminently transferable? Obviously, the utterance could be part of a mathematical problem, explicitly conditioned by a series of cotextual astronomical observations and interpretative codes such that the reply should always be the same; or, as we have noted, the explicit cotext of Les Gommes uses similar principles to structure clock-time in terms of an Oedipal code, able consistently to provide the question with far more than its isolated meaningfulness. In such cases, transferability is significantly increased by textual presentation not only of a general semantic world, but also of the asymmetric "we" for whom the utterance is initially meaningful. Untransferability can thus be understood as a result of the absence of such cotextual presentation; it ensues from dependence on unexpressed or implicit context.
More metaphorically, if a text cannot be taken away from its owners, it is sometimes possible to convert the owners into signs and to transfer them along with the text.
Several objections might be raised here. It could be said
that the isolated question "What is the time?" is relatively untransferable
because it refers to a fixed reality belonging to a world external to the
discourse itself, whereas the possible mathematical or literary cotexts
make the question transferable simply because they are fictional representations
of that world. Are we then confusing reality with fiction? But surely the
information sought is in neither case external: it is merely an encoding
of the I-here-now internal to the utterance itself. In relatively untransferable
cases like the naked "What is the time?", the appropriate codes remain
unsaid and implicit; the I-here-now cannot be moved and the implied receiver's
response must be specific. In relatively transferable examples like Les
Gommes, the codes are inscribed and explicit; both the I-here-now and
the implied receiver's response remain constant and can be moved with the
text. From the perspective of translation studies, the apparently greater
reality of codes like clock-time is simply a function of their common use
and acquired naturalness within certain cultures. Their transferability
does not concern any assumed fictionality - clock-time is just as much
a convention as is Oedipal fate - ; it is simply a question of what is
implicit or explicit in the text concerned.
Transfer may call for explication
In principle, the more explicit the codes, the more transferable the text and the weaker the belonging to an original I-here-now. This might be understood in terms of Bernstein's distinction between universalistic and particularistic meanings:
"Universalistic meanings are those in which the principles and operations are made linguistically explicit, whereas particularistic orders of meaning are relatively linguistically implicit. [...] Where orders of meaning are particularistic, [...] then such meanings are less context independent and more context bound. That is, tied to a local relationship and to a local social structure [...]. Where meanings are universalistic, they are in principle available to all because the principles and operations have been made so explicit and public." (1970, 164-5)
Bernstein's use of this distinction is broadly based on observations of class language: working-class speech tends to use implicit codes and is more context-bound dominant-class language is generally more explicit and context independent. This sociolinguistic projection is highly pertinent to translation studies in that it breaks down the all too frequent assumption that transfer starts and ends with "given" societies or "given" tongues. Since societies and tongues are very heterogeneous and badly delimited entities, principles of transferability should be quite subtle and sensitive to complex local conditions. Heterogeneity does not stop at class barriers: upward and downward social mobility create numerous internal modes of transfer by which it is possible to go from one order of meaning to another. The context-dependent "What is the time?" might be rewritten as a context-independent mathematical problem; the shorthand codes used in scientific theories could be rewritten as explicit prose descriptions in works of divulgation; a literary text may use multiple explicit codes to recuperate and regenerate the implicitness of unelaborated spoken language... And if all else fails, the transferability of a text can be enhanced by attempts to make its implicit codes explicit through monumental and reactionary academisation in an extra-textual space, in the way that standardisation of the French language was partly an attempt to ensure the transferability of neo-Classical drama written in that language. Through all these movements, the same basic principle holds: the more explicit the codes, the more transferable the text.
Absolute explicitness is rarely transferred
Absolute explicitness can be understood as textual completeness, as the ideal presentation of all the axioms, definitions and descriptions needed to understand the exact meaning of each proposition made. This ideal would require the creation of a specific self-contained world wholly inscribed in the text, assuming no prior knowledge and making no presuppositions of extra-textual material. Such a text would remain meaningful and stable throughout even the longest cultural trajectory. But it would be fair to say that there exists no such text, or at least, that a text as wilfully explicit as Russell and Whitehead's Principia Mathematica (1910-13) must effectively defeat its transferability by becoming so painstakingly dense as to be relatively unimportant beyond fairly immediate cultural horizons. If everything is put in the text, what participation is left to the actual lives of non-textual readers?
Some distinction must thus be made between that which
can be transferred and that which has no reason to be transferred. Unlike
the perverse density of the Principia Mathematica, the relative
explicitness of Euclidean geometry has enabled its stability to be maintained
throughout a long series of intercultural transfers and adaptations, basically
because it remains an important tradition for pedagogical purposes (Barnes
1977, 42-43). So there are certain practical limits to textual explicitness:
the transferred text must remain of some importance to the social interests
promulgating the transfer, and excessive density or textual length can
sometimes destroy this value.
Belonging may be a tone of voice
The movement from implicit to explicit meanings, from the particularistic to the universalistic, is also a movement from first-person and second-person bonds to the neutrality of the third person. This is why the very fact of transfer forces translators to seek and elaborate the anonymity of third-person worlds, doing away with the I-here-now as far as is humanly possible. But some particularistic meanings still refuse to be transferred. The Bavarian "Mia san mia" ("we are we"), for example, is no doubt best rendered as "Bavarians are Bavarians" (or even "Bavarians will be Bavarians", on the model of "boys will be boys"), but this third-person utterance has obviously lost its Bavarian particularity. The words no longer accomplish the act of being Bavarian. Indeed, one senses that the words are no longer instrumental once transferred and translated.
How is it that a phrase with no explicit performative and no obviously restricted I-here-now can nevertheless so strongly resist transfer? Is it merely because of the first and second persons inscribed in the semantic "we" of "mia", or is it because of the particular quantitative form of the term itself, the peculiar shape of the mouth required to pronounce it, a special intonation and certain sliding vowels which only slide so and are intoned thus in a highly localised part of the world? The words themselves belong to the "we"; they will probably continue to belong to participative Bavarians whatever their referential or discursive status. In this sense, their distinguishing traits carry, implicitly, a sociolinguistic "we" defined by untransferable familiarity. The utterance activates this familiarity; it activates all that remains unsaid because already understood, all that is felt to be performative, understanding but not saying "I declare myself to be Bavarian" and "I declare the 'you' of my inclusive 'we' to be Bavarian", as well as "I declare the 'you' excluded from my 'we' to be non-Bavarian." And so to the hills.
As much as language might appear to be public property
- the res publica of the Republic of Letters - , most specific uses
of language indicate some degree of belonging to a "we". This collective
first person could be a culture, a social group, a profiled sender and
receiver, and sometimes an individual sender like the Joyce who is both
the implied author and reader of Finnegans Wake. Belonging can depend
on a tone of voice, an accent, or an endemic choice of words. And in theory,
the stronger the belonging, the less transferable the text.
Belonging may work on implicit knowledge
The following is an Australian aboriginal chant recorded at Groote Eylandt in 1948:
"Nabira-mira, Dumuan-dipa, Namuka-madja, Aï-aïjura."
The text has been rendered as follows:
"At the time of creation, the Nabira-miras, father and son, tried unsuccessfully to fish with spears. Their spears were changed into the cliffs called Dumuan-dipa, and they themselves became the cliffs known as Namuka-madja, close to the island of Aï-aïjura." (Mountford 1956, 62-63)
As a translation strategy, this could be described as absolute equivalence plus expansion, since all the surface elements of the Y text have been transliterated in the TT. The added quantity is used to explain what the four proper names, each of them repeated in the translation, could have meant in the projected original. That is, ST context has become TT cotext; there has been a radical movement from the implicit to the explicit. Everything appears to have been rendered. Indeed, the TT allows one retrospectively to read the four proper names of the aboriginal chant as forming an English-language text, as a kind of back-transliteration. Such would be maximum translation as cultural appropriation. But has everything really been transferred?
Since the chant has no explicit I-here-now, it is difficult to analyse its belonging in terms of performative verbs. And since its surface language can be read as forming an English text, there is no reason to assume that these sounds in themselves convey any sacred or specifically aboriginal sense of belonging. But the chant unmistakably belongs more to the aboriginal tribe than to the people of the translator Mountford, if only because it is through the repetition of such ritual texts, through repetition of the spiritual time when humans and land features were interchangeable, that the participating aborigines affirm their institutional status as the tribe of what the Dutch named Groote Eylandt. It is not difficult to see the proper names as representing ancestral dreamtime characters whose continued natural presence gives the tribe at least spiritual authority for its existence on the land. The translator and his readers can no doubt understand and explain the basic structure and social function of the chant; some translators and readers might even believe in the narrative content thus attributed; but only the absent aborigines of the recorded text could fully participate in the discursive action by which they reaffirm what we might see as their ownership of land, and what they would prefer to see as their belonging to the land. In a sense, without any recourse to pronouns or specifically performative linguistic forms, the cultural knowledge presupposed by the chant enables its participants to sing their specially instrumental version of "Mia san mia".
Gladys Corunna on the transfer of untransferable categories (in Morgan 1987, 306):
"I suppose, in hundreds of years' time, there won't be any black Aboriginals left. Our colour dies out as we mix with other races, we'll lose some of the physical characteristics that distinguish us now. I like to think that, no matter what we become, our spiritual tie with the land and the other unique qualities we possess will somehow weave their way through to future generations of Australians. I mean, this is our land, after all, surely we've got something to offer."
In opposition to Mauss's belief that "the categories live and die with
their peoples" (see note 1), dying peoples must place hope in the fact
that their "we" can always become something else.
Belonging may work on forgotten knowledge
One further point should be made here. The textual implicitness associated with an element like a proper name does not necessarily mean that all participants in the corresponding performance are able to say exactly what the proper name signifies. Many elements are left implicit because, although they have gained familiarity through repetition - and perhaps because of excessive familiarity - , they are no longer fully understood. In the case of Australian aboriginal songs, von Leonhardi found that "most of the men sing the chants without understanding them, and women and children witness sacred rites without having knowledge of their meaning; but the old men, who carry the traditions, know precisely what these acts mean, in all their details, and are able to explain them" (in Lévy-Bruhl 1935, 114). The initiated can thus guard the secrets of belonging, but the texts will still belong simply because they are felt to be familiar or are associated with childhood, with the family or with tribal tradition. Although European nursery rhymes have similarly had their meanings confined to specialist works of reference and research, they are still felt to be "ours" and are still passed on to future generations.
Such belonging without knowledge is very much a problem of translation. It might explain why I experience considerable difficulty when trying to translate hackneyed Australian terms like "the bush" into Spanish: I simply do not know what these words really mean in Australian culture (the culture of which I nevertheless consider myself a "native informant"). As a general term, "the bush" has been repeated so often and everyone is supposed to be so aware of what it means as a mythologised anti-city that no one bothers to explain its exact boundaries. Does the bush include desert? Does it necessarily exclude farmland? If one "goes bush", does one necessarily go to the bush? Answers vary. If there ever was an exact meaning, it has been forgotten or covered over with myth.
The bonds of belonging thus depend not only on the collective knowledge that enables certain elements to remain textually implicit, but also on the forgotten knowledge that has enabled implicitness to become familiar and tend towards ignorance. In this second case, the translator's task is all the more difficult because there is textual indeterminacy well before the moment of translating. The best the translator can do is often to reproduce this same indeterminacy, the same cultural lacunae, as a trace of the other's belonging (cf. Pym 1992b).
Indeterminate belonging in Patrick White's "Down at the Dump", into Spanish:
"Do we [Australians] know exactly what our words mean? Can we identify 'the stench of crushed boggabri and cotton pear'? Which years are associated with 'Customlines' and 'Breath-o'-Pine'? What is meant by 'There's some boys Sherry knows with a couple of Gs'? Translation is often a negotiation of details that are mysteriously only half-known in the source culture. It is sometimes necessary to resort to strategies rarely repeated in the textbooks: when in doubt, bluff or leave it out. I must admit that I was not sure enough of 'cotton pear' to reject 'acanto' (prickly thistle); 'Breath-o'-Pine' became 'Centella' (a common detergent product remembered by Spaniards who also remember 'customes', the Spanish version of customlines); and as for 'Gs', they have become 'coches' (cars), but will probably disappear if not confirmed." (Pym 1989a, 667)
Two years after publishing these questions in Australia, I still have
serious doubts about "Gs", and Patrick White died without replying.
The tongue carries forgotten belonging
The analysis of belonging is important not only as a way of approaching a dynamic conception of cultures, but also as a partial explanation of the peculiar powers retained by language systems as tongues. Although translation is not an exclusively linguistic phenomenon, it is very difficult entirely to ignore the special power that tongues have over the way translators work. As important as it is to know how many kilometres or centuries lie between a Y text and its place of maximum belonging, affective possession is usually more immediately altered by a jump from one language to another. Tongues would thus appear to have the potential to become symbolic places in themselves.
One explanation of this peculiar power lies in the external symbolisation of tongues as entities subject to collective ownership. If a tongue can be taught and thus transferred, then it can be sold, bought and possessed, as is implicitly recognised in the question "How many languages do you have?". But the possessive aspect of the symbolisation and transfer of tongues nevertheless remains a mostly metalinguistic affair, concerning what can be said about the tongue rather than what the tongue itself might have to say. Such overdetermination can equally affect ways of dressing, colour patterns, music, warfare, or indeed any other communication system. External symbolisation is thus more likely to express existing bonds - power relationships which exist outside the tongue - rather than explain why the tongue as such should occupy a privileged place amongst communication systems.
Of the numerous attempts to explain the power lodged within the tongue itself, at least two are worth retaining. The first is a principle of translatability, as found in Hjelmslev:
"In practice, a language is a semiotic into which all other semiotics may be translated - both all other languages and all other conceivable semiotic structures. This translatability rests on the fact that all languages, and they alone, are in a position to form any purport whatsoever; in a language, and only in a language, can we 'work over the inexpressible until it is expressed'." (1943, 109)
Hjelmslev's position should not be associated with naïve axioms of unlimited expressibility and creativity, which not only fail to explain why languages are difficult to translate, but necessarily attract at least a few voices of reasonable dissent: BakhtinVoloshinov, for one (or two), believed that the semantic fields of music and painting could not adequately be structured in any tongue (1930, 62). Emphasis should instead be placed on the value of what Hjelmslev cites from Kierkegaard as the possibility of "working over the inexpressible until it is expressed". A labour-value principle of effability would enable the statement to be relativised in the following way: "...in a tongue, translation from other communication systems is always possible and involves considerably less labour than that required to translate from a tongue into any other communication system". The formulation is perhaps not as lapidary as Kierkegaard's, but seems less likely to invite facile evasions of its consequences.
If translation into a tongue is easier than translation into any other communication system, a simple principle of least effort would suggest that the contents of all media, all means of expression - and thereby all the various fields of human activity - are most likely to find their common ground expressed in the space of the tongue. Despite synaesthesia, despite cinematographic montage, despite an age supposedly given to visual communication, it is in language that a colour can most easily enter into syntagm with a mathematical function, that rocks can become contiguous with metaphysics, and that the past can become worry about the future. Like a vast marketplace, the tongue enables exchange relationships to be established between the most diverse materials, between the most disparate levels. The result of this massive convergence is a certain fusion and confusion, not only of the contents thus brought into contact, but of the lexis itself, whose corresponding principles of thrift and modulation lead to homonyms and synonyms, the fundamental instruments of semantic contamination generating fleeting metaphor and ever more subtle and intricate connotation. Part of the privilege and power of the tongue thus ensues from its capacity to tap numerous different aspects of human activity and to overlap, superimpose and even bury them within more complex wholes. The general result of these processes may be described as embeddedness.
The second possible explanation of the tongue's peculiar power has already been touched upon in our consideration of the I-here-now: only tongues, it seems, systematically incorporate the basic instrument necessary for the asymmetric expression of subjectivity:
"It is remarkable that amongst the signs of each tongue, whatever its type, era or region, there are always 'personal pronouns'. A tongue without expression of the person is unthinkable." (Benveniste 1966, 261)
This is by no means an idle observation, especially when aligned with Benveniste's explicit restriction of active pronouns to asymmetric first and second persons, and his recognition of this same asymmetry in performatives requiring "authority of the speaking subject" (1966, 273). If the basic structure of asymmetric subjectivity is one of the intrinsic elements of languages, it is surely no accident that the tongue should be symbolised as a privileged means of expressing belonging.
The tongue thus at once binds together countless aspects of social life and incorporates the essential structure of that which can potentially move between different aspects of social life, the discursive subject.
Transfer and translation themselves might not be especially linguistic, but their opposite or opponent - the belonging of texts - certainly is.
We have seen several of the many ways in which bonds of textual belonging can operate, creating resistance to transfer and translation. A text can belong to a specific "we" through its status as an effective performative, through surface markings indicating specific group ownership, through incorporation of specially semanticised implicit material, through the indetermination of certain implicit material, or through its formulation in one or several symbolically overdetermined tongues. All texts activate some if not all of these mechanisms to one degree or another. Which is to say that all texts to some extent belong to a complex social "we".
However, few texts are so extremely owned as to be untransferable.
The modes of belonging outlined above do not normally reach the level of
"cultural specificity"; they do not depend on immutable non-corresponding
linguistic categories or impenetrable epistemological precepts; there can
be no question of associating the fact of belonging with any sterile dichotomy
between relativism and universalism. It should be stressed that the degrees
of attachment to a specific "we" vary considerably and are open to considerable
transformation and manipulation. This is especially true in politics and
publicity, the discourses most directly concerned with the formation and
identification of inclusive/exclusive first-person plurals. A theory able
to mistake belonging for epistemic relativism is also likely to trust the
fundamental myths of politicians and publicists. This is one reason why,
as a humanist wary of exclusive collectivities, I prefer to analyse belonging
in terms of the mechanisms outlined above.
Cultural embeddedness conditions translational difficulty
The problem of the relative difficulty involved in translating different texts concerns two quite distinct domains: the training of translators, in which texts have to be selected for courses at various levels, and the remuneration of translational work, where different rates are sometimes presumed to correspond to different degrees of difficulty.
Unfortunately, these two domains are commonly confused. Since technical and specialised translations are deemed the most valuable in terms of the supply-and-demand economy within which translators are paid, it is assumed that these same texts must be the most difficult from the pedagogical point of view, where supply-and-demand has little to say.
But analysis in terms of the relative belonging and embeddedness of different texts suggests that the real scale of difficulty is more likely to be precisely the reverse, since technical and scientific texts are in principle relatively explicit and untouched by complex social embedding.
Using this kind of analysis, translational difficulty might partly be defined as a quality of the Y text - not the mythical ST - immediately prior to the moment when the translator begins to translate. A text to be translated is obviously not more or less difficult simply because the translator is good or bad or because the translator's client is liberal or demanding. The kind of difficulty we are concerned with here has nothing to do with one-off shortcomings. But it has everything to do with how much the text presupposes its place of production and how easily it can enter the proposed place of translational reception. The kind of difficulty we are interested in is thus a function of the directionality and relative completeness of a text when in the Y position.
The hypothesis to be investigated is that translational difficulty is determined not by the the actual operations that take place in the transition from Y to TT - the number of such operations is better seen the manifestation of difficulty - , but by the relative ability of a text to escape embeddedness and belonging.
If the problem has been formulated correctly, the most transferable texts should also be those which are most easily translated. One way of drawing out the consequences of this hypothesis is to compare it with the observations and conclusions of other analyses, particularly those applied to pedagogical situations.
A syllabus project prepared by the Barcelona translation school (Neunzig et al. 1985, working from Reiss 1975 and Arntz 1982) presents a "text difficulty profile" based on no less than fifteen criteria, with four categories for each criterion. Most of these criteria are fairly self-evident, including the reference to high information density as a feature of difficult texts. But there is no criterion for legitimate reduction or expansion, no consideration of ways of creating or overcoming embeddedness, and thus no reference to the strategies by which embeddedness might be altered in translation. This significant absence - based on the assumption that translational equivalence always requires quantitative equality - underlies some fairly dubious principles like the following:
- Subject matter: "general" texts are held to be easier to translate than "specialised" texts.
- Social location: communication between a non-specialist sender and a general public is easier than communication between specialists.
- Register: "everyday" language is easier to translate than "technical" or "literary" language.
- Distance: "cultural and temporal proximity" presents less problems than does extreme cultural and temporal distance.
Although only the last of these points refers explicitly to intercultural transfer, failure to consider embeddedness can be seen as a common denominator leading to shortcomings in all four criteria.
What is a "general text" if not a culturally embedded text, of considerable translational difficulty to the extent that it is used in or draws on numerous different contexts? The Barcelona analysts are not alone in their failure to understand this. Jean Delisle, for example, openly recommends the use of such texts in the teaching of translators, since "initial training in the use of language is made unnecessarily complicated by specialised terminology" (1984, 25). This sounds quite reasonable. But in saying this, Delisle falsely assumes that "general texts" are automatically free of terminological problems, as if magazine articles, publicity material and public speeches were not the genres most susceptible to embeddedness, textually bringing together numerous socially contiguous and overlapping contexts in their creation of complex belonging. A specialised text may well present terminological problems - the translator might have to use dictionaries or talk with specialists before confidently transcoding the English "tomography" as French "tomographie" or Spanish "tomografía" - , but this is surely far less difficult than going through the context analysis by which Delisle himself takes seven pages or so to explain why, in a newspaper report on breast removal, the expression "sense of loss" - superbly embedded and indeterminate in English - cannot be translated (for whom?, why?) as "sentiment de perte" (1984, 105-112). No truly technical terms are as complex as this most vaguely "general" of examples! The extreme difficulty of such texts involves negotiation of the nuances collected from the numerous situations in which an expression like "sense of loss" can be used and which, for reasons which escape purely linguistic logic, have never assumed the same contiguity with respect to "sentiment de perte". In foregoing the use of specialist dictionaries and collaboration with specialist informants, believers in the ease of "general texts" risk condemning students to the far greater and more mysterious difficulties of perpetual indetermination through embeddedness. The outcome tends to be not only professional guilt in those apprentice translators who are not quite as well embedded as are the ST and the TT, but also a severely inaccurate impression of the translator's profession, since truly general texts, which must be recreated afresh if parallel embedding is to be produced, are rarely translated, except in the training of translators.
The above argument against presuppositions of general subject-matter can be repeated for the Barcelona school's similarly dubious projections of general readers or the false naturalness of an "everyday" discursive register. At base, these are simply different manifestations of the same lack of realist thought, since the very generality of the general mostly denies any reason for transfer and thus cannot adequately explain translation as a purposeful activity.
However, Barcelona's fourth criterion, which attempts to articulate difficulty according to distance, does indeed address the fact of transfer and the problem of belonging. At first glance it is only common sense to say that it is easier to translate from a familiar context than from distanced modes of belonging about which virtually nothing is known. As has been argued above (with respect to "What is the time?"), there are theoretical reasons why increases in the distance YST tend to break the performative capacity of texts and result in relative untransferability. But is good common sense always borne out by common practice? Is performance theory always pertinent to translational difficulty? The answer is perhaps not as obvious as it seems, especially since it has been suggested - by George Steiner (1975), Seleskovitch and Lederer (1989, 137) and indeed in the above discussion of contradictory equivalence - that certain modes of translational difficulty may in fact be inversely proportional to the distance crossed in transfer.
Proximity can be more problematic than extreme distance. As we shall see in further analysis of "La Movida" - which continues to move in the next chapter - , the French translation of the Spanish term, although geo-culturally the closest, was by far the most complex and drawn out, and thus presumably the most difficult to produce. This might be understood in terms of a basic home truth like "familiarity breeds contempt", or a less common economic truth like "common borders breed agricultural competition". It is mostly forgotten that the cultures best known by the translator are also likely to be the cultures about which TT receivers have the most preconceptions and often misconceptions. The closer you are to a culture, then the more you are likely to be aware of the complicated mechanisms of belonging that make that culture's embedded texts difficult to transfer. If one knows almost nothing about such embeddedness, texts may well radically change their value in transfer, but they will effectively become far less complicated to translate. In questions of cultural distance, a certain ignorance might be bliss. Despite performance analysis, complexity would thus seem more likely to result from cultural proximity than from extreme distance.
Does this conclusion contradict our comments on performatives and belonging? I think not. Consider for a moment the essential difference between short-distance transfer ("La Movida" from Spain to France) and longer-distance transfer ("La Movida" from Spain to Newsweek). In the first case, complexity results from the numerous levels on which the two cultures are connected and interact both before and after the particular transfer concerned; the transfer situation itself has something of the heterogeneity of intracultural movements; the receivers of the translation are likely to adopt a participative attitude towards the text. In the second case, however, where there are far fewer contacts and the transfer is directed towards a more restricted and possibly more ignorant public, receivers of the translation tend to be positioned as passive consumers and the concealed problems of embeddedness and belonging become almost irrelevant. There can be no doubt that a text like "La Movida" will retain more of its original belonging in cases of short-distance transfer, but this is precisely why proximity creates more problems for translation. In the case of long-distance movements, where transferable belonging tends to zero and purposes are more highly focused, translated texts may be grossly exotic, inappropriate and appropriative ("The Happening"), but they will be considerably easier to produce.
The question of relative distance can thus be interpreted as another version of the false dichotomy between general and specialist texts, general and specialist publics and general and specialist language. There is no substantial reason why proximity should fall on the "general" side and long-distance transfer on the "specialist" side. The more correct conclusion is that maximum translational difficulty is associated with maximum generality.
These are not merely perverse arguments. According to my own experience - and allowing that I work in active collaboration with clients and specialists - , it is easier to translate a technical text than a newspaper article; it is easier to address a closely profiled receiver rather than a general public; it is easier to translate Australian aboriginal texts into Spanish than to describe Protestantism to Catholics; and it is strictly impossible to find any text written in everyday language.
If one seriously wants to analyse difficulty, it is more useful to consider degrees and modes of transfer and belonging than common-sense illusions of naturalness.
The liberty of extreme distance according to Borges:
"The language of Whitman is entirely contemporary. It will be centuries before it is dead. Only then will we be able to translate in with complete liberty." (Spanish translation of Leaves of Grass, 1969, 22)
But then, when the English language has died, who will be left to own
Embeddedness or the collective belonging of texts is ultimately resistance to the movement of texts. The way to overcome this belonging is then textual elaboration or explicitness, the bringing to the surface of presuppositions and the reasonable cleaning away of connotations. It follows that, in order to make a text more transferable, additional work must be invested to make it more explicit, just as an additional initial charge is required if electricity is to be transferred over a considerable distance. It also follows that, if this additional work has not been carried out at the point where transfer begins (ST), it can to some extent be put in at the point of translation (Y). That is, translation itself can heighten the transferability of texts, not only for immediate TT receivers, but also for the future receivers of all further TT versions.
If texts belong, transfer and translation work against belonging, in favour of increased potential movement.
But what then is the actual movement of texts? And why,
seriously, should anyone want to take away something that belongs somewhere
Embeddedness is complex belonging
As we have seen, implicitness, and thus a good part of belonging, can be related to repeated use and accrued familiarity. According to Zipf's law of abbreviation, the most frequently used linguistic items will tend to be the shortest, with a high semantic density per quantity of textual material and thus a high degree of implicitness. But just as implicitness is only one mode of belonging, in-situ repetition over time is only the most simple of the many ways in which textual material can historically achieve abbreviation and resist transfer. Different modes of belonging usually occur simultaneously, in different parts of the same text or to varying degrees for different participants, such that resistance to transfer is commonly of a complexity that bears little resemblance to mechanisms of simple repetition.
By in-situ repetition I mean cases like the aboriginal chant repeated generation after generation on Groote Eylandt, or perhaps the circulation of schoolboy jokes, which constantly revolve around the same taboos. Such repetition tends to produce results like a restriction of the circle of receivers, certitude, genre stability and thus a certain homogeneity within the participative "we". This is the simplest way in which a text can become attached to a group of people. But can belonging can also be the result of complex, changing and fundamentally heterogeneous social relationships? It should be noted that a text that is often repeated by the same participants in the same situation can gain a degree of implicitness without necessarily becoming untransferable: the chant can be translated; schoolboy jokes travel around the world and across generations. But when a text has been repeated in numerous different situations, when it has circulated across contiguous or overlapping social contexts and accrued different but related layers of meaning, the embeddedness ensuing from the variety of possible participants surely makes that text much more difficult to transfer. In the first case - restricted repetition - the singularity or stability of the situation usually means that universalistic meanings are recoverable. In the second case - repetition in different but contiguous or overlapping situations - meanings tend to be particularistic because only half understood by many participants. It is thus quite wrong to associate maximum non-transferability with cultural closure or social homogeneity, especially of the kind presupposed by exponents of "cultural specificity".
Continuing this argument, relative embeddedness can be associated with a diffusely collective and profoundly ideological "we", since it depends on modes of participation which cut across the division of labour. As much as the specialisation of the sciences and the internationalisation of markets progressively work against embeddedness, there are other discursive organisations - politics and publicity, but also religion, the family and Taylorist production - which work to open and maintain passages from one field to another, creating no more than the illusion of a closed semantic space in which all are compelled to use the same terms and to speak the same vaguely understood language. Such, perhaps, is the type of complex embeddedness which best resists transfer.
Significant examples from the significantly limited transfer of AIDS:
The abbreviations AIDS (acquired immunodeficiency syndrome), HIV (human immunodeficiency virus) and TMP-SMX (trimethoprim and sulfamethoxazole) can all be found repeated in various articles in the New England Journal of Medicine over the period 1988-1990. Repetition and abbreviation suggest that the terms in some sense belong to participants in these articles. And yet the degree of belonging is clearly not the same for all three terms. "TMP-SMX" rarely goes beyond medical situations, except when there is public debate about the cost of these drugs; the term might be said to belong to specialists, but this very restriction means that it is readily transferable to specialists around the world, most of whom feel no need to translate the abbreviation (the Spanish form is "TMP-SMX"). "HIV" does have more widespread usage but tends to be restricted to participants who know what the letters stand for (although news reports are full of the semantic redundancy "the HIV virus", as if the "V" stood for something else). This siglum thus tends not to accrue extensive non-specialist layers of meaning, but is nevertheless usually transformed ("VIH" in Spanish). The term "AIDS", on the other hand, is used and repeated well beyond strictly medical articles; it has value in many social groups where people have forgotten or are unable say what the letters stand for; it may mean the curse of God, a CIA plot gone wrong, the persecution of minorities, a tragedy, a challenge, an expense... or all these at once, since the term circulates from social group to social group, taking on layers of meaning whose very complexity is in each case a measure of embeddedness and thus untransferability. In its specialist pathological sense, "AIDS" is no doubt "AIDS" wherever it occurs. But in the sense in which the term can culturally belong to its participants, "AIDS" is one thing if you have it, and quite another if you do not; it is one thing in Britain, where circulation of the term is conditioned by the fact that most sufferers are male homosexuals, and quite another in Spain and Italy, where most sufferers are intravenous drug users. The specialist sense can be transferred and thus translated (significantly, as "SIDA" and not as "AIDS"); but the circuits of cultural embeddedness are more difficult to transfer, and thus of a belonging that is far more difficult to understand, to explain, or to translate.
There is an apparent paradox here that requires a brief note: How is it that the intracultural circulation of texts, which is surely a mode of transfer, can create cultural embeddedness, which is by definition a mode of resistance to transfer? Can this paradox be explained without having to assume cultural closure or intersocietal discontinuity?
Clearly, there are two different kinds of transfer involved here. The first movement is from a restricted context to a heterogeneous series of contexts, from specialist usage to general usage, from relative independence to relative embeddedness (e.g. what "AIDS" means across American society). The second movement then has as its starting point a situation of greater heterogeneity and relative embeddedness, and will probably seek reception in a more restricted context (e.g. what "AIDS" means to Susan Sontag). This second movement thus in principle involves much greater resistance to transfer. There is no overwhelming paradox or contradiction between these two kinds of transfer. The first simply complicates things for the second.
Michel Bréal on the acceleration of embeddedness and two kinds of transfer:
a) "[One of the causes of the modern acceleration of semantic change] comes from industrial production. Thinkers and philosophers have the privilege of creating new words of impressive amplitude and intellectual allure. These words then enter the critic's vocabulary and thereby find their way to the artists. But once received in the painter's or sculptor's studio, it is not long before they extend to the world of manufacture and commerce, where they are used without constraint or scruple. Thus, in a relatively short time, the language of metaphysics feeds the language of advertisements." (1897, 116)
b) "Metaphors do not stay forever chained to the tongue in which they are born. When judicious and striking, they travel from language to language and become the heritage of humanity." (1897, 146)
As one of the most perceptive linguists of the modern city, Bréal
saw not only that particularistic meanings tend to develop from universalistic
meanings, but also that the movement of terms from field to field can lead
to rapid embeddedness, to culture as fashion, and thus to an acceleration
of semantic change. This same movement may then be such that some new terms
- Bréal naïvely assumed them to be the metaphors of most value
- will travel beyond the tongues of any one city. In this way, rapid embeddedness
could possibly work to the advantage of a certain intercultural transferability,
the term gaining a kind of centrifugal force that eventually shoots it
out into another culture, although the value of what is thus transferred
is quite another question. Example, obviously: "La Movida".