ETHNOLINGUISTIC NOTE 34

Links to pages: 495, 497

Comments Welcome

George W. Grace

October 27, 1988

ETHNOLINGUISTIC NOTES

Series 3, Number 34

THE IDEA OF A THEORY OF TRANSLATION:
THE OBJECT OF THE VERB "TO TRANSLATE"

In Grace 1988b, I talked about the importance of idealized models in our thinking. It is now apparent to me that when I first began thinking about a theory of translation, I was conceiving of translation in terms of an idealized model. It is also apparent that the model in question strongly influenced my conception of the task. In this model translation (1) begins with a source-language linguistic expression (a "text") and seeks to find (or design) (2) a corresponding target-language linguistic expression (text) such that (3) the correspondence between the two is sameness (or at any rate equivalence) of meaning.

In one way I was aware that the model represented an idealization. That is, I would have acknowledged that there are some kinds of operation which go under the name "translation" which do not conform precisely to my idealized model. I would also have acknowledged that most, if not all, of the different operations going under that name seem to be related--or at least we might say that they show "family resemblances"--so that it would not be easy to find a place to draw a boundary which would exclude some, but not all, from the scope of the theory. In sum, I was aware that my idealized model did not accurately represent all forms of translation, but assumed that it did represent what we might think of as a kind of prototypical form--or translation reduced to its essentials.

(For reasons which are not now apparent to me) I assumed that translation of this prototypical form-- translation meeting only the requirements listed above--was logically prior to any other. I assumed that all other kinds of translation were logically derived from it and that the derivation involved the addition of complications. I think I also imagined that it must have been chronologically prior (How else could the others have developed from it?).

Given these assumptions, it seemed that the proper strategy for a theory of translation would be to begin with this pure, prototypical translation, attempting first to provide an account of it. Once we had a theory which accounted for the prototypical translation, we could then undertake to determine how that theory could be generalized to account also for everything else that deserves to be called translation.

However, this strategy has run into trouble.

Note that there were two key assumptions about this kind of translation which I took to be prototypical: (1) that the nature of the correspondence between the original and its translation was equivalence of meaning, and (2) that the equivalence of meaning obtained between linguistic expressions (or "texts").

First, it became apparent (as I mentioned in Grace 1988a: 480) that in translating a particular linguistic expression, where the expression with the same meaning (where, NB, "meaning" is understood as the resultant of the meanings of the constituent lexical items and the "meanings of the constructions"; cf. the discussion in Grace 1987b: 455, 471-72) in the target language is different from the expression that target-language speakers would use in the same situation, it is in ordinary circumstances always the latter expression that should be chosen over the former as the translation equivalent. Consequently, my assumption that the prototypical correspondence between translation equivalents is one of meaning seemed clearly to be wrong.

Further evidence that the basis of the correspondence may not be meaning is provided by literary translation. There is a long tradition of translation of literary works, of both prose and poetry. There is also a substantial theoretical literature (i.e., literature devoted to the principles underlying such translation) on the subject. In the case of poetry especially, it is apparent that maintaining equivalence of meaning is very often not the primary objective. In fact, what are called "translations" (and I have no alternative designation to propose for them) of Lewis Carroll's "meaningless" poem "Jabberwocky" have been made in French, German, and Latin, at least. If the source-language text is meaningless, the target-language text cannot be equivalent in meaning (or else, if you prefer, any meaningless text at all would be equivalent in meaning, but that would not explain why these particular texts are more apt as translations than any other conceivable French-, German-, or Latin-like nonsense). In any case, many translations of poetry, and translations of "meaningless" poetry in particular, seem to provide further evidence that the correspondence involved in translation cannot be simply equivalence of meaning.

But what of the other assumption: that the correspondents--what the correspondence relation (whatever its nature) holds between--are linguistic expressions (texts)?

I mentioned that there is a substantial theoretical literature on the subject of literary translation, but it is evident that my idealized model of translation owed little to that literature. Rather, it was influenced by another literature, that of linguistics and philosophy (particularly logical-positivist philosophy; the language with which this philosophical tradition was primarily concerned was scientific language--or more exactly, with an ideal scientific language which the philosophers were attempting to design), and the translation dealt with was primarily of scientific writing and such materials. Now one often hears references to "scientific-technical" translation, and I will use that label, although translation according to approximately the same principles is also used for other kinds of subject matter.

But now it appears that my idealized model is actually an idealization based just on the kind of translation that I am referring to as "scientific-technical". It also appears that the tradition of literary translation is at least as long and well-established as that of scientific-technical translation, and this fact immediately puts into question my assumption that the (scientific-technical-type) translation of my idealized model is phylogenetically prior to literary translation. But if there is no reason to assume it to be prior in time, there is no reason (except for the other gratuitous assumption that it is translation-reduced-to-its essentials) to assume it to be logically prior.

Michel Foucault tells us that literature assumed a separate existence--separate from other uses of language-- only from the end of the eighteenth century with (Foucault 1977: 60) "the works of Sade and the tales of terror". It is associated with the Romantic period in which Foucault (1973: 300) says, "literature becomes progressively more differentiated from the discourse of ideas". Before that, presumably, no distinction existed between literary texts and scientific-technical texts (or any precursor thereof), and therefore the translation of scientific-technical texts, on the one hand, and literary texts, on the other, could not have constituted separate traditions.

But if that is the case, what was translation in its prototypical form? What has translation in its modern forms evolved from? What, in short, was the earliest form of translation like?

There are no records to turn to, but there seem to be sufficient grounds for some limited speculation. First, is there any reason to doubt that some sort of translation has occurred for as long as different, non-mutually- intelligible, languages have existed in close enough geographical proximity for encounters between their speakers to occur? I can think of none. Second, if this is true, then the earliest translation must have antedated writing by a long time--in short, the earliest translation must have been oral translation--what would now be called "interpretation".

And that is what makes me wonder about the second assumption of my idealized model of translation: the assumption that what the correspondence (whatever its nature) holds between are linguistic expressions (texts).

As Charles Hockett has pointed out in various writings, one of the "design features" of human language is "rapid fading". That is, the artifact produced in a speech act (a linguistic expression) consists of nothing but perturbations in the surrounding air. These subside very rapidly, and once they have subsided, nothing remains. That, of course, means that the text does not exist except during a fleeting moment. And that in turn means that any comparison between source-language and target-language texts would have to be completed with great speed (and, moreover, that any subsequent evaluation of the accuracy of the results would be impossible).

I am left with the conclusion that whatever translation occurred before the invention of writing cannot have attached the same prominence to the source-language text that my idealized model did. And the same argument would extend much further: to much translation subsequent to the invention of writing--to translation by illiterates in any society, of course, but also to any oral interpretation where no physical record (written, taped, or whatever) of the source-language text existed.

Therefore, I submit that frequently translation cannot be a matter of the comparability (in terms of some kind of equivalence relation) of source-language and target-language texts because there is no way to establish their comparability, or even to subject the texts themselves to careful scrutiny.

The earliest translation must have occurred in conditions of "primary orality" (as Walter Ong, for example, has called the orality of cultures untouched by literacy). What can translation have been like in those conditions? What basis have we for speculating? I am not sure what might be found in a careful search of the anthropological literature, but I am aware of one paper by the anthropologist Richard F. Salisbury (Salisbury 1962) which describes translations which he observed during his visits to a preliterate community in the New Guinea Highlands. In most of the cases that he reports (although I should emphasize that these were not intended to constitute a representative sample of the translations actually performed during his stay) it was apparent that the main purpose of the translation was symbolic rather than (as he [1962: 4] puts it) "to ensure unambiguous transmittal of information". Most cases occurred in more or less ceremonial circumstances where most of the audience clearly were capable of understanding the original, source-language, speech act. Translation seemed to have the function of giving overt recognition to the target language's status as the "home" language of the locality or that of "emphasizing the importance and public nature of the discourse" (1962: 4). In the latter case, translators, who were generally prominent individuals, sometimes took the liberty of elaborating on the original.

That last point deserves emphasis. We have been thinking of translation as creating an equivalence relation of some kind (which apparently isn't meaning--or at least it can only be meaning in the absence of anything better) between something connected with the source language and something connected with the target language. According to the idealized model with which we began, those "somethings" were texts. However, we have seen that in oral translation-- i.e., translation in strictly oral conditions, where nothing is written down--there is very limited possibility of making an exact comparison of the two texts because the possibility of subjecting them to simultaneous examination is fleeting if it exists at all. But now we have an additional point. Salisbury reports translation which is not under a constraint even to attempt to be faithful to the original text, but is at liberty to produce a target-language text which is more elaborate.

We have been speculating on translation under conditions of primary orality, but the contemporary scene seems quite different. The professionalization of translation and interpretation has occurred largely, if not wholly, within the framework of the modern world monoculture (for the notion of the "monoculture", cf. Grace 1988b: 487ff). I believe that one may confidently assert that most professional translation (and interpretation) today is intra-monoculture translation (although this statement requires some qualification where literary translation is concerned).(1) By "intra-monoculture translation" I mean translation where the functions being performed by both the source-language and target-language speech acts are monoculture functions, and functions which both languages are adapted to perform.

Where translation involves cultural functions for which both source and target language are regularly used, and where both languages have linguistic expressions dedicated to (or otherwise suitable for) those particular functions, translation of the kind represented in my idealized model can likely be approximated. In short, my idealized model becomes less preposterous if its application is restricted to intra-cultural translation (of which intra-monoculture translation is, of course, a particularly prominent instance).

It is worth pointing out, however, that the better suited the case is to be represented by my idealized model, the less understanding is required of the translator. One can imagine the ideal case of a text such that the two languages' ways of talking about the subject matter of the text are so perfectly matched that translation can be isomorphic. By this I mean that each source-language sentence can be translated by a target-language sentence having the same syntactic structure and having each lexical item in the source-language sentence translated by an equivalent lexical item in the target language. In such an ideal case the target-language expression could indeed be said to be equivalent in meaning to the source-language expression, and the process of translation would not require any intervention of an understanding translator. That is, it could be carried out by an automaton.

But even in the modern world--even in translation between monoculture-dedicated languages--translation is not always a relation of strict fidelity between texts. Hildegund Bühler (1986) reports that today's translators and interpreters are sometimes expected to produce translations which are--by design--not faithful to the details of the source-language text. She writes (1986: 107), "So-called literary (aesthetic) translation, the translation of persuasive texts or of texts in a multi-media environment will call for a creative-communicative operation where the translator may assume the role of a second author, and where (in order to achieve a corresponding effect) recasting of larger text segments or of an entire text may be required. In some instances, adapting, expanding, or abridging the original will be expected of the translator-expert; such processes may also be part of simultaneous or consecutive interpretation. How much of paraphrasing is "permissible" before the activity no longer qualifies as translation (as well as the various types of translation) has been discussed by Reiss..."

So far I have proposed two objections to defining translation as requiring a correspondence between texts. The first was that, except where a record is made of the utterance (something which was not even possible before the invention of writing), actually comparing the source- language and target-language texts is difficult or impossible. The second was that often--even, as Bühler points out, in the exceptional conditions of the modern world--exact fidelity to the original text does not suit the purposes of whoever commissioned the translation.

Now I would like to propose two other hypothetical cases in a pre-literate setting. The first case is one in which there is an interval of some time between the source- language speech act and the translation. For example, the person who is ultimately to play the role of translator is told, or simply overhears, something said in one language (that which is eventually to be assigned the role of source language). S/he subsequently reports (the content of) the speech act to someone else--this time by means of the target language. Even if it was understood from the beginning by all parties that the translator was going to deliver the message (in translation) to its eventual recipient, it could not reasonably be expected that it would correspond "sentence by sentence, preserving truth conditions" (I take this formulation from Lakoff 1987: 322) to the original text.

The second case is one where there cannot be said to be a single source-language text. For example, a woman from the source-language community has married into the target- language community. She tells some of her people's stories to her children and others in her new home. Of course, she tells them in the target language. She originally learned the stories from hearing them told repeatedly--not, of course, always in the same words. Her target-language narration does not attempt to follow faithfully ("sentence by sentence, preserving truth conditions") any single source-language telling of the story which she has heard. It seems to me that she has done something which is at least very similar to translation. Would it make sense to think of this as translation in which the object of translation is not a single text but a set of texts?

Then, what are we to say of the anthropologist, Paul Bohannan, who considers the principal problem of ethnography to be one of translation? He writes, for example (Bohannan 1963: 10), "The ethnographer is, in short, a translator of strange ideas, customs, and things into familiar language." But he does not mean translation of particular texts. He says (1963: 8), "The completion of the ethnography has as its major problem the accurate translation of data from one cultural milieu into another without serious warping of facts and interpretation." (Cf. also Bohannan 1959).

This takes us still farther from the assumptions of the model with which I began. But that model was really based on the logical positivist view of the nature of language. A main reason why I became interested in translation was that I thought it held the key to a large part of the linguistic changes that result from so-called "language contact". However, the phenomena involved here are, I think, much more general than those admitted by the "sentence by sentence, preserving truth conditions" notion of translation. I think I probably best expressed what I have in mind in the following passage (Grace 1985: 9), which I will take the liberty of quoting:

"I would like to suggest that what is really significant in so-called 'language contact' situations is the fact that some of the utterances in the language in question are produced by people who, at least part of the time, think in another language. By 'thinking in a language' I mean roughly what I understand the lay public to mean by it. I mean formulating utterances directly in that language rather than formulating them first in another and then translating them. People who are thinking in a certain language may be said to experience reality in terms of that language, or at the very least to interpret their experiences in its terms. Anyway, I propose that if a language is sometimes spoken by people who have done some of their thinking in another language, that fact is a significant factor in its environment. A language must be under pressure to acquire the means to express the thoughts of those who speak it, and it is thus that one language influences another."

Concluding Remarks

The main conclusion seems to be that translation cannot be conceived of as essentially a correspondence between texts. So to conceive of it would require us to exclude all instances in which there is some deliberate departure from strict fidelity to the original text (at least unless the nature of the departure can be given a precise formal characterization). It would also require us to exclude all instances in which the exact original text was not available to the translator. And it would require us to exclude many other kinds of relations between things uttered in one language and somehow-corresponding things uttered in another where the correspondence is very like what anyone would call translation. I believe that, for many purposes, the phenomenon of translation needs to be defined more broadly.

I tried to show in Grace 1986 that a speaker characteristically has multiple motivating purposes which govern his/her speech act. And in Grace 1987a (esp. pp. 440ff.) I tried to make the point that the linguistic expression actually produced cannot be thought of as precisely defining the speaker's purposes (i.e., the effect the speaker wishes to produce)--that these purposes may underdetermine the linguistic expression. Therefore, for that reason alone the translation which would best approximate the speaker's intended effect might not correspond faithfully to the linguistic expression which s/he used. And if the object to be translated (what in my title is called the object of the verb "to translate") is some direct function of the speaker's purposes--of the effect that the speaker is attempting to achieve--then it may well not correspond very precisely to the linguistic expression which s/he used.

But I also pointed out in Grace 1986 that, among all of the people involved in a communicative transaction and its subsequent translation, the speaker is not the only one who has purposes of his/her own. It is not necessary that a translation attempt to serve the speaker's purposes; it can be commissioned to serve the purposes of any of the people involved, or of anyone else.

What, then, is the object of the verb "to translate"? The answer is, I believe, that there is no single answer-- that the answer must vary with the purposes for which the translation is made.

There is another point that emerges here. I was working with an idealized model of translation that now appears to me to be applicable only to languages which are written. That seems entirely inappropriate if we are concerned with translation in general, or with language in general. In fact, so much is now coming to light about the profundity of the influences of writing on our perception and use of language that one wonders how, in our investigations of languages which are regularly written, it will ever be possible to distinguish those of their characteristics which are attributable to the fact that they are languages from those attributable only to the fact that they are written.

NOTE

1. Although I suppose most (especially most non-literary) professional translation today to be intra-monoculture translation, this cannot always be so. One example would be interpretation in the courtroom. There is nothing which would prevent individuals who do not speak a monoculture-dedicated language from showing up as defendants or witnesses in a case.

I have no information on this, but I imagine that it is rare for professional interpreters or translators for non-monoculture-dedicated languages to be certified and on call anywhere.

However, the problem does not end there. Some interpretation between monoculture-dedicated languages may actually involve an individual speaker of one of the languages who participates only in a very limited way in monoculture functions, and therefore has only a very limited ability to use his/her language to perform such functions--either in speaking or understanding. Thus, for example, although several major Asian languages would count as monoculture-dedicated, that does not mean that every peasant who speaks such a language should be expected to understand everything said in that language in a courtroom. (No more should all Americans be expected to understand everything said to them in English). (Back up)