Links to pages: 448, 450

Comments Welcome

University of Hawaii

August 27, 1987

ETHNOLINGUISTIC NOTES

Series 3, Number 30

WHY TRANSLATION WORKS (TO THE EXTENT THAT IT DOES)

I have tried in various recent writings (e.g., Grace, forthcoming) to make the point that what I call "the intertranslatability postulate"--i.e., the proposition that anything whatsoever can be expressed in any language whatsoever--is false (at least insofar as it can be given any meaning at all). However, I get the feeling that some people are not much impressed with that point, and that they are not impressed for what appears to be a good reason: namely, that translation seems to work. It is used extensively every day, and used with great confidence that it can, and indeed that it consistently does, produce precisely accurate results. In fact, institutional practices which depend upon the assumption that satisfactory translation will always be possible are quite commonplace, and go almost entirely unchallenged. It is not surprising, then, if people come to the conclusion that the intertranslatability postulate is valid, or that at worst any invalidity can be nothing more than a mere technicality.

How are these facts to be explained if I am right and the intertranslatability postulate is indeed false? I think the answer is that where the conditions under which a translation task is performed are sufficiently favorable, the results can be quite satisfactory, and that most of the translation tasks performed in the world today are performed under relatively favorable conditions. However, I also think--and that is the essential point--that the reasons for this fact have nothing to do with the nature of language and have almost everything to do with the cultural situation which exists in the modern world.

Now, it seems to me that the conditions for translation are favorable to the extent that the cultural presuppositions underlying the subject matter of the speech acts are shared. The most unfavorable conditions would exist when what is being talked about in the source language expression is utterly foreign to the experience of any speakers, past or present, of the target language. An example might be the task of translating Quine's sentence, "Neutrinos lack mass", into a New Guinea highlands language. Such a case would presumably provide the nearest possible approximation to perfect untranslatability.(1) The most favorable conditions permit perfect isomorphic translatability (cf. Grace, forthcoming). This can occur when the way of talking (ibid.) about the subject in question that was employed in the source language is shared by the target language. Such seems to be the case for all translations between local varieties in Kupwar village in India, for example. However, as a crude rule-of-thumb, we can say that it will be approximated to the extent that the subject matter of the source language speech act is shared with the target-language culture.(2)

I have proposed, then, (1) that the reason that translation appears so consistently to work is that most of the translation tasks performed in the world today are performed under relatively favorable conditions, (2) that the conditions for translation are favorable to the extent that the cultural presuppositions underlying the subject matter of the speech acts are shared, and(3) that the reasons why the conditions are so frequently favorable have nothing to do with the nature of language. I will attempt to explain those statements.

The reason why the conditions under which most present-day translation tasks are performed are favorable is that they, at least most of those that receive attention from linguists or professional translators or interpreters, involve translation between what in Grace 1987 were called "languages of universal translation" (LUTs). Of course, an LUT in becoming an LUT has undertaken the obligation to be able to translate any other language, and a solution is likely to be found quickly for any recurring problem in translation from one LUT to another.

But furthermore, LUTs are a phenomenon of that modern world culture which had its roots in Europe, but now belongs to a large and rapidly increasing part of the world's population. Participation in this culture requires the use of an LUT. And surely most of the translation tasks which come to the attention of the translating and interpreting professions involve the affairs of modern culture: of modern political nation-states and the agencies of their bureaucracies; of their law codes, law enforcement agencies, and judicial systems; of their commercial institutions and the operations of these; of the multifarious products of their technology; even of their recreational activities, etc., etc. Surely, for a member of modern world culture, what there is to talk about is made up mostly of the things of this culture.

Finally, although there is surely much informal translation which goes on even today between non-LUTs in various parts of the world, little of this comes to the attention of the people in our society who are concerned with the business of translation or interpretation. More likely to come to their attention are cases which involve one LUT along with one non-LUT, such as court cases in which a non-LUT is used by a defendant or witness. However, the asymmetrical power relations in such cases make it doubtful that we can accurately gauge the accuracy of the translation. it seems likely that, as long as a coherent LUT transcript results, failures of the translation process could escape notice.

The conclusion, then, is that the most prominent languages of the world today have undertaken to be able to translate one another--i.e., to be LUTs. Furthermore, their business is the same--the things they are used to talk about, and the things they are used to say about these things, are to a very large--and steadily increasing--extent the same. This is not a fact about the nature and potential of language, it is a fact about what has happened to the cultures of this world.

NOTES

1. In the literature pertaining to what I am calling the intertranslatability postulate, two alternatives are generally considered to be available to someone assigned the task of producing a translation under such conditions. The first is to change the target language so as to equip it to express the content of the source-language expression. According to the assumptions usually associated with grammar-lexicon model of languages, the only changes which would be required would consist in the introduction of vocabulary because the facts of the grammatical system of a language are regarded as irrelevant to anything outside that grammatical system and vice versa. In this view, it is the lexicon that serves as a buffer between the grammatical system and anything outside, and accordingly, the lexicon is the only part of the language which could be changed in response to needs originating outside the grammatical system itself.

However, at the same time the grammatical system is regarded as constituting the very essence of the language. Therefore, lexical changes are often regarded as not being real linguistic changes because they leave the language's essence unaffected, and in fact, it would be regarded as inaccurate to speak, as I did above, of the addition of vocabulary as "changing the target language".

Whether or not adding vocabulary to a New Guinea highlands language in order to permit it to translate "Neutrinos lack mass" would count as changing the language, it would not offer much of a solution to a translator in an ordinary real-life translation situation, unless he/she supplemented it by teaching some or all of the speakers of the language to understand and use the new vocabulary. The introduction of vocabulary might be offered as an argument that anything can be translated into any language, but hardly as an argument that the language could then be used to communicate "neutrinos lack mass" or the like to its speakers.

I said above that two alternatives are generally considered to be available to someone assigned the task of producing a translation. The first is to change the target language; the second is to resort to what is thought of as circumlocution, but which I think is something different, which I have called "perlocutionary translation". I have discussed this elsewhere (cf. Grace 1986, for example). (Back up)

2. The literature of lexicostatistics envisaged a universal "basic vocabulary"--a small set of meanings which were assumed to be represented in the lexicons of all languages. This basic vocabulary was sometimes described as "non-cultural". Of course, there is probably no one now who would want to claim that such vocabulary items are truly non-cultural--their meanings clearly also depend on conceptualization.

However, at the same time it is true that the amount of interpretation required to arrive at concepts of this kind is unusually small. They come closer than most concepts to being empirical givens--to being concepts provided directly to the senses through observation. Although I have not tried to investigate this systematically, it also seems likely that where all of the content words in an expression belong to basic vocabulary in this general sense, something approximating truth conditions could be used as a basis for translation. In any case, such expressions surely make up only a very small fraction of the discourse of the modern world. (Back up)

REFERENCES

Grace, George W. 1986. Perlocutionary translation. Ethnolinguistic Notes, Series 3, Number 26. Printout. (Back up). Also internet World Wide Web page (Click Here).

Grace, George W. 1987. The Post-Linnean world-view and languages of universal translation. Ethnolinguistic Notes, Series 3, Number 29. Printout. (Back up). Also internet World Wide Web page (Click Here).

Grace, George W. Forthcoming. The linguistic construction of reality. Beckenham, Kent: Croom Helm. Printout. (Back up)

First put on the Web on 10 August 1998