Translation Studies Forum: Universalism in translation studies

Abstract

It is sometimes said that the way to develop current translation theory is to look at specifically non-Eurocentric and non-Western approaches and learn from them. Against such a position I take a Popperian view. I argue that this proposal is flawed because it commits the genetic fallacy, where an idea or hypothesis is assessed according to its origin. Rather, any hypothesis should be tested as widely as possible, regardless of where it comes from. This includes taking account of the context of discovery. I illustrate my main point with reference to some basic conceptualizations of translation (such as the transfer metaphor), so-called translation universals and the debate about whether translation studies should have a standardized terminology.

Keywords:

• hypothesis testing
• metaphor
• universals
• generalization
• terminology
• Popper

There is a view in contemporary translation studies (TS) that our field is too Eurocentric, or too Western, and therefore needs to expand to incorporate non-Western approaches. Such an expansion is seen as offering solutions to some current trends that challenge the development of translation theory. These include the rapid spread of translation technology and the increase of non-professional and group translation practices – seen, for instance, in crowdsourcing and fan translation. It is also argued that current Eurocentric views cannot account adequately for some translation practices in non-Western cultures. Underlying this general view there seems to be the assumption that TS as a research field is particularly affected by cultural relativity, and has been dominated for too long by European culture; it needs shaking up. I shall argue that the terms in which this relativist view is commonly presented are wrong and misleading, although there is a good point to be made from the evidence alluded to in the discussions of this issue in the literature.

Kuhn vs Popper again?

At a philosophical level, the situation outlined in the previous paragraph recalls the famous debate in the 1960s between Thomas Kuhn and Karl Popper. Fuller (2003) argues that although Kuhn seems to have won this debate, he should have lost. I agree. According to the relativist view we seem to have a kind of Kuhnian “paradigm”: a “Western”, socioculturally determined conceptualization of translation, let us say, which (so the argument goes) Western scholars have assumed is universal but is not, because non-Western conceptualizations are different (see e.g. Hermans 2006, Tymoczko 2007b, Wakabayashi and Kothari 2009). This paradigm has arisen (partly) as a result of the colonial dominance of the West and the hegemony of some of its major languages, notably English. The paradigm persists (partly) because of the entrenched and institutionalized nature of this hegemony, and ultimately reflects a continuing cultural imperialism. On this view, TS will advance by overthrowing its Western paradigm, or by expanding it to incorporate non-Western concepts and theories. We in the West therefore need to know more about Chinese or Indian theories, for instance, in order to make TS more universally applicable. By including non-Western ideas TS will also improve its ethical standing (see e.g. Tymoczko 2007a). Note that this relativist view seeks universality, but argues that this cannot be achieved with our current Western-centric theories.

Against this position, I set a Popperian view. Here, the initial assumption is that all scientific endeavour is intrinsically universalist. Hypotheses and theories can emerge anywhere, in a given context of discovery, but they are assessed anywhere they can be assessed, in a universalist context of justification. There is no added value in assessing them specifically in country or culture X, any more than there is added value in the fact that they were first formulated in country or culture Y. Applied to translation, this implies that while a local theory of translation – a putative Finnish theory, say – might contribute to an understanding of translation practices in Finland – for instance, as a case study in the history of translation – it would only have wider value insofar as it could also shed useful light on translation practices elsewhere. For universalists, the above-mentioned relativist position commits the genetic fallacy, according to which an idea, hypothesis or theory is valued (or not) according to its origin. In other words, to claim that TS would be enriched by ideas, etc. that originate in non-Western cultures because they arise there, is a bad argument.

It is of course valuable to take a given context of discovery into account when examining a hypothesis, etc., since we know that all our thinking is affected by our culture. It thus makes sense to check out the cultural origins of hypotheses, in order to evaluate the influence of these particular origins on the hypotheses in question. For instance, as Pym (2011) has recently argued, several of the main Western theories of translation can be seen as having arisen as responses to “Western” problems. Nida's concept of dynamic equivalence was evidently influenced by the (largely American) evangelical concern to propagate the Bible in new languages; Skopos Theory was probably motivated by the need to raise the academic status of translator training in German universities; and translation corpus studies were stimulated by the growing numbers of overseas students studying translation and/or applied linguistics in the UK. These motivations were not necessarily the only ones, of course, but they were surely important contributory factors in the development of these ideas. A Popperian would add that a critical examination of the context of discovery of a given hypothesis follows naturally from the critical attitude that is basic to scientific methodology in general.

Now, the relativists are surely right in pointing out that some of our current Western assumptions in TS look doubtful in the face of evidence from translation practices that have so far been little studied, whether in the West (e.g. the study of non-professional translation) or elsewhere. But then the Popperian response would be: let us then test these assumptions on whatever relevant data we can find, wherever we find some, and the more widely the better. Conversely, ideas that originate in non-Western cultures should also be tested, on any relevant data available, in whatever culture, to check whether they apply more generally. If the assumptions and ideas are supposed to be valid for translation universally, we can test them anywhere. To repeat: there is no particular geographical virtue in privileging an idea because it comes from the West or the non-West. To do so would take us back to Nazi or Stalinist science (see further Fuller 2003, 182–183).

It is obvious that groups of scholars within a particular country or area do sometimes influence each other, so that ideas of a certain kind sometimes merge around particular approaches or major projects. It is well known that much of the literature on functional approaches in TS stems from scholars in German-speaking countries. Brazilian ideas about translation as cannibalism have nourished both translation and its research especially in that country (see e.g. Vieira 1994). In Finland, the quest for translation universals has stimulated many projects (see e.g. the papers in Mauranen and Kujamäki 2004), and still does. But this is not evidence against what I am calling the universalist, Popperian position. Any fruits of local theoretical schools can, and should, be tested wherever possible. And this is where we have often appeared to be premature: there is a tendency in TS to make claims about universality without actually testing these claims adequately.

It should also be mentioned that some TS scholars seem to resist all attempts at generalization or universalism, seeking instead to uncover the context-bound contingencies of particular translations rather than make claims about general tendencies. For instance, Hermans (2007, 145–150) calls not for general hypotheses but “thick”, detailed analyses of individual cases; this would serve, he says, to keep in check the “universalizing urge of theory”. I agree that any theory of anything – if it aims to be a general theory – must claim to “universalize”; but I do not agree that this is a tendency that must be kept in check. A theory that did not aspire to universalize – that is, to generalize as far as possible – would not take us beyond particular cases, or particular sets of cases. What matters is that any general claim needs to be tested, on particular cases of all relevant kinds. Conversely, thick case studies can generate hypotheses whose generality can be tested on other data. We can work top-down (testing general claims against particular cases) or bottom-up (generating potential generalizations from particular cases); both directions are surely needed in the field as a whole.

A non-universal metaphor?

Another point concerns the terms and metaphors we use to conceptualize our subject, how we interpret it. It has been argued, for instance, that because of the etymology of the English term translation and its Indo-European cognates, TS scholars (many of whom have been in the West, yes) have traditionally over-valued the cognitive metaphor of transfer, carrying something across, and the concomitant notion of equivalence, at the expense of other conceptualizations of translation (see e.g. Chesterman 2006; Tymoczko 2007a, 2007b). The result has been an overgeneralization: the transfer metaphor has been assumed to be universal, which may have influenced the selection of texts in translation corpora in different languages, and also the quality assessment of these texts. The solution, in my view, is not to compare the Western notion of translation to non-Western ones because these others are non-Western, but because they are, quite simply, different. One tests the usefulness and applicability of a conceptualization (among other ways) by comparing it with competing conceptualizations. It matters not a hoot where these competing conceptualizations come from. So it makes good sense to look at alternative etymologies of words that seem to mean something like “translation” in different languages, in an attempt to arrive at a concept that might be more general than one rooted only in the idea of transfer.

One might, for instance, work towards a cluster concept, as argued by Tymoczko (e.g. 2006). She argues that translation can be usefully conceptualized together with the notions of transference (involving material movement), representation, and transculturation (the transmission of cultural characteristics, such as art forms). In this conceptual cluster, the idea of semantic equivalence (i.e. transfer) is not foregrounded. Another possibility is suggested in Chesterman (2006), based on a semantic cluster of the three central notions of equivalence, difference and mediation, following Stecconi (2004); evidence from admittedly limited evidence from a small number of languages suggests that these three central notions can be manifested with varying degrees of relative priority in the terms for “translation” in different languages. It remains to be seen, however, whether any such cluster concept will prove to have sufficient added value to become widely adopted.

The point here is that if Western scholars have been too hasty to assume the universality of the transfer metaphor, this assumption can be tested and refined or rejected according to the usual criteria for testing a hypothesis. In this case, we are dealing with an interpretive hypothesis. (An interpretive hypothesis claims that X can be usefully interpreted as Y; see further, e.g. Chesterman 2008.) Maybe its range of useful application will be found to be narrower than we have thought. The consequent adjustment to the theory would thus reflect a very typical progression in science: a general, or even universal hypothesis (of any kind, empirical or interpretive) is proposed, then tested and found to be less general than was previously thought, so conditions are determined which define its more limited scope of application. The contrary procedure, which is also standard, is to start with a limited, conditioned claim and then discover that its range of application is wider than assumed. Top-down and bottom-up, again.

Given that conceptual metaphors are interpretive, not empirical hypotheses, it might be thought that they are particularly culture-bound; it would therefore be naive and unjustified to extend them to universal application. I think there is some support for this view, especially because there is a tendency to neglect the testing of interpretive hypotheses: they tend to be proposed and defended, and illustrated with selected examples, rather than critically tested. (Interpretations cannot in principle be falsified – for an interpretation is always possible – but they can certainly be assessed in comparison with alternatives, and their usefulness can be tested in practice. See further Føllesdal 1979.) Assessing the potential universal applicability of our interpretive hypotheses requires not only wide-ranging data of all kinds but also an imaginative openness to alternative hypotheses, alternative interpretations, a willingness to see our “X” (translation) in terms of other “Ys”, seeing it differently. This is a considerable epistemological challenge, but it does not mean abandoning a critical attitude to any of the alternatives. A cultural evolutionist might observe that the larger the population of available conceptualizations, and the more variation there is in that population, the better the conditions will be for the eventual emergence of a robust theory.

Universal features of translations?

The problems of finding the appropriate scope of an empirical hypothesis, on the other hand, and also some of the different attitudes to universalism, are illustrated in what is perhaps the most obvious recent manifestation of universalism in TS: the research on translation universals, already mentioned above.

The idea that there are (or might be) “universal” features of translations is an old one, although it is only during the past couple of decades or so that serious research has been done on this topic. It has long been realized that there are always differences between translations and their originals (i.e. there are always shifts). Typical differences between source and target texts include interference, explicitation, or increasing standardization. In addition, translations often show evidence of “translationese” – that is, non-natural target language. Differences between the target text and non-translated texts in the target language include simplification and the under-representation of target-language-specific (so-called unique) items. (For references and more details, see e.g. the papers in Mauranen and Kujamäki 2004.)

At first sight, this line of research makes intuitive sense, although the general hypothesis that “translations are different” makes only a cautious, unsurprising claim. However, it is a claim that has given rise to some heated debate. In a polemical piece against the whole notion of universals, Juliane House argues that “the quest for translation universals is in essence futile” (2008, 11), mainly because she sees the only valid universals as those pertaining to language pragmatics in general, not translation in particular.

So what is so controversial about the idea? In the first place, critics have not liked the term “universals” itself, which was popularized (though not invented) by Mona Baker (1993). It does indeed seem misleading, if it is interpreted as being analogous to the sense in which it is used in “linguistic universals”. In language typology, the search for universals, later influenced by Chomsky's theory of Universal Grammar, has meant formulating generalizations about language structure that are claimed to be based on the constraints of human cognition, and underlie (or are manifested in) the structures of individual languages. Claims about such universals have been tested on hundreds of languages, and could in principle be tested on all known languages, since these amount only to 6,000 or 7,000. Some such universals are claimed to be absolute, occurring in all possible natural languages; others are probabilistic, for example, of the form: if a language has feature A, it probably also has feature B (for a general account, see e.g. Croft 2002). But with respect to translation, the term must be interpreted differently, more loosely, for several reasons.

The notion “translation” is categorically different from that of “language”. Translations are instances of parole, not langue, and as such they come in very different forms and sizes, from a single spoken word to a whole book, for instance, not to mention all kinds of multi-modal variants. There is enormous variation of usage and interpretation here, both culturally and historically; and we do not even have access to much of the relevant evidence from the past, or from oral translation (i.e. interpreting). Maria Tymoczko (1998) is one scholar who has drawn particular attention to this problem, and to how seriously it affects the construction of corpora in projects that set out to generate or test hypotheses about translation universals.

In my view, what all this implies is that the term “universal” has been an unfortunate and misleading choice in this context, as applied to linguistic features of translations. Some scholars have preferred to speak of tendencies, or probabilistic translation laws, such as those of interference and growing standardization (Toury 1995). But ultimately, the quest for universals is no more than the usual search for patterns and generalizations that guides empirical research in general. Our mistake has been to assume too quickly that the tendencies or regularities we have found might really be “universal”. True, to state a hypothesis in a maximally general form exposes it rapidly – and usefully – to the risk of counter-evidence, which should then lead to the increasing limitation of the scope of the hypothesis, or indeed to its rejection. Claims about the tendency of translators to explicitate, for instance (going back to Blum-Kulka 1986), obviously cannot apply so much to translation types such as subtitling or the translation of comic strips, or to gist translation, owing to the constraint of available space, so it cannot be a uniformly “universal” tendency.

On the other hand, hypotheses have been proposed that may not be general enough. All translations may well tend to show signs of interference, and of shifts; and causes for these signs can be inferred to lie somewhere in human cognition processes (see e.g. Halverson 2003). However, such tendencies might not be specific to translation and may also be found in other forms of discourse mediation, as argued, for example, by Ulrych (2009). Along similar lines, Lanstyák and Heltai (2012, 100) make a good case for the view that (potential) translation universals are a form of “language contact universals”, which are also manifested in the language performance of bilinguals. These language contact universals are in turn a subset of more general “universals of constrained communication” (2012, 100).

It is thus essential to specify the scope of a hypothesis, and to be open to the need to extend or restrict this scope. Conditioned generalizations, at a lower level of generality, specific to particular languages, genres, translator profiles, working conditions and so on, may bring a wealth of new information about translation, although they appear less ambitious than grand claims of universality. As our knowledge accumulates about how these conditioned hypotheses fare in different tests, we may be able to generate more general correlative or causal hypotheses. It is less the individual translation features themselves that might justify this kind of universality claim, I suspect, but rather the relations between types of features and types of contextual conditions. We are not there yet, but it may well be these hypotheses that can claim to be generally valid, applying to any case where a given set of conditions is met. For instance, we might develop hypotheses about the way in which space or time constraints affect translator decisions under different circumstances. But it's always the testing of the hypotheses – the context of justification – that matters.

Standardized terms?

At bottom, as Tymoczko (1998) has stressed, we still lack a generally accepted definition of “translation” itself. Although there are also fuzzy borders in linguistics between what we want to call a language and what we want to call a dialect, this is nothing compared to the problematic terminological mess around translation, version, adaptation and so on. How free, or indeed how bad, does a claimed translation have to be before it is felt (by whom?) not to be a translation at all? Isn't a really bad translation nevertheless still a translation, of a kind? Should “bad” (or indeed non-native) translations be included in a representative corpus of translations in general, on which hypotheses about universals could be tested? I think the answer to this last question must be yes, unless the scope of the hypothesis in question is restricted to competent, published professional native translations only, not claimed to apply universally to all translations. (Is there a corpus somewhere that explicitly includes “bad” translations?)

One solution to the definition problem might be to resort to formalization, in an attempt to get beyond the limitations of a single natural language and take refuge in the universal language of mathematics. Very few TS scholars have taken this route, however. A notable exception is Garcia-Landa (1990), but he has not been followed, as far as I know, perhaps because even if one formulates one's theory in mathematical symbols these still need to be defined and interpreted in natural language(s). If anyone comes up with a solution that gains wide acceptance, I think the TS field would only benefit, at least as a field of empirical research.

A more important issue is the need (or not, according to your view) for a standardized terminology. “Universalists” have been arguing for more agreement on TS terminology (see e.g. the special issue of Target, volume 19, number 2, 2007), and some practical steps have been taken in the work on the Benjamins online Bibliography of Translation Studies (see http://benjamins.com/online/tsb/). But there are still many who find that such a project limits their freedom to interpret terms and concepts in ways they find most appropriate, who feel that a standardized terminology would be a straitjacket for TS, not an advantage. Here I think the difference is between those who see TS as an empirical human science (which therefore needs some standardized terminology, like any other science), and those who see it as a hermeneutic discipline (where conceptual argument about meaning and interpretation is more central). If we agree that it can be both, the consequence may be even more of a split in the field than we have at present.

The terminology problem is one that still remains without a generally agreed solution. The debate on the issue has of course raised matters of language policy, institutional power and democracy, and the argument will no doubt continue. If any standardized terms are to be agreed for, say, English, this still leaves open the problem of their standardized translations in other languages. There are several multilingual glossaries of the field available, but none has yet been generally accepted as standard.

Concluding remarks

I have argued that the proper assessment of ideas, theories and hypotheses has nothing to do with their geographical origin in the West or non-West. As Pym puts it, “no idea is going to be superior simply because it came from a particular direction” (2011, 60). As translation is a cultural phenomenon, we can expect a huge variety of views and approaches. The more the merrier, indeed. But this does not mean that we have to subscribe to a naive relativism (where all we see is differences, and similarities are ignored), any more than we need to subscribe to a naive universalism (where all differences are ignored). After all, is this not what translation itself is all about: creating similarity alongside difference?

So if we are working towards a general theoretical framework that can describe and explain translation in its widest sense, all potential components of such a framework need to be tested as widely as possible, regardless of their culture of origin. What matters is whether a concept, model, hypothesis or theory turns out to be useful or not, in solving a given problem and/or in bringing deeper understanding. Or, indeed, in generating further fruitful questions.

Note on contributor

Andrew Chesterman was born in England but moved to Finland in 1968 and has been based there ever since, mainly at the University of Helsinki, where his main subjects have been English and translation theory. In 2010 he retired from his post as professor of multilingual communication, but continues to be active in translation studies, refereeing, writing, supervising theses and giving occasional lectures. His main research interests have been in contrastive analysis; translation theory, translation norms, universals, and ethics; and research methodology. He was CETRA Professor in 1999 (Catholic University of Leuven) and has an honorary doctorate from the Copenhagen Business School.