Andrew Chesterman


1999a.         The empirical status of prescriptivism. Folia Translatologica 6, 9-19.




1. The gap between theory and practice


Translation Studies studies translation, translations, and increasingly also the process of translating. There is therefore some cause for concern when the people who actually do the translating seem to find Translation Studies irrelevant, metaphysical nonsense.

         This, for instance, is the view of Lars Berglund, an industrial translator working in Germany (1990). He is especially critical of the German academic tradition of Übersetzungswissenschaft, which he describes as “an irrelevance, a discipline without much significance to translators and the people they serve” (145). According to Berglund, this is mainly because academic theoretical studies have an idealized view of source texts and because “they reflect an inadequate and misguided view of the translator’s obligations and loyalties, and seem to ignore the practical requirements that arise from these obligations” (146). For practising translators, this is a serious legitimacy problem.

          Or consider the critical response among professional translators to the recently published Routledge Encyclopedia of Translation Studies (1998), which is surely a reasonable attempt to crystallize the current state of the art. A review in Language International (10,1, 1998) comments: “language professionals involved in the day-to-day business of translation will find it largely irrelevant to their immediate concerns” (39). The work lacks entries on e.g. dictionaries, translation tools, LISA (the Localization Industry Standards Association), translation markets or quality certification.

         Or this, from a review of the same work in the ITI Bulletin (Cross 1998): “What [the Encyclopedia] does not include at all is any reference to the largest body of translation work being done in the world today, that is “technical” translation. [...] Will it help one become a better translator? I doubt it. [...] Does it help to give the translation profession a feeling of self-esteem and worth? Hardly. [...] From the point of view of my working life, it is interesting but irrelevant.”

         There are several possible reasons for reactions such as these. One is that practising translators have particular expectations about what kind of a discipline Translation Studies is supposed to be, expectations that are very different from the assumptions held by many translation scholars. Translators (and many translation teachers too) expect it to be an applied science; some scholars might agree, but others would see their field very differently, either as a hermeneutic humanistic discipline like literary theory or culture studies, or as a human science like psychology or sociology or linguistics. Different conceptions of the field are also reflected in the kinds of translations that tend to be studied: predominantly literary ones or not, for instance.


2. To describe and explain


I suggest that a further reason for this gap between theory and practice is, paradoxically, the stress given in much recent work to a descriptive approach, as opposed to a prescriptive one. I say “paradoxically”, because the whole point of inaugurating a descriptive approach (see e.g. Toury’s early work, summarized in Toury 1995: see especially pp. 7-19) was to push the discipline onto a proper scientific basis, a basis formed of empirical facts about what translations are actually like, not wishful speculative thinking about what they should be like. We study the norms, descriptively; it is the norms that prescribe, not the scholar.

         Toury (1995: 9) cites Hempel (1952:1) on the fundamental aims of any empirical science:


Empirical science has two major objectives: to describe particular phenomena in the world of our experience and to establish general principles by means of which they can be explained and predicted. The explanatory and predictive principles of a scientific discipline are stated in its hypothetical generalizations and its theories; they characterize general patterns or regularities to which the individual phenomena conform and by virtue of which their occurrence can be systematically anticipated.


At this point, some scholars might object that Hempel’s philosophy of science is more suited to the natural sciences than to a humanistic discipline like translation, in which (these scholars might say) the aim is not to explain and predict phenomena in the way that this is done in, say, physics, but to simply to understand them. I shall not discuss this philosophical objection here, except to say that the borderline between explaining and understanding is very fuzzy, and that “predicting” can also cover a weaker sense of “not being surprised at”. I shall assume that Hempel’s basic view can also be extended and/or adapted to cover hermeneutic approaches. At its core is simply the observation that scientists and scholars in general seek to describe and explain — and hence to understand — and that they do this by means of proposing, testing and refining or rejecting hypotheses, or generalizations of different kinds (not necessarily explicitly, it is true).

         The bulk of Toury’s book focuses on the aim and methodology of describing what translations are like, but it also encompasses the explanatory aim: why are they like that? — Hence the need to study norms. Norms can be studied descriptively, but they have prescriptive and thus explanatory force. Recall Toury’s statement of the issues that Translation Studies must tackle (1995: 15): all that translation can involve; what it does involve (not: should!) and the reasons for this (explanation); and what it is likely to involve under given conditions (i.e. prediction). The ultimate theoretical aim is then to propose hypotheses which can be tested, and which may become established as laws (not prescriptive laws, but abstract explanatory generalizations concerning the behaviour of translators).

         The rejection of prescriptivism, explicitly stated in many contexts by descriptive translation scholars, has not been taken on board by everyone. Scholars who see Translation Studies in more applied terms have continued to be concerned with issues of translation quality, translator training, the improvement of translator aids, etc. Newmark (e.g. 1988) continues to hand down prescriptive common sense, and Viaggio (e.g. 1994) has insisted that concepts of quality are inherent in a community’s concept of translation itself. Says Viaggio (1994: 97): “prescriptivism is inescapable. If there is no right, or at least better way of translating, then we are all wasting time, breath and money: there is nothing but language to teach.”

         Viaggio’s view here can be interestingly linked to the philosopher Alasdair MacIntyre’s discussion of virtue (1981). MacIntyre argues that the classical concept of virtue can be best understood as relating to a practice: virtue is (or was understood to be, in classical antiquity) excellence in a social practice. Practices are things like carpentry, soldiering... translating. One of MacIntyre’s points is that functional concepts which describe practices naturally imply a qualitative, ethical aspect. To say (my example) that “X is a translator”, for instance, implies “X ought to do what a translator ought to do”, and furthermore that “X must know what a good translator is like, i.e. one who does what he/she ought to do”. At least, these implications hold under the default reading of the utterance “X is a translator”. If the implications are known not to hold for X (a non-default reading), we would probably rather say “X is a bad translator” or “X is a translator, but I would not recommend him” or the like. (Thus does MacIntyre derive “ought” from “is” — cf. Chesterman 1993.)


3. Hypotheses of cause and effect


Translations are not merely phenomena that are determined or caused or influenced by other phenomena; they are also phenomena that have effects, they themselves are the determining causes of other phenomena. Translations change things, and so translators themselves are also agents of change, not just of preservation. Translations affect readers in multiple ways, they affect target and also source cultures, and they affect intercultural relations and perceptions. Indeed, the cultural turn in Translation Studies has given quite some attention to these issues (consider Venuti’s arguments about the causes and effects of a domesticating translation strategy, for instance: see e.g. Venuti 1995, passim).

         Toury’s approach includes effects implicitly, in several ways. This is evident for instance in notions such as norms themselves, tolerance of interference, or degree of conformity to a norm. Because norms have prescriptive power, norm-breaking may lead to sanctions, which are one obvious kind of effect. We can also seek to explain effects (such as sanctions) by appealing to aspects of a translator’s decisions, or aspects of the commission, or aspects of the sociocultural conditions (norms) prevailing in the source or target cultures.

         It might nevertheless be worth setting the aims of empirical translation studies more broadly than Toury’s formulation, with a more explicit inclusion of translation effects. Let us say that empirical translation studies aim to discover


         (a) what translations are, what characteristics they have;

         (b) why they are like that; and

         (c) what effects they have on readers and cultures.


The first aim is descriptive: we describe the features that translations of various kinds have, e.g. in linguistic terms, and/or in terms of translation strategies, shifts or procedures, and/or in terms of a translation typology; such a description gives us a translation profile. The second aim is explanatory: why do we have this profile, why these features etc.? We can look for reasons in the translator’s state of mind, in the decision-making process (at the level of what Toury calls the translation act); and we can look for causes in the external circumstances of production (at the level of the translation event: Toury 1995: 249), or further afield in the sociocultural conditions of the source and target  languages. The third aim can also be understood descriptively: we can describe (roughly) what effects translations actually appear to have on various kinds of readers and cultures, and we can then try to interpret these effects. (Defining precisely what an effect is, of course, is not easy; nor is an effect something that is always easy to measure.) We can also try to predict certain broad effects: we can propose hypotheses about what effects will follow from a translation that has such-and-such features. If our predictive hypotheses are corroborated, we are then in a position to explain these effects. Let us consider how such predictions fit in with empirical translation research as a whole.

         In Translation Studies, various kinds of explanatory and predictive hypotheses can be broadly classified into four classes. Each of these classes has to do with a relation between a particular translation profile, i.e. features or strategies evident in a given translation (or type of translation), and either its antecedent conditions or its effects. These relations may be causal, such that the translation itself is either determined (the dependent variable) or determining (an independent variable). But the relation may also turn out to be a weaker one, of correlation only; or it may of course even be coincidental. The four classes of hypotheses can be illustrated as follows (where causes, profile and effects obviously represent no more than labels for complex systems and hierarchies):


                  Conditions (causes)


                                     (i)  ß                     (ii)  Ý


                  Translation profile


                                     (iii  ß           (iv)  Ý





         (i) Given certain antecedent conditions concerning the translator and the circumstances of translation, we can hypothesize (predict) that these will give rise to particular translation profiles, features, strategies or even general laws of translation behaviour.  Example: if literary texts are translated from small (dominated) cultures into big (hegemonic) cultures, translators will tend to use a domesticating strategy (cf. Venuti, passim). Or: if a given literary work is translated successively at different periods into a given target culture, the first translation will be plain prose, then later there will be a free translation, then later still a closer translation (inferred from Goethe, see e.g. Lefevere 1992: 75-77; cf. also Gambier 1994). Or: if the translator is a first-year student on a translator training course, the translation will be more literal than if the translator is a more advanced student (inferred from personal experience).

         (ii) Given certain features of a translation profile, we can propose hypotheses about their causes (or, more weakly, about correlations with their conditions of production). Example: given certain shifts in Desfontaines’ translation (in 1743) of Fielding’s novel Joseph Andrews, we can hypothesize that this was because of the translator’s explicitly stated desire to make the translation conform to French stylistic and rhetorical norms, and to make the novel morally more uplifting (Taivalkoski 1998).

         (iii) Given certain translational features, we can propose hypotheses about their effects. Example: if a translation of a children’s story uses a lot of complicated syntactic constructions, it will be less appreciated by adults reading the story to children than a translation that uses simpler constructions, and the simpler translation will be easier for the children to understand (cf. Puurtinen 1995). Or: if texts are translated with a domesticating strategy, the result will be an increase in cultural prejudice and the suppression of difference (Venuti) — not so easy to test, this one! Or: if EU translations have certain features (partly because of the constraints of certain regulations), the effect on EU citizens reading these texts will be a distancing or alienating one, inducing a critical attitude to EU bureaucracy, as evidenced e.g. in irritated letters to newspapers.

         (iv) Given certain effects, we can make hypotheses about the translational features that caused them. Example: if a play that does well in one country is a flop when translated into another culture, the nature of the translation itself may be the cause (Leppihalme 1998). Or: if a translation is rejected by the client, we can hypothesize that this is because it had such-and-such features.

         Hypotheses of types (i) and (iii) are thus predictive, and those of types (ii) and (iv) are explanatory. Comprehensive research structures can obviously link both causes (e.g. client’s goals in commissioning the translation) and effects (are the client’s goals realized? are there unintended or undesirable side-effects?). The immediate effect of given translation conditions is certain features of the translation profile; but these in turn then have their own effects — the chain of causation continues forwards. Similarly, the immediate cause of observed effects is of course the translation profile itself, but this profile is itself determined by the conditions in which the translation was done, at the levels of translation act and translation event, and the chain also extends further back: one can always continue asking “why?”  Ultimately, the extensions are presumably infinite, in both directions: think of the much-cited example from chaos theory, of a butterfly flapping its wings and causing a hurricane on the other side of the world...

         So where does prescriptivism come in? Precisely here: prescriptive statements are none other than hypotheses about translation effects. They fall into class (iii) above, as one kind of predictive hypothesis. True, they are not often presented explicitly as hypotheses, but that is what they are. Statements like “In principle, in authoritative and expressive texts, [original metaphors] should be translated literally” (Newmark 1988: 112), or “translations should aim to have the same effect on their target readers as the source texts had on the source readers”, or “translators should translate transgressively, not fluently”) can be paraphrased approximately like this: “I predict that if translators do not translate in the way I prescribe, the effect will be that readers will not like their translations / that the publisher will reject the text / that intercultural relations will deteriorate” or the like.


4. Problems with hypotheses of effect


The main methodological problem here is this: often, implicit hypotheses of effect are not tested. It is simply assumed that they hold good, that they are common sense or that they follow logically from a given argument. In terms of research methodology, such statements are incomplete, for if they have not been tested they remain speculative, however plausible they may seem, and they cannot contribute to empirical progress.

         The good news is that there is ample scope for useful research here. We might first test prescriptive claims by checking them against what (professional) translators actually do: do they actually translate original metaphors in the way Newmark prescribes? All such metaphors or just some of them? Which ones? Do all translators seem to do this, or only some of them? Which ones? And we can then try to test the implicit claim concerning the assumed effects of not translating metaphors in this way: an experimental study might be able to show how clients, or publisher’s readers, or indeed any other kind of readers, react to other kinds of translation.

         A related methodological problem has to do with sampling. Prescriptive pronouncements (i.e. implicit hypotheses of effect) are usually made by teachers, translation scholars, critics, sometimes by clients: figures who have, or assume they have, some authority. These authority figures thus predict that unless the translation is like this, the effect will be undesirable. But: undesirable to whom? To what readers? To the teacher, scholar or critic? — Yes, presumably, at least to the people making the pronouncements. What about other readers, the readers for whom the translation was actually intended, or the client? Can we assume that the teacher, scholar or critic is a typical reader? I.e. is this sample (usually of a single reader) representative? Obviously not. So on what grounds can a teacher, scholar or critic implicitly claim to represent other readers, to be a valid test-case? Presumably, on the grounds of experience, acquired expertise and imagination, they can put themselves empathetically in the position of other, more typical readers to some extent. But this is no substitute for empirical research testing their prescriptive claims.     Translation effect analysis is still in its infancy. Skopos theory may have a contribution to make here: after all, the skopos of a translation could be understood in terms of its intended effect on its intended readers. The intended effect is thus a condition prior to the translation process, which then regulates (or should regulate) this process. Skopos theory has not yet focussed on ways of assessing actual effects after the translation has been delivered, published and read (see e.g. Vermeer 1996). Vermeer (1998: 52) actually appears to have grave doubts about the feasibility of studying concrete effects in any detail; however, some kinds of general effects are easy enough to observe, and might be studied with the aid of simple parameters such as intended vs. unintended (see Chesterman 1998). The relevance-theoretical approach advocated by Gutt (1991) also seems a fruitful one here.

         In this context, it is surprising that studies of translation quality often rely on the kinds of highly untypical readers I have mentioned. There has been less research on the reactions of real-life clients, of real-life, typical, intended readers.


5. Narrowing the gap between theory and practice


If we include the empirical study of translation effects in our research field, prescriptive approaches thus fall naturally into place as one form of normal hypothesis formation and testing: they do not need to be banned. A second advantage would be that this would narrow the gap between Translation Studies and the quality concerns of practising translators, of clients, of translation teachers, and of society at large.

         Empirical research of any kind often starts with a problem. What, then, are the problems which Translation Studies might be expected at least to tackle, if not to solve? Here are some, grouped in terms of cause and effect.


         Studying causes in general

         — Why is this particular translation like this? Why did the translator use these strategies?

         — Why do translations of a particular type, at a particular period, tend to be of such-and-such a kind?

         — Why do translators (of a given kind) tend to translate as they do? These questions (like several of the others in this list) obviously presuppose that we have an adequate descriptive apparatus available, allowing us to specify the characteristics of a given translation (type) with respect to all relevant parameters.


         Studying effects in general

         — How do translations of certain kinds affect cultural relations? How do they affect readers of different kinds?

         — How can translation effects actually be described and studied? With what conceptual and empirical tools?

         — How do we test the validity of a scholar’s or critic’s assessment of a translation? After all, an assessment is itself (the result of) an effect of a kind: it represents the effect which the translation had on the assessor; but the latter is usually not a typical reader.


         Searching for causes of, or conditions favouring, or strategies leading to,  “undesired” effects

         — Why are there so many bad translations around? As receivers of translations we note that they have a certain effect on us, often an undesired effect: we are irritated by errors and defects. Who produces such translations, under what conditions, for what pay, with what qualifications and training, for clients with what kinds of expectations? The need for sociological fieldwork here is evident.

         — Why are source texts often so badly written?  What can we discover about the people or committees that produce such texts, how can we influence them? (Cf. the “Fight the Fog” campaign currently under way within the EU: for details, see the website <>.) This question thus focuses on one of the assumed causes of bad translations.

         — Why are (covert) translations often nevertheless recognised to be translations? What can we learn from corpus studies of parallel texts about translationese?

         — What are the causes of translation errors and irritating features? Yes, we can make some guesses... Do we then test these guesses?


         Searching for causes of, or conditions favouring, or strategies leading to, “desired” effects

         — How to improve the quality of translations? Yes, we have a lot of research on training methods, and also a good deal on producing useful technical and other aids. There is less on, say, the optimal way of organizing a translation agency, or selecting candidates for a training course.

         — How do competent translators improve on badly written source texts? What strategies do they use, what are their tricks of the trade? How do they organize their work processes? (Mossop 1998)   

         — How do translators successfully solve other kinds of problems, such as translating particularly long sentences, or allusions, or jokes?

         — How to program machines to translate? Yes, much research and progress has been made here. The machine translation project has perhaps been the one that has kept closest to Hempel’s empirical methodology: constructing a machine translation system means proposing hypotheses about what a machine will produce under given conditions; the translations produced are means of testing these hypotheses. Each machine-produced translation has an effect (on readers, also on the programmers): you find it intelligible, or not, you accept it, or you don’t, you wonder how to fine-tune a subcomponent of the program in order to improve some aspect or other, then you test it again...

         — How to educate clients and the general public, to change their attitudes and prejudices about how translators work and what translations can be? Would more enlightened clients contribute towards better translations? Not much research here yet...

         — How to formulate ethical principles of translation that will improve intercultural relations? How then to test these principles, to see whether cultural relations are in fact improved?


         So: let us by all means promote empirical research, let us continue to formulate hypotheses that can be empirically tested, let us seek descriptive and explanatory adequacy for our theories, looking at both causes and effects of translatorial action. But even if we aim to rid translation research of non-empirical bathwater, we do not have to throw out the prescriptive baby as well. The effect of doing so may be undesirable.


Note: This article is based on a paper read at the EST Congress in Granada, 23-26.9.1998.




Berglund, Lars O. 1990. “The search for social significance”. Lebende Sprachen 35/4. 145-151.

Chesterman, Andrew. 1993. “From 'is' to 'ought': translation laws, norms and strategies”. Target 5/1. 1-20.

Chesterman, Andrew. 1998. “Causes, translations, effects”. Target 10/2. 201-230.

Cross, Graham. 1998. Review of Encyclopaedia [sic] of Translation Studies. ITI Bulletin, Feb. 27, 1998.

Gambier, Yves. 1994. “La retraduction, retour et détour”. Meta 39/3. 413-417.

Gutt, Ernst-August. 1991. Translation and Relevance. Cognition and context. Oxford: Blackwell.

Hempel, Carl G.1952. Fundamentals of Concept Formation in Empirical Sciences. Chicago: University of Chicago Press.

Lefevere, André (ed.). 1992. Translation / History / Culture. London: Routledge.

Leppihalme, Ritva. 1998. “Foreignizing strategies in drama translation: the case of the Finnish Oleanna”. Paper read at the EST Congress, Granada, 23-26.9.1998.

MacIntyre, Alasdair. 1981. After Virtue. A study in moral theory. Notre Dame, Indiana: University of Notre Dame Press.

Mossop, Brian. 1998. “Four questions about the work habits of translators”. Paper read at the EST Congress, Granada, 23-26.9.1998.

Newmark, Peter. 1988. A Textbook of Translation. New York: Prentice Hall.

Puurtinen, Tiina. 1995. Linguistic Acceptability in Children’s Literature. Joensuu: University of Joensuu.

Taivalkoski, Kristiina. 1998. Fielding travesti. Étude comparée descriptive de la traduction faite par l’Abbé Desfontaines de Joseph Andrews de Henry Fielding. Unpublished licentiate thesis, Department of Romance Languages, University of Helsinki.

Toury, Gideon. 1995. Descriptive Translation Studies and Beyond. Amsterdam: Benjamins.

Venuti, Lawrence. 1995. The Translator’s Invisibility: A history of translation. London: Routledge.

Vermeer, Hans J. 1996. A Skopos Theory of Translation. Heidelberg: TEXTconTEXT.

Vermeer, Hans J. 1998. “Starting to unask what translatology is all about”. Target 10/1. 41-68.

Viaggio, Sergio 1994. “Theory and professional development: or admonishing translators to be good”. Dollerup, Cay and Annette Lindegaard (eds), Teaching Translation and Interpreting 2. Amsterdam: Benjamins. 97-105.





Back to [Publications list]                    Back to [Homepage]