A note on norms and evidence


Andrew Chesterman


[2006b. In J. Tommola and Y. Gambier, eds, Translation and interpreting – training and research. Turku, University of Turku, Department of English Translation Studies. 13-19.]




There are two senses of the concept “norm”: one is descriptive and weakly explanatory, and the other is causal or prescriptive and more strongly explanatory. Studying norms in the causal sense means looking for plausible links between observed regularities and evidence of normative force: this may be found in belief statements, in criticism of norm-breaking, or in norm statements. Norms are explanatory hypotheses, and can be tested in various ways.




1. Introduction


Norms have become a key concept in Translation Studies, at least since Toury (1980). But there is still disagreement about how best to define them, and also how to study them. One stage in the conceptual debate was summarized in the contributions to Schäffner (1998), a collection of papers which debated several important issues concerning the interpretation of norms within Descriptive Translation Studies (DTS), such as the relation between norms and conventions, and between norms and expectations.

              The starting-point for the present paper was a conference organized in 2004 by the School of Languages and Area Studies at the University of Portsmouth, on the topic: “Translation norms. What is ‘normal’ in the translation profession” (see Kemble 2005 for the Proceedings). As the conference title indicates, the perspective here was not exclusively that broadly taken by scholars working within DTS; the scope of the term “norms” was taken in a wider sense, to include “what is ‘normal’”. However, this wider scope was the cause of some conceptual confusion, which was evident not only at the conference itself but also in the written proceedings (as pointed out e.g. by myself in the same Proceedings). This leads not only to misunderstanding but also to methodological difficulties: different interpretations of the norm concept imply different methodological decisions.[1]

If we can clarify the concept of norms more sharply, and agree more explicitly about what would count as evidence for norms, it might be easier to answer some of the problematic questions that arise when we consider appropriate methodologies for studying norms.


2. The two senses


Let us start by stating the two interpretations as explicitly as possible.

              Under interpretation (a), a norm is defined as a tendency; it denotes an instance of typical behaviour. In this sense, a norm is a descriptive notion, and is close to the concept of a convention and to the idea of what is “normal”. In this sense, I could claim that payment norms for subtitlers seem to be gradually falling, as more and more work is done by less qualified translators for lower rates, and clients appear to be satisfied with poorer quality: this is becoming “normal”. This claim does not imply anything about what should be the case.

              Under interpretation (b), on the other hand, a norm is understood as something rather more complex. According to Bartsch (1987: xiv), norms in this sense are interpreted as “the social reality of correctness notions”. This definition can be further specified as implying the following three elements (paraphrased from Bartsch 1987: 76):


              X manifests / is caused by a norm if

              (i) most people (in a given society or group at a given time, under given conditions) regularly do X; and

              (ii) they think they should indeed do X; and

              (iii) they can justifiably be criticized if they do not do X.


This is the general sense in which norms have been understood in most DTS. Unlike interpretation (a), interpretation (b) is a causal one, not just a descriptive one. It implies that people behave in a given way because of certain norms. Norms in this sense carry a normative (or prescriptive) force: they thus affect behaviour.

              Both interpretations are social, but in different ways. Interpretation (a) is social in the sense that it describes a general tendency covering more than a single instance distributed across time and/or place. The discovery of only one badly paid subtitler would not justify a description of this practice as being “normal”. Interpretation (b) is social by definition, in opposition to purely personal principles which may also motivate behaviour. I can claim that I always brush my teeth after breakfast “because I think I ought to”, but this “ought” is not normative in our sense (b), unless it can be shown that it is shared by a significant number of other people who also brush their teeth after breakfast (e.g. rather than before). Norms are not personal (although attitudes to them may be personal).


3. Descriptive and explanatory


Before we proceed to some methodological implications, a reservation is in order. The distinction I have just suggested between the descriptive sense (a) and the explanatory sense (b) is not an absolute one. If we know that translators have a tendency, for instance, to reduce repetition, and we are wondering why a given translator has reduced some source-text repetition in a particular translation, we can offer this explanation: she did it because that is what translators typically do. We have explained this translator’s behaviour by relating it to a general tendency. True, this explanation is a rather weak one, in that it does not show that we know anything about why translators in general tend to reduce repetition. Nor does it show that a given translator did this because she was aware that this is what translators often/generally do. But our realization that this translator is behaving just like many other translators certainly lessens any surprise we may feel at her reductions: they are, in this sense, quite normal, quite expected. (For further discussion of generalization as a first step towards explanation, see also Halverson 2003).

              It is in this sense that Toury’s proposal for translation “laws” relates to the title of his book: Descriptive Translation Studies and Beyond (Toury 1995). The “beyond” refers partly to the explanatory power of the proposed laws (of interference and growing standardization). If these general tendencies do indeed exist, they do serve to explain some aspects of translatorial behaviour in some way. The same is true of the increasing body of reseach on translation universals (see e.g. Mauranen and Kujamäki 2004).

              But Toury’s book also offers explanatory concepts of a stronger kind: norms, in our sense (b). Norms in this sense can of course be described (i.e. not prescribed, not laid down by the researcher), and so we can take a descriptive approach to them, just as we can take a descriptive approach to ethical values for instance, or the language of instructions. Having identified and described norms, we can then appeal to them in order to formulate explanatory hypotheses e.g. about the occurrence or distribution of certain translational features. A translator may, for instance, naturalize source-text “miles” to target-text “kilometers” because she is following a domestication norm to this effect, or perhaps because she is deliberately resisting an opposite norm according to which she should keep the foreignizing “miles”. However, it is relevant to underline that this formulation of a causal link is strictly speaking imprecise: norms do not affect behaviour directly, because their influence must be filtered through the translator’s mind as decisions are made during the translation act. Translators adopt attitudes to norms, if they are aware of them: to follow them or not, as the case may be. A translator follows a norm because she chooses to, for whatever reason; or, she does not follow it because she does not choose to (or of course because she does not know about it).

              Norms, especially in sense (b), are thus abstract, intangible. How can we study them? What counts as evidence for the existence of a norm? These questions lead to the methodological implications of the two interpretations.


4. Regularities


If data analysis shows up regularities, repeated patterns, tendencies, this of course constitutes evidence for norms in our interpretation (a). Indeed, to say that a regularity reveals a norm, in this sense, is no more than a pleonasm.

But an observed regularity does not constitute sufficient evidence for a norm in our interpretation (b), the DTS sense. This sense is a causal one, as mentioned above. The cause of an observed regularity may be the existence of a norm, but it does not have to be. Other possible causes include cognitive constraints, time and task constraints, or factors concerning the translator’s background knowledge and proficiency – and of course chance. If amateur translators regularly produce poor-quality translations, for instance, we do not assume that they are following a norm to this effect. Interference, too, is a regular tendency, but it is not often likely to be directly influenced by norms (although norms may of course affect people’s reactions to interference).

              On the other hand, we cannot argue that something is a norm without some evidence of a regularity which would manifest this norm (or, as mentioned above, which would manifest the translator’s attitude to this norm). (Cf. condition (i) in Bartsch’s definition, above.) An isolated case, without the support of other similar cases, cannot be adequate evidence of a norm; nor can non-systematic instances which do not suggest a tendency of any kind. Observed regularities are thus necessary conditions for norms in our sense (b), but not sufficient ones.

              The challenge of research into norms is in fact to show plausible links between observed regularities on the one hand and evidence of normative force on the other.


5. Evidence of normative force


Evidence of normative force comes, I suggest, in three forms.

              Belief statements. These are statements which people use to justify an action. They have the general form: “I think people should (not) do X (under conditions ABC)”. (Compare condition (ii) in Bartsch’s definition.) Translation research can find this kind of evidence e.g. in interviews with translators, in translator notes and prefaces (translator’s statements), in essays or letters written by translators about their work, in texts by critics or clients or consumers of translations. If such evidence is not found in the form of explicit belief statements, a researcher may be able to argue that there is at least implicit or circumstantial evidence of this kind: there may be evidence that people seem to believe that they should (not) do X. However, the mere existence of belief statements is not a sufficient condition for postulating a norm, because of the gap between what people say they believe and what they actually do believe and/or what they actually do. If people do not in fact follow a given norm N, although they say they believe they should do so, we do not have strong evidence for the existence of the norm. Maybe there is another norm, a competing norm M, which people actually follow.

              Explicit criticism. Criticism against behaviour that does not conform to a given norm can be evidence of the existence of the norm. (Recall condition (iii) in the defintion above.) Bartsch points out (1987: 76) that such criticism is considered to be justified by other members of the society concerned, in that the criticism itself is not criticised. Criticism of norm-breaking in Translation Studies can be found in teachers’ comments and feedback, in translation reviews, and also in clients’ and consumers’ reactions, perhaps also in the reactions of other translators. However, criticism may also have other causes. A critical reaction may also be due to a personal preference or opinion, bad temper, rivalry, or some other reason.

              Norm statements. A norm statement (or norm codification: Bartsch 1987: 177) is an official statement by a norm authority of the content of a norm. They have the implicit form: “we hereby declare that (under conditions ABC) people should (not) do X”. A norm authority is a person or institution authorized by a given society to issue and maintain norms. In the field of translation, norm authorities may be trainers, publishers, patrons, literary critics, clients, cultural institutions, governments, and so on. A publisher or a large company, or an international body such as the EU, for instance, may issue style norms which translators have to follow.

There may also be explicit norm statements about changes permitted or forbidden, such as the EU norm about not changing sentence breaks in translations of legal documents. In this category we can also include official statements about what may or may not be translated: translation censureship. (See e.g. the special issue of TTR: 2, 2002.)

              Norm statements by norm authorities can make a norm valid. But not all norms have associated norm statements: a norm may become valid simply by being followed. I have never seen a notice in a lift forbidding eye-contact with strangers, but there does seem to be a norm to this effect: people simply avoid prolonged eye-contact in these circumstances, because it feels uncomfortable. There is a shared feeling that one ought not to do it; it would be rude. Anyone breaking this norm may provoke irritation or embarrassment.

              On the other hand, if (most of) the relevant people do not actually conform to a given norm statement, it is hard to argue that the norm in question actually exists. A norm that is not followed is not really a norm at all, because it has no social existence. So norm statements in the absence of evidence that a norm is being followed (regularities, tendencies) do not make for persuasive evidence.

              A researcher who wishes to propose that a given norm N exists therefore needs to produce as much evidence of normative force as possible, and to link this plausibly with evidence of observed regularities. Both kinds of evidence are necessary; neither suffices on its own.

              But norms are slippery, abstract things. If they exist at all, they are somewhere in the social consciousness, in Popper’s World 3 (see e.g. Popper 1972: 106f). Norms themselves lie hidden behind regularities and beliefs and norm statements. In other words, norms themselves are best considered as explanatory hypotheses rather than observable facts.


6. Testing hypotheses about norms


The main point of sections 4 and 5 above can be summarized by saying that this is how hypotheses about norms (in the DTS sense) can be generated. One can either start with observed regularities and look for related signs of normative force, or one can start with some evidence for normative force and check for corresponding regularities; or even work both ways at the same time. In translation research, both textual (e.g. regularities, norm or belief statements, written criticism) and extratextual sources are relevant (observation of translator work procedures, interviews etc.). But like all hypotheses, these claims then need to be tested. How? Here are some suggestions.

Triangulate the evidence. Triangulation (in research) means using different methodologies or data to explore the same research question, and then looking for connections or correlations between the results, in the hope that the results gained by one method can corroborate the results gained by another method. If research can show, for instance, that different kinds of evidence of normative force can be found concerning the same regularities, the norm hypothesis in question becomes more convincing. A good example of this is some research by Brownlie (2003), in which she correlated data from questionnaires sent to translators with data from their translations themselves. She asked the translators about their attitudes towards literal translation, cutting up long sentences in the source text, foreignizing, clarifying the source text, the treatment of technical terms, and so on. These data are then related both to publishers’ instructions and to textual features. Brownlie also discusses several of the issues raised in the present paper.

              Norm-breaking. A researcher might deliberately break a hypothesized norm, and see what happens. Does this provoke criticism? Sanctions? (Try eye-contact with strangers in a lift and see what happens.) This means in fact formulating a predictive hypothesis.

              Counter-evidence. Alternatively, one can look for counter-evidence (always a good idea) and honestly report what one finds. By this I mean looking not only for evidence of norm-breaking which is not followed by criticism, but also evidence of competing norms that may be being followed under the same conditions.

              Norm-conforming. Test for the results of conforming to a given norm. How do people in a foreign culture react if you try to follow their norms? (I.e. try another predictive hypothesis.) I read recently of a western journalist attending a Hindu religious rite in India, in a small boat on a river. All the other westerner observers had cameras and filmed everything excitedly, but she behaved quite differently, sitting quite still and respectfully, with no camera, like the local believers taking part in the rite. Her boat was the only western one allowed to float undisturbed to the centre of the area where the rite was taking place: she had conformed to the behaviour norm.

              Elicit a belief statement. Ask a member of the society concerned whether it is true that (under conditions ABC) people should do X.  Or ask openly whether a given observed regularity is a norm.

              Elicit a norm statement. Ask a norm authority the same questions.

              Expand the data-base. If we see regularities in one text, or set of texts, which we suspect to be norm-governed, we can check for the same regularity in other texts which might be assumed to be within the scope of the same norm. Translation scholars may also be able to check paratexts and extratextual sources.

None of these tests might be conclusive, of course. But they might well strengthen or weaken a norm hypothesis.





Bartsch, Renate 1987. Norms of Language. London: Longman.

Brownlie, Siobhan 2003. Investigating explanations of translation phenomena. A case for multiple causality. Target 15, 1, 111-152.

Halverson, Sandra 2003. The cognitive basis of translation universals. Target 15, 2, 197-241.

Kemble, Ian (ed.) 2005. Translation Norms. What is ‘normal’ in the translation profession? Portsmouth: University of Portsmouth, School of Languages and Area Studies.

Mauranen, Anna and Pekka Kujamäki (eds) 2004. Translation Universals: Do they exist? Amsterdam: Benjamins.

Popper, Karl R. 1972. Objective Knowledge. An evolutionary approach. Oxford: Clarendon Press.

Schäffner, Christina (ed.) 1998. Translation and Norms. Current Issues in Language and Society 5, 1&2.

Toury, Gideon 1980. In Search of a Theory of Translation. Tel Aviv: Porter Institute for Poetics and Semiotics.

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



Back to Publications List

[1] My thanks to the members of the MonAKO research seminar, where many of these ideas were discussed and developed in two workshops on norms in 2005.