2000c What constitutes “progress” in Translation Studies? In Birgitta Englund Dimitrova (ed.), Översättning och tolkning. Rapport från ASLA:s höstsymposium, Stockholm, 5-6 november 1998. Uppsala: ASLA, 33-49.
(Note. This www-text is slightly edited from the printed version: the distinction is clarified between descriptive and interpretive hypotheses.)
Notions of progress in Translation Studies (hereafter TS) depend largely on what kind of academic discipline you think TS is. There seem to be three basic views, each of which is based on certain philosophical assumptions. These different assumptions give rise to a good deal of misunderstanding, as well as disagreement. Individual scholars are not necessarily to be associated exclusively with one or another of these views, but I find the three positions useful in drawing up a map of the current state of affairs.
First, there is the view that TS is an applied science, like engineering or technology, or perhaps economics. On this view, the goal of the discipline is to solve practical problems: specifically, problems that are seen by society as being important. Seen in this light, the discipline has an agenda that is set by external needs: the need for a high-quality machine translation system, for instance; or the need for good translators that can be trained rapidly and at low cost; or the need for a multilingual term bank of financial and banking terminology. The keyword of this view is for: TS is for society, for translators, for international communicators.
This is the view held by most professional translators I know. The kinds of questions that interest them are, for instance: How to translate faster without sacrificing quality? How to improve translation quality? How to improve source-text quality? How to educate clients, so that they have a more realistic idea of what translating involves? How to train future translators more effectively? How to set up a translation typology that can be used to specify translation contracts more realistically? Where are the guidelines or principles that professionals can turn to in order to get ideas about how to solve particular kinds of recurrent translation problems?
In other words, these professionals assume that TS is there to help them. They want to “look up” to theoretical work, to find solutions there to their own problems. To some extent, however, their concerns conflict with one of the recent developments that is seen by many as a sign of progress: the change from prescriptive to descriptive work. Descriptive scholars do not see it as part of their job to hand down useful guidelines; for them, professional translators and their products are the object of research, that which is to be described and explained.
Applied TS has of course addressed some of the issues that professionals are interested in — much has been written on training methods, on practical problem-solving, on assessment criteria, etc. There are even general national and international guidelines on good translation practice, too (if these can be counted as applied TS). Skopos theory can also be mentioned here: it aims (or appears to aim — see Chesterman 1998) to describe the factors leading to an optimum translation, one that satisfies the skopos, and the theory thus preserves an inherent prescriptive aspect that is invaluable for instance in teaching. In principle this could be invaluable for practising translators, too.
The whole of the machine translation project, with all its by-products, also deserves to be mentioned. Whether it is fair to include this work under TS — it was not originally developed by translation scholars but by mathematicians and computer engineers — is a point I will not discuss here, except to note that it certainly plays a role on the contemporary translation conference scene.
The second view sees TS as a hermeneutic discipline, one of the humanities, like literary theory or philosophical conceptual analysis. To the extent that this kind of discipline is also conceptualized as “problem-solving”, its problems mainly originate not in external society but within the discipline itself; they are problems that are deemed scientifically interesting for their own sake. Examples are: How to define translation? How to define equivalence? What do we gain, or lose, if we see translation as cannibalism, or as manipulation, or as performance, or as decoding? What insights does postmodernism bring to our understanding of translation? How do certain types of translations betray the Other? (For a recent representative example, see Bassnett and Lefevere 1998.) The keyword symbolizing this view is as: translation, or the translator, is see as something. This hermeneutic as lies of course at the root of many theories of understanding: one understands an unknown by seeing it, or framing it, as a known: thus we understand (or think we understand) light by conceiving of it as a wave, or as particles.
The cultural turn taken by many TS scholars in the 1980s, with its interest in translators, clients and patrons as manipulators of power, also brought with it a line of research with an explicit ideological agenda. Studying the relations between translation and culture meant not only exploring the patterns of these relations, but also uncovering their hidden causes and effects (e.g. Baker 1996). Critical cultural studies in TS seek to reveal the mechanisms of translatorial manipulation, of bias, in order to increase people’s awareness and hence prevent such manipulation in future. Feminist scholars, for instance, want not just to increase knowledge but to change society. Venuti wishes to change translation practices and thus change the ways in which one culture perceives another.
These latter, critical scholars thus share one particular feature with my first group, those for whom TS is an applied science. For both, TS is a means to an end — an end that may be practical (better quality translations, better translators) or ideological (more emancipated citizens, a better kind of society). However, very few professional translators are concerned with translating the kinds of texts that these scholars tend to study; and they are not primarily worried about as questions, although they may well be motivated by a desire to facilitate intercultural relations more generally.
With respect to practising translators, proponents of this view seem to be mostly interested in such themes as the translator’s self-image, and also the translator’s image in society. The scholars themselves tend to see their role as that of interpreters, interpreting translators to themselves and to society at large, interpreting the causes and effects of various kinds of translations, for instance in terms of prevailing ideologies or ideological conflicts.
Empirical human science
The third view is that TS is an empirical human science, like sociology or psychology; a behavioural science, whose object of study is a particular kind of human behaviour. On this view, TS has the same kinds of aims as any other science: to describe, explain, and predict. Here is the classical statement by Hempel (1952: 1, cited by Toury 1995: 9):
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.
On this view, TS proceeds by proposing well-justified hypotheses and testing them against empirical data. Interrelated and corroborated hypotheses may then lead to theoretical laws. A recent example of this approach is Toury’s work on descriptive translation studies, norms, and laws of translational behaviour (1995). My keyword for this view of TS is if: this kind of research seeks to establish the conditions under which particular kinds of translations are produced, or under which translations have particular kinds of effects.
Scholars working along these lines do not see themselves as people to be “looked up to” by practising translators; on the contrary, these scholars are trying to describe and explain the wonderful and extraordinary things that translators do. Their job is not to suggest guidelines, to prescribe. They have in fact been nearly unanimous in rejecting prescriptivism — understood as the focus on what translators should do, what translations should be like — because they see such a focus as unscientific. Applied translation studies are accepted within this approach, but kept distinct from descriptive or theoretical studies proper (recall Holmes’ map, in Holmes 1988). However, the rejection of prescriptivism has widened the distance between translation theorists and professional translators.
There are considerable disagreements between proponents of these three visions. Some scholars working within a hermeneutic, culture studies approach seem suspicious of attempts to establish general laws of translation. Rosemary Arrojo, for instance (1997), not only criticizes the attempts of earlier scholars such as Dolet to set up prescriptive laws of translation, but is sceptical of the more recent attempts to put translation studies onto a more “scientific” basis. She also rejects the view that the task of translation theory is to propose appropriate translation methods and guidelines for different text-types: precisely the applied science view. What she wishes to preserve is the idea of translation as an art, something that cannot be reduced to a system of rules. Both the applied science view and the empirical science view represent, according to Arrojo, “a conception of theory as a form of privileged legislation that can rule over the practice it selects as its object, regardless of ideology, history or local interests and peculiarities” (11). Possible translation laws are construed by Arrojo as having legislative, presciptive force, and thus as restricting the translator’s freedom. Laws, says Arrojo, need to be enforced, and who will do that? Translation specialists? Theoreticians proposing such laws are merely imposing “their own (inevitably) local perspective which thus becomes that which should be considered as “universal,” “scientific” or “objective” and which should be followed by every translator” (12).
In other words, there is a very different reading here of the nature of empirical laws. Those discussed by empirical scholars in translation theory are no more “legislative” than the kind of laws a sociologist might propose in order to explain why people entering a café tend to choose certain seats — e.g. by the window, by a light, in a corner, as far as possible from other customers so as to share the available territory fairly, and so on. Such general principles are not prescriptive, of course: you can in fact choose any vacant chair you like. The point is that, given certain conditions, most people tend to make certain kinds of choices.
On the other hand, the empirical scientists are suspicious of the hermeneutic literary scholars, who are felt to be merely speculative, not properly scientific, engaged only in metaphysical philosophizing. They are also critical of anything that seems to be prescriptive: in their view, prescriptivism is prescientific, old-fashioned and inevitably subjective.
The practising translators, of course, who want prescriptive guidelines and expect an applied science paradigm, are scornful of anything that does not appear to be of practical relevance. They tend to respond with a “so what?” when introduced to the latest achievements in non-applied Translation Studies — see the critical response to the Routledge Encyclopedia, for instance (e.g. Cross 1998).
Notions and criteria of progress
External progress (or social progress) refers to the institutional status of a discipline, as measured for instance by the number of academic posts, conferences, journals, publications, scientific associations, etc. All these factors reflect (to some extent) the status of the academic field in society, and also (to some extent) the status of professional translators. Measured in this way, the external progress of Translation Studies over the past couple of decades has been remarkable; but I shall not examine this aspect further here.
Internal progress, on the other hand, is evaluated by the members of the academic discipline concerned, or by neighbouring disciplines; it is measured as scientific progress, progress towards greater understanding, or the like. Our question is: how is this internal progress to be understood? Progress might be thought of as gradual approximation towards a goal — but what goal? Each of our three basic positions frames its goal in a different way, and thus has a different notion of progress.
For our first group, the applied scientists, progress is a pragmatic issue, concerning product quality. Basic questions here are: Is this product better than the previous one? Does this new tool work well enough? Does it have more applications? Is this process better? Is the client more satisfied? Is the translator more satisfied? Is there a saving of time and/or money? Is the process now more ecologically viable? Does this way of training translators make for better translators? Progress is something better for someone, or something better for more people.
What progress has TS made in this respect? Actually, quite a lot, especially if you accept the machine translation project as being part of TS. We have many useful computer aids, termbanks etc. Our translator training curricula are developing all the time. However, there are still too many poor translations around, still too many clients who do not understand the translator’s job. Many translators suffer from stress and frustration, and are far from satisfied with their conditions of work. There is still ample room both for translation quality improvement and for establishing better and more realistic methods of assessment.
If you see the field more as a hermeneutic discipline, progress means more understanding, more awareness, more self-awareness; more understanding of historical and cultural influences; a more critical attitude to prevailing ideology; ethically better practices. Progress means a bigger picture of the whole, more abstraction, more generality; but it also means a closer view of the details, more elaboration, more specialization, more articulation. Theoretical progress means a better representation of the object of research, in macrocontext and microdetail. Progress is a better as.
Signs of this progress? Well, we have an ever-increasing accumulation of metapors and images for translation. We have a greater understanding of the translator’s historical role, and of the ideological and other pressures constraining translators’ choices. We can see something of the causes and effects of particular translations or kinds of translations. We can see that certain kinds of translation may conceal hidden agendas, of which the translators themselves may or may not have been conscious. And we can see something of the whole web of power relations amid which the translator sits. So we now have a much broader view of our object of study than scholars did, say, thirty years ago — including ethical aspects. On the other hand, the “we” in these conclusions refers mainly to the scholarly community itself; less so to society at large, to clients, or even to translators. Also, it is not always clear what the added value is of a new metaphor for translation, or whether the adoption of such a metaphor would have any empirical consequences.
For those who see TS as an empirical human science, progress means better hypotheses: those that are more general, more productive of other testable hypotheses, better corroborated; those that suggest empirically justified laws; those that appear to approximate more closely to the truth. It therefore also means more valid explanations and closer predictions; and of course better theories. It also means more facts, upon which hypotheses and laws can be based. In sum, progress is more accurate links between conditions and associated translational phenomena: better ifs, you might say.
Progress? Well, we have lots of empirical case studies, both historical and contemporary. We have several empirical research methods that have been well tested, ranging from descriptive analysis to protocol studies. We have lots of significant facts, and lots of apparent interconnections between translations and surrounding conditions of various kinds. And we have several ways of getting at the causes of translational features, notably the concept of norms. But we have very few explicit hypotheses or laws, very little research that is so explicitly reported that it can be replicated, and indeed very little testing of hypotheses. We have not yet made much progress towards building a general empirical theory of translation, in the empirical sense of “theory” as an explicit body of related axioms and laws. We do not even have a generally accepted translation typology, having to make do with a confusing mass of overlapping terms. Here too, the professional translator is disappointed.
Theories and paradigms
All in all, it is perhaps the applied science approach that has made most visible progress so far, on its own criteria. This is somewhat paradoxical, as it seems that we have been rather less successful in creating anything much to “apply”. The second and third approaches have not yet made much progress towards creating a general theory of translation. True, there are some “theories” in TS: we do use the term “theory” as a countable noun, but in rather restricted senses. Current in TS are at least the following:
• Catford’s linguistic theory of translation
• the interpretive theory of sense
• skopos theory / the functionalist school / translatorial action
• polysystem theory / the manipulation school
Catford’s linguistic theory seeks to be a description of the translator’s potential textual solutions, based on interlinguistic comparison. It is really a taxonomy of textual replacement, accompanied by a translation typology that may be of interest of linguists but is largely irrelevant to the needs of real-world clients. His notion of conditioned equivalence is important — and obviously relevant to machine translation — but the theory itself proposes no hypotheses that can be tested and takes little account of non-textual cause or effect.
The theory of sense, based on the notion of deverbalization, proposes an answer to the question: how do interpreters interpret? It has been much applied in Paris, in interpreter training. However, its main tenet does not appear to be directly testable empirically. In essence, it remains descriptive, and largely speculative.
Skopos theory also centres around a single idea. And it too appears most relevant to the applied science view, as a pedagogical tool. By focussing on one particular kind of cause, it usefully detaches translator trainees from staring too exclusively at the source text; and it also serves to loosen the narrow concern with equivalence. Its empirical status is less clear (see Chesterman 1998), as is its testability. Its focus on one kind of cause may also mean that other causes are neglected. It has so far paid less attention to effects, except in the sense that translations can be usefully assessed in terms of how well they meet their intended skopos. The functional approach as a whole has taken as its starting-point the position of the translator faced with a source text to translate: how to do it? What factors to take into account? However, this approach has been less successful in pinpointing the effects of the skopos on the translator’s actual textual choices, and in proposing testable generalizations about the relationship between skopos and translation-text.
Polysystem theory has been concerned with the causes and the effects of mainly literary translations, at the level of whole cultures rather than specific clients or readers. It has perhaps been most successful in outlining the various major socio-cultural forces that impinge upon a translator’s textual choices, forces that exist mainly in the target culture but also include the client and aspects of the source text and culture. The starting point has been the target text itself, not the position of the translator. Toury has proposed two hypothetical laws (of interference and growing standardization), accompanied by a number of variants in empirically testable form. With respect to our three basic views of TS, polysystem theory represents a blend between literary studies and empirical science, but the latter approach has become more prominent in recent work.
Other “theories” have appeared occasionally, but have remained isolated cases not taken up by many other scholars. In addition, other concepts such as “model” or “school” have been used to label particular approaches, perhaps with more modesty. None of the theories mentioned above, nor other models or schools, have yet attained the status of a general paradigm. All are partial in scope.
In terms of the progress of TS as a scientific discipline, then, we seem to be at what Kuhn (1970) called the pre-paradigmatic stage. We have competing “schools” or “approaches”, sometimes known as “theories” and sometimes not. We have not yet arrived at a period of “normative” science, in which practitioners all share a basic methodology, a set of goals and an understanding of what constitutes an important and interesting problem. We do not even agree yet on what kind of thing might properly be called a theory of translation — in this sense, we do not seem even to be looking for the same kind of object. (Recall the title of Toury’s 1980 book: In search of a theory of translation.)
We find ourselves in this pre-paradigmatic position partly because we have different ideas about what TS should be like as a discipline, as I have illustrated. But partly also because we are coming into Translation Studies from very different angles, with different philosophical backgrounds. Some of us come from literary or cultural studies, some from computer engineering, some from language teaching, some from linguistics, some from history, some from sociology or psychology, even some from professional translation or interpreting. My own generation is mostly self-taught in TS itself. The next generation of translation scholars will be the first that has actually been trained specifically in this discipline. A good research training should expose this coming generation to a wide range of approaches, so that the strongest one can be selected from a varied population as the one best fitted to survive. This too is one way of looking at progress: as the survival of the fittest theories and ideas.
If Kuhn is right, the next stage will be the emergence of a shared paradigm that will hold sway for a while, before other competing paradigms appear to challenge it and the field shakes down to a new (temporary) consensus.
What now follows represents my prediction about how the coming paradigm will look. (Note: this prediction is falsifiable!) In my view, if it does indeed look like this, then that will represent progress. I will single out two particular features: the first concerns explanatory adequacy.
Understanding and explanation
I think that the coming paradigm of TS will centre around building an explanatory theory, not just a descriptive one. This move towards a concern with explanatory adequacy is of course implied in the title of Toury’s 1995 book Descriptive Translation Studies and Beyond: the “beyond” refers to the first attempts to propose explanatory laws of translation.
All our three positions might accept the general point that progress equals solving problems (albeit of different kinds), or at least improving a situation. They might also accept that you make scientific progress if you understand how a phenomenon is related to its wider context or to more general phenomena or ideas; or if you understand more of what happens on the micro-level; or if you can explain the phenomenon better. All three positions also seem to appeal, either explicitly or implicitly, to notions of cause and effect (albeit in different ways).
The differences between the three positions are only partly those dividing an applied science (position one) from a descriptive or theoretical discipline (positions two and three). There are also differences with respect to the traditional divide between explanation and understanding: explanation as being appropriate to the hard sciences and understanding to the soft, more human ones. (See von Wright 1971, upon which I draw in the rest of this section.) Indeed, what we are currently witnessing in TS is reminiscent of the battle in the social and psychological sciences, between positivism and hermeneutics. (The Sokal affair is a recent example.) Both “understanding” and “explanation” are complex notions; furthermore, they largely overlap: if I understand something, this normally implies that I can explain it; and vice versa. Moreover, if someone offers me an explanation of something, and I understand the explanation, I usually also understand the something. Both understanding and explanation reduce ignorance and/or surprise. Both work by relating the unknown to the known, so that the unknown is understood/explained in terms of the known. Both thus increase the known, although the known unknown (that is, the unknown that we know we do not know) also expands, as we see more that we do not understand and cannot explain.
Explanation itself is a broad concept. I can explain (or understand):
— what X is
— how it relates to other things
— how to make it
— how it works
— what’s wrong with it
— how to use it
— why it exists
— why it has the form it has
— what its effects are
— what X is for
— what will happen to X if Y
— what it means
— that it has a meaning
Traditionally, understanding has been linked most closely to the questions “what X means” or “that X has a meaning”, but this does not seem to be a necessary restriction on the acceptable use of the term.
Philosophers of science distinguish between various kinds of explanation (and I will add here: understanding), corresponding to some of the aspects of progress mentioned above. The following points seem especially relevant to TS.
(a) Explanation can be subsumptive (relating X to a higher, more general law or framework) or reductive (relating X to lower-level phenomena). In the former case, we have explanation in terms of covering laws. So: subsumptively, X is explained as a natural subclass of the more general law or phenomenon Y; or, reductively, X is explained by showing how it is made up of A, B and C.
(b) Explanation can be descriptive (explaining what X is or how X came about or how X works) or causal (explaining why).
(c) Causality is not a homogeneous concept. Causal explanation comes in many forms: one such is teleological/final (in its functional and/or intentional senses); others are material, formal, efficient (Aristotle’s set).
(d) Some explanations (and laws) are probabilistic, not universal. They therefore explain, or make intelligible, why something is to be expected, rather than why it must happen.
(e) Some causes are internal to a human agent, mental, “reasons”, for instance beliefs and intentions, innate or acquired propensities to act in a certain way; others are external.
(f) Causes may be necessary, sufficient, or merely contributory. Even mere influences are a kind of cause.
(g) Causes themselves also have causes: we can always continue asking why. For instance, initial causes of features of a translation profile can be sought via protocol studies of the translator’s decision-making process (Toury’s translation act: Toury 1995: 249). Causes of these decisions can then be sought in the translation situation (Toury’s translation event: factors such as the skopos); and further back again in the norms of source and target cultures, etc.
Translation Studies has already made some use of these kinds of explanation. Consider the following kinds of research, for example. We start with a translation (or a translation type), and ask why it is like it is: who translated it, under what conditions, under what pressures from the client, from the source or target socio-culture. (Cf. the work of Lefevere, e.g. 1992.)
Or: we start with the way a given translation is received in the target culture, with reader reactions or assessments (e.g. rejection by client, irritated readers, acceptance in a central or peripheral position in the target culture polysystem). We ask why these effects have occurred, why the translation was received in this way; for answers we can look both at the target culture norms and at features of the translation itself. In the same way, we can compare receptions to different translations of the same source text.
Or: we start with a target text, a source text and a skopos, and investigate how the intended function of the translation has affected the translator’s textual choices. We might then compare our results with other studies using the same source text and target language, but a different skopos; or the same skopos but a different translator, or a different kind of translator.
Or: we start with a source text, a skopos and a translator, and examine the translating process with a think-aloud protocol. We ask how the translator arrives at his or her decisions, how solutions are justified, how work processes seem to affect translation quality.
We can even start with a set of conditions that we believe to be causal: given these conditions, and given this source text and this skopos, the resulting translation will probably have such-and-such features, we predict; and then we can test our prediction, to see whether our assumed causal conditions have indeed had the effect that we thought. Or we can start with a translation with given features and predict how readers or cultures will react, and then test the prediction. If our predictions hold good, then we can assume that we have found a possible explanation.
All these research types involve some notion of explanation: we are trying to explain or understand the causes of translational features or of translational effects. It will rarely be possible to discover a uniquely determining cause, but we are nevertheless focussing on causation in the broad and heterogeneous sense outlined above. We are asking why.
The second, and related, feature that I predict the coming paradign will have is that of explicit hypothesis formation and testing. Like that of explanation, the concept of a hypothesis is also a broad one. The following basic kinds of hypothesis all seem relevant to TS:
• Explanatory hypotheses: these are retrospective, as I have illustrated above. They look back in time from some phenomenon and propose an explanation for it, such that certain antecedent conditions are claimed to cause the phenomenon in question, or to make it intelligible, non-surprising. We can use explanatory hypotheses in studying why translations have certain features, or why we find certain translation effects. A simple example: this translation was rejected by the client because it was written in a clumsy style.
• Predictive hypotheses. These are the strictest sort, because they are easiest to falsify. They are used to test possible explanations. We can predict, for instance, from socio-cultural conditions or translation task conditions to translation type, or from translation features to translation effects. An example: it has been suggested that subsequent translations of a given source text into the same target language tend to be closer to the original than earlier translations (see Gambier 1994). It is worth noting here that prescriptive statements are also implicit predictive hypotheses: you should translate like this, because if, under these conditions, you do not translate like this, I predict that the effect will be that the client or the readers will not like your translation.
• Descriptive hypotheses. These make claims about the generality of descriptive features: for instance that translations tend to use more standardized language than their originals.
• Interpretive hypotheses. These are the kind of hypotheses that are found in literary studies, for instance, and in conceptual, purely theoretical analysis. I can hypothesize, for instance, that this image is a good way of describing or understanding X, or that this is a useful conceptual tool for the analysis of the distance between a souce text and a target text (e.g. Leuven-Zwart 1989-90), or that this is a useful taxonomy of translation error, or a good way of assessing translations.
We can find implicit examples of all these kinds of hypotheses easily enough in the TS literature. However, they are rarely stated explicitly, and hypothesis-testing research is rarely explicitly replicated: this is a point where progress is certainly desirable. Hypotheses are sometimes proposed, but there is often little evidence that they have been tested — either because they are not stated explicitly enough to allow testing, or because they are not even testable in principle. In this case, they remain speculative, however interesting they might seem.
Another reason for the lack of testing and explicit hypothesis-formation might be the emphasis in recent descriptive research on case studies: this work is of course necessary, but unless we can build on it, developing generalizations and hypotheses, we shall remain at this preliminary stage.
Hypotheses are of course tested primarily against empirical data, via observation or experiment. Even interpretive hypotheses can be tested in this way, some of them at least: we can for instance test empirically the hypothesis that translation is a prototype concept (see e.g. Halverson 1998). But we can also test in other ways too. We can compare two hypotheses with each other, in a joint test against data, to see which is better supported. We can test a hypothesis for its own internal consistency, logic, elegance, and its relation to other linked hypotheses. And we can test a hypothesis (for instance, an interpretive hypothesis) for its added value more generally: what do we gain, if we employ a particular concept or analytical method? What will the consequences be? Of course, hypotheses are also tested “socially”, in the academic community, in terms of whether they become widely accepted or not, perhaps even regardless of how well they fare in the above-mentioned kinds of tests: questions of status and power complicate the picture here, at least in the short term.
The most stringent demand on a hypothesis is that it should be empirically falsifiable: for instance, an explicit prediction that turns out not to come true is falsified. Prediction-testing is thus the strongest form of hypothesis-testing. A weaker requirement is that a hypothesis should be testable: test results may suggest support or doubt for the hypothesis. At an even weaker level, a hypothesis might not be testable directly, but it should at least have some testable consequences. If not even this requirement is met, we remain in the realm of speculation.
Corroborated hypotheses might then be related by means of general laws, and we can start building a proper theory... (Laws are not necessarily causal, of course. They may be correlational, or developmental.) When we have such a theory, we can start applying it in all kinds of ways: perhaps in promoting causal conditions which, we confidently predict, will lead to certain kinds of translations, for instance those that best further given ideological or ethical intercultural goals.
By way of illustration, here are some hypotheses that were more or less explicitly mentioned in the abstracts of the recent EST Congress in Granada (1998).
• The hypothesis that interlingual translation is the most central member of the prototype category of translation (Halverson). This is an interpretive hypothesis, which was empirically tested with informants, and supported by results.
• The hypothesis that recent socio-cultural development in Poland leads to greater tolerance of foreignness from a dominant language culture, i.e. English (Kwiecinski). This is an explanatory hypothesis, supported by the evidence. (There might also be other causes involved, of course.)
• The hypothesis that interpreters working with a written text in front of them will occasionally omit long passages because of a growing time-lag (Gile’s hypothesis, tested here by Lamberger-Felber). A predictive hypothesis, tested experimentally and supported.
• The hypothesis that translations of mass culture will be characterized by such features as homogenized style, narrative and size; elimination of some sexual or scatalogical elements, etc. (Milton). This is phrased in the abstract as a predictive hypothesis (under these conditions, translations will have these features). It seems actually more like a descriptive hypothesis: a general claim about characteristic features of a certain kind of translation.
• The hypothesis that in a recall and recognition test, there will be more correct responses after mere listening than after interpreting, i.e. that interpreting inhibits memory (Tommola and Karunen). A predictive hypothesis, from conditions to performance, tested experimentally to replicate other research, with rather inconclusive results.
• The hypothesis that translation constitutes an act of interpretation based on the ideological presuppositions of the target society (Lianeri). Phrased in this way, this looks like a vague kind of interpretive hypothesis, rather difficult to test as such. In intention, it is rather an explanatory hypothesis of effect, in that the research referred to in the abstract explores how certain translations affected the target culture of Victorian Britain.
Two methodological comments are worth making at this point. First, there seems to be little reported research on hypotheses that were tested and not supported: such results are surely also valuable, particularly if the hypotheses seem reasonable. Second, hypotheses have tended to be tested on their own, not in relation to other, perhaps competing hypotheses, which would surely be a good idea. The vague conclusion that hypothesis H was supported (let alone the rash claim that it was “confirmed”) is hard to evaluate in the absence of other possible hypotheses which might have been even better supported.
A focus on explanation, on both cause and effect, thus provides a convenient bridge between our three initial positions. The applied scientists are interested in effects; in particular, effects that are perceived to be practical and positive: good quality translations, better tools, etc. Quality assessment itself is a kind of effect produced by the translation. So we need to know what kinds of conditions lead to translations that have desired positive effects, which are assessed as good. In fact, the best way in which pure theory can made relevant to professionals is surely this: if we can systematize a wide range of textual choices made by other translators both present and past, relating them to their background conditions, together with an analysis of the various effects of these choices, then we can provide information of great value to translators pondering their own choices and imagining their potential effects.
We need to know more about useful ways of analysing the effects of translations in general, too. The literary scholars are interested in such matters as the socio-cultural and ideological causes of the form (profile) of a translation, and in its reception (i.e. effects) in the target culture. And the empirical scholars are interested in norms as causes, and in how translations affect target norms; how translatorial decision-making affects translation quality; and how translatorial decision-making is itself affected by various factors.
It seems that many of us are looking at the same elephant, but talking different languages. Maybe what we need is some mutual translation between our different frames of reference.
Translation Studies is a multifaceted interdiscipline, drawing on different research traditions. Because of this, and because of the very nature of translation as a complex human (and machine) activity, we need broader and more differentiated concepts of explanation, causality, and hypothesis-formation and testing than those prevailing in narrower fields. This broadening may enable us to build a general empirical theory of translation that is both rich and robust, one that can make the best use of all three of our initial positions. And that will be real progress.
Arrojo, Rosemary 1997. Asymmetrical relations of power and the ethics of translation. TEXTconTEXT 11, 5-24.
Baker, Mona 1996. Linguistics and cultural studies: Complementary or competing paradigms in Translation Studies? In Lauer, A. et al. (eds), Übersetzungswissenschaft im Umbruch. Festschrift für Wolfram Wilss zum 70. Geburtstag. Tübingen: Narr, 9-19.
Bassnett, Susan and André Lefevere 1998. Constructing Cultures. Clevedon: Multilingual Matters.
Chesterman, Andrew 1998. Review of Hans J. Vermeer, A Skopos Theory of Translation (Some arguments for and against). Target 10, 1, 155-159.
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.
Halverson, Sandra 1998. Prototype effects in the ‘translation’ category. Paper presented at the EST Congress, Granada, 1998. Also part of her PhD thesis Concepts and Categories in Translation Studies. 1998. Department of English, University of Bergen.
Hempel, Carl G.1952. Fundamentals of Concept Formation in Empirical Sciences. Chicago: University of Chicago Press.
Holmes, James S. 1988. Translated! Papers on Literary Translation and Translation Studies. Amsterdam: Rodopi.
Kuhn, Thomas S.  1970. The Structure of Scientific Revolutions. Second, enlarged edition. Chicago: University of Chicago Press.
Lefevere, André 1992. Translation, Rewriting and the Manipulation of Literary Fame. London: Routledge.
Leuven-Zwart, Kitty M. van 1989/1990. Translation and original. Similarities and dissimilarities, I and II. Target 1, 2, 151-181 and 2, 1, 69-95.
Toury, Gideon 1980. In Search of a Theory of Translation. Tel Aviv: Porter Institute.
Toury, Gideon 1995. Descriptive Translation Studies and Beyond. Amsterdam: Benjamins.
Wright, Georg Henrik von 1971. Explanation and Understanding. London: Routledge and Kegan Paul.