266
DESCRIPTIVE TRANSLATION STUDIES AND BEYOND
TOWARDS LAWS OF TRANSLATIONAL BEHAVIOUR
267
variables are far easier to control and the studies themselves have a much higher level of replicability, at least in principle.
This is not to say that every descriptive study is, or even should be carried out with a view to revising a theory. In fact, as 1 see it, one of the weaknesses of Translation Studies in the present phase of its evolution lies precisely in the fact that descriptivism as such is often looked down upon, driving almost every scholar to theorize, very often in a highly speculative manner. It will be recalled that a main point in regarding the discipline as a divisible territory was to cater for a better division of labour between practitioners locating themselves at different points on the 'map'. It is to be hoped that acknowledgement of the centrality of descriptive-explanatory work of all kinds will indeed bring about a healthier distinction between the tasks of researchers and theoreticians.
At the same time, the findings of a well-executed study will always bear on the theory in whose framework it has been performed, thus contributing to the verification/refutation/modification of this theory, whether theory-relevant implications are drawn by the researchers themselves or by empirically minded theoreticians. A theory thus refined will, in turn, make possible the execution of yet more elaborate studies, which will then reflect on the theory and render it even more intricate; and so on and so forth, towards an increasingly better understanding of the ways translation and translators, as individuals and members of societal groups alike, manoeuver within the manifold constraints imposed on them, and produce texts which look and function the way they do.
In this process of refinement it will become necessary to do more than just accumulate isolated variables and state the relations between pairs of them, which would have led to formulations of the type
if Xj, and/or X2, and/or ... Xn, then the greater the likelihood that Y, whereas if Zv Z2, and/or ... Zn, then the lesser the likelihood that
Y,
which are basically linear. It will also be necessary to weigh the individual factors and their bearing on translation against each other, as well as establishing their interconnections. The ultimate objective is thus to give every law a multi-conditional format, i.e,
if X1 and Zv then the likelihood that Y is greater than if Xj and Z2, and even greater than if Xj and Z3.
Proceeding this way, translation theory will ultimately become a series of truly interconnected hypotheses, which is the only kind of theory which would offer a possibility of supplementing exhaustive descriptions and viable explanations with justifiable predictions. Incidentally, it would also make it possible to explain the occasional failure of a prediction, by entailing the option of searching for variables which have presumably remained unknown, or changing the respective positions and relationships of [some of] those which have been known all along; in other words, by continuing to perform the helical movement.
A progression of this kind is of course infinite: On the one hand, there will always be something further to account for, so that future studies will become more and more focussed, more and more targeted; on the other hand, the theory of translation can always be further refined. And most fortunately so. After all, one would hate to be accused of foreseeing the end of one's discipline at the very outset... Looking at it the other way around, Translation Studies can be viewed as an inherently optimistic kind of discipline, ever seeking to refine the descriptions and explanations performed within its framework while improving the probability of its predictions, as boldly suggested by Chesterman (1993). At this point, the presentation of a couple of exemplary laws of translational behaviour is called for. Unfortunately, the little deliberate effort invested so far in the establishment of such laws will not enable me to do much more than present two of them in their raw form and trace parts of their conditionalization: the law of growing standardization and the law of interference, which will, furthermore, be found to be largely interconnected.
3.   Two exemplary laws
I will start with a gradually unfolding and partly exemplified exposition of one basic law which decades of text-based research into translational products, in many different cultures, have been able to come up with; namely, the law of growing standardization, which can also be presented as the law of the conversion of textemes to repertoremes.
3.1. The law of growing standardization \
In its most general form, where no conditions have been specified, the tentative law of growing standardization may read as follows:
268 DESCRIPTIVE TRANSLATION STUDIES AND BEYOND
in translation, source-text textemes tend to be converted into target-language (or target-culture) repertoremes.
The rationale underlying this formulation is simple enough:
In every community, phenomena of various types, linguistic and non-linguistic alike, which have semiotic value for its members, undergo codification. Sets of codified items form repertoires, i.e., aggregates governed by systemic relations, which determine the relative availability of items pertaining to such an aggregate for any particular use within the community's culture. In other words, a repertoire amounts to the range of choices which makes cultural functions realizable through real products and practices (see Even-Zohar 1990: 40-43). Any sign, irrespective of rank and scope, which forms part of such an institutionalized repertoire would be defined as a repertoreme.
When a repertoreme is retrieved from the repertoire it is part of and is put to actual use (i.e., inserted in a particular text, in the most general sense of the word), it enters into a unique network of internal relations, peculiar to that act/text. These relations lend the retrieved item ad hoc textual functions, by virtue of which it is rendered a texteme. One and the same item — or, more precisely, different aspects of it — can, of course, partake of several textual structures. As a result, its textemic status is enhanced through syncretism, i.e., the simultaneous coexistence of a number of functions on one single carrier.
On the basis of this brief account, our law can be reformulated to read as follows:
^ in translation, textual relations obtaining in the original are often | modified, sometimes to the point of being totally ignored, in |      favour of [more] habitual options offered by a target repertoire.
Implied is the claim that the dissolution of textual relations, inevitable in translation, is by no means temporary, and is far from characterizing the initial phase of the process only. Rather, the disintegration of the patterns exhibited by a source text is hardly ever reparable in full: its traces can still be observed after the final phase of recomposition has been completed, especially under a comparative observation of translated vs. source texts. This is one of many reasons why translations so often manifest greater standardization than their sources: in practice, decomposed textual patterns are normally reconstructed to a lesser extent than is initially possible.
TOWARDS LAWS OF TRANSLATIONAL BEHAVIOUR
269
Regard the English phrase 'drifting up', whose initial collocability with either 'lawn' or 'vines' is rather low. In the following text, this phrase was made part of a figurative network implying a race over hurdles (see the italicized portions), thus being elevated to the textemic status of a (relatively) living metaphor:
The lawn started at the beach and ran toward the front door for a quarter of a mile, jumping over sundials and brick walks and burning gardens — finally when it reached the house drifting up the side in bright vines as though from the momentum of its run. (Fitzgerald 1972 [11926]: 12-13; all emphases added)
And this is how 'drifting up' in this ad hoc use was translated into three different languages; independently, to be sure:
TT,:    (German): emporrankte
TT2:    (French):  comme emportee
TT}:    (Hebrew): (kedei) letapes [= (in order) to climb]
All three translations thus used items of their respective linguistic repertoires whose collocability with the replacements of 'lawn' and/or 'vines' (which occur in all three of them) is much more habitual, rendering the overall textual structure of each of them considerably looser than that of the original.
The following is another alternative formulation of our law: w0.
in translation, items tend to be selected on a level which is lower than the one where textual relations have been established in the source text.
Regard, for instance, the "Rocky" passage used by Anderson et al. (1977) to illustrate the different possible interpretations of one piece of text in terms of knowledge structures or 'schemata':
Rocky slowly got up from the mat, planning his escape. He hesitated a moment and thought. Things were not going well. What bothered him most was being held, especially since the charge against him had been weak. He considered his present situation. The lock that held him was strong, but he thought he could break it.
Each of the following German translations, supplied by advanced translation students at Heidelberg (Germany) and presented by Kurt Kohn in a recent Conference (see Neubert et al. in press), is no doubt much simpler than the source text, which — devoid of background information — allows for both readings — in full accordance with the law:
270
DESCRIPTIVE TRANSLATION STUDIES AND BEYOND
TOWARDS LAWS OF TRANSLATIONAL BEHAVIOUR
271
TT,: "Ringkampf"
Rocky gelang es, langsam von der Matte wieder hochzukommen. Er versuchte, sich zu befreien. Einen Augenblick lang zögerte er und überlegte. Es stand nicht gerade gut. Am meisten ärgerte ihn, daß er sich in einem festen Griff befand; dabei war der Angriff eher schwach gewesen. Rocky konzentrierte sich auf seine Situation. DerCriff, mit dem er gehalten wurde, war zwar stark, aber er war überzeugt, dali er ihn würde brechen können.
TT2: "Gefängnis"
Langsam stand Rocky von seiner Pritsche (Matraze) auf. Er plante seinen Ausbruch. Für einen Moment zögerte er noch und dachte nach. Die Dinge hatten sich nicht gerade gut entwickelt. Das schlimmste war, daß er jetzt im Gefängnis saß; dabei war die Anklage eher schwach gewesen. Er konzentrierte sich auf seine augenblickliche Lage. Das 5cMoß war zwar stark, aber er war überzeugt, dali er es würde knacken können.
Under certain circumstances, the ambiguity inherent in the source text may itself count as functional; that is, it may form part of the overall semantic structure of the text. Under such circumstances, any disambiguation^a,feature the two German translations share) would result in greater simplification 'still, as it would involve an irreparable dissolution of more, and more intricate textual relations.
Having Kohn's experiment duplicated in a controlled way — that is, having subjects of different backgrounds translate the "Rocky" passage — may well shed more light on the issue at hand. Thus, for reasons beyond the scope of this discussion, the "Gefängnis" option seems to be the unmarked among the two modes of disambiguation. However, within a milieu where the name Rocky has come to be habitually associated with boxing, this last fact may well tilt the balance in favour of the "Ringkampf" option. Then again, it would not be unreasonable to assume that there would also be translations — e.g., those produced by inexperienced translators, who often resort to rather small, and low-rank translation units — which would be impure, in terms of that disambiguation; these would present an admixture of elements belonging to both organizing frameworks.
All this, and much more, would of course add weight to the claim that in real-life translation situations, decisions tend to be made on a level which is lower than that of (the network of relationships which constitutes) the text. What is more, other variables (such as biological and bilingual age, or previous experience in translation of different kinds and for different purposes) may all turn out to be in a position to influence translational behaviour, and may hence
ft
be rendered as 'conditions' in a more elaborate, multiconditional formulation of the law of the conversion of textemes into repertoremes.
This fact notwithstanding, the law can also be conditionalized beyond the level of previous knowledge, or experience, of translators as individuals; for instance, by recourse to one of Even-Zohar's cultural-semiotic hypotheses, first put forward in 1978. Thus, there seems to be a discernible correlation between the degree of flexibility (or rigidity) with which the law is adhered to in a particular (sub)culture and the position assigned in it to translation, both as a type of activity and as an aggregate of texts, such that
the more peripheral this status, the more translation will accommodate itself to established models and repertoires.
That is to say, only when centrality is assigned to translating and/or translations will the law show signs of cracking — within one culture (e.g., with respect to texts of different types), or cross-culturally (e.g., literary translation in England vs. Israel), including different periods in the history of one culture (e.g., translation into Hebrew in Germany during the Enlightenment period and in today's Israel) .s
If this condition holds true, then the operation of our law, or its failure to operate, may serve, conversely, as an indication of the position assumed by translation, or by a certain section thereof, in the target system — which is an important part of any individual study. Since descriptive research has amply demonstrated that this law is only seldom broken, and even then only to a rather limited extent, this can be taken as a verification of another of Even-Zohar's 1978 hypotheses; namely, that
It seems that the most common historical move has been towards greater reliance on the verbal formulation of the source text, even at the price of a lesser extent of acceptability (which was actually connected with the establishment of different concepts of acceptability as a text in the target language/culture and a translation into it). This is one way of connecting the present law with the one which will be discussed later on, the law of interference. Needless to say, this move has not occurred at the same time, or evolutionary phase, in all cultures either, nor even with respect to all types of translation and/or translation of all text-types within one and the same culture. And see, e.g., the claim made by Robyns (1990) that in the 1950s-1960s, Anglo-American detective novels were translated into French according to the tradition of translations knwon as belles infideles, which was said to have disappeared in the 19th century; for canonized literature, to be sure.
272 DESCRIPTIVE TRANSLATION STUDIES AND BEYOND
translation tends to assume a peripheral position in the target system, generally employing secondary models and serving as a major factor of conservatism.
More, and better focussed studies are obviously needed to determine what other factors there are which may reinforce (or attenuate) the operation of our law as well as establish the relative force of these factors themselves and their interdependencies. Thus, there are probably basic cognitive factors which enhance the validity of the conversion law; for instance, limitations of memory and/or other difficulties in drawing abstractions over longer stretches of text, a sine qua non for being able to construct 'mental maps' of either the source or the target text (as posited, for instance, by Holmes 1988: 81-91 ['1978]) and to compare them to each other. Consequently, it would seem that special socio-cultural circumstances are not enough for breaking loose from the bonds of the conversionlaw^eciaTcoRnitive efforts are also probably required, if one is to approximate the reconstruction in target-language material of the network of relationships constituting a source-language text. Moreover, even though basic cognitive capacities are probably universal, it may well be that the way they manifest themselves in individual instances is socio-culturally constrained too, or else they would immediately be marked as deviating from dominant patterns of behaviour, with all that this implies.
To be sure, it is clear that in translation, repertoremes are converted into textemes too, inasmuch as the end product is identifiable as a text to begin with (which implies the existence of an internal network of relations). Textemes occurring in a translated text can of course result from either reconstruction (of source-text relations and their resulting functions) or construction (of new webs).
For instance, a thorough comparative analysis by Even-Zohar (1975) has demonstrated in great detail how a semantic pattern running through Baudelaire's "Spleen" ("Quand le ciel bas et lourd"), where the sky, likened to a cover of a vessel, pours a black day on the world, has disappeared from the Hebrew translation and a developed image of a grave appeared instead. Both translation and source text are thus highly intricate and rich in elements of textemic status, yet their constitutive principles are very different.
To be sure, care should be taken not to couple the obliteration of a source textual pattern and the introduction into the translation of a different set of textual relations in such a way as to regard them automatically — that is, on the basis of absence vs. presence alone — as a pair, allegedly testifying to what is
TOWARDS LAWS OF TRANSLATIONAL BEHAVIOUR
273
sometimes, often rather naively, referred to as compensation. The two practices may well be independent and reflect two completely different, and totally _ unconnected series of translation decisions.6 This seems to have been the case in the Baudelaire poem: there is absolutely no reason to assume that the substitution as such of the 'grave' pattern for the 'vessel' one was intentional. It was more likely a result of a universal of the kind expressed by our law of conversion reinforced by certain needs of the recipient culture of the time.
The relations between obliterating source textual patterns and introducing new ones into its translation are not a given quantity then. Rather, they should always be submitted to investigation; and not from the point of view of the end product alone (as a possible explanatory hypothesis of the shifts which may manifest themselves), but also from the point of view of the process which has yielded it (i.e., as a possible translation strategy). Also, interestingly, descriptive studies have been able to show that even if new structures have been introduced into a translation, and even if there is reason to regard them (or some of them) as compensating for dissolved source-text structures, translations still tend to reveal reduced rates of structuration (that is, simplification, or flattening) vis-avis their sources; which is justification enough for our selection of the law of converting textemes into repertoremes as a privileged law of translation behaviour.
The gradually unfolding theory of translation emerges as optimistic in yet another respect, due to the possibility of taking the laws formulated within it as a basis for conscious manipulation, e.g., in translator training. Thus, while a theory should certainly not be concerned with bringing about changes in the world of our experience, it is precisely one of the advantages of laws of the kind envisaged here that they can be projected onto the applied extensions of Translation Studies too. After all, once a law has been uncovered and formulated, it can be passed on as a piece of knowledge, conditioning factors and all. From this point on, one can be taught how to behave; not only in accordance with the law (which is what one tends to do anyway, otherwise it would hardly have emerged as a law in the first place), but also contrary to it, if this is deemed appropriate, be it for the sake of sheer exercise or for any other reason — fully aware, however, of the deviation from prevalent patterns of behaviour, and hence ready to take the consequences.
A decision to consciously adopt 'compensation' as a strategy (e.g., Hervey
6.    Andcomparejones'briefdiscussion of a target-language ™ extra level of structuration" which did not exist in the source text (1989:192-193).
274
DESCRIPTIVE TRANSLATION STUDIES AND BEYOND
TOWARDS LAWS OF TRANSLATIONAL BEHAVIOUR
275
r-
and Higgins 1992: 34-40) would thus count as a manipulation of the tension between textemic and repertoremic status, and the principle itself could be imparted to translation trainees with relative ease.
Coming back one last time to our metaphor example, one could thus be trained to give up any attempt to reconstruct each and every living metaphor on the spot, just because that is where it is, and resort to establishing other target-language metaphors, located elsewhere, instead. In this way, the very occurrence of living metaphors, even metaphoric density, would be retained as a textual feature, which would amount to giving priority to 'global' over 'local' considerations while paying an obvious price in terms of small-scale matching. (And see my detailed treatment of metaphor as a translation problem in Chapter 3, Section 5.)
A much more general manipulative step in the same vein might be to promote 'textemic translation', in spite of the fact that it has proved so marginal — the exception rather than the rule — in actual practice. In such a case, the recommendation would read:
replace source-text textemes with ad hoc combinations of textual relations equivalent tp-those found in that text and target-language items capable of fulfilling these functions,
even if, as a result, certain deviations from target repertoires would ensue, with possible bearing on the acceptability of the target text; either as a target-language text or even as a translation into it. Or else it would be realized that the law of converting textemes into repertoremes has proven more powerful again, despite the conscious attempt to prevent it from becoming operative. The latter situation would incidentally contribute to greater awareness of the basic nature of the law and to acknowledgement of the crucial fact that every translational decision has its price elsewhere, which — to me — constitutes the most appropriate conceptual framework for translator training anyway (see Excursus C, Section 5).
3.2. First steps towards a law of interference
Let me conclude with a gradual unveiling of a second law which I have been contemplating for years, a so-called law of interference. Nor is this law totally unconnected to the law of conversion, which should be taken as a first welcome step towards the establishment of translation theory as an intricate system of laws.
In its most general form, the law of interference would read:
in translation, phenomena pertaining to the make-up of the source text tend to be transferred to the target text,
whether they manifest themselves in the form of negative transfer (i.e., deviations from normal, codified practices of the target system), or in the form of positive transfer (i.e., greater likelihood of selecting features which do exist and are used in any case). This general formulation implies that interference is a kind of default, so that the establishment of an interference-free output (or even of an output where interference has been relegated to less disturbing domains) necessitates special conditions and/or special efforts on the translator's part.
Basically, what I have called 'discourse transfer' (see Excursus C) has to do with the basic mental processes involved in translation, especially the series of rapid switchings between source and target codes, alternating in both directions. It is thus our mental apparatus which is probably at the root of the universality of transfer in translation. (See also Danchev 1982; James 1988.) In this sense, discourse transfer is the external manifestation of a general cognitive law (see, e.g., Albert and Obler 1978: 209-212). However, were it a matter of pure cognition, all translators, in all cultures, would be expected to behave in just one way while translating anything and everything. Similarly, all cultures, in whatever phase of development, would have to accept interference, at least in the same way/to the same extent, and there would be little point in waging war against it. The fact that none of these consecryences_.hold should lead to the conclusion that tolerance of interference — and hence^the realization of interference itself — have to do with the socio-cultural conditions in which translation is performed and consumed as much as they have to do with our cognitive machinery. There are therefore good reasons to count socio-cultural factors again among the important conditions of the law. This will be done soon enough. First let us stay a little longer within the cognitive paradigm and carry out a tentative process of conditionalization, starting from the following, rather trivial observation concerning discourse transfer.
The extent to which interference actually shows in a translation has to do with whether the source text was approached and processed as one entity, a holistic message in an act of communication, or as an organization of lower-level linguistic entities; for every text is obviously both. The real question is, of course, one of balance. After all, performing translation (and establishing a translational product) with absolutely no recourse to, and no concern for the source text as a
276
DESCRIPTIVE TRANSLATION STUDIES AND BEYOND
higher-level organization of linguistic items — which might have yielded no interference, not even of the 'positive' type — is hardly conceivable; not even in the case of very short, very simple texts which could presumably be retained in short-term memory. On the other hand, attempts to produce translations which would represent their sources 'as closely as possible' are in constant tension with the attempts to establish in a target culture and language native-like texts, especially if these are designed to perform for the target audience a job which is more or less similar to the one performed by the source texts for their own addressees (which has been a recurrent requirement in many cultural domains). From this it should be clear that
r
the more the make-up of a text is taken as a factor in the formulation of its translation, the more the target text can be expected to show traces of interference,7
no matter (for the time being) what it was that brought about the dependence on the make-up of the source-text. Suffice it to say that, popular beliefs notwithstanding, the distance between languages, textual traditions and/or entire cultures seems to have no automatic bearing on the extent of interference. Similarity may well have its implications for the proportion of 'negative' vs. 'positive' transfer, but as, psycholinguistically speaking, there is only one procedure which yields both, interference as such will always be present. It may just be more or less readily seen.
As we have seen in the case of the train notice example (Chapter 4), even with regard to short, relatively simple texts, which are, moreover, codified entities in their respective cultures, translational behaviour cannot be expected to be automatic; and most texts in our world of experience — both originals and translations — are not of that kind anyway. Rather, they are more or less unique acts of linguistic behaviour, which would add grounds for processing them as organizations of entities pertaining to lower levels.8
7. Of course, from the researcher's point of view, what is actually at work is the reciprocal theorem:
The more a translation shows traces of interference, the more closely the make-up of the source text can be hypothesized to have been leaned upon in the translation process.
8. And see again the experiment reported in Chapter 12, Section 2.1, where advanced students of translation were found to replace the German simile strahlenförmig by the
TOWARDS LAWS OF TRANSLATIONAL BEHAVIOUR
277
This fact notwithstanding (and here we are moving towards a possible conditioning factor), tackling larger and/or higher-level constituents of a text has normally been presented as a function of professionality, so that
even when taking the source text as a crucial factor in the formulation of its translation, accomplished translators would be less affected by its actual make-up.
Such a correlation is of course far from wrong. However, there is much more to it, and therefore it seems insufficient even as an explanation of the difference of approach between more and less experienced translators, or even between experienced translators and novices.
Even if it is true that "Experten und Novizen unterscheiden sich nicht in der kognitiven Hardware ihrers Gedächtnisses, sondern eben in der kognitiven Software und der Organisation der Datenbasis" (Esser 1990: 85), the transition from text-, let alone language-orientedness to communication-orientedness, or, alternatively, from source-bound to target-bound decisions, cannot count as self-evident in any sense. This is amply demonstrated by huge portions of the history of translation, even that part of it which was carried out by persons considered 'accomplished translators' in their own time and place: they were no more ignorant of what could be achieved than are today's translators; they simply operated within different socio-cultural settings and hence had different norms as guidelines for their translational behaviour.
The alleged undesirability of interference is thus not 'natural' in any sense. Rather, if and when it is rejected, its undesirability is always a function of a host of socio-cultural factors, which may therefore be said to condition our law. Here it would be quite safe to start by arguing, very generally, that
communities differ in terms of their resistance to interference, especially of the 'negative' type.
Strong resistance to interference may indeed lead to a considerable reduction of its manifestations, especially in the translational output of professionals, shaped
English expressions 'like the rays of the sun' or 'like rays of a star'. Th.s tend «y repeals a high rate of interference, especially in view of the pracUces of speakers of English under non-translational conditions, where 'the spokes of a wheel' was normally used to express the same visual contents.
278
DESCRIPTIVE TRANSLATION STUDIES AND BEYOND
TOWARDS LAWS OF TRANSLATIONAL BEHAVIOUR
279
as it is by the environmental feedback (see Excursus C). Thus, resistance quite readily leads to the activation of purification, or other censorial mechanisms, whose influence, however, can hardly ever be absolute, due to cognitive as well as behavioural factors. These mechanisms are often resorted to post factum, after the act of translation has been terminated, by way of [post]-editing, whether by the translator him-/herself or by some other agent, who may have had a different kind of training and was charged with other responsibilities. Often, such a revisor is not even required to know the source language, and even if s/he does, it is not necessarily the case that s/he also falls back on it. Censorship can also be activated during the act of translation itself though, inasmuch as the translator has internalized the norms pertinent to the culture, and uses them as a constant monitoring device.
However, even the correlation of interference, and tolerance of it, with the overall normative structure of a particular target culture should be submitted to further modification. For it is not as if, within one culture, interference of all 'foreign' languages, all textual traditions, all cultures is always equally tolerated (or rejected). What should be brought in as another conditioning factor is thus the relative prestige of cultures and languages (as seen from the vantage point of the prospective target system) and their power relations with the latter. The rule here seems to be that
tolerance of interference — and hence the endurance of its manifestations — tend to increase when translation is carried out from a 'major' or highly prestigious language/culture, especially if the target language/culture is 'minor', or 'weak' in any other sense,
'majority' and 'minority', 'strength' and 'weakness' being relative rather than fixed, let alone inherent features of languages and cultures.
The sweeping generality of the correlation just established between socio-cultural circumstances and discourse interference in translation may well have to undergo further modifications still, along various additional axes; for instance, with respect to text-type. Thus, towards the end of the 20th century, one would hardly be surprised to find — within one target culture, and with respect to the same source culture — differences in the tolerance of, leading to differences in recourse to interference between, say, instructions for the use of an electric appliance and a lyrical poem; not only, not even mainly on the (lower) linguistic levels, but also on the (highest) level of the textual model.
For instance, the claim may well be that instructions for the use of a VCR translated from Japanese into German are likely to be recast into a textual model favoured by the German culture (which may well entail a movement away from the organizational principles of the source text) — whereas a translated haiku would often be acceptable, maybe even more acceptable, when the Japanese model is reconstructed to capacity. To be sure, this has not always been the case, which means that more work is warranted on the correlations between socio-cultural factors and text-type. Thus, technical translation has not always involved an overall change of model, whereas the dominant tendency in the translation of haiku poems into western languages has long been to remove elements of the original model and adopt conspicuous elements of domestic textual models instead in order to make the translations look more like European poems (see Excursus B).
There are no doubt many more conditioning factors which need — and deserve — to be established, along with their implications for our law. Thus, the following theorem, which stems from the fact that interference on one level, or in one domain, does not necessarily entail interference on all other levels, or in all other domains, may well represent the beginning of yet another process of conditionalization:
even for one and the same text, neither interference nor tolerance of it are necessarily the same with respect to all linguistic and textual levels.
This fact also makes it possible to manipulate interference on the basis of discriminate treatment; e.g., work towards its reduction where its manifestations are considered most annoying while ignoring it in domains where it is considered less problematic. On occasion, this would even make it possible to deliberately adopt interference as a strategy; e.g., in an attempt to enrich the target culture/language, in domains regarded as needing such enrichment, in an act of cultural planning. (See Toury 1985b.)