Article body

1. Preliminaries

Two themes that recur in translation studies are translation process and translation product. The first is concerned with a course of performance during which translation is carried out, the second with the end product. Thus, while a process-oriented concern focuses on how a translation is produced, a product-oriented one would address itself to what has been produced.

Concerns of the latter type manifest themselves, for example, in traditional works that attempt to idealize the product setting up requirements that a translation has to meet. One classic example in this respect is Nida (1964). This is, of course, understandable in a field of inquiry where translation, as an enterprise, is goal-directed. That is, a product should be produced at the end of the day. Newmark (1983: 1) speaks about translation as an “industry” and Sager (1983: 121) argues that translation “is a commercial product of the information market, to which certain price can be attached.” So, a major concern is how to make this product suitable for consumption.

The process orientation is a more recent enterprise that seeks to investigate translation while it is in the making (e.g. Séguinot 1989). This seems to have demanded that the enterprise has to be well equipped to address the question: How do different translators perform the job? The expectation is that performance analysis may contribute to a theory of translation by illuminating our understanding of the nature of the activity. But a query that immediately pops into one’s mind is what kind of data such an investigation would be drawing on. Surely, one can always arrange for an investigation in which a translator is required to translate some kind of text. But what kind of evidence on processing one should be looking for as he observes the translator at work. Most importantly, how he would interpret his observations and make use of them.

There have been attempts (e.g. Gerloff 1987 and Krings 1987) to investigate practically different aspects of translation process. In such works, the authors tell us about the design of their investigations and how they intend to carry them out. But they do not provide us with the basis for data collecting and interpreting within a clearly defined pattern of investigation that could cater to different types of observational data and permit future application and refinement of the adopted procedures.

Such works, however, do point to some impressive prospects. Séguinot (1988: 106) makes it clear that

Empirical studies can provide evidence about how people translate, can help us understand how languages are stored and accessed in the brain, and provide clues as to how meaning is received and encoded in language and transferred between languages.

But again, the author looks into some research done in the area drawing attention to “general problems with the interpretation of… empirical studies” (p. 107, my emphasis).

Nevertheless, interest in process-oriented research continues motivated by a feeling that analytical investigation into how language material is comprehended and reproduced into another tongue can illuminate our understanding of the activity. This growing interest, however, has continually shown that this orientation is largely psycholinguistically disposed. It is also felt that this area of investigation can be handled only through empirical analysis of actual translation performance. Lörscher (1992: 146) affirms that

only on the basis of empirical studies of translation performance using a process – analytical approach can hypotheses on what goes on in the translator’s head be formed.

The author, however, maintains that this “is largely unknown and uninvestigated” (Lörscher 1992: 146). Therefore, any enterprise that attempts to handle this orientation has to formulate suitable theoretical constructs and tools of investigation to be well qualified for the job.

Recently, the problem of a unified framework of investigation according to which translation process can be accessed and described has been brought to the forefront. Jääskeläinen (1996: 61) notices that “The first process-oriented research projects were started in isolation, independently of each other, therefore, the studies reflect different backgrounds.” The author admits that a variety of research interest is in a sense a healthy phenomenon. However,

The overall goals of [former] research have made it more difficult to test the methods employed in previous studies[…] Although the problem is to some extent understandable in a field that is still in its infancy, not enough attention has been paid to testing and refining the methodology.

Jääskeläinen 1996: 61

Clearly, with the absence of clear and well-defined methodology of investigation, directing translation studies towards a process view may, perhaps, remain wishful thinking. It is in the light of the above needs that the present work intends to contribute. It is a response to the growing interest in the study of translation as a process. It attempts to propose a methodology that could make this kind of study a manageable business.

2. A Theoretical Framework of Investigation

An early question we have to answer before attempting to formulate a descriptive and interpretive framework of investigation is: What tools to employ in the formulation of such a framework? Alternatively, in more general terms: Should translation theory always develop its own tools of investigation or rely on the theoretical constructs of related disciplines? Indeed, the domain of translation is language, and a state of interdependence already exists between translation theory and other linguistic studies. Catford (1965), for example, bases his approach to translation on Halliday’s (1961) Systemic Grammar. Moreover, in the field of foreign language (FL) learning, translation is used as a procedure for eliciting learners’ linguistic abilities (cf. Oller 1979: 50). It is, therefore, difficult to see how any attempted development in translation studies can proceed without drawing heavily on a theory of linguistics.

James (1980: 4) places translation theory in interlingual linguistics, which belongs to applied linguistics. Interlingual studies is an area of research concerned with languages in contact (cf. Jakobson 1959: 233). According to James, in this area of research “Although the point of departure is the two languages concerned (…SL or ‘Source Language’ and TL ‘target language’ in the case of translation), the focus of attention is on the intermediate space between the two.” James endorses Mel’uk (1963) who points out that the ‘language’ that evolves along this ‘space’ is called an ‘interlingua.’ The concept of interlingua seems to be of interest to a processual study of translation as it is closely related to the process of target text (TT) development. We intend to expand on this concept as we proceed to formulate a framework for a process-based study of translation.

3. The Interlingua

One area of research that makes use of this concept is machine translation (MT). The way it is used there can perhaps tell us something about it. According to Mel’uk (1963: 61) the concept is introduced at the early stages of the development of MT where the process is divided into two major stages: analysis and synthesis. In the first stage, specific data is extracted from the ST. It includes language information about the translation of words, their morphological forms, how they are connected in sentences, etc. These are the analysis features of the SL. They are accommodated into the TL through certain characteristic features either specific to it or which it shares with the SL. These are called the synthesis features of the TL.

Because the amount of language-information units that the machine has to deal with would be very huge across the two languages, an “intermediate language” is sought in the most economical way for convenience of operation. This entity is called an interlingua. According to Mel’chuk, an interlingua, therefore, is an entity “created by determination of the correspondences among natural languages” (p. 67) and an “‘intermediary’ (transitional) language” along the way between the source text (ST) and TT (p. 64).

This concept is still a recurrent theme in MT. While discussing computer-assisted translation, Bühler (1990: 33) for example, speaks about interlingua as an “Intermediary phase… from which a new target text is created.”

We intend to make use of the concept to postulate the existence of a resemblant entity that develops during the process of human translation. This postulation is justified by the existence of the transitional formulations and reformulations translators make while at work. These formulations seem to be the surface substance of a developmental process enacted in the translator’s mind. During this process, attempts are continually made to deconstruct the ST into its basic units and relay them into a different language usually intended for a different culture. The postulated entity seems to be evolving along this process.

We are about to formulate a proposition representing the central thesis in this section. But before doing this, mention should be made of the closely related concept of “bi-text.” Introducing this concept, Harris (1988: 8) states that we usually think of the ST and TT as two separate entities. But as far as the translator is concerned, they are “simultaneously present and intimately interconnected in his or her mind.” They “co-exist” there during the process of translation in the form of a “bi-text.”

This psychologically motivated concept seems to further support our postulation of a transitional entity that develops during the process of translation. It puts us in a better position to make our first proposition on the way to formulate a framework for a processual study of translation.

Proposition No. 1: When translation is attempted, a growing linguistic entity comes into being along a process of re-creation stretching from the ST to the TT.

4. Continuity of the Process

It is perhaps important to point out here that this developmental process of translation creation (considered from the point of view of the translator) is generally non-terminable. It may lose momentum but often does not terminate satisfactorily. This is reflected by statements like translation “is never finished” (Newmark 1983: 1), and translation is a “futile pursuit of a non existent perfection” (Weaver 1989: 124).

The dynamic nature of translation process seems to be closely related to the unstable nature of the translator’s vision during task performance, i.e., the way he understands ST information and the way he relays it into a TT. In a translation task, the translator has a ST to read and, more often than not, re-read to extract information from. But “each reading of a text is a unique act…, a process subject to the particular contextual constraints of the occasion…” (Hatim and Mason 1990: 224). So, one aspect of the instability of the translator’s vision lies in the fact that there is usually new information each time the ST is considered.

The translator often settles on a certain type of ST information. But when he moves to the job of relaying this information in a TL, a developmental process is initiated where another aspect of instability could be exhibited. During this process, TL forms start to evolve but always subject to change and refinement as the translator continually refers to the original to ensure the best possible degree of accordance with ST elements. So, there seems to be a kind of ebb and flow of information going on in the translator’s mind as he moves to and fro between the ST and its dependent newly emerging TT. This kind of developmental instability is often reflected in translation production in the form of false starts, omissions, corrections and the like.

The process of TT development sometimes continues even after the translator has decided that it is over. Weaver (1989: 123-124) expresses this clearly when he states:

[…] My “fair” copies are never completely free of x’d-out words and penciled-in emendations; and even on the proofs – braving the publisher’s reproaches – I make a few, last minute changes. Once a translation of mine is published, I never re-read it. I know that, if I did, I would soon be reaching for a pencil, to make further additions and subtractions.

In the light of this argument, the expression final version seems to be a misnomer. It has to be redefined or, perhaps, modified, as translation is not always whole-heartedly finalized. This is particularly true in non-scientific texts where absolute SL-TL equivalence is not a usual phenomenon. The expression final draft, therefore, seems to suit a translation better, if by this we mean it remains a draft. (This is why the word ‘final’ in expressions like ‘final version’ is put between inverted commas in this work).

Translation production, therefore, seems to be governed by the law of diminishing returns in economics. The effort being made at improving a TT could be stopped beyond a certain point when it ceases to produce proportionate results.

It is time now to consider another issue on the way of formulating a framework for a study of the process of translation. We shall also be considering the fact that translators’ successive formulations are transitional, but this time in a different sense of the word.

5. Translation Process and the Status of the Pair Languages

What is meant by status here is whether either of the pair languages is the translator’s native language (NL) or FL, an issue often neglected by translation theory assuming that the translator is equally competent at both languages, i.e., an ideal bilingual. If such bilinguals are the norm, then this assumption is justified and consequently translation process is not expected to be affected by uneven linguistic ability in the pair languages. However, James (1980: 51) seems to be skeptical of the idea of ideal bilingual:

To be a balanced bilingual is to have solved the problems of L1 [first language]: L2 [second language] mismatch and of the dominance of one of these languages over the other […]

The fact that the translator is usually a non-native speaker of at least one of the languages (and in a sense a learner of that language) may cast doubt on this ideal assumption of equal linguistic ability. Being a learner of the FL, the translator is not expected to be as competent at it as he should be at his NL. This is because he is, generally, not expected to have a native-speaker ability in the FL. Strevens (1977: 47) supports this claim when he points out that “the aim of achieving native-speaker ability in the foreign language is usually a myth, a platonic ideal at best…”

Therefore, we assume that translating from or into a FL contributes to make the translation process as indecisive as it looked in the previous section. We shall expand on this making our second move in the direction of formulating a framework for studying translation process but without losing sight of translation product.

Translating from a FL is generally expected to affect translator’s internalization of the ST and mark the process with signs of uncertainty about ST elements and in certain acute cases to bring about misrepresentations of these elements.

Of course, one might come across a case where the translator is better at his FL than at his NL. However, such a case would not be the norm, knowing that in translating we are not only concerned with comprehending a ST as a surface substance tied up by certain formal rules. A translator would not only need this to be able to interact with a ST. We are also concerned with his mental representation of the ST as a language event embedded in a certain culture and produced in accordance with certain socio-cultural conventions.

We shall say no more on ST comprehension here. It is a subject that belongs to psycholinguistics (cf. Gleason and Ratner 1993 for example). Expanding on it would take us beyond the scope of this paper. We are moving now to the issue of translating into a FL where the impact of the translator’s FL is clearer since this language surfaces in the form of printed material.

6. Translating into a FL

We have alluded in the previous section to the fact that a FL user is in a sense a learner of that language. In this sense, therefore, a translator translating into a FL is a learner of that language and consequently he uses “language-learner language” (Coder: 1978). An earlier term that has been proposed to describe this kind of ‘language’ is the one introduced by Selinker (1969) namely, the Interlanguage (IL) Hypothesis.

Selinker (1972: 214) identifies IL as “a separate linguistic system based on the observable output which results from a learner’s attempted production of a TL norm.” Clearly, this points to the postulation of a linguistic entity that emerges during a FL learning process. A “system” is governed by its own rules and, therefore, its features tend to recur in the output of a learner or a group of learners generally sharing the same learning situation.

Selinker (1972: 214) mentions five L2 learning processes central to the IL Hypothesis, all or parts of which may contribute to bring about IL. As we are only interested in their implications on translating into a FL, we shall only mention them. (The interested reader is referred to the original works and to Tarone (1980) and James (1980) for an extensive coverage of the issue). Selinker’s five processes are language transfer, transfer of training, strategies of second language learning, strategies of second language communication, and overgeneralization.

Selinker (1972) mentions other minor processes like hypercorrection, an exaggerated attempt to correct one’s own deviation, which may lead to another kind of deviations, and many other processes responsible for incorrect pronunciation. According to Selinker, the above processes shape IL and mark it with features incongruous with L2. But this, of course, does not mean that a learner’s IL is only a collection of deviations. It may include parts which are isomorphous with the L2, i.e., those parts that the learner has learned successfully.

A feature often associated with IL is the linguistic “fossilization” phenomenon: “The persistent failure of the majority of adult learners to achieve complete mastery of a second language…” (Selinker and Lamendella 1979: 363). The idea is that as L2 learning proceeds; learners fail to adjust parts of their IL to correct language use. Such parts remain stubborn inadequate linguistic behaviour.

Having looked at Selinker’s pertinent views on IL, it is perhaps important to stress that if we accept the IL Hypothesis (which has continually influenced linguistic research since its introduction) and the idea (to which we alluded earlier) that a translator translating into a FL is an IL user, then the above processes are expected to affect this kind of translating. This kind of argument poses a fundamental question that translation theory has to answer: Would the above types of processes influence IL used in a translation task the same way they influence IL used in a non-translation task (as in free composition writing for example)? Would a translator, for instance, tend to avoid a topic (strategies of L2 communication above) in a translation task when he does not have the linguistic means to express it?

Corder (1978) endorses Selinker’s conceptualization of the concept of IL, but prefers to think of it as being extended along more than one type of continua. To him Selinker visualizes the learner as being continually engaged in restructuring and readjusting his IL to fit in TL norms. In this case, IL, in Corder’s words, develops along a “restructuring continuum” (p.75). This type of continuum accommodates the fact that IL shows traces of L1 interference.

But to Corder, IL also appears along a “developmental continuum.” Although restructuring may also imply development, what is emphasized in the second concept, however, is that IL starts with simple TL forms and systems that develop as more language is acquired, i.e., like a child learning his NL.

Whether we accept the first type of continuum or both to understand IL, one thing seems to be always present. The ‘language’ FL learners use is a transitional entity that develops along an intermediate ‘space’ between the NL and FL. We are now, probably, in a good position to present a proposition representing the main argument pertinent to the manipulation of a FL in translation:

Proposition No. 2: The translator is not expected to manipulate a non-NL (receptively or productively) always in a native like manner. He will be using a transitional language, parts of which usually undergo development.

It is important to notice here that we are talking about translators’ ability in pure linguistic veins, while in our first proposition we were intrinsically concerned with translation as a skill. Truly, this separation is arbitrary in a sense, since translation subsumes language use, the two are naturally simultaneous. However, while considering the highly complex subject of translation process, the two can be kept apart. After all, translation strategies are not exactly the same as linguistic strategies.

7. The Impact of the SL and Culture on the TL and Culture

So far, we have been considering translation process by focusing on the transitional ‘space’ between the ST and TT. It is time to get nearer to the TT by considering the effect the ST may exert on the TT during translating, and thus conform to our objective of accounting for translation process but not losing sight of what this process may bring about. This will be our third and final move in the direction of formulating a framework for a study of translation process.

It seems axiomatic to say that ST writers draw on the linguistic resources of the SL while producing texts. They may follow the usually accepted norms or choose to be creative. Likewise, translators make ample use of the TL system when they re-create a ST into a TT, either following the usual track of conservatism or choosing to be creative. Nothing seems unusual about these two simple facts.

A great deal of sophistication, however, can be introduced into this simplicity when we acknowledge the fact that a TT is ST-bound and that TL selections are not independent of SL restraints. After all, a translation is possible only when ST forms and meanings lend themselves to the TT. We believe that these forms and meanings may leave imprints of various quantitative and qualitative natures on their dependent new medium.

Linguists of different tongues have acknowledged this fact. The Prague School linguist Renský (1972: 228) makes it clear that “even good translations reflect (to a certain extent at least) the linguistic structures of their originals.” James (1980: 117) reflects the same sentiment when he states that “Since the translator must be give access to the original, there is no way of preventing him from transforming features of its texture onto his TL rendering.”

One major form of SL pressure is embodied in the conflict between ST meanings and TL forms. Meaning is basically the content of the language event and form is the linguistic vehicle that carries this content. What seems to happen in translating is that the translator extracts the content out of its ST forms. This content is supposed to be rendered (as unchanged as one can) in TL forms.

The relationship between content and form is natural in the original, “they form a certain unity, like a fruit and its skin…” (Benjamin 1992: 76). The problem starts to evolve when the translator notices that it is not always so in the TT, the content being alien to the new forms in many respects. The new forms may empty it of some important characteristics and recharge it with new ideological and attitudinal elements. Of course, there is always the possibility of having a content common to both SL and TL forms due to the common core of human experience. Nevertheless, this is not a common practice in translation.

Content, therefore, tends to pass over into the TL along with some of its original formal contours. Translators usually try to resist this pressure but SL contours may infiltrate creating a medium of incongruous elements representing both SL and TL.

It is perhaps important to emphasize that the emergence of linguistically incongruous elements in translation is not liable to occur only in the case of translating into a FL as one may think, encouraged by the fact that a translator’s FL competence is usually weaker than his NL competence. In fact, such elements could also be present when the translator is translating into his NL as well. Investigating the TTs of a group of translators who translated from English into their native German, Wilss (1982: 206) notices that some of them produced “something which they definitely would not have done in a strictly monolingual communication situation….” This, undoubtedly, stands to reason, because in a monolingual situation there would be no background language (i.e., SL) to project itself on TL production. In a translating situation, however, “the SL textual segment imprints itself on the translator’s native language competence which fails to fulfill its natural function…” (Wilss 1982: 207). It is this projection of one language on the other that marks TTs with incongruous features, and this may well happen in both NL FL and FL NL directions.

But if we accept the hypothesis that stronger linguistic habits (i.e., those of the NL) usually affect weaker ones (i.e., those of the FL, cf. James 1971: 61), then interlingual effect in the direction NL → FL will be clearer than in the opposite direction. This is, perhaps, why linguistic science has generally occupied itself more with NL FL interference than with FL → NL interference:

L2 / L1 error analysis has so far been less dominant in applied linguistics than L1/L2 error analysis. This is attributable to the fact that L2/L1 error analysis is a young subdiscipline of the science of translation.

Wilss 1982: 196

SL → TL infiltration should not be thought of only as violation of TL norms like the formation of strange grammatical structures. They could be SL textual conventions like those of religious Latin and Greek introduced through translation into modern European languages. They could also be scientific and technical terms of the language of developed parts of the world introduced into the language of less developed parts. In cases like these, what seems to be at the outset new and unexpected verbal means would be accommodated into the TL and later become part of its linguistic habits.

Neubert (1990) considers the issue not at the level of individual translations but on a larger scale where in particular fields (say journalism) translations consistently bring innovations. The writer affirms that in cases like these “one can speak of a translational discourse integrated into the ‘normal’ (non-translated) discursive practice of native TL users” (p.97).

This is also significant in literary translations where translational hybridization (when considered over long periods of time) becomes not a case of violating TL systems but a process of enriching them and, thus, a life giving force to languages. De Beaugrand (1978: 27-8) supports this when he states that “The introduction [through translations] of a foreign literary work has often contributed to new developments in the literary conventions of a culture….”

This, however, is not to deny the fact that there are always nationalist views that favour language purity and regard this carry-over of linguistic means as a process of corrupting a language. No matter what stand we take against such infiltration, the phenomenon is likely to happen as long as there are attempts to translate. Indeed, it sometimes preferably happens as when the translator deliberately keeps a SL item in his TT (see below). Translation theory, we feel, has to accept this and develop systematic means for a descriptive analysis of this kind of carry-over. If it restricts itself to theories of idealizing translation, it will overlook important facts.

8. The Socio-Cultural Dimension

The above kind of SL → TL cross-breeding can also be viewed as a carry-over of values, attitudes, etc. specific to the SL culture. This happens, for example, in the translation of culture specific terms (e.g. Jazz, the music of US Negro origin). Two common solutions are to borrow the SL term or introduce it with some explanation (Fraser 1996: 89). The important thing to notice here is that the SL term does not simply carry conceptual meaning, but along with it additional socio-cultural values that may leave imprints on the TL culture.

The SL, TL channelling of customs, mental achievements, moral values and the like can take many other forms and we do not propose to tabulate them all here, as we are chiefly interested in the possibility of their existence. A few more examples, however, may illustrate this possibility.

When translated, an ideology present in the ST could alert the TL reader to a human conduct peculiar to the SL culture. Translating crime and violence stories, for example, could affect the behaviour of young TL readers. Another example is the introduction of SL items in translated cooking recipes. Such items are often introduced to preserve a local flavour, to maintain the jargon of the job or simply because their equivalents do not exist. They often bring with them new interests and could initiate social and sometimes industrial concerns.

Many other SL political, artistic and technical terms may find their ways into translations simply because the concepts or products they denote are not well established in the TL culture, or because of a certain accepted ideology. According to Newmark, (1981: 82) French diplomatic terms are preserved in translations because of the “supremacy of French in diplomatic language.”

Of course, there is always the possibility of having the cultural imprints resulting from this carry-over naturalized in the TL to become part of its repertoire. But this often happens over long periods of time, and the fact remains that such novel features are often regarded, at least at the early stages of their introduction, as infiltrators of the socio-cultural sphere of the TL. Such new comers, like those discussed in the previous section, bring new colour to the medium where they start to habitate. This medium is no more the original pure one. Duff (1981: 10) draws this neatly:

The translator who imposes the concepts of one language on to another is no longer moving freely from one world into another but instead creating a third world and a third language.

Now we can present our third proposition, but after we summon the previous two:

Proposition No. 1: When translation is attempted, a growing linguistic entity comesinto being along a process of re-creation stretching from the ST to the TT.

Proposition No. 2: The translator is not expected to manipulate a non-NL (receptively or productively) always in a native like manner. He will be using a transitional language, parts of which usually undergo development.

Proposition No. 3: The language employed in a translation can be an entity banking on the SL and TL and their cultures.

9. Formulation of the Hypothesis

So far, we have worked out three propositions. No claim is being made here that these propositions are new discoveries. They are simply statements that may come out after an examination of the very nature of translation activity. For our own purpose, what is important is that each proposition can account for a different facet of translation process. Still more important is that it seems arbitrary in a sense to keep these facets apart, since a great deal of overlap exists between them during the process.

Therefore, we intend to superimpose the three propositions one on top of the other and propose the Translanguage (TrL) Hypothesis, which can encapsulate the three of them. The proposed hypothesis can do that by virtue of three senses, which the trans-part of the term has. Each sense corresponds to a different proposition and thus highlights a different facet of the hypothesis.

First, trans- can be used as a contracted form of the word transitional, and thus the first facet of the term TrL indicates a transitional ‘language,’ pointing to the successive formulations translators make during translation process. This corresponds to proposition No.1.

Second, as translating usually involves the use of a non-NL (receptively or productively), a transitional language will be in action, but now in an additional different sense, that of ‘language-learner language.’ This is the second facet of the term and it corresponds to proposition No.2.

Third, trans- can be used as a contracted form of the word translation. The final facet of the term TrL, therefore, highlights the emergence of a hybrid entity we are calling translation‘language.’ This corresponds to proposition No.3.

So, given the arguments that have helped to work out the above three propositions, TrL is a transitionally unstable linguistic entity that evolves during acts of translation along intersecting stages in a ‘trip’ stretching from the ST towards the TT during which hybrid ‘language’ comes into being banking on the linguistic and social potentials of the SL and TL.

10. Application of the Hypothesis in a Processual Study

It has to be recalled here that we have originally set ourselves the goal of working out a framework for the practical investigation of translation process. The TrL Hypothesis, we believe, can provide such a framework in what may be called translingual investigation.

Being process directed such an investigation has to draw on a special kind of data. It should have access to three sources of information:

  1. The formulations translators make while at work. These include their alternative formulations, corrections and the like, i.e. the physical process.

  2. The mental processes that have generated the physical process. Access to such information can be attempted at by eliciting translators’ judgments and intuitions about their formulations. Such mentalistic elicitation can also be useful in a different respect, detecting accuracy of ST comprehension.

  3. What the translator has ‘finally’ settled on, i.e. his ‘final’ version or what is generally considered a TT.

The first and third types of data can be called textual data while the second is mentalisitc (cf. Corder’s (1973) distinction between ‘textual’ and ‘intuitive’ data).

In a translingual investigation we start with textual data and support our investigation by insights arrived at by examining mentalistic elicitation. The whole corpus of translational data is accepted and dispersed into three modes. Each represents a facet of the TrL Hypothesis, just as in the analysis of light where the transparent prism separates white light into a spectrum of colours. This can be shown diagrammatically as follows:

Figure 1

Translingual analysis of translational data

Translingual analysis of translational data

(On the idea of dispersing data in this way, we are inspired by Firth; 1957: 192, who proposes to disperse meaning into modes.)

-> See the list of figures

Facet 1 tells about the intersecting and developmental stages of the process. It addresses itself to issues like translation strategies employed during the process of generation. Facet 2 opens a window on the manipulation of FL. It is concerned with translators’ transitional ability to handle the FL receptively or productively. The last facet makes use of relevant information gained up to this point and tries to see what kind of translation ‘language’ the TT employs.

We make no apology for repeating that these facets are not watertight compartments and, therefore, this separation is arbitrary in a sense. Nevertheless, it is maintained in a translingual analysis so that one can concentrate on one facet of process at a time, a process that seems to be shrouded in mystery and complexity.

11. Conclusion

The theory of observing translation performance has to be guided by some kind of a metatheory. An examination of the very nature of translation acitivity may point to a framework of investigation that could guide performance observation on three interlated facets: (a) the development of interim linguistic constructions into a translation, (b) the interplay between the pair languages during the development, and (c) the type of language that the course of performance produces.

Approaching translation activity within a pre-set metatheory can ensure a clear view of objectives and, consequently, precision of judgements. It can also allow for future examination and refinement of the adopted methodology of investigation.