Interpreting and Translation – Two Sides of the Same Coin?

This paper discusses an umbrella project which compares the end results of interpreting and translation in order to chart similarities and dissimilarities between the two crafts. 1 The article places the project in context, presents its objectives and describes the corpus of the empirical data for specific studies which will, in turn, allow for assessement of differences and similarities between interpreting and translation. The third study focusses on establishing whether translation and interpreting are basically identical or completely different tasks. It therefore intends (a) to look at the characteristics of the two processes, their differences and similarities, in order to identify those factors in the situational context which influence the work of the translator/interpreter and thus also the end-product; (b) to study the tools used by professionals for their rendition of the message of the source text, with spe-cial focus on information loss and the appurtenant compensation strategies applied (consciously or unconsciously) by the translator/in-terpreter: their nature, frequency and – whenever possible – the reason for their application. 7


Types of interlingual transfers
The difference between translation and interpreting may not be known to all readers, and accordingly a few words of definition are in place. On the other hand, a mapping of all types of interlingual transmission falls beyond the scope of the present article, so it will suffice to relate 'our' two modes to two others which are highly visible in Denmark.
There is summarising. Summarising is a brief rendition ('the gist') of messages from the source language to a target-language audience. We suggest that, globally speaking, this is the most frequent type of interlingual communication: think of the information from and about other countries which floods target languages and cultures in newspapers, on TV, in the radio, and in books.
There is subtitling. As defined by Gottlieb (e.g. 1992), subtitling is the collateral transfer to the written medium into another language than the one audibly spoken. It involves summarising, although not as much as ordinarily believed. In small countries and minor language communities (like Denmark) many television programmes are of foreign provenance, and accordingly subtitling is certainly in the public eye.
Then there is translation. For the purposes of the present paper we shall only define it briefly as a transfer from one language to another in a written form.
Interpreting is, roughly speaking, an oral rendition in the target language of an oral message in the source language.
Our study will focus on translation and interpreting. Therefore, it must also be mentioned that the latter comprises various forms, the most important types being (a) consecutive interpreting, in which the interlingual transmission takes place after the 'whole' text has been delivered in the source language, and (b) simultaneous interpreting, in which the target language rendition takes place at the same time as the delivery of the 'original message' in the source language.

Basic assumptions
In our intitial work in the study, we operated on two basic hypotheses: (a) that there are similarities between interpreting and translation, and (b) there are also differences between interpreting and translation.
We believe that the similarities tie up with the general nature of the two tasks: both are transfers of messages from one language to another. Therefore, there is reason for believing that the mental exercise performed by mediators of the two types of transfer is, by and large, similar.
Conversely, the most obvious differences between interpreting and translation are relatively well-defined (see e.g. Lenstrup / Zoëga 1987;Padilla / Martin, 1992). They connect with the mediators' communicative context, the accessability of the texts transferred, and the temporal aspect, which factors we shall now briefly address:

The interlingual mediator's integration in the communicative context
In translation, the performances of sender, translator, and receiver take place in physical separation, i.e. the parties are not at the same physical location at the same time: in other words, first the sender (author) prepares the written text, which is then (possibly by mediators ('initiators')) sent to the translator for translation. The tasks involved (writing in the source language (encoding), translating (decoding and encoding), and reading in the target language (decoding)), are performed separately and over a prolonged period of time (days, weeks, months, years). Once translated, the text is read by the receiver. In principle at least, translators are thus neither aided nor disturbed in their performance of transfer tasks by immediate extralinguistic factors in the situational context.
In interpreting, conversely, all participants are integrated in the communicative situation: speaker, interpreter and listener are present at the same time and, usually, in the same room; the speaker's delivery, the interpreter's rendition and the listener's reception of the message, take place within a limited time span: a few, brief seconds. The intepreters' physical integration in the communicative context implies that their performance of the transfer task may be influenced, positively or negatively, by factors in the immediate communicative context (the speaker's and listener's body language, the speaker's tone of voice, the use of overhead transparencies, noise in the room, etc.).

The accessibility of texts
Interlingual communication always involves two texts, 'two versions of the same message': the source text and the target text.
In translation, both texts are written and are therefore (like most written texts) 'planned'. As pointed out by Stubbs (1988: 109), this means that they are usually characterised by e.g. high information density, few repetitions, little redundancy, and few grammatical errors, which is due to relatively careful drafting (possibly even redrafting). Written texts are also usually well-organised, and the layout, the punctuation, and the orthography (underlining, capitalisation, etc.) serve to drive home the message.
Furthermore, the written text is permanent: in case the translator fails to understand something on her first reading, she can read it again.
Interpreters work with spoken texts and with what Ochs (1979: 51-78) aptly terms 'planned' and 'unplanned' discourse: Planned discourse (discourse that has been thought out and organised prior to its expression) is, for instance, reading out of manuscripts, whereas unplanned discourse (discourse that lacks forethought and organisational preparation) is represented in speeches delivered off the cuff, spontaneous conversation, discussions, questions and answers, etc. It will be appreciated that planned spoken discourse bears much similarity to written discourse, while unplanned spoken discourse relies less on style and syntax and more on context and nonverbal means for conveying the message.

Temporal aspects
In principle, translators can interrupt work on a translation any time and put it off until sometime later, for instance, because they must answer a telephone call, don't feel like working, or because they want to do terminological or background research on the text in hand.
Interpreters never have this option: they are required to do a 'hereand-now' performance. It goes without saying that the quality of an interpreting job depends on the interpreter's language and professional skills. Yet few non-practitioners realise how closely interpreters' performance is dependent upon the sender: the better the speaker's delivery, the better the interpreter's rendition. If, for instance, the speaker speaks too slowly, the interpreter's task becomes more difficult since it takes more time before the interpreter identifies and analyzes the line of argument. When the speaker's delivery is fast, there is no real difficulty in interpreting unplanned discourse. But with planned discourse, the interpreter will have problems because of the syntactical and informational density.

Compensation
We are still only in the intial stages of our project: in terms of our final analyses, the above points are therefore only hypotheses. On the other hand, they fit in with the reality of translation and interpreting, so they should not be written off lightly.
If we look at the common denominators at a higher level, it appears as if we can tentatively suggest that the general similarities between the two modes lie in their claim of producing (approximations to) 'identical texts' in two language systems. Differences connect with constraints imposed by the specificity of the texts and the transfer situations, by the communicational contexts.
This leads to the hypothesis that the differences also connect with the compensatory techniques which mediators resort to in given contexts: unlike translators, interpreters cannot brood over their 'texts', they cannot check carefully on unknown terms, they cannot scrutinize the sourcelanguage text, nor are they at leisure to improve the product once it is finished. However, interpreters compensate for this by having a fairly specific idea about the topic discussed, by checking terminology and by going over relevant documents before meetings, and by decoding and encoding the message in its situational context, where there is an instan-taneous control since all communication parties -sender, transmitter and receiver -are present and will spot errors and irregularities right away.
In terms of time, there is anticipation and instantaneity in the work of interpreters whereas the main characteristics of translation are retroactivity and passage of time.

The corpus
Other factors, such as orality vs literateness (cf. Shlesinger 1990) may be fruitfully investigated with a large corpus of interpreting and translation. But, like the above observations, they must be tested against the corpus we intend to set up.
The corpus we are setting up must meet the following criteria: 1. In the main, the corpus will be limited to two languages: Danish and English. We realise that this is a severe limitation to the immediate general applicability of our findings. However, research must start somewhere, and our familiarity with both languages should, conversely, permit us to draw valid conclusions. Furthermore, we believe that the corpus may even stand to benefit from a modest beginning, insofar as it will enable us to improve the quality of the corpus in case the data collected and our methods for data collection turn out not to be sufficiently comprehensive. It may be, for instance, that studies of interpreting must chart more situational factors than one should think of at first glance.
2. The data are authentic. The texts are produced by would-be translators and interpreters in recognised and institutionalized translator and interpreter programmes and by professional practitioners for actual use.
3. The corpus must comprise various types of messages. It is unrealistic to believe that we shall ever attain fully global representation of all text types, but the awareness that onesidedness should be avoided is important.

The data
We have collected the data, and -as planned -we have got source as well as target language texts at three levels of proficiency: student, trainee, and professional level.

Student level
Data at this level derives from students whose proficiency is sufficient for them to be taken officially serious as would-be practitioners. As far as student translators are concerned, this permits us to be precise: student translators must be English majors who have passed the first exam in which the craft of translation or interpreting is tested, i.e. the Danish '1.-del' exam (after 2 or 3 three years of study) at either University or the Copenhagen Business School. 2 There is no corresponding formal exam for student interpreters. Yet there is an admission test for the Brussels 'stage' as well as at the Business School conference interpreting programme which checks applicants' potential for interpreting. Accordingly, we have decided that performances from the first weeks of these six-month courses are comparable to those of the translation students.

Trainee level
Trainees are students at an advanced academic level. The translator trainees are students at '2. del', preferably towards the end of their university careers. The corresponding interpreter trainees are in the latter half of the conference interpreting programme.

Professional level
These informants are practitioners who have worked professionally for at least one year. 49 2 There are formal translation programmes at the Danish Business Schools in Aarhus and in Copenhagen, and at the Center for Translation Studies and Lexicography, University of Copenhagen. Our data do not derive exclusively from these institutions, because many professional translators in Denmark translate from languages in which they have no formal training. The point about insisting on a '1.del pass' is that once this exam has been passed, graduates can find employmentas translators or teachers using translation. By international standards foreign language proficiency is high in Denmark in the secondary school system and this, of course, affects teaching at University. Furthermore, foreign language majors (and minors) in Denmark are devoted exclusively to foreign language culture-and language-specific courses.

Corpus management
It is our intention is to make all source and target texts from both transmission types machine-readable. In the long perspective, we plan to make the material accessible to interested scholarly parties.
The data from the translation groups are handwritten and then typed out, or typed texts; 3 they will be transferred to computers either manually or by means of optical scanners.
Data from the interpreting group are taperecorded and subsequently to be transferred into a computer. In transcribing these data, we may well have to establish markers for e.g. intonation, pitch, stress, pausing, variation in volume, etc.

Status
At the time of writing, status for the data is as follows:

Beginners' level:
Translations: collected, most of it typed, but not machine-readable. Consecutive interpreting: collected and taped. Not typed, not machinereadable. Simultaneous interpreting: not practised at this level.

Trainee level:
Translation: half of the material collected. All of it typed, but not machine-readable. Consecutive interpreting: collection in progress Simultaneous interpreting: collection in progress Professional level: Translation: collection in progress. These data are machine-readable. Consecutive: We originally planned to get some material here, but consecutive interpreting is rarely practiced at meetings we would have access to. Common sense dictates that we must abandon the idea: in order to get truly authentic material of this type we must get the prior 50 acceptance of all three parties and, hence, their tacit understanding that any conflicts may be taped, typed out, and even printed for posterity to see. Simultaneous: the material has been taped. 4

Studies
At the beginning of this article we had occasion to mention how, once complete, the corpus will provide a basis for various studies. Three specific studies are already in the offing: The first study centres on translation, notably on errors and strategies leading to success (or error) in translation, using beginners' level as the point of departure and subsequently analyzing higher levels of proficiency. It will also attempt to set up a taxonomy of 'right' and 'wrong' ('adequate' vs 'fully understandable reversion') in relation to their import on the message in the target language. This implies that the results may be used for assessing the general characteristics of bad versus good translations as well as progression in translational competence. 5 The second study plans to examine the frequency and character of word-class substitution in translation and interpreting from Danish into English. It is generally accepted that in order to achieve adequacy both in translation and interpreting, translators and interpreters must develop strategies which permit omission or substitution of the source text word-class for other word-classes in the target language. It is the purpose of the study to establish to what extent word-class substitution is effected for communicative purposes and/or esthetic euphonic purposes (idiomatics). The hypothesis of the study is that as the translator's (interpreter's) professionalism increases, there is a marked tendency to disregard the source text choice of word-class and to make for better idiomatic appropriateness and euphony in the target language version. This hypothesis can be confirmed or falsified by means of material which reflects advance in routine and skill.The results of the study can, hopefully, be used in the training of translators and interpreters to help them implement well-considered word-class substitution as a strategy towards achieving adequacy of substance as well as form. 6 The third study focusses on establishing whether translation and interpreting are basically identical or completely different tasks. It therefore intends (a) to look at the characteristics of the two processes, their differences and similarities, in order to identify those factors in the situational context which influence the work of the translator/interpreter and thus also the end-product; (b) to study the tools used by professionals for their rendition of the message of the source text, with special focus on information loss and the appurtenant compensation strategies applied (consciously or unconsciously) by the translator/interpreter: their nature, frequency and -whenever possible -the reason for their application. 7 Once completed, each of the studies will check its findings against the other studies as well as data from the other types of linguistic transmission in the corpus. For this is the only way in which we can, meaningfully, establish similarities and differences between interpreting and translation.
Firstly, it is our experience that introspection is time-consuming and demands masses of data before assumptions about specific features can be made -and it takes even longer to set up hypotheses about generalities (cf. e.g. Dollerup 1991).
Secondly, we believe that as teachers and language experts we are probably better than beginners at setting up hypotheses about why a translation or an interpreting performance is good or bad.
We have argued elsewhere that this is a sound enough basis for hypotheses (Dollerup 1982): Whenever we come across something good (or bad) in a translated or interpreted message, we set up a hypothesis about its cause. When we meet with another good or bad transfer, the hypothesis previously set up may be confirmed, falsified or supplemented by a new one. Other phenomenon may be explained by the same model and, if so, the hypothesis gains strength. If, conversely, the model does not serve to explain any other feature in the corpus texts, it is unsubstantiated and should (usually) be considered null and void.