THE ROLE OF TRANSLATION IN TESTING ENGLISH FOR ACADEMIC PURPOSES

by GIANFRANCO PORCELLI (Università Cattolica di Milano)

In the light of some new, recent orientations in foreign language teaching (see, for example, Widdowson 1978: 158 ff.; Widdowson 1979, chapter 8), it seems appropriate to have a look at translation to see whether the highly critical attitude against it that has prevailed over the last twenty years or so is still justified. As a summary of the structuralists' position on this topic, Lado's concise and forceful account is perhaps the most useful (Lado 1964: 158). Let me quote it in full:

As a test of ability in speaking, listening, reading and writing, we notice the following limitations in the use of translation:

1. The most proficient students do not translate when they use the language.

2. There are various ways to translate and to judge a translation: for artistic purposes, for accuracy of information, for grammatical exactness, or for vocabulary equivalence. A translation can be judged from these and other points of view. If the student is forced to translate for vocabulary or grammar, his literary appreciation may suffer.

3. The grading of translations tends to be unreliable because of the various ways to translate and the variations that the scorer may or may not allow.

4. Translation is a special skill different from speaking, listening, reading and writing.

5. Translation is slow as a test. Unless he has had special training, a good student takes longer to translate a letter than to write one. la the time that it takes him to translate a passage, he can cover more material using other techniques.

6. Translation is slow to grade, since the examiner has to weigh each response to see if it is allowable.

7. The use of translation in tests encourages the abuse of translation in the classroom.

Perhaps the only favorable things to be said for translation as a test of language proficiency are that translation questions are easy to set and are compact. The price paid for these is high.

Before debating each of these points, there are three important preliminary questions to answer:

a) What is translation, exactly?

b) What do we use translation (as a test) for?

c) What sort of learners (in terms of level, interests, motivations, goals, etc.) do we intend to test using translations? The three questions are closely interrelated, insofar as any object is also defined in terms of its use and field of application. However each of these aspects deserves specific attention. Defining translation perse is, of course, a formidable task (cf. Catford, 1965; Mounin, 1965); for our purposes it will be enough to distinguish between translation proper, from one language into another, and the various forms of "information transfer" (in Widdowson's terms), including those involving L1-L2 code-switching. Lado's remarks, for instance, definitely apply only to the former type of translation.

Even within translation proper there are a few important distinctions to make. First of all, translating from a foreign language into one's own mother tongue calls for specific abilities that are somewhat different from the ones that are required for the reverse process. Only that rare exception, the "balanced bilingual" would make no difference between the two, but if we think in terms of the students we normally teach, we realize we must keep these aspects quite distinct. And, of course, Pavese's translation of Joyce, or Quasimodo's translation of Shakespeare's sonnets, have very little in common with the routine, everyday translations of business letters in an office.

The second preliminary question is also important, because we can attempt to define the pros and cons of translations as language tests only if we attack the issue from the side of objectives. In language teaching, no testing activity exists without a specific purpose; and whatever translating actually takes place in everyday life, it responds to some specific need. In spite of all this, it seems to me that this side of the problem is often overlooked by" teachers. There is one deliberately ambiguous question I sometimes ask when the problem of translation arises during seminars or workshops on language testing: why translation? The answers I usually get almost invariably reveal that the participants have no difficulty in saying why they have recourse to translation, while they are often at a loss to say what they use it for. The general attitude seems to be that since translation is there (because the rules or Statutes require it, or just because it is the traditional way of assessing proficiency), you simply have to make the best of a bad job, both before (by trying to prepare students adequately) and after (by developing effective marking schemes and procedures). This is of course a most unfortunate state of affairs, and I am sure we all hope it will be soon thing of the past.

This indiscriminate approach to translation also makes it difficult - if not altogether impossible - to select those groups of students for whom translation is a valuable tool (both during and after university) and to whom, accordingly, it can be an examination with a satisfactory degree of validity. As has been remarked (Widdowson 1979) there are now ever wider areas of scientific language involving a 'superposed' knowledge of certain universal concepts and methods. The concepts constitute the grammatical deep structure and the methods the rhetorical deep structure of scientific discourse, whether this be superficially realized by Japanese, Russian, French, English, or any other language... Semantic and pragmatic translation can be used as a teaching device for learners who need the TL as an additional medium for scientific communication.

I am convinced that Widdowson's approach to scientific discourse can be extended to other areas of communication, and that each of us can easily find aspects in the teaching of EAP where translation is applicable both as a teaching and as a testing device. I shall say something else about this later on; the point I want to make here is that we can decide on the advisability of using translation as a test type only if we take into consideration the actual needs of our students. In some cases they can be not only acceptable but altogether necessary, and if this happens, Lado's objections can no longer be held to be valid.

In my opinion, his strongest point is the last, concerning possible and undesirable backwash effects'; no translation (especially into the second language) should be used when this may lead to unwanted grammar-translation procedures in the learning process. We must always be wary of this risk, and restrict the use of translations as tests to the very few instances where translation proper has its legitimate role as an object and/or as an instrument.

Speed, reliability, and scorability are no longer such crucial factors as they were a few years ago. Examiners have realized that speedy, compact, discrete-point tests can do only part of the work, while more complex and slow procedures have to be employed when they want to evaluate production (or even comprehension) as a whole - the so-called "integrated skills', So, although what Lado wrote about these aspects is still true, the importance of such issues is now being seriously questioned: the criticism against the technical aspects of translation appears far less important than the criticism of its general validity as a test.

This leads us to a fundamental question: what does translation actually test? Lado mentions artistic values, accuracy of information, grammatical exactness, vocabulary equivalence and "others", including a variety of things, from spelling and punctuation to circumlocutions that may be necessary to render certain idioms or 'untranslatable' words. The mere discussion of what equivalence is could take us very far (cf. Newmark 1981), so we had better confine ourselves to actual instances of translation as it is (much too often) still used in our universities. One recent paper included this sentence:

leri un mio amico mi ha detto che non fuma più da anni.

We might discuss the value and meaning of this sentence as an actual example of use (as opposed to usage, in Widdowson's terms): when, if ever, would an Italian presumably utter this sentence, and what for? If we think in terms of illocutionary acts and of the implications of a functional analysis of language, such questions are not trivial at all. The "correct" translation often depends on how we interpret the speaker's intention. Anyway, let us assume - for a moment - that such problems are not relevant, and let us analyse the sentence in terms of language usage. What does it intend to test?

The most immediate answer is, "verb tenses", and specifically the so-called 'duration form' in the second part of the sentence; the Italian passato prossimo corresponds to a simple past in English (ieri/yesterday being the key-word), and the presente is translated with a present perfect tense. If the examiner only wanted to test this, however, a blank-filling technique ('modified cloze procedure') could be more efficient:

"My friend (tell)............................ me yesterday that he (not smoke).......................... for years now."

Here the difference in usage between LI and L2 is not explicitly recalled but it is still there, as a possible source of interference. But the main point is that the sentence to be translated contains far more than just two problems related to verb tenses. Even if we leave spelling aside, and assume that such lexical items as friend, yesterday, tell, smoke, year pose no difficulty to our students, we still have to cope with word-order and levels of appropriacy. Is "my friend" an acceptable equivalent of un mio amico? Since we are considering an isolated sentence, we cannot be sure of the answer; generally speaking, it would probably be acceptable in most instances of use, whereas "a friend of mine" might sound pedantic or unnecessarily precise. There is no difference, in this respect, between Italian and English: when I say mia figlia I do not normally imply I have only one daughter.

If usage is emphasized, on the other hand, the "rules of the game" may require that the students provide the more accurate and directly related form "a friend of mine" or "one of my friends". This was the common situation with the grammar-translation approach; the student had to show that he was familiar with the rules of grammar concerning possessive pronouns. The acceptability of solutions in a translation paper, accordingly, is connected with the examiner's views on language and language learning, as well as with the objectives of the examination. As a consequence, there are as many possible criteria as there are teaching and testing situations, and it is very difficult to reach consensus on how a translation should be marked.

It is hardly necessary to point out that if paragraphs or passages are used instead of isolated, unconnected sentences, the problem of acceptability is not solved but merely shifted to a higher level.

A doubtful and weak tool as an achievement or proficiency test, a translation paper can be an efficient diagnostic instrument. Its complexity can be an asset when we do not want to make any final evaluations but simply find out how well a student can master certain areas of the target language and, most of all, what his or her weak points are. While discrete-point items (multiple choice questions, blanks, etc.) single out one problem at a time - which implies both a selection and a limitation of the possible difficulties - the translation test is "open" and poses no constraints on the students' creativity (that unfortunate but revealing sort of creativity that leads students to make all kinds of mistakes). This has led one researcher (Pickett 1968) to declare the superiority of translation over blank-filling. The figures he gives show that the translation scores are much closer to the half-yearly scores than those of the blank-filling test. Unfortunately the paper does not say how all those scores were obtained, so it is hard to decide how reliable the mean deviations given can be; besides, there appears to be a very strong bias against blank-filling as a testing technique, which makes one doubt the validity of the findings.

In any case, however useful translation may be for spotting mistakes and problems of learning, we can only have recourse to it if a paper based on it is coherent with the students' approach to the target language. The principle of "construct validity" rules out any form of examination that is not consistent (both in theory and in practice) with the teaching method adopted with the particular students that are being tested. And, again, can we safely assume that the presence of translation will not have any undesirable effects in terms of excessive emphasis on translation in teaching?

I have dwelt on translation proper so far because the questionnaire revealed that it is the most widely used test type, included in written exams by 70% of the faculties, and also present, in some form or another, in most oral examinations. The data appear to indicate a great variety in the materials used for translation: authentic, simplified or constructed; from English and/or into English; previously seen or unseen by students; with or without dictionaries; and ranging from isolated sentences (of the type discussed above) to newspaper articles or business letters. Such a variety of approaches and techniques could provide ample subject-matter for a discussion among the participants as to which of them appear to be the most useful and valid - if, of course, the existing regulations prevent us from looking for something better.

If, however, changes are possible, Widdowson's discourse-to-discourse scheme can provide a suitable basis for a new type of examination paper involving translation. Under this scheme, translation is viewed rather as "information transfer" than as "code switching". In most areas of EST (English for Science and Technology) we frequently find non-verbal representations of discourse: formulae, diagrams, flow-charts, tables, drawings, mappings, blueprints, etc. If we take them into account, the common concept of translation - converting one structure into another - is modified into a three-way translation type:

A
non-verbal device
/	\
L1 use -- English use
B C

When A and C are given, the task of finding the suitable equivalent in L1 is mainly a comprehension activity, while if A and B are given, C is achieved through a composition process.

In order to be applicable in a testing situation, the scheme must respond to a few basic requisites:

a) an adequate number of non-verbal devices must be in common use in the subject-area we are concerned with (in this case, Economics, Commerce, and Political Science);

b) the problem-solving activities must be relevant to the learners, in terms of understanding scientific discourse, or with a view to their future work. This implies they have to be realistic, and based on up-to-date theories and actual data;

c) there has to be the possibility of grading the levels of difficulty;

d)the evaluation criteria must be identifiable and definable without excessive difficulty. The first condition is amply met: formulae, diagrams and tables are widely used in economics, statistics, mathematics, sociology, geography, accountancy, and in most other subjects (excluding perhaps juridical fields).

As to the second point, it is easy to think of activities that can be developed on the basis of the scheme: for example, drawing up commercial offers, orders or invoices based on catalogues and price lists in the other language; using conversion tables for measures or currency; interpreting/verbalizing the information provided by companies by means of graphs (internal structure and organization, etc.) or such financial documents as balance sheets, budgets, and so on; planning itineraries based on airline maps and timetables (they are all in English); and many, many more. I am convinced that whatever field of activity we are interested in, from nuclear physics to sports and games, we have no difficulty in finding non-verbal devices that can be suitable (perhaps with minor adaptations). The specialized press and all the publications issued by companies and organizations usually provide interesting and useful material. Generally speaking, identifying data and retrieving information are much easier than describing processes or giving detailed instructions in the second language. This gives us a first, sketchy indication on how to grade difficulties. We might simplify the students' task, by providing key words or phrases (or any other verbal or non-verbal clues) or by asking them to complete a table that has already been partly filled in. However, I think that the sooner we let the students perform authentic or realistic tasks, the better it is both for their motivation and for the proper development of the learning process. But whichever course of action we choose, it is definitely possible to grade the levels of difficulty and test beginners as well as intermediate or advanced students. < P ALIGN="JUSTIFY">The definition of criteria is connected with the type of task to be performed. It is usually simple to decide whether the information conveyed by a non-verbal device has been understood and used correctly; in some cases, one can even count the pieces of information that have been retrieved, and work outscores from those figures. With more complex problems, where quality is more important than quantity, it may be necessary to analyse and define the implications of the task: is it a semantic, a pragmatic, a communicative or a cognitive translation we want (cf. Newmark 1981)? Is the diagram to be interpreted in terms of definition, or instructions? The acceptability of any given solution depends on whether the students have complied with these requirements, using the appropriate register, type of discourse, lexical items, etc. This may sound overly complicated, but we should not forget that each branch of science and technology normally restricts the choice to a limited number of possible alternatives, connected with the level of formality and the purpose of the communication. If the students have been given adequate instruction on these points, they should be able to select the most appropriate response in each case.

In conclusion, since the four basic requisites seem to be adequately present, the three-way approach to translation appears to be promising as a testing - as well as teaching - technique. Although a lot of practical work still has to be done, the way is open to us and is worth exploring. It is my wish and hope that these ideas may be put to the test in controlled experimental situations, so that we all can get reliable data for the further development of testing techniques and evaluation procedures.

REFERENCES

Catford, J.C. (1965) A Linguistic Theory of Translation. Oxford, Oxford University Press.

Mounin, G. (1965) Teoria e storia della traduzione. Torino, Einaudi.

Newmark, P. (1981) Approaches to Translation. Oxford, Pergamon Press.

Lado, R. (1964) Language Teaching. A Scientific Approach. New York, McGraw-Hill.

Widdowson, H.G. (1978) Teaching Language as Communication. Oxford, Oxford University Press.

Widdowson, H.G. (1979) Explorations in Applied Linguistics. Oxford, Oxford University Press.