The achievements of the communicative approach, especially its methodological innovations, are indisputable. Communicative tasks have become an unalienable part of present-day language teaching culture. No one would wish (or dare) to advocate a return to the grammar-translation method: it is universally accepted that the most important objective of language teaching is to develop communicative competence in the learners.

Language testing has followed in the footsteps of language teaching, since developing communicative competence becomes impossible where testing remains based on testing linguistic competence alone. A situation of this type developed in Hungary in the late eighties, when communicative language teaching in many schools did not match the more or less code-centred state examinations (Dörnyei & N.Tóth, 1987; Heltai, 1989).

At the present moment there is an explosion of language test development in Hungary, with most prospective examination providers wishing to design communicative language tests. While this is to be applauded, there is a lot of loose “communicative” talk about: some teachers and testers seem to have forgotten Swan’s (1985a, b) criticism of the excesses of the communicative approach, and may be committing the same mistakes. 

What Swan criticised was not the communicative approach as such, but certain dogmas attached to it. The present paper has a similar objective, that is, to warn against dogmatic attitudes. My warning concerns two points: authenticity and use of the mother tongue. Authenticity in language testing is not a new issue in the literature: the journal Language Testing devoted a whole issue to the problem in 1985 (Language Testing 2, 1). In spite of this, I feel that the term continues to be misinterpreted and/or used in a rather loose fashion. In this paper therefore I wish to argue against misinterpretation of the concept of authenticity and unjustified abhorrence of using the mother tongue in language testing.

Authenticity

The communicative approach has always laid great stress on authenticity. The principle that naturally occurring texts (“used language”, to use Brazil’s term, 1995) should be used in language teaching and testing is justified many times over. In real-life situations the learner will meet authentic (non-simplified and non-made-up) texts and will have to solve authentic tasks by using authentic language, so the strategies that will equip them for such situations and tasks ought to be developed from the very beginning. 

However, the term authentic is rather problematic: it has several senses and is often used in rather a loose manner. Let us then examine in some detail the different senses of the word authentic.

The first and original meaning of authentic refers to texts: a text is authentic if it is or has been used for actual communication. In this sense the term ‘authentic’ contrasts with ‘made-up’ or ‘doctored’. For this meaning Widdowson (1978) recommends the term genuine.

Used in relation to test tasks, Bachman (1991) makes a distinction between situational and interactional authenticity. He defines situational authenticity as “the perceived relevance of the test method characteristics to the features of a specific target language use situation” (pp. 90-91). In plain English this means that a task is authentic if it occurs in real life and is performed in the classroom or examination room in a way similar to that in real life.

Interactional authenticity is “a function of the extent and type of involvement of task takers’ language ability in accomplishing a test task” (Bachman, 1991, p. 691, citing Widdowson, 1978). According to a less technical definition, authenticity is “the simulation of real-life texts or tasks [=situational authenticity] and the interaction between the characteristics of such texts and tasks and the language ability of the test takers [=interactional authenticity] (Douglas, 2000, p. 90). The key word here is involvement or engagement, which makes authenticity dependent on the learner or test taker (and consequently relativises it to some extent). Bachman also notes that authenticity is a graded concept: instead of authentic and inauthentic we should think in terms of high and low authenticity.

In present-day loose communicative talk the term ‘authentic’ is used in yet another sense, communicative. Communicative teaching uses authentic texts and situationally authentic (life-like) tasks to generate authentic communication, so whatever is communicative is authentic and the other way round. Here we cannot stop to define the concept of communicative as against authentic; I should simply point out that an authentic cloze test, as shown by the name, is an authentic task, but not necessarily a communicative one.

Summing up, the adjective authentic is used in the following senses:

•  textually authentic (genuine),
•  situationally authentic (life-like),
•  interactionally authentic (involving/engaging the learner), and
•  communicative (generating meaningful communication).

In analysing authentic and inauthentic we must bear in mind that all testing situations are inevitably simulations, and even the most authentic examination tasks are only similar to authentic tasks (Spolsky, 1985). The kind of authenticity that is easiest to achieve is text authenticity, but even a fully authentic (genuine) text is not used in an examination in an authentic function. Candidates do not read authentic texts in the examination for information or pleasure, but because they are expected to demonstrate comprehension of the text. Neither do they have to actually carry out instructions given in an examination task, and few of them would actually answer the particular job advertisement that is used as input material for a writing task. A truly authentic task would have the same purpose as a real life task, which it clearly cannot have, since the aim of all examination questions is to get the candidate to display knowledge. Paradoxically, the most authentic task in an examination situation is one that is explicitly aimed at gaining information about the candidate's language ability. From this point of view role-play tasks are much less authentic than traditional examiner-candidate interviews where both participants are their real selves and act out a test situation.

We must accept then that authentic task simply means successful approximation or a high degree of similarity to real-life tasks. However, satisfaction of this condition is not sufficient in itself. A common pitfall encountered in designing communicative language tests is forgetting that situational authenticity may not be accompanied by interactional authenticity. A situationally authentic task may not engage the target group in the way intended. Taken in itself, the task "write a report about the performance of the sales department last year"1may seem very authentic, since it does occur in real-life situations, yet it will be interactionally authentic only for a target group whose members work in companies with sales departments and/or have relevant experience to help them perform the task, and who are or will indeed need the ability to perform that task and have the motivation to get involved in the task. These conditions are absent in many Language for Specific Purposes (LSP) examinations where the candidates are students who will meet tasks similar to those in the examination only much later in their lives, and at the time of the examination the task may prove inauthentic for them. In this case low interactional authenticity can be expected (although it is difficult to predict all the motivational factors, and an apparently low authenticity task might engage candidates in spite of all expectations).

Low authenticity in itself may not be a serious problem. Indeed, "certain test tasks may be useful for their intended purposes" even if they are low in authenticity (Bachman 1991, p. 692). What is less acceptable and indeed irritating is false pretences, that is, tasks low in authenticity trying to pass themselves off as highly authentic. It is for such test tasks that I have used the term quasi-authentic.

Due to the neglect of interactional authenticity tests that are intended to be `authentic' and `communicative' often contain tasks based on authentic texts and requiring life-like communication in life-like situations which are inappropriate to the purpose, target audience and/or other circumstances of the examination, and in this way remain quasi-authentic. To use Swan's expression, there is "communication without the reality" (Swan 1985b, p. 83).

The trouble with quasi-authentic tasks, however, is not only that they may cause irritation, but also the fact that they oust other task types that are regarded as less `authentic' and less `communicative'. Test developers and item writers do well to remember that there are no test tasks that are authentic or inauthentic in themselves: a test task that in general is less authentic may be more authentic for a given target group in a given situation. Therefore, the test developer/item writer, as well as the test taker will lose by adherence to a pre-established categorisation of tasks into authentic and inauthentic.

Many quasi-authentic tasks are due to the non-correspondence between the task and the test taker's proficiency (cf. Widdowson's (1978) discussion of the interaction between the characteristics of texts and tasks and the language ability of the test takers). Communicative examination developers often underrate the importance of linguistic competence and insist that a given task can be completed even if the test taker has limited linguistic competence. As a result, candidates do indeed perform tasks in the examination, and may even pass, without the confidence to attempt the same task in real life. Knowing the limits of their proficiency, they will avoid such tasks in real life or resort to solutions not envisaged by the examination rubric2.

A case in point is writing formal letters at elementary level - letters of complaint, answers to advertisements and the like. Writing formal letters, however, is not an elementary task in real life: everyone knows about the pains of writing even the simplest formal letter in their native language. I will come back to this problem in the next section.

Favourite tasks

Because of the deeply ingrained belief in some communicative circles that some tasks are par excellence communicative and authentic, communicative teaching and testing show a marked preference for certain topics and tasks without assessing their suitability at the given level of proficiency and without assessing how likely the given task taker is to encounter similar tasks in real life and whether he or she will have the confidence and/or willingness to attempt such tasks in real life at all. The favourites in the teaching and testing of writing skills seem to be letters of complaint and answers to advertisements. Although personal experience carries little weight, I must note here that as a teacher of English living in Hungary with sporadic visits to Britain I have not yet had the chance to perform either of these tasks in real life.

In listening comprehension the favourites are understanding directions and railway station announcements over the loudspeaker. The high incidence of such tasks in textbooks would suggest that they are particularly important and thus represent the peak of authenticity. Textbooks contain a number of very complex directions, as if assuming that the person asking for directions was invariably a pedestrian with unlimited memory capacity. Driving in various European cities I found that I was able to follow directions containing one or two bits of information, then I had to stop and ask again. If I understood directions at all, since I understand only standard English. As for railway station announcements, they are notoriously difficult to understand even in one's native language: your best bet is to look at the electric information board or ask a guard.

Causes of quasi-authenticity

Communicative language tests are not easy to develop. They have to meet many requirements and may present the test developer with many difficulties. Detailed analysis of the inherent difficulties of writing communicative language tests falls outside the scope of the present paper, but the magnitude of the problem is indicated by the fact that in assessing receptive skills communicative language tests prefer to use non-communicative, low authenticity objective tests.

In many cases, however, quasi-authenticity is not simply a consequence of the inherent difficulties of writing communicative tests. As I see it, there are two additional causes: lack of experience and/or bias against certain tasks.

Lack of experience

Inexperienced item writers may have the naive belief, referred to above, that there are tasks that are authentic and communicative in themselves and may forget about checking interactional authenticity. Such problems are common in tests of speaking and writing. Let us see a few examples.

Writing letters is regarded as an authentic task: most of us do write letters. Of course, the task is constrained by the fact that the letter we write in an examination has no real purpose and is addressed to an imaginary addressee. A further constraint on many letter-writing tasks is that in most examinations there is some sort of guidance: for comparability, candidates have to write about certain topics in certain ways. The authenticity of the task is further reduced by the fact that writing letters in the traditional way, either informal or formal, has lost much of its importance through the use of electronic communication.

In spite of these limitations, in some broad sense guided compositions might be regarded as a kind of semi-authentic task, provided that the instructions are appropriate, which is not invariably the case. If the instructions require the candidate to write a letter to their parents who live in a provincial town about their first day at university, the question arises why a Hungarian student should write to their parents in a foreign language? Situationally, then, this task is inauthentic. If the instruction is more sophisticated and the purported addressee is a speaker of English (and non-speaker of Hungarian), the question still remains why on earth you should report to them "what support the state provides for young people to acquire a flat", which seems to fall short of interactional authenticity. Naturally, communicative language testing as such should not be blamed for problems like this: they stem from lack of expertise and/or experience in designing examination tasks.

The situation is somewhat more complicated when the task seems to be situationally fully authentic, i.e. one that may and does indeed occur in real life. "You left your watch in the bathroom in the hotel where you stayed during your holiday in Britain. Write to them and ..." The question then becomes whether the successful candidate who has passed their examination and left a watch in the hotel will indeed write a letter at all? Won't they try and phone the hotel instead? One also wonders how the prescribed length of test writing affects the "authenticity" of the product: in the situation referred to above, the following letter would do:
 

Dear Sir,

I left my watch in the bathroom when I left your hotel on December 31, 2000. Could you send it back to me, please? I will of course pay for the postage expenses.

Yours etc.

This letter contains all the necessary information that the hotel needs if they find the watch - in 35 words, which falls short of the 150-word length required by the task rubric. Of course it does not contain the reference to the "emotional value" of the lost object that the instructions require, but why should it? Do you have to explain why you want it back? In this case, then, prescribed length makes the task inauthentic, and reveals that the real purpose of the task is to elicit a sufficient sample of language.

Another task I have met had the following instructions. "Write a letter to the mayor about an environmental problem you have experienced in your town. Describe the causes of the problem and suggest ways to solve it. Make an offer of help." To eliminate the absurd requirement to write to a Hungarian mayor in English we can change the instructions and say that all this happens while you are resident in Britain or the US (and you write your letter to your mayor there), the question still remains: how often do we write a letter to a mayor either at home or abroad, and, in the latter case, would we insist on writing when we know the limits of our language ability? The likely answer is that we would not, so the task remains quasi-authentic, and the question arises whether it is more authentic at all than one where you are simply asked to write a composition without dragging the Mayor into it? (Describe an environmental problem in your home town and suggest ways to solve it. What do you think are the main causes and how could you help to solve it?)

In the last analysis, guided compositions whose instructions are designed to confer an aura of authenticity may prove no more authentic than compositions that do not care to appear authentic, and they can be even less authentic (and irritating) when the instructions are infelicitous.

Role-play tasks used in testing speaking skills may create problems similar to those experienced with guided compositions. A role-play task where the examiner and the candidate have to assume roles in which they are on fairly intimate terms cannot achieve a high degree of authenticity. "Discuss with your wife/husband (=the examiner) how late the children can stay up and watch TV." It feels kind of funny to regard a girl of 16 as my wife with whom I share a couple of children, but again, the main question is what language do you speak with your wife/husband? How many of us are likely to have English-speaking spouses? No comment is required for the following - probably apocryphal - role-play task, which allegedly appeared in an examination preparatory book some years ago (I have not seen it, but have heard it quoted several times): "You are in Sweden on a holiday with your family. You enter a restaurant. Order borjúpaprikás and túrós rétes. Complain that you had to wait too long and the food was cold, and tell the waiter you will not give a tip." [Borjúpaprikás (veal stew) and túrós rétes (cottage cheese strudel) are typical Hungarian dishes]

Biases and feelings of guilt

More experienced item writers and examiners do not make mistakes like the above, yet they may commit mistakes due to their uncritical attitude to the fundamental principles, or rather, the established practice of the communicative approach. In other words, they are heavily biased against anything that has not received the communicative stamp of approval. A widespread bias concerns the use of translation and, in general, use of the mother tongue. After years of communicative teaching and testing many teachers have, perhaps subconsciously, come to the conclusion that anything that is related to the use of the mother tongue in not authentic and hence cannot be communicative, and vice versa. One cannot help remembering Swan's comment in 1985, that teachers always feel guilty about something. This feeling of guilt may get communicative teachers and testers into the curiously contradictory situation where they accept reduced authenticity rather than resort to use of the mother tongue. One could say that some of the problems are of the test writers' own making. This is especially common in testing the receptive skills.

Use of the mother tongue in language testing

Whatever their disadvantages, test methods involving the use of the mother tongue are often more convenient and direct than other test methods. Let us then first examine the issue of directness and indirectness in language testing.

Authentic and indirect

All tests are indirect to some extent: the underlying language abilities of the candidates can only be studied indirectly through samples of their performance. However, the degree of indirectness may vary, and test methods that make use of the mother tongue seem to be relatively less indirect than many other testing procedures.

I should note here that I am using the term direct and indirect in a sense slightly different from that used by Bachman (1990, p. 323). In discussing direct testing Bachman indicates that the terms direct and indirect are often used synonymously with authentic and inauthentic, and he defines direct testing, as "a match between the test method facet, `expected response channel', and the language ability channel". On my interpretation `direct' means that the operation of the ability or abilities being tested can be observed directly and the interpretation of the data for the underlying ability is relatively straightforward, while in indirect testing interpretation of the data will require elaborate inferences. One might say that in direct testing we have eye-witness reports (not always to be trusted, to be sure) while in indirect testing we have only circumstantial evidence.

Thus, although there is certainly some overlap between `authentic/inauthentic' and `direct/indirect', the two pairs of terms are far from synonymous. If a candidate is asked to talk for two minutes about the topic of money, for instance, the task may not be very authentic, yet certain aspects of his/her speaking ability will be accessible to direct observation, and less circumstantial evidence will be needed to make inferences about the underlying abilities than would be needed if we wanted to test speaking ability through dictation or a vocabulary test or oral translation of isolated sentences into the second language (L2).

The operation of receptive skills is always less accessible to direct observation than the operation of productive skills and assessment has to be based on more circumstantial evidence. The most direct tests of listening comprehension are those that involve carrying out instructions, which may be impracticable in a test situation and the range of text types, the comprehension of which can be tested in this way, is limited.

Comprehension of a text (presented in either the oral or the written mode) can be tested in the relatively least indirect way by asking questions about it, or by asking the test taker to summarise or paraphrase it. This can be done in either the foreign or the native language, and perhaps it is summary in the native language that provides the least indirect approximation. Asking candidates to answer multiple choice questions on the text, sort out jumbled passages, insert or match subheadings and paragraphs will give a more indirect and at the same time less authentic test method.

The greatest drawback of the less indirect methods involving the use of the mother tongue (L1) is that rating must be subjective, and many would also add that such methods have undesirable washback effects. That may very well be true, and these two disadvantages alone might provide sufficient reason for discarding them in a given test situation and for preferring other test methods. What I am arguing against is not the rejection of test methods related to use of the mother tongue, but the gut reaction "it's not communicative, it's not authentic, it's not in".

Advantages and disadvantages

By now it has become a commonplace that every test method has its advantages and disadvantages, and in selecting a test task one must look at both. One must consider questions like the following.
 

•  How is learning affected if translation is used in language examinations? Which method is more authentic, more direct, more reliable and easier to mark?
• Do the questions used in some reading comprehension tasks give the answer away, or facilitate it too much?
• In which task can you guess blindly, and which task aggravates the test taker's job by giving complicated instructions that are more difficult to understand than the text itself?
• Where can you give an answer by just repeating part of what you have read without understanding it?
• Where is it easier to isolate sub-skills? 

In most cases the advantages and disadvantages will balance each other, and the selection of the most appropriate method for a given examination will be crucially dependent on an external requirement, i.e. practicality: the test task used must be relatively easy to (mass-)produce, administer and mark and at reasonable costs.

To illustrate the balance between advantages and disadvantages, the table below compares translation with other types of test tasks often used in communicative tests of reading comprehension (information transfer, matching, jumbled passages, etc.).

Table 1. Features of translation and `communicative' reading comprehension tasks

The table shows that test tasks not involving use of the mother tongue have a number of advantages over translation (5, 6, 7, 8, 9), while translation as a test task is in general more authentic (1, 2, 3): it is more often needed in real life than sorting out jumbled sentences. This does not mean that it is interactionally authentic for a given target group, but the possibility also exists that it is, and this feature alone might make it the best choice under given conditions. Features (10) and (11) can be interpreted from different points of view.

In selecting test tasks, text availability (4) may also be an important consideration, both from the theoretical and the practical point of view. Some reading comprehension tasks can only be used with a limited range of texts. Admittedly, not all texts are amenable to information transfer or matching paragraphs with subheadings. As a result, the texts used in communicative reading comprehension tests tend to be confined to a few well-worn types. Thus the validity of the test will be reduced, since fewer text types will be covered in the test. In addition, since the limited range of text types makes the job of finding a text suitable for testing purposes more difficult and time-consuming, the costs of producing reading comprehension tests will increase.

Dictionary use is a further consideration. It is open to debate whether candidates should have skills in using both monolingual and bilingual dictionaries and if so, when and how should they acquire them, how and if they should be tested, and so on. Since most communicative tests do not require or allow dictionary use, we may conclude that dictionary-using skills do not rank high on the priority list of communicative testing.

The effect of proficiency level

The insistence of the communicative approach on authentic materials and the (over-) emphasis on scanning and skimming and global and selective comprehension respectively, and the de-emphasis on detailed and accurate comprehension has led to a tendency to play down the importance of proficiency level. Testing global comprehension is indeed very important and life-like. We deal with most texts using a top-down approach, establishing first what the text is about and what the main points are, what the pragmatic function of the text (the `text-act') is, and so on. However, this is not invariably the case, and even when we start from the top down, we may have to go down to the very bottom and try to extract all the information that the text contains. Imagine a reading comprehension task in a basic level LSP examination with a text describing safe storage of pesticides. There is really no point in asking for global comprehension: it is just irrelevant in this case. It is simply not a valid measure of comprehension in the given context. With special-language texts comprehension must often be all or nothing, and consequently at the lower proficiency levels it becomes very difficult to assess detailed comprehension using the preferred methods of general communicative language tests. Strangely enough, while it may prove difficult or indeed impossible to write a viable test using the technique of information transfer or jumbled paragraphs at this level, the text itself may very well be within the candidates’ abilities and detailed comprehension could easily be tested by asking the candidate to translate into L1.

Use of the mother tongue in testing listening comprehension

Similar difficulties may arise in testing listening comprehension, where most of the methods used by communicative language testing do not call for the performance of an authentic task. This, again, is often due to a biased attitude against use of the mother tongue. To come back to our example of the railway station loudspeaker, in real life we are not likely to have to answer True/False questions on the announcement made, while it is perfectly possible that a member of our family or a friend who does not understand the language of the announcement will ask us what the message was. We may have to repeat or explain the message in the same language to another foreigner who seems to have less English than ourselves. Repeating and summarising or giving a full paraphrase in L1 or L2 are situationally authentic tasks of the less indirect types, while the tasks preferred by "communicative" testing (filling in tables, following directions on maps, writing in missing words, etc.) are less authentic and much more indirect.

If we compare the method of testing listening comprehension in the examinations of the State Foreign Languages Examinations Board (SFLEB) before 1991, which called for the summarising of a recorded text in Hungarian, and the objectively marked methods (e.g. multiple-choice and true or false questions) used in most other examinations, the result will obviously be that the former was more authentic and less indirect, but also less reliable due to subjective evaluation, while the latter are very indirect, though, in principle at least, more reliable. The former method also had a much higher degree of authenticity than the latter. In deciding which task to use we must then consider the following questions. 
 

• What level of objectivity and reliability is required by Hungarian examinees?
• Is objectivity (or face-objectivity3) and reliability more important than task authenticity?
• Are Hungarian examination providers prepared to write good objective tests?
• Will examination fees cover the costs of writing such tests?
• If subjectively marked tests are used, can the examinations centre arrange for a sufficient number of standardising sessions to harmonise evaluation among its examiners?
• Can the centre provide a sufficient number of listening comprehension texts of equal difficulty?
• Can the centre ensure satisfactory technical conditions for the task?

Over ten years ago a TOEFL staff member visited SFLEB4and, somewhat ashamed, we described to him the primitive ways we used in testing listening comprehension (summarising a recorded text in Hungarian). To our surprise, he exclaimed with enthusiasm, "That's pure comprehension", meaning that it was a test with a minimum of test effect. Of course, this may or may not be true, yet the spontaneous reaction indicates that this method may indeed produce fewer test effects than the methods used by TOEFL. It would have been (or would be) worth trying to develop the method of oral summary in L1 so that it can retain its advantages while getting rid of some of its disadvantages. The task used in SFLEB at present (answering questions in Hungarian about a recorded text in a language laboratory) is still similar in some respects, but while eliminating some of the technical problems it loses some of the authenticity of the former test. It is a pity that in discussing the advantages and disadvantages of different test methods most debaters base their arguments on principles they know to be true or on (anecdotal) empirical evidence, while systematic comparative studies are lacking. I am not aware of any research aimed at exploring, e.g., whether questions in Hungarian yield better results than questions in English, or how a score of 8 out of 15 marks in an objective listening test compares with the subjective rating scale used with the oral summary.

Sporadically reports on testing methods involving use of the mother tongue appear in the language testing literature (e.g. Lee, 1986; Shohamy, 1984; Shohamy & Reves, 1985; Shohamy et al., 1986; Sunderland et al., 1988). Most of these reports claim that there is indeed nothing wrong with the use of the mother tongue, and it may offer advantages over monolingual methods. This undercurrent in the literature seems to have made no impact and has not been followed up consistently.

Monolingual and bilingual examinations

The 71/1998 governmental decree on the accreditation of language examinations divides examinations into two categories, `bilingual' and `monolingual'. The bilingual examination must contain a translation task, or to use a broader term, a mediation task. For practical purposes a distinction between bilingual and monolingual examinations may be useful, but basically the dichotomy is false, since even the so-called bilingual examinations do not claim they are testing the mediating skills proper: they cannot even aspire to do so. From a theoretical point of view we could say instead that some examinations make use of the mother tongue in testing, while some other examinations do not. The artificial division of examinations into monolingual and bilingual may have the undesirable consequence that examinations that are basically monolingual may avoid the instrumental use of the mother tongue in order to remain monolingual, even though use of the native language would simplify matters in some cases.

It is my firm belief that the type of task that is best to use under given conditions should be selected after careful consideration of the advantages and disadvantages and the given circumstances, and not on the basis of beliefs about authenticity and communicativeness. To put it quite explicitly, use of the mother tongue is not a matter of principle but a methodological issue. If we want to check comprehension of particular parts of a reading comprehension text, the question is not really whether the mother tongue is used in the task, but this: everything considered, which is more efficient, a short question in Hungarian or, let us say, a multiple choice item with four distractors. `Monolingual' tasks are not `authentic', `communicative' etc. by their very nature and `bilingual' tasks are not necessarily inauthentic, non-communicative and so on.

LSP examinations

As described above, the preferred tasks of `communicative' reading comprehension tests tend to measure global comprehension plus specific details. This might work very well for some occupations; however, most of the LSP examinations currently being developed in Hungary aim to cater for would-be professionals. In their case, global comprehension is simply not enough: they must be able to fully comprehend the texts in all their detail. Indeed, the very notion of `basic level LSP examinations' has been challenged several times in the past. While basic communication skills may be needed by waiters and taxi-drivers, the value of such an examination for a biologist would be very low indeed.

However, if we accept the idea that basic level LSP examinations are needed by doctors, engineers, economists and so on (the rules for accreditation require three levels for an accredited examination system) it may appear that translation at this level may provide a more viable and more authentic task than what you can find in the usual communicative kit. Once the test taker has the necessary background knowledge, he can translate into or summarise a special language text in L1 without undue difficulty. In contrast with the diverse functions of general and literary texts, in special-language texts the referential function is dominant, and within a specific domain the number of grammar constructions and lexical items to be acquired is relatively small. To a great extent meaning is dependent on technical terms, which - compared to general words - have a more straightforward meaning and are easily identified. Comprehension and translation of a simple LSP text at the elementary level examination is not such a hard test as translation of a newspaper text in a general language examination, is easy to administer and gives sufficient, though not always quantifiable, information on comprehension. With special language tests linguistic competence is more directly related to successful comprehension, and heavier reliance on linguistic competence would be quite legitimate.

Conclusions

All examination situations are simulated, and the authenticity of every test task is limited. There is no test task that is `authentic' or `communicative' in itself, by its very nature. In selecting tasks for examinations one should assess without prejudice the advantages and disadvantages of various test tasks together with careful consideration of the issue of practicality. Test methods related to the use of the mother tongue (translation, summarising in L1, answering questions in L1) may in some cases be more practical, more authentic and even more communicative than monolingual test tasks. Use of the mother tongue in language tests should not be a matter of theoretical correctness but a matter of practical methodology.

Notes
 

1. The examples in this article are drawn from various examination preparatory materials or past examination papers that I do not wish to name since it is not my intention to criticise any particular author, publisher or examining body. None of the examples represents a task in current use by any examinations centre. Hopefully the anonimity of my sources will not damage my credibility. [back]
2. Rubric: a set of rules telling one what to do in an examination paper.[back]
3. I use this term as a parallel to face validity to signal that objectivity is only skin-deep: almost everything in an objective test is subjective (Bachman 1990). [back]
4. The present author worked for SFLEB between 1983 and 1991.[back]

References

Bachman, L. F. (1990). Fundamental considerations in language testing. Oxford: Oxford University Press.
Bachman, L. F. (1991). What does language testing have to offer? TESOL Quarterly 25, 671-704.
Brazil, D. (1995). A grammar of speech. Oxford: Oxford University Press.
Douglas, D. (2000). Assessing languages for specific purposes. Cambridge: Cambridge University Press.
Dörnyei, Z., & N.Tóth, Zs. (1987). Megméretett... A magyar állami nyelvvizsgarendszer áttekintése. Pedagógiai Szemle 37, 1119-1132.
Heltai, P. (1989). Elméretett... Hozzászólás Dörnyei Zoltán és N.Tóth Zsuzsa cikkéhez. Pedagógiai Szemle 39, 63-76.
Hughes, A. (1989). Testing for language teachers. Cambridge: Cambridge University Press.
Lee, J. F. (1986). On the use of the recall task to measure L2 reading comprehension. Studies in Second Language Acquisition 8, 201-212.
Shohamy, E. (1984). Does the testing method make a difference? The case of reading comprehension. Language Testing 1, 147-170.
Shohamy, E., & Reves, T. (1985). Authentic language tests: Where from and where to? Language Testing 2, 48-59.
Shohamy, E., Reves, T., & Bejarano, Y. (1986). Introducing a new comprehensive test of oral proficiency. ELT Journal 40, 212-220.
Spolsky, B. (1985). The limits of authenticity in language testing. Language Testing 2, 31-40.
Sunderland, J., Yixing, Z., & Barr, B. (1988). `To be an explainer': a reporting test of oral proficiency. ELT Journal 42, 28-33.
Swan, M. A. (1985a). A critical look at the communicative approach. Part I. ELT Journal 39, 2-12.
Swan, M. A. (1985b). A critical look at the communicative approach. Part II. ELT Journal 39, 76-87.
Widdowson, H. G. (1978). Teaching language as communication. Oxford: Oxford University Press.