An important distinction in language testing is made between 'direct' and 'indirect' tests, that is between those that involve learners in actually performing the skill that is being assessed (direct) and those that assess the language elements which are thought to support or underpin the four main language skills (indirect).

The first theories of language testing were based on structuralist linguistics (Lado, 1961) and assumed that to know a language meant no more than having a mastery over aspects of the language system. Tests therefore tended to separate out these aspects and test candidates' knowledge of discrete areas such as grammar, without much surrounding linguistic context. It was felt that indirect testing was the only reliable and consistent means to measure language ability. Examples of such tests include:

The advantage of indirect tests is that they are thought to be a more precise and objective method of assessment, allowing linguistic features to be tested discretely or separately, and ensuring assessment is transparent and highly reliable: answers are either right or wrong and markers do not need to apply much judgement. Moreover, such tests are simple to administer and are thus very practical. Consider the following:

The student here simply has to ring the correct word (accusative or dative direct article) to go in the gap; there is no writing involved and only minimal reading.

However, such test items are not without their problems. In the above example, with only two possible answers, there is potential for guessing, so one might extend the question as follows:

Although this reduces considerably the scope for guessing, it introduces a further complication in the shape of the gender of the noun so that the student does not just have a simple distinction to make between the use of the accusative or dative case, but now also has to decide whether the noun is masculine, neuter or feminine. This is a simple illustration of the potential disadvantages of such indirect tests. It can, in fact, be very difficult to produce a simple discrete test of a single linguistic point.

Another problem is that no one really knows if indirect tests are valid tests of the skills they claim to measure. Is someone with a good knowledge of FL vocabulary necessarily a good FL reader? And does being able to reproduce sounds and intonation necessarily mean you are a good FL speaker? Most tutors will have experience of students being able to spot errors in the FL when they are asked to identify them in exemplar material (a form of indirect testing) but then proceeding to make the same errors in their own FL writing. Experience suggests there is often a link between indirect tests and the skills they purport to assess, but one cannot say with real certainty that there is.

Furthermore, there are implications for teaching in indirect tests. As Bailey (1998: 75) notes: 'Indirect tests […] may result in negative washback. For instance, if learners spend time studying bits of decontextualized grammar in preparation for an indirect test of writing, they may spend less time actually writing in the target language.' (See Module 13, section 13.1.4.5, for more on 'negative washback'.) This is a real concern in programmes that focus excessively on grammar and translation, where the teaching of FL writing can be seriously neglected (see Module 9).

Growing recognition of the fact that language performance involves the integration of these discrete aspects of linguistic knowledge led to the development of skills testing, ie of separate tests of, speaking, writing, listening and reading. This was given impetus by the spread of communicative language teaching and the need for tests that could measure integrated performance and assess the ability to apply knowledge of the linguistic system in the achievement of communicative goals. (For a fuller discussion of these developments, see McNamara, 2000: 13-21.)

Since language teaching in all areas of higher education has been influenced to some degree by elements of communicative methodology, the present module adopts this basic communicative approach to skills testing. However, it does not ignore indirect testing or the enduring use of translation testing, nor indeed more recent innovations in assessment (see 14.5.2 -14.5.6). For the sake of simplicity and clarity, we deal with each of the four skills in turn, but that is not intended to recommend this approach as some sort of norm: just as much language teaching nowadays features tasks involving more than one skill, so in assessment you may well want to employ integrative tests that assess students' ability to use a combination of skills.