The 1970s saw much discussion of the so-called 'Unitary Competence Hypothesis' (Oller, 1979), in essence the belief that language ability could not be broken down into component parts. According to this theory, it was not possible to measure reliably and separately a learner's writing, reading or grammatical ability. Any such test would simply measure a unitary language ability. However, it was not long before the theory was shown to be false; as most language learners are aware, there frequently are differences between, for example, one's ability to read and speak in a FL.

All the same, these differences are usually quite small, since most language training ensures that skills develop at a broadly uniform rate, and there would seem to be some learning transfer between skills. Consequently, while rejecting the Unitary Competence Hypothesis, many test developers still felt that owing to the relatively minor differences between language skills, it was in fact possible to devise measures of overall language ability. Cloze and C-tests fall into this category.

They are two examples of an approach to language assessment that makes use of our capacity for predicting how an utterance will develop, for filling in missing information by using limited linguistic stimuli. (The most common such technique is gap-filling, employed very often, of course, in the practice and reinforcement of FL grammar.) Cloze and C-tests rely on a combination of 'syntagmatic' and 'paradigmatic' competence (cf Bailey, 1998: 59f). The syntagmatic competence refers to our ability to work out what part of speech is likely to be missing; for example, in the sentence 'The policeman stepped out into the road and _______ the traffic', we can assert pretty confidently that a verb is missing. Paradigmatic competence allows us to work out elements of the missing word's meaning, ie the sorts of action a policeman might perform once he has stepped into the middle of a road.

The great advantage of this approach is that it allows the tutor to focus the testing on a single element (eg knowledge of appropriate verbs or nouns, or else some grammatical feature such as inflected adjectives or the use of articles). Moreover, marking is straightforward as there is usually just one correct answer, although with non-grammatical and thus open-ended items, it is essential to try the items out with colleagues or, ideally, students, to identify possible ambiguities and to help compile an appropriate mark scheme.