Language testing and methods of assessment

What are discrete-point tests and tests of integrative skills?

Methods of Assessment

• 10. Discrete-point and integrative testing – What are discrete-point tests? What are tests of integrative skills? Discuss their relative advantages and disadvantages. Is the distinction valid?
• 11. Factors influencing language learning – What factors influencing language learning should be tested? How?
• 12. Changing views of language and language learning – How has language testing been influenced by varying contemporary views on language and language learning?

10. Discrete-point and integrative testing

What are discrete-point tests? What are tests of integrative skills? Discuss their relative advantages and disadvantages. Is the distinction valid?

Spolsky (1975) identifies three stages in the recent history of language testing:

1. the pre-scientific stage;
2. the psychometric-structuralist stage; and
3. the psycholinguistic-sociolinguistic stage.

Psychometric-structuralist testing: Lado's approach

The psychometric-structuralist approach breaks down the complexities of language into isolated segments. This influences both what is tested and how it is tested.

What is tested? A structural contrastive analysis of the first language (L1) and the language being tested (LT) is used at different linguistic levels, from syntax to phonology.

How is it tested? The same approach is applied to the construction of the test. Discrete items are constructed which ideally reveal a candidate's ability to handle one level of language in terms of one of the four skills.

Advantages of discrete-point testing

The clear advantage of discrete-point tests is that they yield data which can be easily quantified.

If language performance is to be described by means of numerical scores, discrete-point testing is useful. The tasks can be unambiguous, marking can be highly consistent, and the candidate's final score is clear. Discrete-point tests can also be marked accurately and objectively, including by mechanical scanning methods.

Disadvantages of discrete-point testing

A major disadvantage is the assumption that proficiency can be adequately measured by breaking language down into separate elements: in other words, that knowledge of the elements of a language is equivalent to knowledge of the language itself.

A vital element is missing from this atomistic approach: the ability to synthesise linguistic elements and use them together. It is also extremely difficult, and probably undesirable, to construct "pure" test items that operate on only one level of language structure. Items of this kind tend to become trivial.

Correct or incorrect judgements may also depend on context. For example, there are communities in which forms such as "I be" and "I were" are accepted.

There is also the question of tolerance. A form may be grammatically non-standard but nevertheless intelligible, as in:

De-contextualised sentences therefore provide a tenuous basis for making judgements about a person's overall mastery of a language.

Quantity versus quality

A further disadvantage concerns the distinction between quantity and quality, particularly in relation to transitional competence (Corder, 1975) and interlanguage (Selinker, 1972), and the idea of successive approximation.

Lado's psychometric tests tend to view language learning as a process of accretion: correct responses are rewarded and incorrect responses are penalised. An alternative view is that answers to test questions should be considered as more than simply right or wrong.

Compare this with Krashen's account of stages of acquisition, including the development of the morpheme system.

Reliability and validity

There is a tension between reliability and validity. Lado's tests may be objective in terms of the actual assessment process, but subjective factors can play an important part in the construction of the test itself and in the interpretation of the numerical score it produces.

In objective tests, students may produce no language at all. In a multiple-choice test, for example, they simply select from a set of alternatives. The ability to recognise appropriate forms is therefore regarded as sufficient. In subjective testing, by contrast, the ability to produce language is crucial.

Integrative testing

Integrative testing attempts to assess a learner's capacity to use many linguistic elements at the same time.

Integrative tests are often pragmatic in that they set tasks which require learners to process sequences of linguistic elements within the normal contextual constraints of the language. They also require learners to relate sequences of linguistic elements to an extra-linguistic context through pragmatic interpretation.

In this sense, integrative tests are often based on criteria of naturalness and realistic language use.

Discrete-point tests, by contrast, cannot easily be pragmatic. There is no ordinary discourse situation or normal language-use context in which a learner would be asked simply to distinguish between ship and sheep, or to perform an active-to-passive transformation in isolation.

11. Factors influencing language learning

What factors influencing language learning should be tested? How?

• Cognitive styles, including differences between children and adults.
• Aptitude and metalinguistic ability, including grammatical sensitivity and inductive ability.
• The role of the Monitor.
• Integrative and instrumental motivation.
• Social group identification.
• Emotional states.
• Personality.
• Cultural factors.

12. Changing views of language and language learning

How has language testing been influenced by varying contemporary views on language and language learning?

Oller advocates a pragmatically oriented approach as opposed to discrete-point-oriented approaches to language testing.

Keith Morrow and Brendan Carroll pioneered communicative language testing. See Brendan Carroll's case studies in communicative testing.

The late 1970s and 1980s saw the advent of communicative testing in public examinations. Examples include the Cambridge examination in the communicative use of English and the British Council course in testing for communicative programmes. There were also changes to the Cambridge examination syllabus.

Books on Language Testing

1. Language Testing Handbook by Harrison (1987) – a clear and concise introduction to theory and practice.
2. Language Testing by J. B. Heaton (MEP, 1982) – a comprehensive review of the important concepts.
3. Writing English Language Tests by J. B. Heaton (Longman, 1975; new edition, 1988) – a practical book which is useful to read before designing tests yourself.
4. The Edinburgh Course in Applied Linguistics, Volume 4: Testing & Experimental Design by Allen (Oxford, 1977) – particularly useful for university research.
5. English Speaking Union Framework by Brendan Carroll and Richard West (1989) – useful for specifying skills and objectives and for defining standards and levels.
6. Language Testing by Baker (1989) – covers the theory and is fairly readable.
7. Marking & Assessment in English by Pauline Chater (1984) – set in the UK secondary-school context and covering assessment of open writing as opposed to closed assignments.
8. Teaching & Testing English by P. B. Ballard and Philip Boswood (Hodder & Stoughton Ltd, 1939) – an interesting reminder that many of the issues involved in teaching and testing English are not new.