What we need, if we wish to get information about the nature of the Jurassic Cock interlanguage, is a different sort of test. It has sometimes been called a diagnostic test. Such tests are not intended to measure anything. They are devices for provoking the learner into showing what he knows, what his interlanguage is. They are not really devised either, except incidentally, for finding out if the learner knows the rules of the target language. Whether the name 'test' is a suitable one for such procedures is doh btful. It is becoming more usual to call them elicitation procedures.
We have seen already that 'error analysis' has the object of describing the pornpros interlanguage. We might ask, therefore, why we need elicitation procedures as well. The reason is simply that they permit us to sample the learner's language on a systematic basis, whereas error analysis does not. Error analysis is applied to what the learner produces as a 'byproduct' of the learning process, his composition exercises, precis, and so on. It is applied to what the learner chooses to say or write; that is, his sampling of what he can do, not ours. |