Linguistics调整后Lecture1617AppliedLinguistics幻灯片.pptVIP

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * The last item (d) is most likely to ensure that all learners write about the same things. This will mean that their ability to do so can be more reliably compared. Scorer reliability is also very important for any test. One way to achieve this has been to write test items that require no judgement at all from individual scorers. Such items in language tests have usually been multiple choice questions (MCQs), which, since they can be scored objectively by either humans or machines, were for some years very popular. They are, however, limited in the types of language ability that they can test and so bring the validity of tests into question. Test items for validly testing learners’ ability to use the language can be difficult or impossible to construct in the MCQ format. The proportion of test questions that have to be scored subjectively has greatly increased. Subjective scoring of course creates problems for scorer reliability. Practicality There is no point in teachers starting to try to design valid and reliable language tests without first considering the practical realities of their own specific L2 learning context. There are four main practical questions to be considered by test writers thinking about a new test. First, will the new test be cheap and easy to construct? Test item writing can be very time-consuming in terms of teacher hours and it can also be costly to print test papers and make recordings. Secondly, will the test be economical to administer? Different tests may mean teachers need training in their administration. The tests may require more or fewer classrooms and teachers. The tests may take a longer or shorter time to complete. Next, test writers need to consider whether the test will be easy and economical to score. Teachers may need training in new marking techniques and extra p