The paper first discusses the complexity of measuring oral proficiency in communicative situations. The difficulty is due to the large number of variables, linguistic and social, which interact with one another. It then reports o n a study which examined the stability of the assessment of oral proficiency on the ora1,interview test.
Students of Hebrew a s a foreign language underwent four administrations of different versions of that test. The administrations differed from one another by the occasion, the interviewer, the speech style, and the topic. Results from the analysis indicated that the different speech style and topic significantly affected students'scores on these tests while the occasion and the interviewer did not. The correlational analysis between pairs of tests pointed to low reliability and lack of stability of the tests, especially when two variables (i.e., occasion and tester) interacted. The results call for use of caution when decisions about individuals are made based on administration of communicative tests, for a need to identify sources of error in communicative tests, and for drawing stringent guidelines for the use of such tests.