After a survey of current methods of studying lexical acquisition, none of which are really suitable for studying how specific words are acquired and used by adult learners and how such usage differs from that of native speakers, a method is proposed for the study of such qualitative aspects of second language lexical acquisition. Learners are required to fill in the blanks in short texts, or "discourses," which as far as possible are so constructed that there is only one lexical item that would be judged an appropriate filler by native speakers. This technique can be tested by means of a questionnaire administered to native speakers. The main advantages of this technique are that it can be used to test specific hypotheses about learners' interlanguage and to compare different populations, including both learners and native speakers. Certain difficulties that arise in item construction and data analysis are also discussed in detail, particularly the problem of limiting both syntactic and situational possibilities, the significance of inappropriate responses, and the wording of questionnaire instructions. (Author/AM)