THE TWO MAJOR PURPOSES OF THIS PAPER ARE: 1) to characterize the historical and present situation in foreign and second language testing in the United States with respect to the major measurement trends represented; and 2) to suggest a number of desirable development activities over the near-and mid-term future that may help to build, in an evolutionary way, on several current initiatives in the fieldinitiatives that already show, as of the early 1980s, substantial promise for facilitating and reflecting effective, performance-based language instruction to an extent not heretofore reached or even closely approximated.
PREVIOUS TRENDSIn order adequately to discuss the current situation in foreigdsecond language testing, it will be helpful to review briefly the history of development of language testing theory and practice over the last several decades. In 1978, Spolsky identified three major historical trends in language testing, which he referred to as "pre-scientific" (roughly prior to the early 1950s), "psychometric-structuralist" (early 1950s through about the late 1960s, and "integrative-sociolinguistic" (late 1960s and following). 1 During the "pre-scientific" period, there was, in general, little concern about or attention paid to the reliability, validity, or other important psychometric characteristics of the testing activities carried out in the course of language instruction. In keeping with the grammar-translation, reading-oriented teaching approaches widely used at that time, classroom instructors found it both acceptable and adequate for their purposes to present the students fairly lengthy passages for translation from or into the target language, exercises on selected ~~ ~ The Modern Languap Journal, 67, iv (1983) 0026-7902/83/0004/431 fl.50/0 01983 The Modern Lanpagv LJournal grammatical points (often dealing with special usages or exceptions to the general rules), and various cultural items, predominantly of the "capital C" type. Teachers acted autonomouslyin preparing and grading their own tests, and it was implicitly assumed that any individual who was qualified to teach a language was, by the same token, fully competent to handle its assessment. Although the development of functional proficiency in listening comprehension and speaking-or, for that matter, of reading and writing competence in other than relatively artificial, academic contexts-was not at issue during the "pre-scientific" period, the testing approaches utilized during this period would obviously not for the most part have served to measure validly the student's ability to make use of the language as a means of functional communication in "real-life" situations outside of the classroom setting. The second period, the "psychometric-structuralist," may be considered to have evolved in very large part from the theoretical work and related empirical studies carried out by Lado in the mid-fifties.* The basic orientation for Lado's approach to testing grew out of, and closely paralleled, the then current structural-analytic approach to...