Two studies at the Defense Language Institute (California) investigated the contribution of several variables to prediction of post-language-training proficiency: (1) scores on a general vocational aptitude battery and a language aptitude battery, both used to screen potential students; (2) scores on other cognitive measures not used in the screening process; and (3) scores and ratings on measures of student motivation, anxiety, and use of learning strategies. Two additional studies continued the effort to add certain types of native language competency measures to the Defense Language Aptitude Battery used for student selection. One competency measure considered was listening assessment, particularly as it accounted for two factors affecting the difficulty of listening tasks: the extent to which the examinee had the opportunity to rehearse the initial stimulus or recode it for later use; and the extent to which the examinee had a pre-existing mental set enabling application of an appropriate schema to select and organize the stimulus input as needed to perform the testing task. The second native language competency measure considered was grammar testing, particularly speeded grammar tests in which the task is to identify grammatical errors in sentences. Findings and implications for language aptitude test battery development are discussed. (Contains 113 references.) (MSE)