Abstract Most research utilizing the Modern Language Aptitude Test (MLAT) has not examined its subtests for prediction of second language (L2) achievement, but instead has used the Composite score. In this secondary analysis of data from Sparks et al. (Lang Test 15: 181–216, 1998a), we follow Dale and Sparks (Stud Second Lang Acquis 46(2): 453–477, 2024) in using regression commonality analysis (RCA) to address the problem of multicollinearity among subtests by decomposing the predictive variance from the MLAT into unique components for each subtest alone and for each possible combination of subtests that have shared variance. Ninety-six U.S. secondary level students of Spanish, French, or German were administered the MLAT at the beginning of 9th grade L2 instruction and measures of L2 word decoding, reading comprehension, writing, and oral proficiency after two years of L2 instruction. The primary importance of the Phonetic Script subtest, alone and in shared variance, is consistent with the major findings of Dale and Sparks. Although there were some differences from the original study results, the results provide robust additional evidence for the role of native language (L1) achievement and L1 literacy-related skills with strong metalinguistic components in all MLAT subtests and in the prediction of all L2 outcomes.
Dale et al. (Wed,) studied this question.