Language learning motivation is a complex, composite construct, and although past research has identified a number of its key components, the interrelationship of these components has often been subject to debate. Similarly, the exact contribution of the various motivational components to learning behaviors and learning achievement has also been interpreted in different ways. The purpose of this study is to use structural equation modeling to evaluate a proposed theoretical model concerning the internal structure of the second language (L2) motivation complex and its impact on motivated behavior. The basis of the analysis is survey data collected in Hungary from 8,593 pupils, 13 to 14 years old, on two occasions, in 1993 and in 1999. Our main finding is that integrativeness appears to be the single most important factor, subsuming or mediating the effects of all the other responses to questions asked. In light of this finding, we analyze what motivational content this core component might represent in various settings.ALTHOUGH MUCH EFFORT HAS BEEN expended in the past on the empirical examination of various measured or latent concepts concerning motivation to learn a second language (L2), less attention has been paid to evaluating the internal structure of L2 motivation. This lack of attention is partly due to the fact that, in order to test relationships in multivariable designs, researchers need to apply a relatively new and rather complex statistical procedure, structural equation modeling (SEM). SEM is a multivariate statistical technique that, like factor analysis, is used to interpret the relationship among several variables within a single framework. Its big advantage over factor analysis, however, is that a SEM model also includes directional paths between the variables and not just information about how the variables The Modern Language Journal, 89, i, (2005) 0026-7902/05/19-36 $1.50/0 C 2005 The Modern Language Journal "hang together." Thus, as Dörnyei (2001) concluded, the technique is appropriate for testing "grand" theories, that is, comprehensive models made up of a number of complex, interrelated variables, which is exactly the case with most factors involved in explaining issues in L2 acquisition. We feel that SEM has been underutilized in L2 research in general. In L2 motivation research, however, there exist some important studies using this procedure, most notably by Robert Gardner and his associates in Canada (e.g., In this article we propose a new, empirically grounded construct of language learning motivation, which is then submitted to SEM so that 20The Modern Language Journal 89 (2005) we can evaluate the extent to which it is compatible with our data by computing goodness-offit measures. An important feature of our SEM model is that we relate the various motivational factors to criterion measures associated with L2 learning behavior rather than with holistic outcome measures, such as L2 proficiency or course achievement. We believe that by doing so we can accomplish increased theoretic...