Applying Structural Equation Modeling to Second-language (L2) Research: Key Concepts and Fundamental Reconsiderations

As conceptual models of language learning, use, and processing mature, it is both natural and necessary for the statistical models we apply to follow suit. One statistical approach with great potential in the field of Applied Linguistics and L2 studies is structural equation modeling (SEM). SEM is introduced in this review paper as a powerful and highly flexible family of analyses. In doing so, the paper outlines (a) the types of variables and possible modeled relationships that SEM is equipped to address and (b) statistical considerations for applying SEM in L2 research, and (c) a number of additional and key considerations for those interested in delving deeper into SEM (e.g., goodness of fit indices, model modification procedures, etc.). This paper also describes the potential of SEM to contribute to construct validation (e.g., convergent and discriminant validity). Throughout the paper, a plethora of examples pertaining to applications of SEM in L2 research are provided.