Metalinguistic Awareness and Word Reading across Languages and Scripts: A Meta-Analytical Structural Equation Modeling Approach
Word reading is foundational to reading comprehension, yet the cognitive-linguistic skills that support it, and how far their contributions are universal versus shaped by language, script, and instruction, remain debated. Individual studies are rarely powered to adjudicate these debates, and conventional meta-analyses, which pool one bivariate correlation at a time, cannot model how these skills operate jointly. In this talk, I present two meta-analytic structural equation modeling (MASEM) studies that take a model-based approach to this problem in two complementary populations.
The first study examined native Chinese-speaking children (k = 446, N = 70,734), modeling rapid automatized naming, orthographic awareness, phonological awareness, morphological awareness, and vocabulary as concurrent predictors of Chinese word reading. All skills contributed independently, but meaning-based pathways outweighed code-based ones. Each effect varied systematically with script (traditional vs. simplified), method of literacy instruction, developmental stage, and whether reading was indexed by accuracy or fluency.
The second study then turn to school-aged dual language learners of English (156 studies, N = 23,371), where the same three metalinguistic awareness skills are modeled both within and across languages to test the transfer facilitation model. Each skill uniquely predicted word reading within language; first-language skills predicted their second-language counterparts, evidencing construct-level transfer; and while only first-language phonological awareness directly predicted second-language word reading, indirect cross-over effects were significant for all three. Moderator analyses on cross-linguistic distance, learning context, and developmental stage showed that transfer facilitation is conditioned rather than uniform.
Taken together, the two syntheses point to both universal and language/script-specific components of word reading, and illustrate how MASEM can test multivariate, theory-driven questions that primary studies seldom can. I close with implications for reading theory, differentiated literacy instruction, and the design of future primary research.
