This study aimed to synthesize the structural relationships among willingness to communicate (WTC) and its high-evidence factors in second language (L2) learning contexts by adopting meta-analytic structural equation modeling (MASEM). The MASEM approach is designed to construct a structural equation model (SEM) to explain correlations between variables by pooling correlation coefficients reported in previous literature. This study integrated 44 independent samples ( N = 12,094) and built a MASEM model to investigate the structural relations of WTC, its three high-evidence factors (learners’ perceived L2 competence, L2 motivation, and L2 anxiety), and frequency of learners’ L2 use. The results of meta-analysis successfully supported our proposed model of L2 WTC. Furthermore, among the three high-evidence factors, it was found that learners’ perceived L2 competence influenced L2 WTC the most. In addition, the results of the moderator analyses indicated that learners’ L2 proficiency and WTC contexts (inside vs. outside classrooms) significantly influenced the relationships between L2 WTC and its high-evidence factors. The implications of these results are discussed in further depth and detail.

In this study, we attempted to measure English learners’ communicative intentions and capabilities (CIC) by adding their English proficiency to their willingness to communicate (WTC) as a predictor of the learners’ English use. In the context of communicative language teaching (CLT), learners’ WTC is considered the key to linking important learner factors such as self-perceived competence, anxiety, and motivation. Given that WTC only accounts for learners’ communicative intentions, we hypothesized that learners’ communicative capabilities (i.e., language proficiency) would be a meaningful addition to measuring overall CIC. To test this hypothesis, we collected a total of 67 samples (N = 17,811) and conducted an analysis using a metaanalytic structural equation modeling (MASEM) approach. The results indicated that the learners’ overall CIC successfully predicted their English use, with a significant path coefficient (β = .525, p = .02). Furthermore, we found that the size of the direct contributions of perceived competence and motivation became non-significantly marginal (ps > .05), indicating that adding learners’ proficiency to the WTC to measure the overall CIC can improve the overall explanatory power of the MASEM model to predict learners’ English use.