Effects of content and language integrated learning at the primary school level: A multi-level meta-analysis

Abstract

This meta-analysis synthesized the effects of content and language integrated learning (CLIL)—an approach in which non-language subjects are taught in a foreign language—on primary school students. The dataset comprised 28 samples (N = 214,103) drawn from 21 (quasi-)experimental studies that evaluated either foreign language (FL) or academic content learning. A multi-level meta-analysis revealed that CLIL was significantly more effective than non-CLIL for FL learning (d = 0.63, SE = 0.21, p = .003), with particularly strong effects observed concerning improving speaking skills (d = 1.24, SE = 0.24, p < .001) and smaller effects for improving other language domains (d = 0.48, SE = 0.18, p = .009). Additionally, publication year was significantly associated with the impact of CLIL on FL learning, with earlier studies reporting slightly stronger effect sizes than later ones, which demonstrated marginally weaker effects (β = −0.07, SE = 0.03, p = .04). In contrast, CLIL and non-CLIL approaches were comparable in terms of aiding academic content learning (d = −0.06, SE = 0.16, p = .72). Moderator analyses suggested that the observed effects of CLIL on content learning might be influenced by pre-existing differences between groups; studies with confirmed group homogeneity indicated a negative effect of CLIL on content learning (d = −0.22, SE = 0.13, p = .09), whereas studies without confirmation of group equivalence showed a positive effect (d = 0.31, SE = 0.22, p = .17).