Abstract

Repeated encounters are widely assumed to support second and foreign language (L2) vocabulary learning, but estimates vary across instructional and assessment conditions. This meta-analysis synthesized 21 independent study-level comparisons (N = 1,227). A random-effects model fitted with restricted maximum likelihood yielded a moderate-to-large pooled post-test effect (g = 0.63, 95% CI [0.26, 1.00], z = 3.36, p < .001), accompanied by substantial heterogeneity (Q (20) = 153.16, I² = 89.51%, τ² = 0.655) and a wide 95% prediction interval [-0.99, 2.26]. Assessment format (QM (2) = 6.26, p = .044) and input modality (QM (2) = 6.43, p = .040) significantly moderated effect sizes; spacing, learning context, educational level, target type, and repetition frequency did not. Reading-only comparisons produced the largest modality-specific estimate (g = 1.21), and combined receptive-productive assessments produced the largest assessment-specific estimate (g = 1.05), although these observational subgroup patterns should not be interpreted causally. Leave-one-out estimates ranged from g = 0.52 to 0.70 and remained statistically significant. Egger's test approached but did not reach conventional significance (z = 1.90, p = .057), and trim-and-fill estimated no missing comparisons. Repeated exposure was therefore associated with improved post-test vocabulary performance, but the small, heterogeneous evidence base and sensitivity of estimates to design and measurement warrant cautious interpretation.

This work is licensed under a Creative Commons Attribution 4.0 License.