Thirty-five years of ISLA on form-focused instruction: A meta-analysis

54 empirical studies, 5,051 second-language learners

Second-language learners in studies published 1980-2015 in six applied-linguistics journals. Again dominated by adult and university learners; school-age classroom trials are a minority.

The most recent large synthesis of form-focused instruction, sampling six journals across 35 years. The headline g = 1.06 is enormous by this archive's standards - larger than anything credible in the L1 literature - and the moderator analysis explains why: outcome-measure mode, learners' onset proficiency, research setting (laboratory vs classroom) and instructional intensity all significantly modulate it. A pooled effect that moves that much with measurement and setting is a statement about the measurement and setting.

Overall effect of L2 form-focused instruction g = 1.06, 95% CI 0.84-1.29. Only a MINOR difference between explicit and implicit instruction - notably smaller than Norris & Ortega's. Statistically significant moderation by mode of outcome measure, by learners' onset L2 proficiency, by research setting, and by intensity of instruction.

Low as a threat to the within-study comparisons; the pooled literature is experimental and quasi-experimental with learners assigned to conditions.

Reproduces Norris & Ortega's basic direction on a larger and later corpus while narrowing the explicit-implicit gap. Its companion methodological synthesis (Sok, Kang & Han 2018) documents which weaknesses of the 2000 corpus persisted.