This quasi-experimental study examined the potential benefits of zero-error dictation, a teaching technique based on dialogue-driven dictation, and the extent to which target linguistic features and proficiency levels mediated its effects on second language (L2) learners' accurate use of French grammatical morphology. Seventy-nine learners of L2 French in four Grade 5 and 6 classes in the Montreal area were assigned to two conditions: traditional dictation (comparison) group and zero-error dictation (experimental) group. The instructional intervention, which consisted of five sessions spread over a 5-week period, targeted four different morphological features. Learners' knowledge of the target features was tested immediately before the experimental intervention started and immediately after it ended. Findings indicated that the experimental treatment group outperformed the comparison group and that learners' gain scores varied across the target structures and learner proficiency levels.