This study examines aspects of communicative classroomdiscourse that may affect the potential of recasts to be noticed as negative evidence by youngsecond language learners. The database comprises transcripts of over 18 hours of interactionrecorded during 27 lessons in 4 immersion classrooms at the primary level. The 377 recasts inthe database have been classified according to their pragmatic functions in classroomdiscourse and then compared to the teachers' even more frequent use of noncorrectiverepetition. Findings reveal that recasts and noncorrective repetition fulfill identical functionsdistributed in equal proportions and, furthermore, that teachers frequently use positivefeedback to express approval of the content of learners' messages, irrespective ofwell-formedness, to accompany, also in equal proportions, recasts, noncorrective repetition,and even topic-continuation moves following errors. The findings suggest that, from theperspective of both learners and teachers, the corrective reformulations entailed in recastsmay easily be overridden by their functional properties in meaning-oriented classrooms.