Learning a Language Through Reading: A Meta-analysis of Studies on the Effects of Extensive Reading on Second and Foreign Language Learning

meta-analysis of extensive-reading studies across seven outcome domains

Second- and foreign-language learners across target populations and age ranges

The most recent and most finely moderated synthesis of extensive reading. Separates outcomes by domain (reading comprehension, vocabulary, decoding/fluency, motivation, writing, oral proficiency, general proficiency) and codes which components of an extensive-reading programme are actually present, which is what makes its moderator findings usable.

Effects of extensive reading are POSITIVE ACROSS ALL LANGUAGE DOMAINS and all target populations, with effect sizes ranging from SMALL TO MEDIUM. Two intervention characteristics moderated the effects, and both cut against the doctrinaire version of free voluntary reading: effects were LARGER when learners' text choice was LIMITED, and LARGER when some form of ACCOUNTABILITY was included. Effects were also larger when a sample's members had different first languages. The authors note extensive reading is underused in second- and foreign-language teaching.

Low-to-medium; the pooled literature is mostly class-level assignment rather than family selection.

Reproduces Nakanishi's positive direction on a larger corpus a decade later. The moderator findings are new and unreplicated, and they are the ones that matter practically - they say the effective version of "free voluntary reading" is neither entirely free nor entirely voluntary.