Syntes: Suzuki (2017) — shorter vs longer gaps for practising L2 grammar

Källa: Suzuki, Y. (2017). The Optimal Distribution of Practice for the Acquisition of L2 Morphology: A Conceptual Replication and Extension. Language Learning 67(3), 512–545. https://doi.org/10.1111/lang.12236

Verified content

• [full text, accepted manuscript] 60 students at a Japanese university (mean age 19.6) learned the simple and complex verb morphology of a novel miniature language for oral production; four training sessions 3.3 days apart or 7 days apart; sample size set by an a priori power analysis; six who could not keep the schedule were excluded, so 54 were analysed (26 and 28; the author reports the same pattern with them included).
• [full text] From the third session on, the 3.3-day group was more accurate, and stayed so on the 7-day and 28-day delayed posttests: vocabulary d = 0.54 and 0.48; rule application d = 0.60 and 0.49; present progressive d = 0.60 and 0.52 (all p < .05). No difference in speaking speed; linguistic complexity did not moderate the effect.
• [full text] The shorter gap also won for vocabulary, not only grammar. The paper itself cites two classroom-type L2 studies with written, receptive tasks where the longer gap won (Bird 2010, d = 2.02; Rogers 2015, d = 0.94), and notes that the 7-day posttest may have acted as extra practice before the 28-day one (corrected 2026-11-04, study-mining).
• [full text] At the 28-day test the 7-day gap was the one predicted to be optimal (gap about a quarter of the retention interval), yet the shorter gap won.

Limits

• [assessment] A miniature language, not a natural L2; university students; grammar, not vocabulary; both groups had spaced practice (no massed arm), so it does not touch C23's spaced-vs-massed prong.

Why it is in the library

Found 2026-10-30 (nightly attacks) against C23's clause "on delayed tests, longer gaps between sessions also beat shorter ones". Held as open question 20 (one study plus Suzuki & DeKeyser 2017 against an abstract-only meta-analysis).

Added 2026-10-30 by the project. [full text] = read in the author's accepted manuscript.