C23 Spaced L2 practice
narrowed 1 attack: 1 narrowed · serves Q1 · last change 5 November 2026
For teenage and adult second-language learners, spreading vocabulary practice out instead of massing it gives clearly better results on tests a day or more later (g = 1.15 in a meta-analysis of controlled studies); for grammar this delayed advantage has not been shown (g = 0.11, not significant). Longer gaps have beaten shorter ones on delayed tests when the gaps fall within one session and in adults, but between separate sessions longer gaps have not been shown to add a reliable benefit (Kim & Webb 2022).
Go deeper
Where it holds
All limits (11)
Inside the circle: the core of the claim. Outside: every bound and carve-out in the claim file, numbered, with the attack that forced or added it where the record shows one.
forced by a narrowing attackin the original wordingwritten at creation, or added without a narrowing
Warm cards in the list were added by later attacks; plain cards were in the original draft or written at creation.
Second-language vocabulary, grammar and pronunciation practice in controlled studies, with delayed tests; not classroom-wide year-long durability (see C11 for spacing in general).
Written when the claim was created (29 October 2026); not forced by an attack
Evidence on file: Kim & Webb 2022 (Language Learning 72(1), 269–319, DOI 10.1111/lang.12479; full text supplied by Jonatan 2026-11-04, read 2026-11-05): 48 experiments from 37 studies, 98 effect sizes, N = 3,411.
Written when the claim was created (29 October 2026); not forced by an attack
Wording changed later: Forced by attack #1 · Kim & Webb 2022 · 5 November 2026.
Spaced vs massed: immediate g = 0.58 (95% CI 0.16 to 1.00; k = 11); delayed g = 0.80 (0.44 to 1.17; k = 15; I² = 82%; prediction interval −0.64 to 2.24).
Forced by attack #1 · Kim & Webb 2022 · 5 November 2026
On delayed tests by learning target: vocabulary g = 1.15 (0.81 to 1.49; k = 10), grammar g = 0.11 (−0.32 to 0.54; k = 5), difference Q = 13.78, p < .001; adults 0.77 (k = 12), school-age learners (grades 1–12) 0.97 (k = 3).
Forced by attack #1 · Kim & Webb 2022 · 5 November 2026
Longer vs shorter spacing: immediate g = −0.15 (ns; k = 17); delayed g = 0.40 (0.16 to 0.64; k = 32).
Forced by attack #1 · Kim & Webb 2022 · 5 November 2026
On delayed tests: adults (university students or older) 0.54 (0.27 to 0.81; k = 24) vs school-age learners −0.04 (−0.52 to 0.44; k = 8), Q = 4.35, p = .04; spacing within one session 0.76 (0.42 to 1.11; k = 11) vs between separate sessions 0.18 (−0.10 to 0.45; k = 21), Q = 6.83, p = .01; grammar 0.56 (0.06 to 1.06; k = 8), vocabulary 0.34 (0.04 to 0.64; k = 22).
Forced by attack #1 · Kim & Webb 2022 · 5 November 2026
Equal vs expanding spacing: no reliable difference.
Forced by attack #1 · Kim & Webb 2022 · 5 November 2026
Publication bias: apparent in the delayed spaced-vs-massed subset; p-uniform judged it negligible, but the authors could not rule bias out and say true effects may be smaller.
Forced by attack #1 · Kim & Webb 2022 · 5 November 2026
Few studies with school-age learners (3 spaced-vs-massed, 8 longer-vs-shorter effects); the claim says 'teenage' because the meta-analysis pools all school ages.
Forced by attack #1 · Kim & Webb 2022 · 5 November 2026
Nakata 2015 (Studies in Second Language Acquisition, DOI 10.1017/S0272263114000825; abstract only): 128 Japanese college students, 20 word pairs; the amount of spacing had large effects; expanding spacing a small edge over equal spacing.
Written when the claim was created (29 October 2026); not forced by an attack
Between-session gaps for grammar: open question 20 (Suzuki 2017; Suzuki & DeKeyser 2017).
Forced by attack #1 · Kim & Webb 2022 · 5 November 2026
8 of 11 bounds and carve-outs were forced by a specific attack.
How it changed
The ledger records 1 attack, but no earlier wording of C23 is on record, so the text cannot be replayed step by step. Only the current wording is shown.
For teenage and adult second-language learners, spreading vocabulary practice out instead of massing it gives clearly better results on tests a day or more later (g = 1.15 in a meta-analysis of controlled studies); for grammar this delayed advantage has not been shown (g = 0.11, not significant). Longer gaps have beaten shorter ones on delayed tests when the gaps fall within one session and in adults, but between separate sessions longer gaps have not been shown to add a reliable benefit (Kim & Webb 2022).
What would change this
What would kill it: A meta-analysis or large preregistered study of controlled L2 experiments finds no delayed-test advantage of spaced over massed practice for vocabulary, or an advantage smaller than about 0.2 standard deviations (sharpened 2026-11-05 with the weaken: Kim & Webb 2022's full text shows no established delayed advantage for grammar, g = 0.11 from 5 comparisons). Added as a claim on 29 October 2026, promoted from a candidate belief in the evidence library.
Attack record
1 attack: 1 narrowed. Counts of attacks, not a score.
Full attack list (1)
The full text of the meta-analysis C23 rests on shows that the delayed spaced-over-massed advantage is carried by vocabulary (g = 1.15) while grammar shows none to speak of (g = 0.11, 95% CI −0.32 to 0.54), and that the longer-gap advantage on delayed tests comes from spacing within one session and from adults, not from longer gaps between separate sessions (g = 0.18, ns) or from school-age learners (g = −0.04), so C23's general statements for "second-language learners" and "longer gaps between sessions" overreach.
Strongest reading and verdict reasoning
Steelmans to: C23 was written from this paper's abstract alone; its own moderator tables split the result. On delayed tests, vocabulary 1.15 (0.81 to 1.49; k = 10) vs grammar 0.11 (k = 5), Q = 13.78, p < .001 (Table 2). Longer vs shorter spacing on delayed tests: within one session 0.76 (0.42 to 1.11; k = 11) vs between separate sessions 0.18 (−0.10 to 0.45; k = 21), Q = 6.83, p = .01; adults 0.54 (k = 24) vs grades 1–12 −0.04 (k = 8), Q = 4.35, p = .04. The grammar estimate sits below C23's own 0.2 kill threshold, and "between sessions" is exactly the clause where longer gaps were not shown to help.
Why: Nightly-attacks 2026-11-05 Europe/Stockholm. Full text, Tables 1–2 and pp. 287–302: 48 experiments, 98 effect sizes, N = 3,411; spaced vs massed delayed g = 0.80 (0.44 to 1.17; k = 15); longer vs shorter delayed 0.40 (0.16 to 0.64; k = 32); grammar longer-vs-shorter 0.56 (0.06 to 1.06; k = 8). Publication bias apparent in the delayed spaced-vs-massed subset (p-uniform: negligible; authors cannot rule it out). Kill withheld: the vocabulary advantage is large and the overall delayed advantage is 0.80; grammar rests on 5 comparisons with a wide interval, so "not shown", not "no effect". Not a duplicate (empty ledger); not a restatement (Bounds held only the abstract and said "exact pooled size has not been read"). Not a preprint. Status alive → weakened. Claim field, Bounds and kill line rewritten in the same turn; brief and diagram core rewritten (C23 has no front-page rule). Also answers part of open question 20 (note added there).
Evidence
5 linked sources: 1 supports, 4 limits (first-pass draft, pending human review). supports: Nakata 2015 · limits: Kim & Webb 2022, Suzuki 2017, Suzuki & DeKeyser 2017, Heidari et al. 2026