Syntes: Heidari, Siyanova-Chanturia & Elgort (2026) — spaced idiom practice helped mid- and high-proficiency learners on delayed tests, not low-proficiency learners

Källa: https://doi.org/10.1177/13621688261468902 (Language Teaching Research, OnlineFirst 2026; open access, CC BY 4.0)

Verified content

• [full text] 199 young-adult Iranian EFL learners (aged 18–25) in 12 intact classes at two language schools, four classes at each of three proficiency levels set by the Updated Vocabulary Levels Test (low = 2,000-word level, mid = 3,000, high = 4,000) (Table 2).
• [full text] Design: quasi-experimental, within-participant; every learner had both schedules (20 idioms each), with the order counterbalanced across classes (Table 1). Spaced = each idiom once per 20-minute session on 5 consecutive days (100 minutes); massed = 5 exposures within two 50-minute sessions (100 minutes). Same slides, L1 translation, context sentence, retrieval tasks and feedback in both.
• [full text] Tests: cued-form recall and meaning recall, immediately, 12 days later and a further 100 days later; learners were not told about the delayed tests.
• [full text] Cued-form recall, long delay (Table 3, % correct, spaced vs massed): low 67.4 vs 71.5; mid 82.6 vs 73.9; high 87.9 vs 78.9. Bonferroni-adjusted contrasts (Table 5): spaced better at 12 days for mid (estimate 0.59) and high (1.04) and at the long delay for mid (0.52) and high (0.69); low-proficiency learners showed no reliable difference at either delay (OR 1.01 at 12 days, 0.81 at the long delay; Table 6). Immediately, massed beat spaced for low (−0.66) and mid (−0.68) learners.
• [full text] Meaning recall, long delay (Table 7): low 69.5 vs 71.5; mid 78.5 vs 72.6; high 82.9 vs 76.1. Spaced better at the long delay for mid (0.34) and high (0.43); low-proficiency learners no reliable difference (OR 0.98), although they did show a spaced advantage at 12 days (0.33) (Tables 9–10).
• [full text] The three-way interaction spacing × proficiency × time was significant for both tests (χ² = 17.78, p = .001; χ² = 14.59, p = .006).

Limits

• [full text] Intact classes, no random assignment; the authors say confounds from classroom dynamics, teachers and motivation are possible. Opaque idioms only. One setting (Iran), young adults.
• [full text] The authors report Cohen's d values of −.05 to .07 and call the effects "very small".
• [assessment] Those printed d values do not match the paper's own odds ratios: the usual logit conversion of OR 2.00 gives d ≈ 0.38, and the long-delay cell means for mid and high learners differ by about 9 points with SDs of about 6–9. The size of the spacing effect is therefore not settled from this paper; only directions and significance are used here.
• [assessment] The same first author's collocation study (Heidari 2026, Applied Linguistics Review, DOI 10.1515/applirev-2025-0329; abstract only) reports the opposite proficiency pattern (low-proficiency learners gained most from spacing, mid-proficiency learners least), so the direction of the proficiency moderation is not consistent across the two studies.
• [full text, Kim & Webb 2022] The meta-analysis behind C23 could not test proficiency because primary studies reported it poorly (p. 306 of the stored full text).

Role for C23

• [assessment] Held as open question 26 (2026-11-10): whether spaced vocabulary practice helps low-proficiency L2 learners on delayed tests. Not a narrowing and not a kill-condition test (single quasi-experiment, idioms only, effect size unclear, opposite pattern in a sister study).

Added 2026-11-10 by the project (nightly attack run). [full text] = read in the full text; [assessment] = the project's own assessment.