C34 Home language in school

survived narrowedone dot per attack, oldest first

1. #1 narrowed 1 November 2026 · Slavin, Madden, Calderón, Chamberlain & Hennessy 2011 · target: claim-text

   In the only multiyear randomised trial of bilingual versus English-only reading teaching, Spanish-dominant children taught to read first in Spanish scored reliably lower in English reading than children in structured English immersion during kindergarten and grade 1 (average effect sizes −0.54 and −0.42), and still lower, though mostly not significantly, in grades 2–4 (−0.20, −0.16, −0.25), so home-language programmes do lower majority-language reading by a reliable amount for a time and are not shown to be "at least as good" at the end.

   Strongest reading and verdict reasoning

   Steelmans to: C34's sentence has no time bound: "do not lower reading by any reliable amount" is contradicted by a randomised, within-school comparison with the same reading programme in both arms, where the bilingual group was significantly behind on every English measure for two years; and the claim's "at least as good" rests on abstract-only meta-analyses of mostly matched, older studies, while the one randomised trial's end-point estimates still lean towards English-only.

   Why: Nightly-attacks 2026-11-01 Europe/Stockholm. Full text (2010 report): six schools, three kindergarten cohorts randomised within schools, analyses controlled for PPVT and TVIP pretests; English reading −0.54 (K) and −0.42 (grade 1), significant on all English measures; after transition −0.20 (grade 2), −0.16 (grade 3), −0.25 (grade 4), one of nine comparisons significant (grade-2 Word Attack); English PPVT significantly lower in grades 2–4; Spanish reading +0.44 to +0.63 in K–3. Kill withheld: after the switch to English the gaps were mostly not significant and shrinking, and the authors conclude English reading ends equal — C34's kill line asks for a reliable disadvantage, which the trial shows only during the home-language phase. Caveats: grade-4 sample small (115); numbers from the report, not the published article (paywalled; added to papers-needed); one programme model (early-exit TBE vs SEI), US Spanish speakers. Not a duplicate (empty ledger); not a restatement (Bounds had no early-phase carve-out; the 2026-11-01 sweep stored this study as an abstract-only limits lead). Not a preprint (peer-reviewed article; the report is its earlier version). Status alive → weakened. Claim field, Bounds and kill line rewritten in the same turn; brief and diagram core rewritten (C34 has no front-page rule).

2. #3 survived 14 November 2026 · Valentino & Reardon 2015 · target: kill-condition

   In a large controlled longitudinal study of 13,750 English learners in one US district, followed on the state English Language Arts test from grade 2 to grade 7–8 with controls for parents' programme preferences, children who started in bilingual programmes would fall reliably behind English-immersion children in English reading after the transition to English — which would meet C34's kill-condition (a reliable majority-language reading disadvantage at grade 3 or later).

   Strongest reading and verdict reasoning

   Steelmans to: C34's "about as good... not shown to lower it by a reliable amount from then on" rests on one randomised trial whose grade 3–4 point estimates still lean towards English-only (−0.16/−0.25, ns, small grade-4 sample) and on abstracts; a multi-cohort district panel with four programme types, school fixed effects and preference controls, followed into middle school, is the largest test of the post-transition years on file and could show the lean becoming a reliable gap.

   Why: Nightly-attacks 2026-11-14 Europe/Stockholm (ordinary rotation). Full text: ELs entering kindergarten 2001–02 to 2009–10 (p. 15). Table 3, Model 4 (student controls, school fixed effects, programme-preference controls; 7,729 students), ELA relative to English immersion: grade-2 level transitional bilingual +0.076* (SE 0.036), developmental bilingual −0.017 (0.044), dual immersion −0.191 (0.059); growth per grade transitional +0.023* (0.010), developmental −0.022 (0.013, p < .10), dual immersion +0.064* (0.018); dual-immersion ELA matches the state average by grade 6 and passes English immersion (pp. 25–26). Latino ELs in all three bilingual programmes grow significantly faster than Latino ELs in English immersion and score above them by grade 7 (pp. 27–28). Kill fails: no bilingual programme shows a reliable ELA disadvantage after the transition; the only reliable lag is dual immersion's early one, inside C34's home-language-phase bound. Caveats: not randomised (estimates stable with and without preference controls, p. 26); one unusually diverse district (p. 35); developmental bilingual's slower growth is marginal overall (p < .10), but for Chinese ELs it is significant (−0.068 SD per grade, SE 0.026, Table 4, while Latino ELs in the same programme type grew faster, +0.056**) — a subgroup result for one programme type in one district, with no level test reported, so not counted as meeting the kill line and noted as a possible open question; raw grade-7 descriptives (Table 2) are lower for developmental bilingual (0.059) and dual immersion (−0.133) than English immersion (0.279) before controls. Not a duplicate (prior entries: Slavin et al. weakened and corrected). Not a restatement of the home-language-phase bound (it tests the post-transition clause). Not a preprint. Claim field and kill line unchanged; Bounds get an evidence line. Status unchanged (weakened). C34 fruitless streak stays 0.