What does background knowledge explain about reading comprehension?

Design: Four experiments manipulating when contextual information is supplied (no context / context before / context after / partial context) for the 'Washing Clothes' and balloons-serenade passages. DVs: subjective comprehension rating and free recall in idea units.

Finding: unverified in primary. Published abstract (seen in search results only): 'each demonstrating increased comprehension ratings and recall scores when subjects were supplied with appropriate information before they heard test passages. Supplying subjects with the same information subsequent to the passages produced much lower comprehension ratings and recall scores.' Per-condition idea-unit means circulating online (3.6/3.8/8.0/3.6/4.0) are unverified.

Detailed limits: Small per-cell N. Passages engineered to be incomprehensible without the topic — maximal internal validity, minimal ecological validity. Speaks to schema availability at encoding, not to durable domain knowledge.

Design: 2 (reader nationality: American / Indian) x 2 (passage: American wedding / Indian wedding). Timed reading, interpolated vocabulary task, free recall, five inference questions. Passages matched for readability (T-scores 12.60 and 12.56), 136 and 127 idea units.

Finding: Verbatim (Technical Report No. 97): reading-time interaction F(1,35) = 10.09, p < .01; gist recall interaction F(1,35) = 39.84, p < .01; culturally appropriate elaborations F(1,35) = 208.67, p < .01; culturally based distortions F(1,35) = 128.24, p < .01. Cell means, Americans/American vs Americans/Indian: 168s vs 304s reading time, 52.4 vs 27.3 gist units, 5.7 vs 0.2 elaborations, 0.1 vs 5.5 distortions.

Detailed limits: Tiny N (~38); all participants recruited in a US university town, so the 'Indian' readers were US-resident and reading in a second language; one text per culture. ASYMMETRY WORTH FLAGGING: Indians recalled essentially the same from both passages (37.9 vs 37.6); the recall interaction is driven almost entirely by the American readers' collapse. The elaboration/distortion split is the cleaner cultural-schema signal.

Design: 2x2 between-subjects quasi-experiment: reading ability (high/low) x baseball knowledge (high/low), 16 per cell. Silent reading of a half-inning narrative, then re-enactment with wooden figures on a scale model field, distractor task, written summarization, and sorting 22 sentences by importance.

Finding: unverified — no F, p, effect size, mean or SD visible in any accessible source. ERIC reports improved recall/retention with greater subject knowledge and NO significant interactions between prior knowledge and reading ability.

Detailed limits: Headline rests on main effects plus a null interaction, not a demonstrated crossover. Sample excluded readers below the 30th percentile, truncating the ability contrast. Single researcher-designed text. Knowledge measure is vocabulary/trivia-weighted. Non-random assignment; knowledge partly confounded with gender.

Design: Random-intercept cross-lagged panel models (RI-CLPM) plus multi-group models testing bilingual/monolingual moderation, on ECLS-K:2011 spring assessments kindergarten through fifth grade. Covariates: working memory, cognitive flexibility, English proficiency, basic literacy, demographics.

Finding: Verbatim: 'the relation between domain knowledge and reading is bidirectional and positive throughout the elementary years'; 'All cross-lagged paths ... were statistically significant'; and 'all coefficients of the cross-lagged paths from science to reading were significantly greater than those from reading to science (18.13 < chi-squared[1] < 49.99, p < .001)'. Individual standardised path coefficients and SEs unverified.

Detailed limits: Authors' own: 'Domain knowledge was operationalized only with science scores'; no information on how proximal the reading-assessment topics were to science knowledge; inhibition not included as a covariate. Correlational — RI-CLPM separates within- from between-person variance but does not license causal claims.

Design: Blocked cluster (school) randomised controlled trial over 12 months, grades 1-2: spring G1 thematic science and social studies lessons, summer wide reading of thematically related informational texts, then fall-spring G2 science lessons. Randomisation blocked on school size and prior reading proficiency. WWC rating: 'Meets WWC standards without reservations'.

Finding: Science content reading comprehension overall ES = .18, p < .01. TRANSFER GRADIENT: near-transfer passages ES = .22, p < .001; mid-transfer ES = .17, p < .01; FAR-TRANSFER ES = .04, p > .05 (not significant). Science vocabulary depth overall .05 n.s. Domain-general reading (MAP) growth: Treatment x Time 2 beta = 1.89, p < .001, with baseline imbalance beta = -1.78, p < .01. WWC independent review scored the general-literacy outcome (MAP Reading Total, n = 2,275) INDETERMINATE. The 2024 spiralled trial (N = 2,870, grades 1-3 plus grade-4 follow-up) reports Grade 3 science reading ES = .14, domain-general reading comprehension ES = .11, mathematics ES = .12, sustained at 14 months at .12 reading and .16 mathematics.

Detailed limits: Authors' own: COVID closed schools in spring 2020, so most Grade 2 teachers could not implement the social studies unit and the planned Grade 2 assessments were not administered; 'more sensitive comprehension measures are needed'. Baseline imbalance favouring control; single district. All three Kim trials draw on overlapping schools in one district — they are extensions, not independent replications.

Design: Process model with simulations, applied to word identification in discourse context and arithmetic word-problem understanding. Kintsch & Keenan 1973 is a reading-time/recall experiment varying propositions at constant word length.

Finding: Not an effect but an architecture: construction is bottom-up and context-blind, using 'weak' rules that fire regardless of relevance, producing an incoherent network; integration is a content-free constraint-satisfaction relaxation over it. Knowledge is the INPUT to the mechanism, not the mechanism. The same machinery is applied across domains.

Detailed limits: Under-specifies how the knowledge net is built. NO VERBATIM PRIMARY TEXT OBTAINED — every full-text route was blocked or unreadable; the characterisation above is a close paraphrase from authoritative secondary descriptions.

Design: Quantitative research review of 16 studies, published and unpublished. (The registry title 'A Review of Nineteen Experimental Studies' is the earlier Technical Report No. 574 — do not conflate.)

Finding: The 1994 journal abstract reports median effects of .32 on standardized tests and .88 on experimenter-developed tests. The full 1993 author report and tables were read; it covers 19 studies, whereas the 1994 abstract reports 16. The study-pool crosswalk remains unchecked.

Detailed limits: Full 1993 author report and 1994 journal abstract read; full final 1994 journal article not read. Heterogeneous study-level session counts and significance categories do not identify a causal dose curve.

Design: Commentary synthesizing eight prior quantitative reviews of reading-comprehension-strategy instruction; not a primary experiment or formal meta-analysis. Its saturation claim is an interpretation, not an experimentally identified dose curve.

Finding: Verbatim: 'none of these reviews show that more practice with a strategy provides an advantage'; 'Ten sessions yield the same benefit as fifty sessions'; 'RCS instruction should be explicit and brief'; 'Far from a let-down, this strikes us as wonderful news'; 'strategy instruction may have an upper limit, but building background knowledge does not'.

Detailed limits: Opinion/commentary, not peer-reviewed. The dose claim rests on reviews that did not manipulate dosage — no dose-response RCT is cited, so it inherits the same weakness as Rosenshine & Meister. Teachers College Record volume/page numbers unverified.