Added Value of Dialogic Parent–Child Book Readings: A Meta-Analysis

16 eligible studies; the headline expressive-vocabulary estimate rests on k = 9, n = 322

Children roughly 2-5 years old, including samples at risk for language and literacy impairment; parents (and in some studies teachers) trained to use dialogic reading techniques.

Early Education and Development 19(1), 7-26. The design feature that makes this source valuable is the control condition: every included study compared a dialogic-reading intervention group against a READING-AS-USUAL control, not a no-treatment control. So the estimate is the added value of turning shared reading into instruction — same parent, same child, same book, same time budget, only the interaction protocol differs. That is the cleanest available isolation of "instructional specificity" from "parental attention".

Making shared reading dialogic — prompting, expanding, prompting the child to produce language rather than listen — buys d = 0.59 on expressive vocabulary over ordinary shared reading (SE = .08, 95% CI [0.44, 0.75], p < .001, k = 9, n = 322). The two moderators are the important part and both cut against generalization: the effect shrank substantially for older children (4-5 years) and for children at risk for language and literacy impairment. The authors' own conclusion is that dialogic reading "can change the home literacy activities of families with 2- to 3-year-old children but not those of families with children at greatest risk for school failure" — i.e. the technique works where it is least needed and fails where it is most needed. Van Steensel et al. 2011 report this meta's overall mean as d = 0.42 with a 95% CI of 0.16-0.54, and criticize it for restricting outcomes to vocabulary and for trimming outliers at 3 SD rather than 2 SD.

Low for the ranking, higher for the level. The comparison is dialogic reading vs ordinary reading within randomized/controlled contrasts, so the family is approximately held constant and the differential is not obviously heritable. But the age and at-risk moderators are exactly where passive gene-environment correlation would predict trouble: the families whose children benefit least are the families least able to implement the protocol.

The direction replicates and the magnitude deflates. Noble et al. 2019 (in this archive) finds shared book reading at g = 0.23 against a passive control but g = 0.03 against an active control; Dowdall et al. 2020 reaches a similar place for shared picture-book reading on language. Van Steensel et al. 2011, covering the same programme family with 47 comparisons, puts the whole class of parent-delivered literacy interventions at d = 0.18 and finds Dialogic Reading in 5 of its 30 studies. The consistent survivor across all of them is the ordering — active, instruction-shaped parent behaviour beats passive exposure — not the size.