Meaningful Differences in the Everyday Experience of Young American Children (the '30 million word gap'), with later re-examinations

42 families observed roughly one hour per month

Kansas families across SES strata.

DOI CORRECTION NEEDED — the previously recorded DOI resolved to Sperry et al. (2019), which is a CRITIQUE of this work rather than the work itself. Hart & Risley 1995 is a Brookes Publishing book and has no journal DOI. The famous 30-million figure is EXTRAPOLATED from ~1 hour of monthly observation of 42 non-randomly selected families, then projected to age 3-4.

One of the most policy-influential findings in early childhood, and among the weakest evidence in this database: 42 families, hourly sampling, an extrapolated headline number roughly 7x too large on modern all-day recording, zero genetic control, and no intervention arm. It should not be cited as a causal basis for anything.

None — purely observational SES contrast with no intervention arm and no genetic control.

SEVERE: parental vocabulary and child language ability are both heritable, and there is no genetic control whatsoever. The correlation cannot support the causal claim built on it.

The headline figure is ~7x overstated on better measurement (Gilkerson 2017). Sperry et al. (2018) is a re-examination, not a clean refutation — it shows the accounting is sensitive to who counts as a speaker, and a Golinkoff et al. rebuttal exists. Grade D on both sides.