The Evidence on Teaching

Capital Gains: Effects of Word Class and Sentence Position on Capitalization Use Across Age

Hawkey E, Palmer MA, Kemp N · 2025

grade Dquasi-experimentindependentunreplicated
Sample
Two cloze studies; primary-school samples n = 99-101, high school n = 59, university n = 78
Population
Australian primary pupils aged 8-12, high-school students, and university students
Design
Two cloze experiments manipulating whether a word requiring a capital letter was a proper noun, was sentence-initial, or both. Cross-sectional age comparison, not a training study and not longitudinal, so it describes what children do rather than what teaching would change. Recorded because it is the best-designed study of capitalisation development that exists and because its YOUNGEST AGE IS EIGHT - which is the honest floor of the capitalisation literature, and directly relevant to a question about five-to-seven-year-olds.
Key findings
Children capitalise more reliably when cued by word class (a proper noun) than by sentence position, and most reliably when both cues are present; adults are near ceiling on both. Capitalisation is therefore not one skill acquired at one moment but a set of cue-driven habits that consolidate at different rates through primary school. For this archive the load-bearing fact is the sampling frame: the best available developmental account of capitalisation begins at age 8, and no comparable study reaches five-, six- or seven-year-olds.
Genetic confound
Medium. Age comparisons are cross-sectional and confound schooling, reading exposure and maturation; nothing here identifies a cause of individual differences.

Effects

OutcomeMetricValueMeasureTimingVsHorizonClass
Correct capitalisation use as a function of cue typewithin-subject comparison across cue conditionsproper-noun cue used more than sentence-position cue; two cues better than one; adults near ceilingresearcher-designedsingle sessionnonenot-applicabledomain-skill
Youngest age at which capitalisation development has been systematically studiedsample age range8 years (primary sample ages 8-12)standardizednot applicablenonenot-applicabledomain-skill

Cited by