Partial Color Word Comprehension Precedes Production

ABSTRACT

Previous studies report that children use color words haphazardly before acquiring conventional, adult-like meanings. The most common explanation for this is that children do not abstract color as a domain of linguistic meaning until several months after they begin producing color words, resulting in a stage during which they produce but do not comprehend color words. Contrary to this account, the current study provides converging evidence from multiple measures that toddlers often acquire partial but systematic color word meanings before production, although adult meanings are acquired much later. Also, we found that whereas children’s interpretation of color words is relatively conservative before the onset of color word production, their meanings become broader and overextended upon the onset of production. These data support the idea that inductive processes of category formation, rather than problems abstracting color, explain the delay between children’s first production of color words and mastery of adult meanings.

Supplemental data

Supplemental data for this article can be access on the publisher’s website

Notes

¹ CIE L*a*b* is an easily quantifiable color space designed to reflect perceptual color space in humans (Mahy, Van Eycken, & Oosterlinck, 1994).

² One participant only used color words in an adult-like way (while providing no response for three colors). This participant’s data were retained for the eye-tracking analysis but are not included in analyses of children’s errors, since they produced no errors.

³ These results hold when observed rates of overextension are calculated for each child and compared using a one-tailed dependent samples t-test comparing behavior to chance rates of overextension, calculated for each child, t (16) = 3.32, p = 0.002.

⁴ Munsell Color Space was designed using higher level judgments of similarity by adults (Long & Luke, 2001; Newhall, 1939, 1940), and is the standard color space used in studies examining color language and other studies of color in which precise perceptual distance between colors is not critical.

⁵ While still low in chroma, brown has a higher chroma than gray, black, or white. The outcome of the analysis does not change if we consider brown to be non-adjacent to gray.

⁶ These results hold when observed rates of proximal errors are calculated for each child and compared using a one-tailed one-sample t-test to the 0.30 chance rate of proximal errors above, t (17) = 2.40, p = 0.01.

⁷ The data from the In-Lab Production task were not analyzed according to the groups determined by Parent Report (see section Eye-Tracking Comprehension task) because by definition there were no Production data for children in the Comprehension-Only and No-Knowledge Groups. Furthermore, it was impossible to separate the two production groups (Production-and-Comprehension; Production-Only) because the Production-Only Group was small and produced an average of only 0.83 color word per child on the In-Lab task.

⁸ At the suggestion of reviewers, we also conducted post-hoc analyses of the words for which the children had no reported knowledge. For all groups, these analyses failed to indicate any evidence of knowledge. We also conducted a post-hoc analysis for the No Reported Knowledge group which focused only on colors that, according to Parent Report, were most likely to be understood by children in the other groups, i.e., red, blue, and green. This analysis found no evidence that children in the No Knowledge Group comprehended these words.

Additional information

Funding

Funded by grants from the NSF (155093) and the James S. McDonnell Foundation to DB.