The noun-category bias phenomenon in 3-year-olds: Taxonomic constraint or translation?

Abstract

Preschool children exposed to a novel noun label (e.g., “dax”) for a standard picture (e.g., a dog) in a match-to-sample task are biased to match the standard with a picture based on taxonomic relations (e.g., another dog) rather than thematic relations (e.g., a dog bone). This phenomenon, called a noun-category bias, has been interpreted as a generalized constraint on word learning that predisposes children to interpret novel nouns in terms of category relations. This experiment tested the generality of this phenomenon across important variations in the match-to-sample task. It also examined an alternative interpretation of the phenomenon called the translation hypothesis, which suggests that the noun-category bias is mediated by the child's translation of the novel label into a familiar word. The data revealed: (a) that the phenomenon generalizes to a three-choice, match-to-sample procedure including taxonomic, thematic, and irrelevant choices; (b) that exposure to either a novel noun label during each choice trial, or to a labelling pretreatment game prior to the match-to-sample task is sufficient both to induce the noun-category bias and to increase spontaneous naming of the standard picture with familiar words; and (c) that there is essentially a zero correlation between the tendency to name the standard picture and the frequency of taxonomic choices during the match-to-sample task.