Poster #9 - Novel Nouns Generalization: Dimensional Distinctiveness and Comparison Formats

• In Event: Poster Session 19

Abstract

Learning novel nouns entails finding out information that characterize entities belonging to a category and distinguish them from other categories (e.g., Murphy, 2002). Discovering conceptually relevant dimensions for categorization is facilitated when children compare multiple stimuli rather than a single stimulus (e.g., Gentner & Namy, 1999). However, not all comparison situations are equal (e.g., Lagarrigue & Thibaut, 2021). We tested the idea that stimuli built along the same set of dimensions may still differ in distinctiveness and hypothesized that carrying out comparisons with high-distinctive objects would help children to identify relevant dimensions more than comparisons with low-distinctive objects. In a comparison versus no-comparison design, we also manipulated within- and between- category comparisons with 3-4 and 5-6-year-olds.

300 children took part in a word extension task with unfamiliar objects. They were divided in two age groups: 150 young (3-4 years old) and 150 old (5-6 years old). They were randomly assigned to one of the six following experimental conditions with 25 children per condition: no-comparison with (1) high or (2) low distinctive objects, within-category comparison with (3) high or (4) low distinctive objects and, between-category comparison with (5) high or (6) low distinctive objects. In the within-category comparison conditions, the two learning stimuli had the same texture but had different shape. In the between-category comparison conditions, the two learning stimuli had the same shape but had different texture. In all conditions, we performed a forced choice between a stimulus with a similar shape and a different texture, and a stimulus with a different shape but the same texture.

We hypothesized that comparison conditions would elicit more correct generalizations, that is texture-based generalizations, than no-comparison formats. Moreover, we hypothesized that high distinctive objects would lead to better results than low distinctive objects because a high distinctive dimension is easier to dismiss as irrelevant and/or notice as relevant than a low distinctive dimension. Younger children should answer more randomly in the low distinctiveness case because of a difficulty to align low distinctiveness stimuli. We conducted a 3x2x2 ANOVA with Comparison (within-category comparison vs. between-category comparison vs. no-comparison), Distinctiveness (low distinctiveness vs. high distinctiveness) and Age (3-4- or 5–6-year-old) as between-subject factors on the number of correct choices.

The ANOVA revealed a significant interaction between comparison and distinctiveness. In both distinctiveness levels, the percentage of correct responses in the within-category comparison conditions was significantly higher than in the no-comparison condition and the between-categories comparison condition. Comparison with highly distinctive objects benefited more children than comparison with low distinctive objects. The ANOVA also revealed a significant interaction between comparison and age. In between-category and no-comparison conditions, there were no differences between age groups, but the former benefited more from within-category comparison than the latter.

These results suggest that within-category comparisons promoted more correct responses than between-categories comparison or no-comparison in both age groups. However, the benefits of comparisons were more important in the high distinctiveness case than in the low distinctiveness case and more important for older children than for younger children.

Authors

• Yannick Lagarrigue, ToNIC UMR 1214

• Jean-Pierre Thibaut, Université de Bourgogne Franche-Comte
  Presenting Author