Poster #40 - Typically and Atypically Developing Children' Generalizations of Novel Names: the Role of Semantic Distance

• In Event: 2-110 - Poster Session 07
  In Poster Session: PS 07 Section - Cognitive Processes

Integrative Statement

Children often learn the extension of novel words with a limited number of “training” exemplars. There is evidence that the opportunity to compare stimuli (by contrast with single exemplars) during novel word learning is beneficial for their generalization in typically developing (TD) children (e.g., Gentner, 2010). However, the role of comparison has not been studied in children with intellectual disabilities (ID) (Chapman & Kay-Raining Bird, 2012). The main manipulated factor was semantic distance within training stimuli and between training and test stimuli and their influence on taxonomically-based generalization. We hypothesized more difficulties for ID children especially with broader semantic distance because they would encounter difficulties to unify semantically distant stimuli.
24 TD preschoolers and 24 children with mild moderate ID with various etiologies were matched for MA with the Raven CPM. There were 7 close training pairs (i.e., from the same basic level category, e.g., two apples), 7 far pairs (i.e., from the same superordinate level, e.g., an apple and a cherry). Each training pair were introduced as “novel name” “these are buxis”. At test, they had to select “another buxi” between a perceptually-similar-but-semantically unrelated item (e.g., Christmas ball) and a same-superordinate-but-perceptually-dissimilar category, either close (e.g., banana) or remote (e.g., meat) (see Figure 1).
We split the ID and TD children into High and Low Raven scores. A 2 (Learning Distance: close or far) x 2 (Generalization Distance: close or far) x 2 (Group: ID or TD children) x Raven score (Low or High) ANOVA was carried out on the taxonomic choices. It revealed that ID children were better than the matched TD children, suggesting functional lexical learning mechanisms. Even ID low-Raven-scores children, surprisingly, obtained better results than high-Raven-scores TD children. ID children, who were significantly older than TD children, could rely on their more developed world knowledge to learn and extend novel names. Close generalization was also significantly better than far generalization. There was no interaction between Group and the other factors. Interestingly, there was an interaction between Raven score and Learning: High-Raven-scores individuals (HR) outperformed Low-Raven-score participants (LR) in the far learning condition whereas they did not differ on the close learning condition. Also, LR participants were better in the close learning than in the far learning case, while the reverse was true for the HR participants. Importantly, this suggests that LR children had more difficulties to conceptually unify dissimilar training stimuli whereas HR children benefited more from broader learning distances. In sum, what these results show is that ID per se was not the crucial factor here, but rather the level of cognitive functioning which interacted with learning distance. We suggest that ID people can extend their vocabulary in familiar conceptual domain. We interpret our results in terms of cognitive constraints associated with comparison activities which might impact LR children in remote conceptual domains. We predict that ID participants should experience more difficulties with less familiar conceptual domains an idea we currently test.

Authors

• Arnaud Witt, Université de Bourgogne Franche-Comté
  Non-Presenting Author

• Annick Comblain, Université de Liège
  Non-Presenting Author

• Jean-Pierre Thibaut, Université de Bourgogne
  Presenting Author