Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Grounded experiences and linguistic regularities jointly contribute to the development of categorical knowledge in preschoolers
Poster Session B, Friday, September 12, 4:30 - 6:00 pm, Field House
Xiaohui You1, Xiaosha Wang2, Xi Yu3; 1Beijing Normal University
Introduction: Human semantic knowledge is organized into conceptual categories (e.g., animals, tools), facilitating efficient processing of complex information. Behavioral studies suggest that children develop categorical concepts during the preschool years, potentially through exposure to shared co-occurrence patterns of words (e.g., dogs and cats often occur with pet). Meanwhile, eye-tracking and imaging studies reveal, as early as in infancy, humans are sensitive to object features, such as shape, that are available through grounded experiences and help distinguish between categories. However, it remains unknown whether and how grounded experiences and/or linguistical statistical information contribute to the development of the categorical concepts that underlie children’s semantic behavior. Methods & Results: To address this question, we tested 42 preschool children (56.3±12.6 months, 36–83 months, 26 female) and 36 adults (21.9±2.7 years, 18–30 years, 22 female) using a spatial arrangement task. Participants arranged 16 objects familiar to children under 3 (four per category: animals, humans, tools, scenes) on a gridded board such that semantically related objects were placed closer together. To quantify categorical organization, we computed a semantic dissimilarity matrix for each participant based on Euclidean distances between object placements. A "categoricalness" score was obtained by correlating each matrix with a binary theoretical category matrix (0 = same category, 1 = different categories), followed by Fisher Z-transformation. We first characterized developmental changes in categorical knowledge. Categoricalness scores showed a significant positive correlation with age (r = 0.58, p < 0.001). When children were analyzed by age group, categoricalness scores were not significantly different from zero in 3–4-year-olds (t₁₃ = 1.93, p = 0.08) or 5-year-olds (t₁₀ = 1.78, p = 0.11), but were significantly positive in 6-year-olds (t₁₆ = 5.60, p < 0.001) and adults (t₃₅ = 13.94, p < 0.001). Next, we examined the contributions of object features and linguistic statistical information to children’s categorical knowledge. The 16 objects were rated on 65 brain-derived semantic dimensions (Binder et al., 2016) using GPT-4o, which were reduced via principal component analysis into 10 components (explaining 79% of variance). Ten feature-based dissimilarity matrices were created from the absolute pairwise differences between objects on each component. Linguistic co-occurrence dissimilarities were derived from the Mandarin CHILDES corpora using 3-, 7-, and 11-token windows. For each participant, linear regression models were used to predict semantic dissimilarity from each of the featural and linguistic matrices. The resulting coefficients of all predictors were then entered into a second-level regression with categoricalness as the outcome. Results showed that coefficients for the 11-token co-occurrence pattern and five featural components--related to positive and negative visual experiences, auditory-tactile, temporal, and numerical features, significantly predicted categoricalness (all p < 0.05). Conclusion: These findings indicate that categorical knowledge emerges gradually during the preschool years. Crucially, both linguistic co-occurrence patterns and object features enriched through grounded experience contribute to the development of children’s categorical knowledge. These results suggest that early semantic development and organization reflects an integration of language-based regularities and experiential knowledge acquired through interaction with the world.
Topic Areas: Language Development/Acquisition, Meaning: Lexical Semantics