Abstract
Children recognize semantic relations between words early in development, yet it is unclear how this ability is affected by children's vocabulary size. We explore this question using an eye-tracking paradigm that manipulated the degree and kind of semantic overlap between labeled target objects and unnamed distractor objects to measure real-time recognition of known word meanings in 18-month-olds with varied vocabulary sizes (N = 136). Additionally, we assess how semantic network structure in children's vocabularies affects disambiguation among semantically related and unrelated words by measuring words' density of semantic interconnections within the semantic networks. Semantic overlap condition differently affected lexical recognition as a function of the kind of semantic overlap, children's vocabulary size, and words' semantic network structure. Across children, word recognition was least robust in the condition involving taxonomic (i.e., categorical) semantic overlap between words. For children with larger vocabularies, denser local neighborhood semantic network structure decreased word recognition under taxonomic overlap. However, for children with smaller vocabularies, denser network structure did not affect word recognition in conditions under taxonomic overlap. These results reinforce the idea that lexical recognition is a dynamic process dependent on the specific configuration of children's semantic knowledge. Further, the results have implications for the trajectories of vocabulary development in late talkers-children with small vocabulary sizes for their age. Beyond delayed word learning, these children are primed to process and learn different kinds of words, which may have cascading effects on later language learning. (PsycInfo Database Record (c) 2026 APA, all rights reserved).