Japanese and American 13-15-month-old infants distinguish between crossing a bounded ground (e.g., a street) versus an unbounded ground (e.g., a field). In English, the same verb – crossing – expresses both types. While language has been hypothesized to guide infants’ progression from language-general to language-specific event perception (Göksun et al., 2011), no prior studies examined this hypothesis. We presented toddlers who no longer perceive this Japanese distinction in events with novel spatial prepositions (N = 24) or nonlinguistic tones (N = 12) to label bounded versus unbounded grounds. Children presented with labels, but not tones, attended to the differences in ground categories by looking significantly longer to the novel ground type at test. This suggests that above and beyond the attention-getting function associated with non-linguistic auditory stimuli, language uniquely facilitates categorization of event components.