Is the mass–count distinction merely a linguistic issue, or is it coded in representations other than language? We hypothesized that a difference between mass and count properties should be observed even in absence of linguistic distinctions driven by the morphosyntactic context. We tested 5–6-year-old children’s ability to judge sentences with mass nouns (sand), count nouns (ring), and neutral nouns (i.e., those that appear in mass and count contexts with similar frequency; cake). Children refused neutral nouns embedded in uncountable morphosyntactic contexts, showing a preference for a count interpretation. This suggests that linguistic features alone are not sufficient to define the mass–count distinction. Additional analyses showed that children’s performance with mass—but not count—morphosyntax correlated with their performance in tasks concerning logical and conservation operations. Altogether, these results suggest that the processing of mass features is not more demanding than count features from a linguistic point of view; rather, mass features entail additional abstraction abilities.