How cognitive salience and cue frequency shape grammar: evidence from animacy

Of all the conceptual distinctions that humans can detect in their environment, only a limited set are encoded in grammars, with striking cross-linguistic consistency. Attributes such as numerosity,animacy, and its subdivisions (e.g., sex) recur across languages, structuring grammatical systems in similar ways. This raises the question of whether human cognition is intrinsically biased towardlearning these contrasts and using them to organize grammar, or whether their presence can be explained by usage and cue frequency in experience. We address this question using categorization and artificial lexicon learning experiments that contrast animacy with color, a perceptually distinct attribute that is never grammaticalized. In a non-linguistic task (Exp 0), both attributes were spontaneously selected as categorization cues. Novel grammatical systems mapping exclusively onto animacy (Exp 1) or color (Exp 2) were learned equally well. In Exp 3, where cues to the two attributes were equally frequent and ambiguously related to learned words, participants selected either contrast with comparable frequency. Crucially, the two semantic oppositions interacted, with a tendency of color distinction to be encoded more robustly within the animate category. This pattern mirrors cross-linguistic tendencies, where higher-order animacy categories host richer grammatical specifications.These observations reveal an implicit bias in emergent grammatical systems, in which animacy functions as a privileged organizing dimension. We suggest that this bias arises from the salience andreal-life frequency of animacy, which provide a reliable foundation for communicative and processing efficiency in grammar and in the broader architecture of language.