Lexical Knowledge Enhances Consistency in Speech Categorization

Speech categorization is a gateway for downstream language processes. Recent work using the Visual Analog Scaling (VAS) task underscores the critical role of categorization consistency (trial-by-trial response variability around the mean response function) as a critical predictor of real-world outcomes such as language and reading abilities. Yet, the mechanisms that contribute to categorization consistency remain unknown. One hypothesis is that higher-level linguistic factors, such as lexical knowledge, may stabilize the percept by cleaning up lower-level perceptual noise. The first aim of this study was to test this hypothesis by examining whether categorization consistency is modulated by lexicality (word vs. nonword). Forty-eight adult American English listeners completed a VAS task involving both word (e.g., batch-patch) and matched nonword (e.g., bazg-pazg) continua. Listeners’ categorization consistency for the word continua was significantly higher than for the nonword continua. This suggests that categorization consistency is, indeed, affected by higher-level linguistic factors. The second aim was to investigate whether individuals’ broader language abilities influence their reliance on lexical information during speech categorization. Although individuals with greater language ability showed more consistent categorization, language scores did not modulate the lexical effect on categorization consistency. Together, these findings demonstrate the roles of top-down knowledge and language knowledge in stabilizing speech categorization.