Asymmetrical effects of speaking rate on Japanese vowels and consonants

Previous research suggests that vowels are more responsive to speaking rate variation than consonants in production. However, this tendency often derives from limited stimuli or broad comparisons that may overlook differences among subcategories like stops and fricatives. Additionally, how listeners adapt to the asymmetrical effects of speaking rate remains largely unknown. We conducted an online experiment with production and perception tasks using Japanese, where many segments have phonemic length. We examined whether different segmental targets, specifically five vowels (/i, e, a, o, u/) and five consonants (t, k, s, m, n/), differ in their sensitivity to speaking rate changes. In the production task, participants produced minimal pairs of nonce words contrasting in length (e.g., /kempina/, /kempiːna/) within a carrier sentence both at fast and slow speaking rates, following production prompts. In the perception task, they categorized duration continua embedded in fast and slow speech as phonemically short or long. Converging evidence suggests a distinction between vowels and stops versus fricatives and nasals, with the former group exhibiting greater sensitivity to speaking rate changes in production. This production pattern was also generally reflected in the perception results. We discuss potential reasons behind this asymmetry and explore its implications.