Logarithmic duration can eliminate speech-tempo effects: Evidence from Polish singleton and geminate consonants

This study investigates whether logarithmic temporal measures can account for the stability of the singleton-geminate consonant contrast across variations in speaking rate in Polish. Twenty native speakers produced non-words containing singleton and geminate plosives and fricatives embedded in a carrier sentence at six experimentally controlled tempos. Closure or frication duration of the target consonant and the average syllable duration of the surrounding utterance were measured and converted to logarithmic values. Regression and discriminant analyses were conducted to evaluate how well a linear combination of these two logarithmic measures predicts and categorizes singleton and geminate consonants. The results showed that singleton and geminate consonants formed parallel distributions across speaking rates in the logarithmic domain. The discriminant functions yielded low classification error rates, and regression models demonstrated stable relationships between segmental duration and overall speech rate. These findings suggest that a combination of logarithmic segmental and syllabic durations serves as an invariant acoustic cue to the singleton–geminate contrast in Polish. The results parallel earlier findings for Japanese, pointing to a potentially general role of logarithmic temporal organization in maintaining duration-based phonological contrasts across speaking-rate variation.