Causal structure in creaky voice acoustics: Evidence for f0 mediation

This project examines why creaky voice varies across speakers by investigating the causal relationships among social factors, f0, and multiple acoustic indicators of creaky phonation. Creaky voice is typically characterized by low fundamental frequency (f0), irregular vocal fold vibration, and reduced airflow, and previous work (including my earlier studies) has found greater acoustic creakiness in men and older speakers. However, because f0 itself differs systematically with gender and age due to both physiological and social factors, it remains unclear whether these group differences reflect direct effects or are largely driven by pitch.

Using spontaneous speech from Canadian English–French bilingual speakers, this study applies causal mediation analysis to quantify how much of the observed gender and age differences in creaky voice are explained indirectly through f0 versus directly through other mechanisms. Across several acoustic measures of creakiness, results show substantial, but incomplete mediation by f0. Some indicators, such as unreliable f0 tracking, are almost entirely explained by differences in f0, while others (including spectral tilt and noise measures) show only partial mediation.

These findings support a primarily physiological account of creaky voice variation, consistent with anatomical and aging-related influences on vocal fold vibration. At the same time, the incomplete mediation indicates that additional factors beyond f0 also contribute, and that the mechanisms underlying gender and age differences are not identical. By disentangling direct and indirect pathways, this study clarifies how commonly used acoustic measures relate to one another and provides a more precise framework for interpreting creaky voice variation in sociophonetic research.