An acoustic investigation into the palatalization contrast in Russian fricatives

Distinct acoustic correlates of palatalization contrast in Russian fricatives are mainly reported for voiceless sibilants using static measures. This study adds to existing research on Russian palatalization by investigating the pairs: [v]-[vj], [z]-[zj], [s]-[sj], [S]-[C], using 18 acoustic correlates quantified dynamically within the frication noise, including spectral, temporal, and amplitude domains. The data consists of 9070 tokens recorded from 59 native Russian speakers (30 females, 29 males), aged 18–30. To detect distinct acoustic characteristics, Random Forests (RFs) were used, and the data was further analyzed by Generalized Additive Mixed-effects Models (GAMMs) to track dynamic changes over time. RFs identified six main discriminative features: dynamic amplitude between low and high frequencies, amplitude in low-frequency ranges, spectral peak and center of gravity, and the number of zero crossings. To track dynamic changes, these measures were subjected to GAMMs. Our results show that palatalization forms a supra-segmental articulatory feature leading to increased air pressure and energy in high frequencies relative to low frequencies, resulting in a more palatal-like realization, with alveolars realized as more back and post-alveolars as more front. Palatalized fricatives in Russian differ from non-palatalized ones, with most changes occurring within the fricative and marginally extending to its edges.