Speech rate effects in Russian onset clusters are modulated by frequency, but not auditory cue robustness

- The interaction between speech rate, phonotactic preference and frequency is probed.
- Auditory cue robustness does not predict cluster behavior under rate changes.
- Low, but not high frequency clusters bar increasing coarticulation with rate.
- Speech rate changes are mostly carried by the internal cluster member.

This study presents data on the durational and timing characteristics of Russian onset clusters and their change as a function of speaking rate. The focus is on Russian due to it being known to have relatively less consonant overlap compared to languages like English and due to its unusual range of consonant clusters. Using articulography, we investigate whether these characteristics have implications for the flexibility of clusters under speech rate changes. In particular we ask whether a cluster's signal modulation profile, taken as an index of auditory recoverability, predicts the degree to which the overlap pattern of a cluster changes with rate. Previous research suggests that stop + stop clusters may be less susceptible to rate change than other, auditorily more robust clusters within the same language. Moreover, even though frequency and phonotactic preference are usually closely aligned, Russian also allows us to probe frequency effects on cluster timing, since for our data these factors are dissociated to a certain degree. Results show that both duration and relative timing of the consonants in a cluster are subject to change. Speech rate effects do not scale uniformly throughout the cluster but are carried predominantly by the constriction formation duration of C2. Clusters show a decreasing rate effect from high to low frequency clusters. Grouping clusters according to their assumed perceptual robustness does not lead to a clear result. We discuss these findings in the context of models of durational control of speech production.