Stability of Second Formant Transitions in Dysarthria: Effects of Articulation Rate

Abstract

The slope of second formant (F2) frequency transitions is an acoustic measure of the rate of change of vocal tract configuration. Previous research indicated reduced F2 slopes for speakers with dysarthria versus healthy controls (Kim, Kent, & Weismer, 2011; Kim, Weismer, Kent, & Duffy, 2009). Other studies suggested a strong correlation of F2 slope and speech intelligibility, indicating that F2 slope is a relatively easily obtained measure with the potential to serve as an index of speech severity (Yunusova, Green, Greenwood, Wang, Pattee, & Zinman, 2012; Martel-Sauvageau & Tjaden, 2017). Whilst acoustic measures usually include transition duration, extent, and slope derived from a small number of averaged values, thus far no studies have considered the stability of F2 transitions over repeated productions. Stability (inversely: variability) indices derived from kinematic and acoustic measures obtained from repeated productions at the sentence-level have been found to be valid indicators of speech motor involvement, and sensitive to different speaking demands, including changes in articulation rates (Anderson, Lowit, & Howell, 2008; Cummins, Lowit, & van Brenk, 2014; van Brenk & Lowit, 2012). The aim of this study was to investigate whether the quantification of variability of F2 transition metrics could be used to characterize dysarthria. Particularly, we investigated whether these repetition-to-repetition variability measures showed similar sensitivity compared to averaged F2 transition measures, and whether they could potentially reveal aspects of speech motor control deficits in dysarthria not fully captured by averaged measures. Methodology and Reporting Audio recordings were taken from 58 participants: 23 speakers with Parkinson’s disease and mild to moderate hypokinetic dysarthria, 9 speakers with various neurological diseases and mild to severe ataxic dysarthria, and 26 age-matched healthy control speakers. Speakers were asked to repeat the sentence “Tony knew you were lying in bed” approximately 20 times, at habitual, fast, and slow articulation rates. The operationally-defined, rising, F2 transition portion of the diphthong /aɪ/ in ‘lying’ was of interest. The following metrics were derived: Transition Extent (TE; in Hz), Transition Duration (TD; in ms), the derived global, Transition Slope (TE/TD; in Hz/ms), and Average Instantaneous Slope (averaged ΔHz/Δms for each data point). The means and coefficients of variation of these metrics were obtained across repetitions. Repetition-to-repetition variability in the shape of the operationally defined F2 transitions was obtained by applying a dynamic time warping algorithm to align the F2 transitions, with the amount of warping necessary for alignment as the primary metric of interest. We will report on the ability of these metrics to differentiate normal and disordered speaker groups as well as to differentiate among the various perceptually-defined dysarthrias, and evaluate the effects of different articulation rates. At the time of this submission, acoustic analyses are completed for approximately 75% of the data.

Frits van Brenk

External consultant in the Motor Speech Disorders Laboratory

My research and teaching interests include motor speech disorders, clinical linguistics, experimental psycholinguistics, speech science, and research methods.