Previous work has established that nasalance (the ratio of nasal to oral + nasal acoustic energy) can be reliably estimated from far-field speech audio, and that this measure provides a reliable index of time-varying velopharyngeal port (VP) opening (Siriwardena et al., 2024). Most languages exploit coupling the nasal tract by positioning the VP in stop contrasts, and about 25% of the WALS database language sample also have contrastive nasality for vowels. While these contrasts are generally binary, with either fully closed or fully coupled VP posture, partial assimilatory nasalization is common in vowel contexts preceding or following nasals in languages where vowel nasality is not phonemic. However, because of the partial nature of nasalization of this type, quantifying the timing and extent of its underlying VP opening gesture is problematic. Here we show that the continuous estimate of nasalance recovered from audio by a trained speech inversion system reliably distinguishes the oral/nasal phonemic vowel contrasts of French, quantifies the partial nasalization common in allophones of American English /ae/, and provides the time course of anticipatory nasalization in vowels preceding nasal stops. Applications to understudied languages and speech pathologies involving VP insufficiencies will be discussed. Work supported by NSF.
Tabatabaee et al. (Wed,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: