Development of a new vowel feature from coarticulation: biomechanical modeling of rhotic vowels in Kalasha

Abstract

Coarticulation is an important source of new phonological contrasts. When speakers interpret effects such as nasalization, glottalization, and rhoticization as an inherent property of a vowel, a new phonological contrast is born. Studying this process directly is challenging because most vowel systems are stable and phonological change likely follows a long transitional period in which coarticulation is conventionalized beyond its mechanical basis. We examine the development of a new vowel feature by focusing on the emergence of rhotic vowels in Kalasha, an endangered Dardic (Indo-Aryan) language, using biomechanical and acoustic modeling to provide a baseline of pure rhotic coarticulation. Several features of the Kalasha rhotic vowel system are not predicted from combining muscle activation for non-rhotic vowels and bunched and retroflex approximants, including that rhotic back vowels are produced with tongue body fronting (shifting the backness contrast to principally a rounding contrast). We find that synthesized vowels that are about 30% plain vowel and 70% rhotic are optimal (i.e., they best approximate observed rhotic vowels and also balance the acoustic separation among rhotic vowels with the separation from their non-rhotic counterparts). Otherwise, dispersion is not generally observed, but the vowel that is most vulnerable to merger differs most from what would be expected from coarticulation alone.