The articulation of consonants in Kinyarwanda’s sibilant harmony

Abstract

Kinyarwanda’s sibilant harmony causes alveolar /s, z/ to become retroflex when preceding a retroflex fricative within a stem. Intervening coronal stops block sibilant harmony, but bilabial and velar consonants are transparent. This study investigates the production of Kinyarwanda’s sibilant fricatives and also examines transparent and opaque consonants in the system. Articulatory kinematic data were collected for a native speaker of Kinyarwanda using electromagnetic articulography. This allowed calculation of the mean angle for receivers affixed to the tongue tip and blade over the target consonant intervals. Average mean angle reliably and robustly distinguished alveolar and retroflex fricatives, with alveolars showing a lower tip relative to blade. No significant difference in mean angle was found for /t/ in contexts where it blocked retroflex sibilant harmony versus ones where it preceded an alveolar fricative, confirming that /t/ does not participate in the harmony. However, in contexts where /m/ and /k/ are perceived as transparent to sibilant harmony, their mean tip-blade angle was significantly different from contexts where harmony did not occur. Furthermore, mean angle during transparent /m/ showed no significant difference from mean angle during retroflex fricatives, suggesting that tongue tip-blade angle conducts strongly and systematically through transparent consonants but without perceptible effect. [Work supported by NIH, USC URP.]