The Perceptual Dimensions of Sonority-Driven Epenthesis

Vowel epenthesis often appears to preferentially target consonant clusters with rising sonority. One explanation for this is perceptual faithfulness (Fleischhacker, 2002; Steriade, 2006): rising sonority clusters are more susceptible to epenthesis because the perceptual distance between the underlying /C1-C2/ sequence and its correspondent output sequence [C1-V-C2] is small, thus incurring a smaller faithfulness cost. This raises the question of how to compute the perceptual distance between two sonority contours /C1-C2/ and [C1-V-C2] in terms of the sonority of C1, C2 and V. In this paper, I propose the metric Sonority Angle, the angle formed by the contours /C1-C2/ and [C1-V]. I apply it in analysing two case studies of sonority-driven epenthesis, Chaha and Irish. A comparison is made to another possible metric, Sonority Rise (Flemming, 2008), the ratio of the gradients of the two contours, as well as to Syllable Contact and Sonority Sequencing, which represent an alternative, markedness-based approach to the problem of sonority-driven epenthesis.