OBSTRUENT-EPENTHESIS AND THE UNITY OF PHONOLOGICAL FEATURES

Burling (1966) proposed an unusual change in which stops /t/ and /k/ w ere added after -i and *-u in the history of the Tibeto-Burman langua ge, Maru. Miller (1970) took Burling to task for what he called his 't heory of the spontaneous generation of final stops'. However, a simila r change involving the spirants /s/ and /x/ is found in certain Austro nesian languages of the Land Dayak group. In both cases the epenthetic obstruent is word-final, and hence increases the markedness of the sy llable type. In addition, the added segment agrees in backness with th e preceding vowel, suggesting a non-arbitrary correlation between vowe l and consonant features which goes beyond the correlation commonly no ted in such assimilatory changes as palatalization. Although the affin ity of high back vowels for velar consonants is theoretically expected in the now traditional feature framework of Chomsky and Halle (1968), the affinity of high front vowels for dental consonants is not. This problem disappears if we adopt a suggestion made repeatedly in recent years that the feature [coronal], which was originally restricted to c onsonants, be extended to vowels.