Coronality and vocalic underspecification

The observation that a close relationship exists between front vowels and coronal consonants dates back to Jakobson, Fant & Halle (1952) and Clements (1976) and the proponents of the feature [grave] (Hyman 1973; Vago 1976). This topic has received revived interest in recent years in the work of Mester & Ito (1989), Pulleyblank (1989), Broselow & Niyondagara (1990), Cheng (1990), Clements (1990), Lahiri & Evers (1991) and Hume (1992) among others. These authors argue that front vowels and coronal consonants should be represented by the same articulator node, Coronal, although additional specifications may differ from analysis to analysis. At the same time, the topic of the underspecification of coronal consonants has received considerable attention (Kiparsky 1985; Avery & Rice 1989; Paradis & Prunet 1989, 1990 and contributors to Paradis & Prunet 1991 b). The argument has been made that, under most circumstances, the feature [coronal] is missing from the underlying representation of coronal consonants, being inserted only for purposes of phonetic implementation. The logical follow-up question is then, if coronals are the least marked and unspecified class of consonants, does the same generalization hold true for front vowels, if they, too, are represented with a Coronal articulator? In this paper, I address this question, examining the role of front vowels in vowel sandhi, harmony processes and epenthesis. I adopt an inventory-driven model of vocalic specification, along the lines of Avery & Rice (1989). This approach accounts for the asymmetric patterning of front vowels in some languages, and their unremarkable behaviour in other languages. In section 1, I summarize some of the arguments for the two areas of research mentioned above and I discuss my approach to vocalic specification. In sections 2, 3 and 4 I examine the role of front vowels in some cases of vowel assimilation, vowel harmony and epenthesis respectively.