Vocalization and Intrusion of R in English

Abstract

Deletion and insertion of r in some dialects of English (Homer left vs. Wanda arrived) have been dealt with by many researchers (McCarthy 1991, 1993; Halle and Idsardi 1997). As pointed out by Halle and Idsardi, there are two non- trivial difficulties in analyzing r-phenomena. One is the arbitrariness of r-intrusion; that is, why only rs, not others, are inserted after a, J, and a. The other is the opacity problem resulting from the interaction of schwa insertion and r-deletion. If schwa insertion takes place between a high glide and a coda r, and the coda r is deleted afterwards, the result of applying these rules to the word fear I /fijr/ will be [fijəl In this form, the schwa insertion rule becomes opaque since there remains no surface environment for this rule. Such an opacity problem has been considered to be problematic in the standard Qptimality Theory (Prince and Smolensky 1993) that is surface-oriented. The primary aim of this article is to provide a unified way of resolving the above two problems. To this end, I adopt an alternative view of r-phenomena, in which r-deletion is treated as r-vocalization (weakening) and r-insertion as glide insertion. Based on these assumptions, a new OT analysis is proposed.