Toward a Nonderivational Account of Epenthesis and Exceptional Stress in Dakota

Recent work in Optimality Theory (Prince and Smolensky 1993, McCarthy and Prince 1993, to appear) has suggested that phonological regularities can be explained without reference to phonological rules, strictly in terms of constraints on representational wellformedness. A unique aspect of Optimality Theory (OT) is that candidates are evaluated in parallel: there is no serial derivation. Given this type of system, a question that immediately arises is whether empirical evidence that has been used to justify ordered sequences of rule applications or ordered lexical levels can be explained without the use of this type of derivational machinery, strictly in terms of constraints on representational wellformedness. This paper will attempt to answer this question by examining a set of epenthesis and stress facts from Dakota that have been presented (in e.g., Shaw 1976, 1985, Kiparsky 1986) as empirical evidence for a derivational phonology.1