A discussion of two aspects of Hindi phonology, schwa deletion and vowel nasalization, compares two theories concerning the processes behind these phenomena. A non-linear analysis is compared with a more traditional, linear notation. Results indicate that in most cases, both sets of rules work equally well but in some, the linear explanation is less "ad hoc" and more insightful. Certain assumptions inherent in the non-linear analysis are also called into question. (MSE)