Constraints on patterns of primary and secondary stress

This dissertation examines the interaction of various phonological phenomena with stress assignment. In some languages primary and secondary stresses behave identically or symmetrically with respect to a particular process. However, in other languages, only primary stress undergoes the process while secondary stress does not. In these languages, stress assignment is said to be asymmetrical. The goals of this study are two-fold. The first is empirical in nature. A cross-linguistic comparison reveals a typology of languages that exhibit symmetrical and asymmetrical stress patterns. Special emphasis is placed on those languages that demonstrate asymmetries in the behavior of primary and secondary stresses with respect to a wide variety of different phonological phenomena. The second goal is theoretical, analyzing these languages within the constraint-based framework of Optimality Theory. It is shown that asymmetrical stress patterns can be accounted for by referring to constraints that are specific to primary stress. A crucial assumption of this proposal is that constraints may not refer exclusively to secondary stress. Ranking a primary-stress-specific constraint in a stringency relation above a general stress constraint, with an antagonistic constraint ranked intermediately between them, yields an asymmetrical pattern. Due to the nature of the stringency relation – in which violation of the specific constraint implies violation of the general constraint, but not vice versa – there is no ranking of these constraints that will yield a pattern in which a phonological process applies only in secondary stressed syllables. This is a desirable consequence, since, with respect to certain phonological processes – including non finality effects, stressed syllable lengthening, and stress-driven sonority – such patterns are unattested. However, with respect to other phonological processes – e.g., quantity-sensitivity and sonority-driven stress – this type of asymmetrical pattern is attested. It is proposed that the difference between those processes that can apply only in secondary stressed syllables and those that cannot rests in whether stress assignment is process-driven or whether the process is stress-driven. This fundamental dichotomy predicts when such an asymmetrical pattern will be attested and when it will not.Ph.D.This research was funded by the Indiana University College of Arts and Sciences as well as by grants from the National Institutes of Health DC00012, DC01694 to Indiana University