Text

theses

This dissertation proposes a unified theory of Place Identity to model interactions of phonological place features along local, long-distance, and input/output dimensions. Empirical support comes from reduction and agreement processes involving complex, or multiple, place on a single segment---processes previously reported to be unattested. Identity constraints (Ident and Agree) are those that demand two segments be alike with respect to the presence or value of some feature, and are constructed here following a schema of Generalized Identity. The resulting constraints vary both in the relation of the segments in question (input/output correspondence, surface correspondence, or strict adjacency) and in the location of place features in the geometry (within-category or cross-category). Sets of stringently-defined constraints are built via Constraint Summation, a constraint-building operation that sums the violation profiles of the original constraints. The resulting system of place identity captures partial class effects (based on Padgett 2002) of place along all identity dimensions, while augmenting the observations in de Lacy 2006 on markedness reduction and inventory structure with respect to complex segments. Empirical investigation of complex segments reveals place behavior unattested with simple stops. In Ngbaka, place co-occurrence restrictions are an example of long-distance major place harmony, which requires place identity over surface correspondence in the Agreement by Correspondence framework (Rose & Walker 2004, a.o.). The patterns are supported by a statistical analysis of a newly-digitized Ngbaka dictionary. Additionally, in Vietnamese and Aghem, a back, round vowel causes a change in consonantal place on an adjacent consonant, forming a labial-velar stop in both instances. This otherwise unattested type of cross-category interaction provides additional evidence for a unified theory of place features (following Clements & Hume 1995), to which place identity constraints crucially refer. Lastly, complex segments shed a place feature when undergoing markedness reduction instead of reducing to the least marked place (cf. de Lacy 2006). These processes are supported by an empirical survey, and show that place identity must count each place disparity. The resulting grammatical system is powerful enough to determine targets of reduction for complex segments without additional representational devices while restrictive enough to not overgenerate patterns for simple place reduction.

ETD_8473

Ph.D.

Includes bibliographical references

by Nicholas Stephen Danis

doi:10.7282/T38055PH

ETD doctoral

The author owns the copyright to this work.