Skip to main content
eScholarship
Open Access Publications from the University of California

UCLA

UCLA Electronic Theses and Dissertations bannerUCLA

Evidence for base-driven alternation in Tgdaya Seediq

Abstract

Standard approaches to morphophonological analysis of paradigms require positing URs from which surface contrasts can be derived. Often, URs are ‘cobbled’, in that they combine information from multiple forms of the paradigm, and do not necessarily correspond to any single surface form (Kenstowicz and Kisseberth, 1977: 33). In contrast, Albright (2002, et seq.) argues for a single surface base approach, where the UR must be based on the surface forms of one slot in the paradigm. In Tgdaya Seediq (Austronesian: Taiwan), all forms of a paradigm (non-suffixed vs. suffixed) suffer from some loss of contrasts, making it a good test case for comparing between the two theories of morphophonology. In this thesis, I conduct an empirical study of Seediq verbal alternations, and find support for Albright’s surface base hypothesis.

This thesis consists of three parts: First, based on a corpus on 340 Seediq verb paradigms, I describe the patterns of alternation in Seediq, and show how asymmetries in the data make it so that the suffixed forms of a paradigm are highly predictable from the non-suffixed forms. In the second part, I use a computational rule-based model to confirm this asymmetry, and show that such an asymmetry is expected under Albright’s single surface-base hypothesis, but unexpected under the traditional cobbled UR approach. Finally, based on these results, I develop a theoretical surface-base model of Seediq verbal alternations, which takes surface non-suffixed forms as inputs. The model uses Zuraw’s (2000; 2010) dual listing framework to model speaker intuitions, and is implemented in Maximum Entropy Harmonic Grammar (Goldwater and Johnson, 2003), a stochastic variant of Optimality Theory (Smolensky, 1986; Prince and Smolensky, 1993).

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View