Speaker: Rachel Walker (University of Southern California)
Title: Locality, Scope, and Sour Grapes Effects in Vowel Harmony
Date: Friday, October 22, 2010
Time: 3:30-5:00PM
Place: 32-155 (PLEASE NOTE NEW ROOM)
Harmony that operates only when it reaches a specific destination, such as a word boundary or other landmark, presents a “sour grapes” effect (Padgett 1995). Unbounded harmony does not show this characteristic, because it can be partial in the domain in which it operates (Wilson 2003). This property of unbounded harmony presents an important test for theories of harmony in different frameworks. Wilson has characterized unbounded harmony as, in effect, local iterative spreading, a phenomenon that is challenging to replicate in the classic version of Optimality Theory but straightforward to generate using autosegmental rules.
This paper investigates the apparent benefits of local iterative spreading through an exploration of systems with nonlocal target scope: an unbounded round harmony in Baiyinna Orochen and a bounded height harmony in the central Veneto dialect. These patterns each show nonlocal target scope but harmony proceeds locally, that is, the operation of harmony between adjacent syllables can depend on information about vowels in nonadjacent syllables. These patterns point to a need to distinguish locality of assimilation from locality for target scope, a separation that presents difficulty for local iterative spreading rules as well as for any constraint that enforces harmony only over adjacent trigger-target pairs. These phenomena call for harmony constraints with nonlocal scope, readily implemented using the global evaluation that is intrinsic to OT.
The bounded pattern of central Veneto, which singles out a stressed target, also bears on a serial version of OT. In this system, unstressed vowels on the path to a stressed vowel undergo harmony as incidental participants but not otherwise. This suggests a need for a fell-swoop derivation, where a stressed vowel target and an intervening unstressed vowel undergo harmony at once rather than in successive steps. However, a derivation of this kind is not generated in the standard theory of Harmonic Serialism (e.g. McCarthy 2008a,b, 2009).
In conclusion, the harmony systems under study highlight empirical advantages of global evaluation and harmony-driving constraints with nonlocal scope rather than a local iterative procedure for harmony. Although harmony-driving constraints with global scope predict certain unattested systems, approaches with only local scope are too restrictive, signaling the need for a fresh look at issues of typological overgeneration.