[Author Login]
Title:Directionality in Nkore-Kiga sibilant harmony: arbitrary or emergent?
Authors:Will Bennett, Douglas Pulleyblank
Comment:manuscript; presented at ABC<->C conference in 2014
Length:20 pp
Abstract:The purpose of this paper is to discuss a problem related to directionality in the theory of Agreement By Correspondence, and propose a simple solution: the problem does not actually arise in the known case that motivates it. The puzzle we focus on relates to sibilant harmony in Nkore-Kiga, a Bantu language spoken primarily in Uganda. Previous work on Nkore-Kiga reports three generalisations:
(a) Anteriority in sibilants is normally allophonic, conditioned by the following vowel
(b) Sibilants deviate from the normal allophonic pattern due to consonant harmony within the stem
(c) Harmony operates strictly right-to-left: the rightmost sibilant is allophonically determined, and other sibilants agree to match it

These three generalisations, taken together, yield a pattern of consonant harmony that defies explanation both in an Agreement By Correspondence framework (Hansson 2001; Rose & Walker 2004) - and more generally in frameworks where assimilation is driven by a need to have agreement or avoid disagreement (Smolensky 1993; Baković 2000; Pulleyblank 2002). We show that on careful re-consideration of the data, the apparent problematicity of Kiga breaks down. In our analysis, right-to-left directionality emerges from a combination of (i) a positional faithfulness constraint for the rightmost member of a correspondence class (CC-Anchor), and (ii) morpheme realization effects that pin down the rightmost sibilant in cases where it is unfaithful.
Type:Paper/tech report
Area/Keywords:phonology, sibilants, ABC, harmony
Article:Version 1