Ewe vowel harmony: implications for theories of underspecification

Deklu, Gladstone (2021) Ewe vowel harmony: implications for theories of underspecification. Doctoral (PhD) thesis, Memorial University of Newfoundland.

[English] PDF - Accepted Version Available under License - The author retains copyright ownership and moral rights in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission. Download (3MB)

Abstract

This thesis examines vowel harmony in Ewe, with a focus on the northern dialect. I show that harmony patterns seen in the northern dialect differ from what has been proposed as a general analysis for the language by Clements (1974). The patterns seen in the northern dialect highlight asymmetries in the behaviour of the vowels /a/ and /ɛ/. These vowels are triggers of harmony in some contexts (i.e., Low harmony), whereas in others, they are targets of harmony (i.e., Height harmony for /ɛ/ and Place harmony for /a/). I argue that this asymmetry can be explained by considering the phonological representation of these vowels as well as relying on the mechanisms of lexical phonology and morphology. I then situate my analysis within Stratal Optimality Theory. Regarding the phonological representation of vowels, I follow Reiss (2017) and propose that contrast is not relevant in the representation of vowels in Ewe; these vowels must be specified for features based on phonological activity. Given this, their specifications reflect what phonological processes each vowel triggers in the language. I provide further evidence supporting this analysis by highlighting the failures of theories of representation that rely on contrast in the face of Ewe data. As part of this discussion, I explore both Radical and Contrastive (under-)specification as well as the Contrastive Hierarchy. I propose that Height and Place harmony are lexical processes, and that Low harmony is a post-lexical process. This proposal is based on the characteristics displayed by these processes, which I also consider in light of cross-linguistic evidence. Within Stratal OT, I assume a strong version of Richness of the Base and argue that underlying representations cannot be restricted by the grammar. Thus, any combination of features is allowed in the underlying representation. Through the interactions of violable constraints, I derive vowel feature specifications based on activity at the stem level of derivation. I observe that constraints that are antagonistic to phonetic richness are ranked higher at this level of derivation, while those favouring it are ranked lowly. The output of the stem level then feeds the word level of derivation, where the lexical processes take place. At this level, the grammar preserves features through a high ranking of faithfulness constraints. Thus, segments are not further impoverished at this level of derivation. At the phrasal level, where Low harmony applies, different constraint rankings capture the observation that /a/ and /ɛ/ are not targets of harmony, and that /ɛ/ is a trigger of harmony.

Actions (login required)

View Item