This paper, in contrast, argues for a re-evaluation of representations in segmental phonology, offering a model of Feature Geometry within the framework of OT. The objective behind this move is twofold. First, theories of strict locality are seen as empirically inadequate and formally problematic. Second, I will propose that an integration of Feature Geometry and Autosegmental Phonology into OT yields an empirically more adequate and formally more elegant model of segment interaction. Surface-based theories run into numerous empirical problems. The so-called "bottleneck effect" (Ní Chiosáin and Padgett 2000) prohibits spreading of consonantal features onto vowels. Vocalic assimilation to consonants is widely attested, however (see e.g. Hume 1992, 1996). Long-distance assimilation cannot be modelled either, and recourse has to be taken to additional processes of segmental correspondence (Rose and Walker 2001), thus yielding two distinct types of segment interaction. In addition, it is unclear how tone-spreading phenomena could be explained in a strictly localist model where spreading cannot proceed between tone-bearing units but affects intervening material as well. Formally, assimilation is modelled by AGREE constraints which demand that neighbouring material be identical. The exact nature of this class of constraints remains obscure, however. How is it different from correspondence? How are agreement domains set up computationally? Moreover, assimilation is seen as an active process which needs to be enforced independently (by agreement constraints) and not as a result of articulatory or structural efficiency (which could be expressed by markedness constraints). In a sense, assimilation is thus computationally effortful. Besides, such a model does not offer a unified analysis of assimilation and dissimilation. While assimilation results from agreement constraints, dissimilation needs to be modelled differently, despite the close relationship between the two (cf. Odden 1994). Due to all these problems, I propose a different model of segment interaction which incorporates autosegments and Feature Geometry into OT. As Feature Geometry provides an accurate description of attested and unattested interactions, I consider it a primitive of UG which functions as a filter on GEN, the generator function in OT: Only well-formed candidates are generated; the EVALuator function then chooses the geometrical configuration which is most harmonic w.r.t. the constraint ranking. Apart from limiting the number of possible candidates (only well-formed candidates enter the competition), this move has a number of consequences. Most importantly, a set of constraints on geometric structure is needed. Many of these constraints can be generalized from existing constraints on prosodic structure, such as PARSE(Cat). As representations are widely accepted in the realm of prosody (syllables, feet etc.), this proposal only extends them to the subsegmental level. These constraints are supplemented by a generalized class of constraints on node b...
This paper looks at Dahl's Law, a voicing dissimilation process found in a number of Bantu languages, in Kitharaka, and argues that it is best analysed within a framework of minimal (contrastive) feature specifications. We show that the standard account of [±voice] dissimilation runs into a number of problems in Kitharaka and propose a new analysis, couched within the framework of the Parallel Structures Model of Feature Geometry (Morén 2003;2006) and Optimality Theory, thereby also addressing the question of the division of labour between constraints and representations. The analysis shows that it is crucial to look at the whole system of phonological oppositions and natural classes in Kitharaka to understand how the process works, ultimately also using loanwords to glean crucial insight into how the phoneme system of Kitharaka is organised.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.