Why Place and Voice Are Different: Constraint-Specific Alternations in Optimality Theory

Abstract

Introduction Both Place and Laryngeal features are often restricted in coda position. The laryngeal distinctions of voicing, aspiration, and glottalization are often neutralized to plain voiceless in coda position. The phonology of Place often shows the Coda Condition pattern of Ito (1986), according to which syllables cannot end in a consonant other than the nasal of a homorganic NC cluster or the first half of a geminate. This is analyzed by Ito as due to a constraint against singly linked Place in a coda. Both laryngeal neutralization and the alternations involving place have been seen as the result of similar restrictions: constraints that restrict features from appearing in coda position, whether these constraints are negative (as in Ito 1986) or positive (as in Lombardi 1991, 1995a, and the Alignment analysis of the Coda Condition in Ito and Mester 1994). However, the types of alternations that languages exhibit in association with these coda restrictions differ. Although both Laryngeal Constraint and Coda Condition languages may show codas in assimilated clusters escaping the constraint, beyond that there seem to be more differences than similarities in the behavior of Place and Voice. Both types of features may be subject to neutralization. But coda restrictions on Place may result in either epenthesis or deletion, while forbidden coda Voice never triggers these alternations. In a theory where phonological rules specify both context and change, as in SPE and much work following it, it is not possible to account for this asymmetry of patterns except by stipulation. For example, an autosegmental rule deleting a voiced consonant from coda position is as well formed as a rule deleting a consonant with Place from the coda, even though the former is unattested. But since it is possible to write all the rules that do exist with the machinery of the theory, we can easily construct analyses of any given language, and the gaps in the crosslinguistic typology are not made obvious.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 140 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.