[Author Login]
Title:Morphological Haplology and Correspondence
Authors:Paul de Lacy
Comment:In Paul de Lacy & Anita Nowak (1998) "UMOP 24: Papers from the 25th Anniversary" GLSA Publications, pp.51-88
Abstract:Morphological Haplology and Correspondence

When the Arabic feminine singular morpheme /ta/ attaches to the verbal prefix /ta/ the resulting phonological form is not the expected *[tata],
but instead just [ta]: e.g. /ta/ + /ta/ + /kassaru/ -> [takassaru] it fem.sg) broke, cf *[tatakassaru]. This and other cases of morphological haplology raise two main questions:

· What is the nature of morphological haplology?
· What triggers morphological haplology?

Two proposals have been made regarding the first question: deletion and coalescence. In the deletion view, the phonological material of a haplologized affix has been deleted or - equivalently - has failed to be realised. In comparison, the coalescence view proposes that no phonological material has disappeared; instead, the two identical phonological strings have merged. This paper argues for the coalescence proposal and adduces evidence for this from Japanese and French.

The adoption of the coalescence view has an immediate explanation within Optimality Theory: the typical coalescence ranking of [Max-IO » Uniformity] (McCarthy & Prince 1995, Lamontagne & Rice 1995). It is demonstrated that the interaction of other constraints with this ranking accounts for the different types of haplology attested in natural language, including total and partial-identity haplology, coextensive haplology, stem-edge and non-local haplology, and reduplicative haplology.

The issue of what triggers haplology has also been given a great deal of attention. Recent works have advocated the use of a generalised OCP that
bans adjacent identical strings. The alternative proposed here simply employs a markedness constraint that militates against a relevant feature. This desire to avoid violating the markedness constraint conflicts with a need to realise underlying features, expressed by IDENT-F (McCarthy & Prince 1995). This tension effectively permits only identical strings to coalesce.
Type:Paper/tech report
Article:Version 1