Untitled

22 The frameworkare not derived at all, but are listed as separate lexical items in the lexicon.This means that phonological regularities occurring among suchforms were once phonologically active processes, e.g. Velar Softening orTrisyllabic Laxing. Synchronically, however, they are no more than historicalrelics. In other words, ‘morphologically related forms which resembleeach other phonologically are not necessarily derived from a commonsource’ (Kaye 1995:313). It follows that the apparent regularitiesof the synthetically derived forms are at best represented as a case ofallomorphy.To sum up, the only morphological information available in phonologyis the analytic domain boundaries. Morphologically complex formswith the synthetic domain structure are equated with the morphologicallysingle underived forms. Finally, note that the distinction betweenthe synthetic and analytic domains brings to mind the distinction introducedin SPE, that is, morphological boundaries ‘+’ and word boundaries‘#’, where the former were assumed to be invisible to phonological processing.However, what distinguishes the Government Phonology accountfrom the previous ones is the postulation of domain final emptynuclei. Thus, in what follows we shall address the question concerningthe presence of domain final empty nuclei and why they are special.3.3.1. Domain final empty nucleiThe significant contribution of GP to phonological theory is the observationthat a word-final consonant does not function as a coda but ratherpatterns with internal onsets. Very briefly, it was pointed out that wordfinalconsonants behave differently from word-internal codas in severalrespects, e.g., they do not trigger Closed Syllable Shortening, they areusually extrametrical with respect to stress assignment, etc. Moreover,from the distributional point of view, word-final consonants pattern togetherwith word-internal onsets (Kaye 1990, Harris and Gussmann1998, 2002). In order to account for this fact, Kaye (1990)proposes the principle of Coda Licensing which narrows down the occurrenceof the coda to a very specific context, that is, it can appear only if itis licensed by the immediately following onset. This was one of the reasonswhy the coda was excluded from the GP constituent inventory. 7Furthermore, since in principle onsets must be licensed by the following7Note that ‘coda’ is shorthand for post-nuclear rhymal complement in GP.