My starting point is the theory of CV phonology as developed by Clements and Keyser (1981, 1983) which, in turn, is one of a number of theories of syllable structure that have been proposed during the past decade to replace the earlier linear concept of generative phonology. These theories have in common that the syllable is recognized as a hierarchical unit in phonological representation and not just a concatenation of segments. Kahn (1976), whose dissertation on English first persuaded generative linguists of the need to depart from the linear model, proposed a tier of syllable nodes (here symbolized as $) linked directly to the segments — consonants and vowels — of the traditional analysis, as in the representation of the word Jennifer in (1) (taken from Clements and Keyser 1983:3).