Introduction

The hypothesis that prosodic mechanisms can compute null vowels or syllables, on which our analysis is based, is comparable to the rather well established claim that syntactic mechanisms can detect empty categories. Like the latter, it is in line with the general thesis that mental representation is rich and abstract, and has properties that elude superficial observation.

Variants of the null-vowel hypothesis have been proposed in the past. As we have already seen, one variant, in the form of a final e that deletes in the course of the derivation, was proposed in SPE, to regularize the stress pattern of words like giráffe, or ellípse, eclípse and others (see esp. pp. 45, 161). SPE further proposed an underlying final glide for words like fréquence, parallel to the y of frequency, as a way to account for the spirantization of t (compare frequen[t]). Spirantization would of course also be induced by the above final e, hence accounting for ellíp[s]e, eclíp[s]e, versus ellíp[t]ic,eclíp[t]ic. Ross (1972, p. 270) extended the final e hypothesis of SPE to all penultimately stressed verbs, such as devélop(e), exámine, whose systematically short stressed vowels contrast with the often long ones of nouns, such as pí:rate, bó:nus. In his analysis, the final e served to reduce the short vowel of the verbs to that of ordinary cases of “trisyllabic shortening.” While bearing some similarity to those early proposals, our view more closely resembles the “zero syllable” thesis of Giegerich (1981, 1985), especially as pursued in Iverson (1990).