On the representation of initial geminates

Abstract

Hume et al. (1997) argue that in Leti, an Austronesian language spoken on Leti Island off the coast of East Timor, geminate consonants are not moraic. In particular, they focus much of their attention on the word- initial geminates of Leti, which are syllable-initial when they begin a phrase. They argue that Leti initial geminates cannot be moraic and are instead best represented as in (1), with a single root node linked to two X- slots. (The example in (1) displays a syllable with an initial geminate; the vowel is moraic. The term root indicates the consonantal root node; the features under it are not indicated.)

While Hume et al. do not explicitly extend their discussion of the representation of initial geminates by examining relevant data in other languages, they do note (pp. 397–398) that ‘given the paucity of discussion in the phonological literature concerning syllable-initial geminates, the evidence from Leti is particularly important not only for further enriching our understanding of these segments but, in addition, for serving as a testing-ground for theories of prosodic structure and the representation of geminate consonants’. From this, one could postulate a strong position in which all initial geminates have the same, non-moraic, representation. The purpose of this squib is to argue against this strong position. While the evidence provided by Hume et al. against the moraic representation of Leti initial geminates is convincing, I present evidence in this paper showing that initial geminates are moraic in other languages. In § 2 I present data from Trukese previously discussed by Churchyard (1991), Hart (1991) and Davis & Torretta (1998) that provide a compelling case for the moraic representation of word-initial geminate consonants in that language. In § 3 I suggest that the different representations of word-initial geminates in Leti and Trukese are supported by the very different phonotactics of word-initial clusters found in the two languages. Finally, in § 4 I relate the discussion on initial geminates to the peculiar patterning of palatal segments in Italian, where palatals always surface as long except in phrase-initial position. I argue that the palatal segments in Italian are moraic even when surfacing in phrase-initial position. I conclude that initial geminates may be moraic in some languages but not in others.