|Title:||Voicing contrast: licensed by prosody or licensed by cue?|
|Comment:||Paper delivered at 13th Manchester Phonology Meeting, May 2005|
|Abstract:||In recent phonological work two approaches to neutralization have been developed, labelled licensing by prosody and licensing by cue (Steriade 1997). The licensing-by-prosody approach is developed by Beckman (1998) and Lombardi (1999, 2001). The central element is the universal difference between onsets and codas when it comes to realizing consonantal contrasts. Contrasts that may be realized in onsets may be neutralized in codas.
The licensing by-cue-approach is developed by Steriade (1997, 1999). This approach bases constraints licensing specific features not just on hierarchies derived from cross-linguistic typologies, but more specifically on differences in perceptual cues to be found in different phonetic contexts.
The licensing-by-cue approach is attractive since it seeks to explain the differential distribution of phonetic properties such as voicing directly in terms of the articulatory and perceptual characteristics of different phonetic environments. Licensing by prosody appeals rather to elements of prosodic organization --specifically, syllable structure-- which is held to be psychologically real, but only indirectly manifest in the stream of speech. Steriade asks (1997: 50) 'what would count as genuine evidence for syllable-final devoicing?' She goes on: 'The simple answer is: any system that allows us to compare voicing maintenance in onset O[bstruent]R [=sonorant] sequences with voicing neutralization in heterosyllabic O.R. Thus the hypothetical language ... distinguishes voiced obstruents in the OR sequences functioning as onsets, but neutralizes voicing in every other obstruent-C sequence, including in heterosyllabic O[.]R.'
I show that Catalan is a language that provides such evidence, as Steriade suspected. The conclusion is that the theoretical option of licensing by prosody, in the case of voicing contrasts, must be retained.