Abstract for Maria Gouskova's talk

Prosody, contrast, and the problem of syllables
Maria Gouskova (New York University)
March 22, 2007 (Thursday)
4:15 PM - ; Room 6417, The CUNY Graduate Center

Not all aspects of prosodic organization appear to be employed contrastively. It is uncontroversial that languages can contrast consonant and vowel length, stress, and tone, but there are no convincing cases of contrast in syllabification (e.g., [pat][ra] vs. [pa][tra]) or being at the edge of a foot. Rule-based approaches and Optimality Theory explain this gap in similar terms: in rule-based approaches, syllabification is assumed to be absent in underlying representations; instead, it is assigned by regular syllabification rules (Levin 1985, Bromberger and Halle 1989). In Optimality-Theoretic work, it is often said that there is no faithfulness to syllabification, so contrasts cannot arise (McCarthy 2003, inter alia). At the same time, faithfulness constraints have been proposed for other prosodic constituents, most notably the foot (Ito, Kitagawa, and Mester 1996, Benua 1997). Even faithfulness to syllabicity has occasionally been used to account for vowel-glide contrasts--an adaptation of the traditional rule-based analysis. This view of glide-vowel contrasts raises a problem for any theory: if vocoids can be contrastively specified for syllabicity, why can't other potentially syllabic segments? My solution to this problem is couched in a particular view of faithfulness in Optimality Theory. I will argue that faithfulness constraints may refer only to features, root nodes (segments), and autosegments (tones, morae, stress grid marks). I will show how the theory explains the behavior of glides in Russian, where they are contrastive, and in Polish, where they are not. I propose that the difference between vowels and glides is featural, for which we have evidence from a variety of segmental processes (Levi 2004, Padgett 2006). Since the larger goal of this theory is to limit the range of possible contrasts to the features and autosegments available in the Universal Grammar, it faces some challenges in dealing with other prosodic contrasts, which I will also discuss.