Edward Sapir Lecture

Donca Steriade (MIT)

"Units of representation for linguistic rhythm"

Thursday, 6 August 2009, 7:30 pm
2050 Valley Life Sciences Building

In the last decade, the role of the syllable as a conditioning factor in segmental processes has been questioned: e.g. final obstruents devoice in many languages not because they are in the coda but because the word-final position deprives them of key auditory correlates to the voicing contrast (Steriade 1999). If the syllable plays no identifiable role in licensing features, and if its use in production or perception is also debatable (Schiller et al. 2002; Content et al. 2002), one can take a second look at its function as a unit in the representation of linguistic rhythm.

In this talk I argue for transferring the rhythmic uses of the syllable to the vowel-to-vowel interval. The latter is a constituent that begins with a nucleus and includes all the segments that separate it from the next nucleus or from the end of the word or larger domain. For instance, the word [sεgmənt] is divided into syllables (by English speakers) as [[sεg][mənt]]; the interval structure for the same word is [s[εgm][ənt]], two VCC intervals prefixed by a word-level onset.

Syllables and intervals are similar in being one-nucleus units; their weight is similarly determined, roughly as a function of their duration. The critical difference between them is that interval boundaries can be detected as the boundary between a C and a V, whereas locating syllable boundaries is a more complex and variable process. Further differences emerge when we consider the details of weight categorization (e.g. how different C-clusters affect the weight of preceding nuclei; or whether medial and final VC(C) strings carry the same weight), the relation between intuitions of syllable division and quantity judgments, and finally the structure of prosodic constituents that go beyond the computation of rhythm. All these differences will support the adoption of intervals.

The talk touches on all these lines of argument but the main focus will be on the role of intervals in the characterization of rhyming domains, the strings required to match in rhyming lines, e.g. sínister-mínister. The structure of rhyming domains varies substantially across poetic traditions but a constrained typology of these units emerges from a set of constraints that mention only intervals, not syllables or subsyllabic units.