Part 1: Units of rhythm: intervals and syllables
Part 2: OT with Ranked Violations
Instructors: Edward Flemming & Donca Steriade
Time: Friday 12-3PM
Place: 32-D461
Course website: here
Part 1. Units of rhythm.
Stress and meter operate on rhythmic units that consist of a nucleus plus some neighboring consonants. These units are currently assumed to be syllables. In the first part of the course, we compare syllables with a different unit, the Vowel-to-Vowel interval. An interval is contains the nucleus plus any segments following it, up to the next nucleus or the end of the domain. Phenomena that depend on counting rhythmic units get identical accounts under both analyses, but the location of unit edges is predicted to differ. This difference has a wide range of consequences for the analysis of quantitative phenomena, for rhyming and alliteration, for segmental phonology and prosodic morphology. We will explore as many of these differences as we can fit in 6 sessions.
Part 2: Optimality Theory with Ranked Violations
In the second half of the semester we will explore a new model of constraint interaction in Optimality Theory. The main motivation for pursuing this variant is to provide a better analysis of scalar constraints in phonology.
Standard OT adopts a mechanism based on constraint ranking to resolve constraint conflicts: in cases of conflict the higher-ranked constraint prevails. In this framework, there can be no compromise between scalar constraints. This is potentially problematic because trade-offs appear to be common. For example, preferences to minimize effort and maximize the distinctiveness of contrasts can be given natural formalizations in terms of scalar constraints, but in an OT setting, one constraint has to dominate the other so it is only possible to derive maximum distinctiveness or minimum effort where we actually observe compromises between the two: moderate distinctiveness in exchange for moderate effort. In standard OT, compromise can only be accommodated by decomposing scalar constraints into multiple sub-constraints.
In the new model of constraint interaction, constraints are weighted and each violation of a constraint incurs a cost equal to the magnitude of the violation multiplied by the constraint weight. Evaluation then proceeds much as in standard OT except that ranking of violations is based on their costs rather than being fixed for a given constraint, so a large violation of scalar constraint C1 can rank above a violation of constraint C2 while a lesser violation of C1 ranks below a violation of C2. This makes it possible to derive trade-offs between constraints.
We will compare these lines of analysis in case studies of markedness and faithfulness scales, including those pertaining to sonority, distinctiveness, and P-map correspondence.