|Title:||Rhythmic Directionality by Positional Licensing|
|Comment:||Handout of presentation, Fifth HIL Phonology Conference (HILP 5), University of Potsdam, 11 January 2001.|
|Abstract:||Gaps in factorial typologies often diagnose imperfections in the constraint set. In this talk I will focus on two gaps in directional stress systems: the lack of leftward iambic systems and the lack of bidirectional systems in which rhythmic stresses run toward a secondary stress specified at an edge. Both gaps will be argued to originate from a single assumption, shared by current metrical models, that directionality involves non-local principles of foot distribution (directional footing in rule theory, Hayes 1980; alignment constraints under long-distance computation in OT).
I will start out from the observation that, in binary systems, directionality can be locally inferred from the position of a ternary interval in words with an odd number of syllables. (For example, rightward footing transpires from a ternary interval directly preceding the penultimate main stress, as in Piro 202020010.) This observation suggests that directionality can be broken down into two separate factors: rhythmically tight parsing, conventionally enforced by *LAPSE or PARSE-SYL; plus some licensing of local ternary intervals in specified positions in the domain (Kager 1994). Cross-linguistically, local ternary intervals seem restricted to two contexts: they are (i) adjacent to the primary stress (Garawa 100202020, Piro 202020010), or (ii) adjacent to the right edge (Pintupi 102020200). Local ternarity seems not to occur, however, at the left edge (as in leftward iambs *002020201), or adjacent to a secondary stress (as in bidirectional trochaic *200202010 or *102020020). In the new theory, foot position is controlled by local factors, rather than by long-distance orientation with respect to an edge.
Consequently, I propose to eliminate ALL-FT-X in favour of a set of constraints licensing rhythmically marked intervals (a lapse or an unparsed syllable) near the right edge or main stress. Both positions are natural licensors of marked prosodic properties (such as syllabic appendices, or vowel length). (Positional licensing of prosodic properties is anticipated by Zoll 1997 and Dresher & Van der Hulst 1998). The new theory will be shown to give a tighter factorial typology, explaining both of the observed gaps. An additional computational advantage is the strictly local evaluation of rhythmic well-formedness, without need for long-distance evaluation by alignment constraints.