[Author Login]
Title:Right-alignment as avoidance of stress lapse and stress clash
Authors:Birgit Alber
Comment:Handout of talk at WCCFL XX, USC - February 24th 2001
Abstract:In this talk I propose that directionality of foot-parsing shows a bias toward the left prosodic word edge. In other words, there is a constraint ALLFTL, but its mirror image ALLFTR is missing from the repertory of possible constraints. Right-alignment, when it occurs, is the result not of an alignment constraint, but of constraints militating against stress lapses and stress clashes (cf. Kager 1993, Kager 1994, Green & Kenstowicz 1995, Elenbaas & Kager 1999). These constraints are stated in purely rhythmical terms:

*LAPSE: rhythm is alternating: no two adjacent unstressed syllables
*CLASH: rhythm is alternating: no two adjacent stressed syllables

The main advantage of viewing right-alignment as lapse and clash avoidance is that it makes the right typological predictions: iambic systems do not right-align because they do not need to \'repair\' lapses and clashes (cf. Kager 2000/2001 for a very similar proposal).
As can be seen in (1a), a left-aligning trochaic system displays a final stress lapse in odd-numbered strings of syllables, while a right-aligning trochaic system (1b) does not. Thus, a right-aligning trochaic system can be interpreted as a means to avoid a stress lapse. In terms of constraints, better left-alignment is obtained at the cost of a final stress lapse in (1a), while right-aligning systems (1b) are those where lapses are avoided through minimal violation of ALLFTL. Left-aligning iambic systems (1c), on the other hand, do not create any final stress lapse since their feet are head-final. The unattested right-aligning iambic system (1d) would rate worse both on ALLFTL and *LAPSE and is therefore harmonically bounded by the alternative left-aligning parse (1c). The typological gap regarding right-aligning iambs thus is a direct consequence of foot form: right-headed iambs don\'t create final lapses, left-headed trochees do.

a. Pintupi: trochaic, left-aligning (Hayes 1995)
b. Warao: trochaic, right-aligning (Kager 1999)
s (Ss)(Ss)(Ss)(Ss)
c. St.Lawrence Island Yupik: iambic, left-aligning (Hayes 1995)
d. unattested: iambic, right-aligning
s (sS)(sS)(sS)(sS)

S = stressed syllable, s = unstressed syllable, _ = stress lapse or clash

In systems that allow for degenerate feet right-alignment is not the result of lapse avoidance, but the re-sult of avoidance of a stress clash. A left-aligning trochaic system with degenerate feet (2a) will parse a subminimal foot at the beginning of the word, since thus feet are closer to the left edge than if the degen-erate foot was parsed at the end. However, in this case ALL-FT-L is satisfied at the cost of creating a stress clash between the first and the second syllable. A right-aligning trochaic system as in (2b) avoids this clash through a minimal violation of ALLFTL. Again, iambic systems can obey both constraints at the same time: the left aligning iambic parse (2c) with an initial degenerate foot will satisfy better ALLFTL and *CLASH than the right-aligning parse (2d). The prediction therefore is that we will find both right and left-aligning trochaic systems with degenerate feet, but that iambic systems with degenerate feet should always be of the left-aligning type. Right-aligning iambic sys-tems with degenerate feet should not occur, and, to my knowledge, systems of this type do not exist.

a. Maithili left-aligning trochaic system (Hayes 1995)
b. Ono right-aligning trochaic system (Crowhurst & Hewitt)
c. Weri left-aligning iambic system (Crowhurst & Hewitt)
d. unattested right-aligning iambic system

Besides the absence of right-aligning iambs this set of constraints predicts also the absence of systems with initial dactyls.
The handout also addresses the issue of whether lapse and stress constraints could be unified under a cover constraint against equal prominence and it shows that the proposed set of constraints can deal equally well with quantity-sensitive systems.

Type:Paper/tech report
Article:Version 1