Deriving *economy: Syncope in optimality theory
This dissertation proposes that markedness constraints in Optimality Theory are lenient: a form can be marked with respect to a constraint only if there is another form that is unmarked. Thus, no constraint bans the least marked thing. The central consequence of this idea is that there are no economy constraints that penalize structure as such. Economy effects follow from the interaction of lenient markedness constraints. Economy constraints are shown to be not only unnecessary but actually harmful: their very presence in C ON predicts unattested patterns that remove structure regardless of markedness.
Chapter 2 develops the theory of CON and argues that various structural economy effects (preferences for smaller structures over larger ones and for fewer structures over more) follow from constraint interaction. Also addressed are economy effects that involve the deletion of input structure, including foot-sized maximum effects in truncation and syllable-sized and segment-sized maximum effects in reduplication. OT's economy constraints of the *STRUC family are argued to produce unattested patterns under re-ranking and are excluded from CON as a matter of principle.
Chapter 3 examines metrical syncope in Hopi, Tonkawa, and Southeastern Tepehuan. Different patterns fall out from the interaction of the same metrical markedness constraints in language-specific rankings. All of these constraints have other, non-economy effects—in principle, they can be satisfied by the addition of structure as well as by removal of structure. Metrical shortening and syncope remove marked structure, not all structure: the well-formedness of an output is determined by the distribution of weight in its feet and exhaustivity of footing, not by the number of syllables, moras, and feet.
Chapter 4 examines differential syncope in Lillooet, Lushootseed, and the Lebanese and Mekkan dialects of Arabic. Under the leniency hypothesis, there are constraints against low-sonority syllable nuclei and foot peaks but not high-sonority ones; likewise, there are constraints against high-sonority foot margins but not high-sonority vowels in general. The interaction of lenient constraints cannot duplicate the effects of economy constraints. There are real crosslinguistic asymmetries in attested differential syncope patterns that can only be explained if we abandon the notion that “everything is marked.”