Serial Markedness Reduction

Abstract

A growing body of research provides evidence supporting Harmonic Serialism (HS; McCarthy 2000, 2008a, 2010; Elfner to appear; Pruitt 2010), a serial version of Optimality Theory in which a single ranking of constraints is used iteratively to construct a derivation. This work has led to new insights into several outstanding problems in phonology and has highlighted problematic over-predictions of parallel OT, showing that HS predicts a more restricted range of interactions that better reflects typology. Although HS makes it possible to capture certain generalizations stated at intermediate levels of representation, it does not provide a general solution to phonological opacity. For this reason McCarthy developed a significant elaboration of HS, OT with Candidate Chains (OT-CC; McCarthy 2007), which evaluates complete HS-like derivations in parallel. This paper proposes a novel extension of HS, Serial Markedness Reduction (SMR), which combines advantages of HS and OT-CC. Like OT-CC, SMR includes a family of constraints that enables the modeling of both counterfeeding and counterbleeding opacity. However, SMR significantly streamlines the machinery necessary to model opacity as compared to OT-CC, relying on the gradual optimization inherent to HS to construct a single optimal derivation. The formal and empirical differences between OT-CC and SMR are discussed.