ABSTRACT

The surface distribution of voiced obstruents undoubtedly has been among the most intensely debated theoretical issues of Spanish phonology in the past three decades. The stop and spirant allophones of underlying voiced obstruents in modern standard Spanish are in complementary distribution: the stops [b, d, g] are found word-initially after a pause, and after homorganic sonorants; the spirants [ß, ð, ɣ] occur elsewhere. Two fundamental issues have been at the heart of the debate: (a) the nature of the underlying segments and (b) the formal mechanism(s) accounting for the surface distribution. This chapter reviews previous approaches to spirantization and proposes a constraint-based analysis of the stop ~ spirant surface distribution of Spanish voiced obstruents within the framework of Optimality Theory (OT), which provides a principled solution to the problems that plagued former approaches to the topic, and at the same time extends its empirical coverage to several facts of dialectal variation. In this approach, the stop ~ spirant distribution follows from the interaction of faithfulness and markedness constraints, as is standard in OT. Specifically, this proposal focuses on the conflict between Ident-[cont], requiring input-output faithfulness of the feature [cont], and two markedness constraints restricting the surface realization of voiced obstruents: Agree-[cont], which demands agreement with a preceding segment for the feature [cont], and VoiObs-[−cont], which embodies the implicational universal that within the class of obstruents (in our particular case, voiced obstruents), stops are less marked than fricatives. The domination of Agree-[cont] over VoiObs-[−cont], and the ranking of both over Ident-[cont], accounts for the core of the stop ~ spirant distribution in a straightforward manner, independently of whether underlying stops or spirants are assumed. Finally, the occurrence of [d] after a homorganic lateral is accounted for by assuming that laterals are phonologically stops before coronals, but continuants before noncoronals, an assumption robustly underpinned by the observation that the articulation of laterals involves a blocking of the airflow in the alveolar region of the vocal tract; otherwise, the airstream flows out through the sides of the mouth.