Exceptionality in Spanish Onset Clusters

Spanish complex onsets have been traditionally described as consisting of a stop (/p, t, k, b, d, g/) or the fricative /f/ plus a liquid. Given that all Spanish varieties have other fricatives (/x, s/), the obstruents that can form part of an onset cluster do not straightforwardly compose a natural class. As such, past studies have argued that /f/ is exceptional in its ability as a fricative to pattern with stops in onset clusters. This paper presents empirical data from a nonce word judgment task that challenges this claim and shows that Spanish listeners rate unattested /xr/ clusters as more acceptable than ungrammatical /sr/ clusters.  These results suggest that /s/, and not /f/, is exceptional in its inability to form complex onsets in Spanish.  As /s/ is the sole sibilant in the Spanish consonant inventory and is uniquely characterized by the feature [strident], this generalization is easily capturable in an Optimality Theory framework. This analysis further predicts that other non-sibilant fricatives should also be acceptable in onset cluster position, such as /θ/, which is supported by data from a follow-up study with speakers of Peninsular Spanish who have this phoneme in their dialect. This analysis also predicts that other sibilants should be unacceptable in onset clusters. This is supported by data from the related languages Portuguese and Catalan that have other sibilant phonemes (/z, ʃ, ʒ/)yet also have similar onset cluster phonotactics as Spanish in that they disallow all sibilants from being in an onset cluster.