Published online by Cambridge University Press: 24 August 2012
In velarising dialects of Spanish, nasal place contrasts neutralise to [ŋ] word-finally. However, whereas velarisation applies transparently in word-final prepausal environments, place neutralisation ‘overapplies’ to stem-final presuffixal nasals and to word-final nasals which resyllabify into onset position across word boundaries. Yet since previous analyses of Velarising Spanish have been based exclusively on theory-led interpretations of impressionistic data, doubts exist as to whether word-final nasals in velarising dialects are consistently realised as [ŋ] (Baković 2000). The first goal of this paper therefore is to submit the claims put forward in these analyses to empirical testing. Experiments using electropalatography confirm that speakers of Velarising Spanish produce robustly dorso-velar nasals in word-final environments; this result refutes the claim that word-final nasals are placeless in velarising varieties. Secondly, because opaque instances of nasal place neutralisation pose challenges for Optimality Theory (OT), I compare two approaches to modelling the nasal alternations in Velarising Spanish, namely Output–Output correspondence in classic OT and a cyclic analysis in Stratal OT. This comparison reveals that classic OT cannot account for the opaque patterns without stipulating fixed OO-constraint rankings. By contrast, the stratal model straightforwardly predicts the occurrence of both opacity effects on the basis of general architectural principles.
I would like to thank Ricardo Bermúdez-Otero, Yuni Kim, Nigel Vincent and three anonymous JL referees for their helpful comments on earlier drafts of this paper. I alone am responsible for any remaining errors.