Pasiego pretonic metaphony revisited

In this paper, I address the issue of full versus underspecification, providing arguments for underspecification. I re-examine the distribution of the feature [high] in Pasiego, a Spanish dialect spoken in N.W. Spain, arguing that the distribution of [high] is predictable from two factors: 1) the syntagmatic process of metaphony, or high harmony; and 2) the vowel inventory present in paradigmatic positions (i.e. positions of prominence and of neutralization). I show that in the latter case, the distribution of [high] is predictable from an optional rule of enhancement that inserts the feature [high] where this feature is non-contrastive. The implications for the underspecification vs. full specification debate are as follows. If some aspects of feature specification are optional, then the distinction between underlying and surface form must be maintained on the grounds that the alternative -- i.e. listing all the possible surface forms for each morpheme -- misses a predictable generalization. The Pasiego data, then, affirms the possibility that underlying forms are underspecified