Against featural alignment

Morphemes are sometimes expressed by elements that are less than full segme nts, and, in a given language, the position of these elements in a word may vary. A recent analysis of these 'mobile morphemes ' claims that their dis tribution is best explained in an optimality-theoretic framework that incor porates a set of featural alignment constraints (Akinlabi 1996) This paper argues that featural alignment plays no role in the realization of 'mobile morphemes'. Instead, it recognizes a set of licensing constraints that expl icitly identifies where featural exponents of such morphemes may appear in a word. Crucially, these licensing constraints, unlike featural alignment, are not morpheme-specific and therefore enjoy cross-linguistic support. Ana lyses of Chaha labialization, Terena nasalization, High tone realization in the Edoid associative construction and Southern Sami vowel lowering in ter ms of licensing are shown to be superior to the alignment-theoretic ones on both descriptive and explanatory grounds.