On the Proper Characterization of 'Nonconcatenative' Languages

Nonconcatenative languages have been claimed to employ a special type of phonological spreading of a consonant over a vowel, which assumes a representation that segregates consonants and vowels on different planes. I argue that this type of spreading can and must be eliminated from the theory, by reducing it to segmental copying as in reduplication. Crucial to this reduction is the notion of gradient violation of constraints in Optimality Theory (Prince & Smolensky 1993), and the notion of Correspondence with its particular application to reduplicative morphology (McCarthy & Prince 1995a). The reduction is demonstrated in detail for Temiar, one of the main indigenous languages of Malaysia, notorious for the complexity of its copying patterns. Extensions of the proposal to Semitic languages are also discussed. Two main theoretical implications of this reduction are then developed. First, the distinction between concatenative and nonconcatenative languages need not and should not be encoded in terms of the special phonological mechanisms of consonantal spreading over a vowel, applying under planar segregation. Second, the locus of the distinction is found to be, instead, in the mode of affixation employed in nonconcatenative languages, namely, a-templatic reduplicative affixation. This type of affixation is predicted, though heretofore undocumented in the typology of word formation.