A prosodic theory of nonconcatenative morphology

Abstract

A Prosodic Theory of Nonconcatenative Morphology Most structuralist accounts of morphological structure overtly or implicitly make a distinction between two formal morphological types.Concatenative morphology, which in the more familiar languages appears almost exclusively, involves prefixation or suffixation only.Thus, morphemes are discrete elements linearly concatenated at the right or the left end of the base of the morphological operation.Morphology of this type is subject to analysis by a relatively simple discovery procedure.Given an adequate phonological representation, concatenative morphemes can be recovered by a left-to-right (or right-to-left) parse of words searching for invariant recurrent partial strings, possibly with constant meaning or function (Hockett (1947)).The other type, nonconcatenative morphology, has remained rather more mysterious until now.Generally, in structuralist treatments we find only a list of the residue, those morphological operations that cannot be analyzed by the method of recurrent partials.These include reduplication, infixation, morphologically-governed ablaut, and suprafixation.All of these terms are in common use except the last, which refers to, for example, the variation in tonal pattern of the stem as a mark of verbal aspect inflection in Tiv (McCawley (1970), Goldsmith (1976)).Although nonconcatenative morphology as a whole has received less attention than concatenative, this is not for lack of exemplification.In a number of languages, processes like reduplication are the primary or sole morphological operations.This residual status accorded nonconcatenative morphology in structuralist analyses extends to generative theories as well.All generative treatments known to me have relied entirely on the extremely rich transformational notation of Chomsky and Halle (1968).What is offered here instead is a new theory of nonconcatenative morphology, one which owes a great deal to Harris's (1941; 1951) notion of long components.It is a prosodic theory in the sense that it uses the devices of autosegmental phonology, This article is a revised version of portions of chapter 4 of McCarthy (1979).Some of the material in section 3.3 was originally presented at the Fifth North American Conference on Afro-Asiatic Linguistics in 1976.I am grateful for the assistance of Lee Baker, Nick Clements, Morris Halle, Jay Keyser, Paul Kiparsky, Alan Prince, Ellen Woolford, and an anonymous reviewer for Linguistic Inquiry.The system for transcribing Arabic used here has its familiar values, with the following exceptions.?and h are the voiced and voiceless pharyngeal glides, respectively.g is the voiced velar spirant, andj is the voiced alveopalatal affricate.A subscripted dot in t, d, s, and d indicates pharyngealization, also known as emphasis.Vowel length is represented bimoraically as VV.