We cannot verify your location
Browse Book and Journal Content on Project MUSE
OR
The Phonology and Morphology of Reduplication (review)

From: Language
Volume 78, Number 4, December 2002
pp. 770-773 | 10.1353/lan.2003.0048

In lieu of an abstract, here is a brief excerpt of the content:

Reviewed by
The phonology and morphology of reduplication. By Eric Raimy. Berlin: Mouton de Gruyter, 2000. Pp. 201. $97.80.

The myriad processes of reduplication have often been romanticized as providing a ‘blueprint’ for phonology—a window into larger mechanisms of morphological expression and phonological computation. Eric Raimy’s book, a revised version of his 1999 dissertation at the University of Delaware, presents an entirely new theoretical framework for the analysis of the various patterns of reduplication attested in the world’s languages. The important contributions of the work are threefold: its explicit formalization of precedence relationships in the phonology, its rejection of the reduplicant as the exponent of a terminal in the morphology, and its characterization of overapplication and underapplication effects, analyzed as the result of the ordering of phonological processes before transfer of the representation to the articulatory interface. R’s blueprint, a representational proposal developed through examination of reduplication phenomena, bears broader consequences if we wish to understand phonological computation as a generative procedure of establishing relations between segments.

The book begins with a fundamental observation: Precedence relations, implicit in the statement of any phonological rule or constraint, can be explicitly encoded in the representation of the segmental string. It often goes without saying that a Chomsky & Halle 1968 (hereafter SPE)- style rule of the form A → B / C __ D can only apply if C precedes A and A precedes D; such left-to-right precedence relations have been implicit throughout generative phonology.1 R employs the graph-theoretic notation of a precedence arrow (X → Y) to explicitly represent that a segment X linearly precedes a segment Y at a level of phonological representation. Motivating the ontological necessity of such relations, R argues that all native speakers know that in their language, the phonological string [t → a → k] is distinct from [k → a → t]; there are no ‘anagrammic’ languages in which these two are not contrastive. Following the formalization of these asymmetric, irreflexive, transitive precedence relations, and the observation that every segment has an immediate predecessor and an immediate successor, R defines two special symbols: the start symbol (#), which has no predecessor, and the end symbol (%), which has no successor.2 The precedence-based representation of a phonological word such as [tak], then, is illustrated as [# → t → a → k → %].

R departs from a morphophonological analysis in which reduplication is a copying operation that results in the exponence of an abstract morpheme, often referred to in the literature as Red. Instead, R proposes that reduplication results from a readjustment rule accompanying zero-affixation. Readjustment rules, originally proposed in SPE, are phonological rules that operate on a base in ‘reaction’ to affixation. For example, Halle & Marantz 1993 analyzes the ablaut in the past tense of English strong verbs (sing-sang) as the result of a phonological rule applying in the environment of the realization of a phonologically null past tense morpheme. R’s novel step is to propose that reduplication is a readjustment rule of the same sort; its effect, however, is not the alteration of particular segmental features but rather the introduction of a precedence relation to the representation. On this view, total reduplication results from the affixation of a zero morpheme (the exponent for the morphosyntactic features of the desiderative, or perfective, etc.) followed by a readjustment rule that adds a precedence relation between the final segment in the string and the initial segment in the string. The process is exemplified in 1.

The representation in 1 is a departure from the simple notion of precedence implicit in all previous phonological representations. That is, R allows for multiple precedence. In 1, [t] is immediately preceded by two segments, # and [k]; likewise, the segment [k] immediately precedes not one but two segments: % and [t] are both its immediate successors. R claims that representations [End Page 770] such as 1 may persist throughout the phonology but that the realization of 1 as [taktak] is due to limitations imposed by the interface between phonology and the articulatory system, through which such a representation cannot persist. Following the lead of Michaels 1989 in the phonology literature and Noam Chomsky and Richard Kayne in the syntax literature, R...