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

Reviewed by:
  • Distributed reduplication
  • Jason D. Haugen
Distributed reduplication. By John Frampton. (Linguistic inquiry monographs 52.) Cambridge, MA: MIT Press, 2009. Pp. xiv, 228. ISBN 9780262513531. $32.

The past few decades have seen a tremendous proliferation of theories specifically designed to handle the intriguing phenomenon of reduplication. With this book John Frampton introduces a novel approach called distributed reduplication (DR). DR is couched within the derivational phonology theory introduced by Chomsky and Halle (1968), wherein phonological ‘surface forms are produced by the successive modification of underlying forms’ (xi). Reduplication in DR is treated as stem modification indicating some kind of inflection and is usually triggered by a null morpheme. Thus, like Raimy’s (2000) theory, DR is distributed morphology (DM)-based and regards reduplication as a variety of stem readjustment rather than as an affix in its own right.

In DR, reduplication is a copying process broken down into separate components; this division of labor is the motivation for regarding reduplication as ‘distributed’. The key to the DR analysis of reduplication is what F calls transcription junctures (t-junctures). These involve inserting juncture boundaries into the timing (x-slot) tier of autosegmental representations; this process is triggered by the morphology and is called juncture insertion. Reduplication involves duplication junctures (notated with square brackets), which demarcate some subset (or all) of a stem for copying, while truncation junctures (notated with angled brackets) mark an additional subset for nonparsing. So, a stem such as hypothetical kata could be marked with duplication junctures at the left edge and after the first x-slot following the first vowel, yielding [kat]a,1 which reduplicates as kat-kata. If the first x-slot (i.e. onset consonant) is marked with truncation junctures, as [End Page 953] in [<k>at]a, it would be nonparsed in the duplicant, thus yielding at-kata. The copying process itself is transcription, and it proceeds to copy the timing tier along with its phonemic associations, ultimately leading to removal of the t-junctures. This is illustrated in the derivation of 1a into 1b and then 1c.

  1. 1. Distributed reduplication (p. 5, ex. 3)

The process in 1 obviously leads to violations of the no crossing constraint (NCC), as illustrated by crossed association lines in 1b, but for F it is crucial that the NCC be an interface constraint between the phonology and phonetics. In this way transcription of an autosegmental representation with inserted duplication junctures can give rise to crossed structures like that in 1b during the course of derivation so long as the crossed structures are eventually eliminated at a later stage of derivation, some time before the final output to phonetics. The amelioration of crossed structures is NCC repair (NCCR), as illustrated in 1c.

Crossed structures, as in 1b, result in ‘long-distance geminates’, that is, multiple associations of phonemes to the timing tier. Various phonological rules can apply before NCCR so they can affect either or both of the twins in the long-distance geminate. Languages differ with respect to whether one member of a geminate is sufficient to satisfy the condition for a given phonological rule or whether both are required. Rules affecting both twins can lead to over-application. Under-application results from geminate inalterability, since one twin does not satisfy the condition for a given phonological rule and thus blocks application of the rule to the other twin.

Inserting duplication junctures delimits which potential subportion of a stem is able to be copied in the reduplication process. This can be either a morphological or a prosodic constituent, but, oddly in my view, F stipulates that prosodic constituents must be limited to subwords of the stem, such as feet (11, 40). This allows him to capture the well-known Yidiny reduplication pattern that restricts copying to the first two syllables of the stem: thus, gin.dal.ba ‘lizard species’ reduplicates the entire first two syllables (i.e. foot) of the stem, yielding gin.dal-gin.dal.ba; but mu.la.ri ‘initiated man’ reduplicates as mu.la-mu.la.ri and not *mu.lar-mu.la.ri. Under F’s analysis, a null affix triggers insertion of a duplication juncture at the edge of the...

pdf

Share