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

  • What is the Base for Reduplication?
  • Jason D. Haugen

Although reduplication has played a central role in the development of various theoretical perspectives, relatively little attention has been paid to the issue of what is able to be copied when a reduplication process applies. While much work has focused on the nature of reduplicants as theoretical objects, the possibility that languages may vary with respect to what may serve as a base for reduplication has rarely been addressed. In fact, comparatively little has been explicitly stated or claimed about the exact nature of reduplicative bases.

The purpose of this squib is twofold: first, to contrast various proposed definitions of the "base" for reduplication and compare their empirical adequacy, and second, to collate various cases from the literature which show that reduplicative allomorphy in certain languages is best dealt with by allowing for the differential assignment of bases for reduplicative copying in different morphological and sometimes even phonological (prosodic) contexts in those languages. Theories that do not allow for this differential base assignment fail to make the correct generalizations about what gets copied in such languages.

1 Definitions of the Reduplicative Base

Some early generative theories of reduplication posited that a morphological stem was copied in full, and then the segmental melody of that stem was linked to either a CV skeleton (Marantz 1982) or a prosodic constituent (McCarthy and Prince 1986). Both of these early theories assumed that what linguists later came to regard as the "base" for reduplication was the entire stem itself, and that whatever reduplicated segments failed to link got erased ("stray erasure"). For example, in [End Page 505] Marantz's theory a hypothetical stem pata could reduplicate and link to an empty CVC skeleton prefix, as in (1).1

(1) Marantz's (1982) approach to reduplication:pat-pata

The arrows in (1) indicate that the association was "melody-driven" in Marantz's theory, with the segments on the melody tier attempting to associate to a consonantal (C) or vocalic (V) slot on the skeletal tier while any nonassociated segments were erased.

McCarthy and Prince (1986) supplanted Marantz's theory by showing in detail that the templates for reduplicative copying are better stated as prosodic constituents rather than CV skeleta. The same data in (1) can be accounted for as in (2), the association being driven now by the prosodic template, a bimoraic (heavy) syllable that associates with as much of the stem's segmental melody as possible. Once again, the "base" is assumed to be a full copy of the stem and nonassociated segments are erased.

(2) McCarthy and Prince's (1986) approach to reduplication:pat-pata

McCarthy and Prince (1986) also discuss certain cases of reduplication where a subset of a stem seems to be selected for reduplication. For example, a perspicuous account of the reduplication pattern in a language like Yidiny (Pama-Nyungan) recognizes that the reduplicant only copies from the first two syllables of the stem, as in (3).

(3) Yidiny disyllabic reduplication

  1. a. mu.la.ri > mu.la.mu.la.ri    *mu.lar.mu.la.ri  'initiated man'

  2. b. kin.tal.pa > kin.tal.kin.tal.pa   *kin.ta.kin.tal.pa  'lizard species'

Although the reduplication patterns differ in (3a) and (3b), a fact that is problematic for a stipulative CV skeleton account, a unified analysis is possible if we allow the base to be limited to a prosodic subset of the entire stem, that is, the first foot, as shown in (4). [End Page 506]

(4) McCarthy and Prince's (1986) approach to Yidiny reduplication

Any theory of reduplication that does not allow the reduplicative base to be limited to the right edge of the second syllable (or, equivalently, the right edge of the first foot) in such cases must offer an alternative explanation of why *mu.lar.mu.la.ri is not as acceptable as kin.tal.kin.tal.pa. The clear generalization is that the former is ruled out because the onset of the third syllable is simply beyond the scope of reduplicative copying in this language—or, in other words, that the base is limited to the first two syllables (or foot) of this language...

pdf

Share