A Review of the Theoretical Studies on Reduplication

【Abstract】Reduplication tends to have prosodic shape invariants which are not necessarily related to a prosodic unit in the base. To learn about the contributions of Correspondence Theory of OT in this respect， we must first focus on aspects of the ancestor theory of templates， and then compare these aspects with the OT viewpoint. This article will make an introduction to the basic concepts of reduplication and outline a general picture of the theoretical development in the research of reduplication.

【key words】Reduplication; CV skeleta; OT

【作者簡介】刘沙沙（1982.02-），女，汉族，黑龙江人，三亚学院外国语学院，英语专业，讲师，硕士，研究方向：外国语言学及应用语言学。

1. Basic Concept of Reduplication

From a purely morphological point of view， reduplication is simply a kind of affixtion， both in its morpho-syntactic contribution （it forms morphological categories， such as plural）， and in its linear position with respect to the stem （preceding it， as a prefix， or following it， as a suffix）. But from a phonological viewpoint， the special property of reduplication is that the reduplicative affix is not fully specified for segmental content. Its segmental content is copied from the stem that undergoes reduplication. Reduplication is therefore by its very nature a phenomenon involving phonological identity between the “reduplicant”and the “base” to which it adjoins. Segmental and prosodic identity of the reduplicant and the base is obvious in the case of total reduplication， which involves copying of a complete word. The following examples of total reduplication in Indonesian show that it is impossible to tell apart the reduplicant from the base：

（1） Indonesian plural reduplication （total reduplication， copies entire word）

a. wanita      wanita-wanita   （women）

b. ma?arakat   ma?arakat-ma?arakat    （societies）

But other languages have reduplication processes that copy only part of the segments of the base. This phenomenon is known as partial reduplication. For example， reduplication in the Wakashan language Nootka copies the first sequence of consonant plus vowel （CV or CVV） of the base， and prefixes this material to the base.

（2） Nootka reduplication （copies initial CV（V） sequence， equalling open syllable）

a. ?ims-i： ?   ?i-?ims-i： ?       （hunting bear）

b. wa：s-?i?    wa：-wa：s- ?i?       （naming whee…）

Observe that the reduplicant prefixes [?i-] in （2a） and [wa：-] in （2b）， both open syllables， do not match similar open syllables in the base， but copy segments that stand in closed syllables： [?im] and [wa：s] in the base.

2. CV Skeletal Approach

Based on the autosegmental representation of morphemes proposed in McCarthy ， Marantz （1982） made a step further in treating reduplication as the affixation of a CV templatic morpheme to base. The CV skeletal template serves as a reduplicative affix in itself， it triggers a melody copying process in which the entire phonemic melodies of the base are copied and attached to the CV skeletal under association convention. The association of melodies to CV skeleta is phonemically driven; pre- and under-associations of features and phonemes are allowed; and unassociated melodies receive no phonetic realization.

To account for such shape invariance， Marantz （1982） assumed a notion of “template，” a morpheme having the defining property of being melodically empty. In his template-and-association theory， all prosodic properties defining shape-invariance of reduplicants are stated in the template. The Marantzian template is a sequence of skeletal segments， for example CVC- in the case of Agta， to which the melodic content of the base is associated from a fully copied string. Unassociated melodic elements are deleted：

The CV skeletal approach provides a formal account of reduplication from both aspects of morphological affixation and phonological copying， but as noted by Raimy （2000）， its reduplicative template does not distinguish between pathological and naturally occurring reduplicational patterns， almost at patterns can be formulated through CV skeletal template as well as pre- and underspecification， so this approach overgenerates， too. 

References：

[1]Marantz， A. Re reduplication[J]. Linguistic Inquiry， 1982，13：435-482.