Abstract
The current paper aims to capture the properties of reduplication within the distributed morphology model (Halle and Marantz 1993, 1994; Embick and Noyer 2007;
;
). Taking Standard Arabic (SA) as a representative, the article shows that SA, like many other languages, has both full and partial reduplication. Full reduplication repeats entire stems while partial reduplication doubles part of it. Rather than the available two analyses, i.e. the readjustment approach (
;
) and the affixation approach (
, 2010, 2011; Haugen and Harley 2010), the current paper provides a novel approach to the phenomenon of reduplication in the world languages. It argues that root consonants and vowels should be decomposed into non-phonetic distinctive features that undergo late insertion at PF. These non-phonetic distinctive features are supplied with sound items at PF in the same fashion that the terminal nodes with morphosyntactic features are fed with vocabulary items. This approach serves three purposes. It accounts for speech errors, captures the non-concatenative morphology in Semitic languages, and allows the reduplicant form to copy all the distinctive features of the roots, yielding instances of full reduplication. Instances of partial reduplication can be generated by root-sensitive impoverishment rules which target and delete some of the features of the stem or the reduplicant form.