|Abstract:||McCarthy & Prince (1995) outline a theory of reduplication in which segments in the reduplicant stand in correspondence with segments in the base. Discontiguous partial reduplication patterns, in which a string of segments in the reduplicant corresponds with a discontiguous string of segments in the base, have been observed in various languages in the Austronesian and Austroasiatic language families. Several such patterns show a preference for the anchoring of the segments at both edges of the base. In this paper, I propose that edge-anchoring discontiguous reduplicants arise as a result of fundamental constraints on phonological properties of particular languages, arguing for what I call the Reduction Model of discontiguous reduplication, formulated in Optimality Theory (Prince & Smolensky 1993/2004). Under this model, discontiguous reduplicants are shown to be derivable from maximal prosodic constituents, which are reduced in size due to language-particular phonological requirements on sonority sequencing, syllable structure, prosodic correspondence, and positional faithfulness. I show how the interaction of these constraints with Contig-BR and constraints on reduplicant size yields discontiguous base-reduplicant correspondence strings in three languages: Type V/VI reduplication in Nakanai, Expressive reduplication in Semai, and Type III reduplication in Ulu Muar Malay. Furthermore, I argue that reference to the right edge of the base is not necessary to yield right-anchoring in any of these patterns (contrary to Hendricks 2001 and Nelson 2003), but that right-anchoring falls out from the same language-particular phonological requirements that limit reduplicant size.