An investigation, analyzing the linking of skeleton and syntactical rules of Madurese, presents counterevidence to Marantz's claims about the nature of reduplication, and to Carrier-Duncan's claim that reduplication must precede all phonological rules. It is proposed that reduplication in Madurese is not affixation, as Marantz claims, and can be determined only after all morphological and phonological rules have been applied. The rules of Madurese must apply in order: allomorphy (morpholexical), phonology, and then reduplication. The rules are considered for forming the most frequent type of reduplication in Madurese which is created by an exact copy of the final consonant-vowel-consonant (C-V-C) of the root after all morphological and phonological rules have been applied. The investigation concludes that it is impossible to derive the reduplication by mechanically repeating the entire stem of the word, and then lending part of its phonemic melody to the skeletal affix. Reduplication is a lexical process and the reduplication form is a copy of the parts of the phonetic representation of the stem; only then is it prefixed to the stem in a position determined by the other morphemes in the word and by the particular derivation. Extensive examples are included. (DJD)