Tree-based data structures are commonly used by computational linguists for the documentation and analysis of morphological and syntactic data. In this paper we apply such structures to phonological data and demonstrate how such representations can have practical and beneficial applications in computational lexicography. To this end, we describe three integrated modules: the first defines a multilingual feature set within a tree-based structure using XML; the second module traverses this tree and generalises over the data contained within it, optimising the phonological data and highlighting feature implications. The third uses the information contained within the tree representation as a knowledge base for the generation of multiple feature-based syllable lexica.