Feature geometry in disordered phonologies

Abstract

Two types of systems are in general use for the description and classification of consonants in disordered phonological systems: conventional place-voice-manner and standard distinctive features. This paper proposes the use of a third model, feature geometry, which is an analysis framework recently developed in the linguistic study of primary languages. Feature geometry allows for relatively independent behaviour of individual distinctive features, but also organizes them into hierarchies in order to capture the fact that features very often act together in rules. Application of the feature geometry to the study of the phonologies of 40 misarticulating children, specifically to the phenomena of apparent cluster coalescence, fricative/affricate alternations, and alveolar stop/glottal stop alternations, reveals that feature geometry provides better explanations for representations and rules in disordered systems than either of the other two frameworks.