Phonological Features

Abstract

Abstract It has been long understood that speech sounds are not atomic, indivisible units, but that they are made up of a number of phonological features. In earlier work, speech sounds were represented as unordered sets or bundles of distinctive features (Bloomfield 1933: 79; Chomsky and Halle 1968: 335 ff ., Lass 1984: 94). Today it is understood that speech sounds are internally structured: certain groups of phonological features behave as units in assimilation processes; certain features appear to be dependent on other features—[distributed] and [anterior], for example, are relevant for coronals and not for velars or labials; in assimilation processes certain spreading features appear to be contingent on the presence of features shared between the trigger and target (Cole 1987). The need to provide a structured representation of phonological features was first properly addressed in Clements’s (1985) paper entitled ‘The geometry of phonological features’ in which he proposed that features are arranged geometrically. Feature geometry rep-resents both the separate and the coordinated aspects of features within a hierarchical structure. In the model of feature geometry that I adopt, phonological features and feature values fulfil the four basic criteria listed in (1).