Working paper Open Access
There are currently many competing feature theories and models of segment-internal representations. Despite differences in detail, however, the general proposals are fairly uniform, each making minor modifications to the feature set of SPE (Chomsky and Halle, 1968) and the geometry of Clements (1985) – with two notable exceptions. First, Clements (1991) proposed an innovative unification of consonant and vowel place features, which greatly economized the set of place features. Second, all proposals for sign language depart radically from spoken language proposals. In this paper, we propose a very different model of feature geometry, in which the insights of Clements (1991) are extended to other areas of the phonology, and structural and featural economy are exploited to the greatest extent possible.
This model not only eliminates a large number of features from the grammar (including the major class features), but it provides a unified analysis for consonants, vowels, place, manner, tones, complex and contour segments in spoken and signed languages.