The Stricture Features

Abstract

The late 1980s and early 1990s saw the publication of a large number of articles treating features and feature geometry, following on the highly influential article by Clements (1985) that had argued for the organization of features into a hierarchically arranged tree. But if one takes a look at the features whose position and organization were so often debated, they fall into only a few groups: first, the vocalic features typically involved in vowel harmony – features of height, backness, rounding, or tongue root position; second, laryngeal features such as glottalization, voicing, and aspiration; and third, the place features labial, dorsal, and coronal and their dependents, especially those of coronal (see chapter 27: the organization of features ; chapter 21: vowel height ; chapter 19: vowel place ; chapter 69: final devoicing and final laryngeal neutralization ; chapter 22: consonantal place of articulation ; chapter 12: coronals ). Such a concentration can be traced back in part to the material covered in Clements's seminal article. But it also reflects those features that seem most often and most unequivocally involved in assimilation, or, to follow McCarthy's (1988) diagnostics, the trio of assimilation, dissimilation, and reduction or deletion. There was also notable interest in the position and behavior of [lateral] ( chapter 81: local assimilation ; chapter 60: dissimilation ; chapter 79: reduction ; chapter 31: lateral consonants ). But most of the traditional manner and major class features – [continuant], [consonantal], [sonorant] – received short shrift. 1 These three, along with the less widely adopted feature [approximant], are sometimes called stricture features (Kenstowicz 1994: 480ff.), because they specify how the articulators are brought together but are not themselves inherently associated with any one articulator. They are “articulator‐free,” in the terminology of Clements and Hume (1995) and must be associated with some articulator to be fully interpreted. The major class features [consonantal] and [sonorant] do not participate often, if at all, in the archetypical phonological processes of spreading and delinking, and they do not form nice bundles of features that beg to be organized under a node of the feature geometry. The other major class features of Chomsky and Halle (1968), [syllabic] and [vocalic] (the latter already tentatively withdrawn in a footnote in Chomsky and Halle 1968: 302), have over the decades been subsumed in representations, with [syllabic] coming to mean head or nucleus of the syllable and not being specified as a distinctive feature of any segment. 2 As early as 1974, Hankamer and Aissen suggested that the major class features might be superfluous, and in a more elaborated proposal, Selkirk (1984) proposed doing away with both [consonantal] and [sonorant] by replacing them with a multivalued sonorancy feature (see also chapter 8: sonorants ). Nonetheless, the consensus view seems to be that these features are still needed. Their necessity is taken for granted in many significant works on feature theory of the last decade or two, such as Clements (2003), to mention one representative example. As an independent issue, the proper definition of [continuant] has always been problematic: it has not been obvious exactly what natural classes should be delineated by this feature, nor how it should be placed on the feature tree (see also chapter 28: the representation of fricatives ).