Introduction: enhancement features

Stevens, Keyser, & Kawasaki (1986) lay out a theory of phonological features in which languages employ features from a universal set in two different ways. Each language employs some features from this set to minimally distinguish lexical items; this is a distinctive use of the feature. Cases often arise in which a particular feature is never employed all by itself to contrast lexical items, but rather, it always occurs in conjunction with another, distinctive, feature. For reasons of compactness of description, various linguists, e.g. Halle (1958, 1964), have proposed that, at some level of the linguistic grammar, only one of the features needs to be specified. The other of the two features is redundant, and can thus be filled in by rule.

While the phonological literature on redundancy of featural specifications is quite large, there has been relatively little consideration of these phonologically based distinctions in feature usage from a phonetic standpoint. Stevens et al add the phonetic property of variability to the phonological, informationally based diagnostic of redundant and distinctive features. They state that, while distinctive features serve directly to convey lexical distinctions, redundant features serve to enhance the salience of the distinctive features. Thus, an alternative term for redundant features is enhancement features. As a corollary to this claim, they suggest that distinctive features are invariantly present in speech, while enhancement features are more variable, being employed especially in situations where the distinctive feature may be obscured by some attribute of the context in which the feature appears.