Distinctive features and acoustic cues

Abstract

Recent findings in speech perception studies have been interpreted as indicating that distinctive features are not ’’real.’’ That is, they seem to be merely convenient descriptive devices which are imposed on the data rather than found in it. This conclusion results from a failure to distinguish between distinctive features and acoustic cues. Distinctive features and acoustic cues are different types of entities and are justified on different empirical grounds. Acoustic cues are properties of the speech signal which relate the acoustic stream to phonological segments. Distinctive features are the elemental units of phonology which govern the behavior of segments within language. Acoustic cues are determined by observation of the speech signal and distinctive features by observation of phonological rules and phonological change. The relation between distinctive features and acoustic cues is not 1 : 1, but rather 1 : many. Therefore, distinctive features cannot function as, nor be equated with, acoustic cues and vice versa. Moreover, confusion matrices, which are generated by masking the input stimulus with noise or filtering, are a direct function of the speech signal, and thus yield information concerning acoustic cues, not distinctive features. Therefore, this research does not disprove the empirical status of distinctive features; it corroborates the 1 : many relation between distinctive features and acoustic cues.