Notes and Reviews

In a recent paper' in this journal, Wang offers an interesting extension of the Jakobson-Hallean universal phonetic alphabet to handle the tones of so-called 'tone languages', particularly the languages phonologically of the Chinese type. Wang makes the same assumptions about the nature of phonological representations of utterances that are made, with respect to the 'segmental' features, by Halle and other practitioners of phonological analysis within the transformational-generative theoretical framework. However, the most questionable of these assumptions-that the universal features are 'binary', in the sense that cells in fully-interpreted phonetic matrices may be filled by any one of three symbols, namely '+', '-', or blank (denoting irrelevance)is defended by Wang (100-3) against one alternative (that the features have more values). The other alternative is that the features be 'singulary' in the sense in which Sydney Lamb uses this term, that is, each feature is either present or absent in any given segment;2 but the system Wang proposes is actually formally equivalent to this, provided we restrict our domain of discourse to tone languages, as it appears (cf. Wang's Table I) that each of Wang's tonal features are relevant to each segment having any tonal features at all (clearly, a matrix in which each cell is filled either by '+' or '-' is not effectively distinct from a Hockett 'stepmatrix' (Hockett, op. cit., chapter 3)). So discussion of Wang's proposals need not