Phonetics, Phonology, and Cognition (review)

Abstract

Reviewed by: Phonetics, phonology, and cognition ed. by Jacques Durand and Bernard Laks Eugene Buckley Phonetics, phonology, and cognition. Ed. by Jacques Durand and Bernard Laks. (Oxford studies in theoretical linguistics 3.) New York: Oxford University Press, 2002. Pp. xiii, 338. ISBN 0198299834. $45. Considerable recent work in phonology has sought connections and explanations for abstract sound patterning in two domains on ‘either side’ of traditional phonological representations: the physical realization and perception of sound that underlie most phonological patterns, and more general cognitive principles that might constrain or even determine the possible encoding of these patterns. As indicated by the title, this book fits squarely in the trend; it collects eleven papers (including an overview by the editors) that emphasize interdisciplinary approaches to the analysis of linguistic sound structure. Most of the papers originated in the conference Current Trends in Phonology 2, held in June 1998 at the abbey of Royaumont; the majority of the contributors (seventeen of twenty-three) are affiliated with French institutions. After a brief survey of the chapters (‘Introduction’, 1–9), Durand and Laks present a substantial overview chapter (‘Phonetics, phonology, and cognition’, 10–50). They discuss trends, primarily [End Page 653] in Europe and the United States over the last century, regarding the relationship between linguistic research on phonetics and phonology. This ranges from the development of narrow and broad transcriptions to more formalist models in which phonetic substance is downplayed but more general cognitive principles might come to the fore. Joaquim Brandão de Carvalho (‘What are phonological syllables made of? The voice/length symmetry’, 51–79) proposes to treat voicing and aspiration—the phonological features responsible for voice onset time (VOT)—as a suprasegmental property like length, rather than as primitive distinctive features. This approach assumes two parallel tiers of sonority (well known as a property of the syllable nucleus) and tension (as found in the greater constriction of consonants). Just as a long segment results from linking a feature set to more than one timing element, so VOT might be represented by multiple linkings between timing elements and the sonority and tension tiers. The details are complex, incorporating diacritics that mark the tier elements as ‘cold’, ‘positive’, and ‘negative’ (cf. Kaye et al. 1990), so it is difficult to assess the specific typological predictions. At a broader conceptual level, more evidence is necessary to motivate the suprasegmental analogy. Thus, while in lenition processes the relation of /tt/ and /t/ is similar to that between /t/ and /d/, it is not clear how the account would handle other cases of lenition, as when /d/ becomes /ð/ or /ɾ/ (cf. Kirchner 1998). Similarly, de Carvalho alludes to evidence from word games for the suprasegmental status of length, but does VOT ever pattern in the same way? To take one example, the foot exchange in the Japanese argot zuuja-go changes jaazu ‘jazz’ to zuuja, but the name daisuke becomes sukedai, not *zuketai with preservation of voicing in situ (Itô et al. 1996). De Carvalho’s proposal is interesting for its potential connections to nonlinguistic cognitive notions, but its linguistic implications remain unclear. John Goldsmith (‘Tone in Mituku: How a floating tone nailed down an intermediate level’, 80–95) presents a rather straightforward example of opacity to argue for a level of representation between input and output. The case involves contour tones in the Bantu language Mituku that surface only when a underlying vowel has been deleted, as in the prefixes /tù-á-/ realized as [tǎ-]. The need for two vowels cannot be expressed on the underlying form because a floating L tone (present before every H-toned verb root) will enter into a contour only if it first is able to link to a toneless vowel, and some toneless vowels arise only after the application of a rule. For example, in /bá-á-L-/ the HH sequence is simplified to H in [bá-a-L], and the ultimate result is [bâ-]; but in /tù-á-bá-L-/, while tone absorption changes [tǎ-bá-L] to [tà-bá-L], we find downstep of the following H rather than a contour because the floating L was never exposed to a toneless vowel. As in similar...