The Effects of Duration and Sonority on Contour Tone Distribution: A Typological Survey and Formal Analysis (review)

One of the touted benefits of optimality theory is that it should encourage typological studies, but studies involving more than a handful of languages have been rare. This published version of Zhang’s 2001 UCLA dissertation is a welcome exception, with at least passing reference to 187 languages, several in some detail. The nine chapters in this book cover two major areas: (i) phonetic and typological data, and (ii) an analysis of these and their implications for broader phonological theory.

Phonetically, Z establishes the importance of rime duration in production and perception of contours. In his typological survey, contours are more likely to be found in syllables that are either long, stressed, final, or belonging to a short word. All of these factors involve increased duration of the syllable rime. He devotes an entire chapter to detailed instrumental studies of duration in five languages, showing that syllables that can bear contour tones are significantly longer than those that cannot.

Building on these phonetic studies, Z proposes a tonal complexity scale, which indicates the phonetic complexity of a tone, and a phonetic index C_CONTOUR, which is the weighted sum of the duration of the vowel and of the rime, and is an indicator of the ability of a syllable to bear a contour tone.

Turning to more formal analysis, both a representational account, in which a tone is associated to a mora, and a general-purpose positional markedness account, in which initial syllables are advantaged, are shown to be inadequate to account for the data Z discusses. A contrast-specific markedness account, which identifies longer rime durations as favored for contours, fares somewhat better, but Z’s preferred approach is a more direct phonetic one, in which duration and sonority can be computed and compared within a word. He does this and develops a direct optimality theory approach to incorporate his phonetic insights. He applies this model to five languages, presenting data for most of them.

The main interaction with broader phonological theory is with positional markedness. With many studies showing initial syllables exhibiting more contrasts than others, the assumption has often been that such prominence on initial syllables applies to all features. However, Z clearly shows that this does not apply to tonal contrasts, and thus positional prominence effects can be feature-specific.

Any time someone tries to propose universals, they leak. At times in this work there is a tendency to shoehorn recalcitrant languages into a box for which the fit is dubious. For example, Bantu languages are simply reported to have penultimate stress (on the basis of long vowels), but it is far from certain that this can properly be called stress. In addition, two languages in which contours are limited to final syllables (even though there are nonfinal long syllables, and final syllables may be short) are treated as exceptional cases which need additional explanation, whereas the pattern is quite robust in West Africa and so should be a natural part of the typology predicted by tone theory. Still, this work is a worthwhile contribution to the literature on tone.