This paper explores a correlation between phonology and phonetics. It first reviews a phonological analysis that proposes that all full Mandarin rhymes are heavy and that all Shanghai rhymes are underlyingly light. Then it reports a small phonetic experiment that attempts to determine whether there is a phonetic correlate for the phonological claim. Four Mandarin speakers and five Shanghai speakers were recorded, each reading five sentences four times at normal speed. Average syllable durations were determined. It was found that the average syllable duration in Mandarin was 215 ms and that in Shanghai was 162 ms. Statistics show that the durational difference is significant. The result thus agrees with the phonological analysis. Implications and limitations of the present study will be discussed.

Perhaps the most important insight in phonological theory since the introduction of the concept of the phoneme has been the role that distinctive features play in phonological theory (Jakobson et al. 1952). Most research since Jakobson's early formulation has focused on the segmental properties of these features without reference to their hierarchical organisation. Recent research, however, has shed considerable light on this latter aspect of the phoneme as a phonological unit. In his seminal article ‘The geometry of phonological features’, for example, Clements (1985), building on earlier work of scholars such as Goldsmith (1976), argues that features are not ‘bundles’ in Bloomfield's sense, but are, in fact, organised into phonological trees with each branch corresponding to what has been called a tier. An overview of the current state of feature geometry can be found in Clements & Hume (forthcoming) and Kenstowicz (1994).

The study of the relation between syntactic structure and phonological representation has attracted the attention of many phonologists in the past few years. One important contribution to this field of study is Chen's (1987) work on Xiamen Chinese tone sandhi domains. He suggests that the syntax–phonology relation appeals to syntactic information such as category types and the edges of syntactic bracketings. This insight has been further elaborated in the general theory of the syntax—phonology relation of Selkirk (1986). In this theory, the relation between syntactic structure and prosodic structure above the foot and below the intonational phrase is defined in terms of the edges of syntactic constituents of designated types. More precisely, this theory incorporates two hypotheses. One is that there are designated category types in syntactic structure with respect to which one end (Right or Left) of the designated category is relevant in the formulation of a prosodic constituent C, which extends from one instance of the appropriate end (R/L) of the designated category to the next. This hypothesis has been called the End Parameter.

One of the major aims of linguistic theory is to determine what is universal vs. language-specific within grammatical systems. In phonology, for example, a number of universals have been proposed and incorporated into the various subtheories that deal with segmental and prosodic aspects of sound systems. In his original autosegmental theory, for instance, Goldsmith (1976) provided a formalism and a set of principles embodying a number of universal claims about how different tiers may link to each other. Most of the support for this theory came from the study of tone: tones (Ts) were said to reside on separate ‘tiers’ joined by association lines to their respective tone-bearing units (TBUs).

Recent work on Articulatory Phonology (Browman & Goldstein 1986, 1989, 1991, 1992a, b) raises a number of questions, specifically involving the phonetics–phonology ‘interface’. One advantage of using Articulatory Phonology (henceforth ArtP), with its basic units of abstract gestures based on articulatory movements, is its ability to link phenomena previously seen as phonological to those which are conventionally described as allophonic, or even lower-level phonetic effects, since ‘gestures are... useful primitives for characterising phonological patterns as well as for analysing the activity of the vocal tract articulators’ (Browman & Goldstein 1991: 313). If both phonetics and phonology could ultimately be cast entirely in gestural terms, the phonetics–phonology interface might effectively cease to exist, at least in terms of units of analysis.

Korean possesses a rich onomatopoeic and mimetic vocabulary that may be augmented by partial reduplication and/or suffixation. This augmented category, which is referred to as partial extension in Jun (1993), is semantically characterised by a lengthening, or temporal extension of the base form:

Ever since Baudouin de Courtenay (e.g. 1917 [1990: 500]), the founder of phonology, it has been customary for every new phonological theory to make claims about the particular level(s) at which rules of versification may refer to phonological representations. Baudouin himself, for example, took rhyming to be evidence of phonemic identity. The birth of generative phonology thus led to a series of claims that rules of versification may have access to underlying representations – or to various intermediate levels of representation more shallow than the underlying but still deeper than phonemic. Starting with Zeps (1963) and for the next twenty-odd years a certain number of such arguments have been published, and it appears that their validity is widely assumed in the field, as reflected in such survey articles as O'Connor (1982: 155–156) and Hayes (1988: 228–229). Presumably it is precisely the widespread acceptance of these arguments that explains the relative lull in the debate recently (but see Malone 1982,1983, 1988a, b).

The structural relationship of the primary/consonantal place features and the secondary/vowel place features has been the focus of much recent work in geometrical and hierarchical feature theories. Sagey (1986), Halle (1986), Selkirk (1988, 1991a, b) and others propose representations in which the secondary/vowel place features are structurally dependent on primary/consonantal place features, while Archangeli & Pulleyblank (1987), Steriade (1987), Clements (1989, 1991), Hume (1990, 1992) and Odden (1991) argue for the structural independence of primary/consonantal and secondary/vowel place features, though to different degrees.

Choctaw verbs form intensives by a complex procedure of (apparent) infixation, gemination and accentuation. Verbs of all shapes have two distinct intensive forms, which Ulrich (1986) distinguishes as the y-grade (involving a geminate yy) and the g-grade (involving gemination of a stem consonant or a falling tone). Lombardi & McCarthy (1991) analyse Choctaw intensives in terms of the theory of prosodic circumscription (McCarthy & Prince 1990). Hammond (1993) gives an alternative analysis within the same theory. While insightful in certain respects, these analyses fail to account for the full range of Choctaw intensives, instead generating a single intensive for any verb. In the present paper, I propose a unified analysis of Choctaw intensives, with two, minimally different, procedures for generating the two intensives of any verb.

In a typical Athapaskan language, consonants divide into three manners of articulation, stops/affricates, fricatives and sonorants. There are generally three major laryngeal classes in the stops/affricates, two in the fricatives and a single class in the sonorants. In this paper, I examine how laryngeal oppositions in stops/affricates and fricatives are best characterised. I argue in §§ 2—7 that stops/affricates (henceforth stops) are phonologically voiceless unaspirated, aspirated and glottalised, while fricatives exhibit a voiced-voiceless distinction. These classes do not account for all laryngeal interactions, however; evidence exists that aspiration and voicing cross-classify the stops and fricatives, with aspiration being necessary for fricatives and voicing for stops in late rules of the phonology; this is the topic of §§8-9. In addition, evidence suggests that while stops/affricates form a single class underlyingly, they are distinct at a late level. Implications of the analysis are discussed in §§10—11. Processes involving laryngeal features suggest a two-level model of phonology in which contrastive features are present underlyingly and predictable features are added postlexically for reasons of phonetic implementation. The absence of predictable features underlyingly argues against recent work that suggests that such features may be present at this level (e.g. Steriade 1992a, b, 1993b); their postlexical presence argues for a model in which the postlexical phonology is identifiable as the phonetic implementation component of the grammar (e.g. Mohanan 1986; see also Cohn 1993).

Basaa, a zone A Bantu language spoken in Cameroon, is only one among many genetically unrelated languages for which the positing of phonetically null phantom consonants facilitates a phonological account of certain otherwise unexpected surface forms encountered in derivational paradigms. Clements & Keyser (1983), Marlett & Stemberger (1983), Keyser & Kiparsky (1984), Crowhurst (1988) and Hualde (1992) propose that phantom consonants exist in Turkish, Seri, French, Finnish, Southern Paiute and Aranese Gascon, for example, syllabifying as onsets or codas where appropriate and in certain cases inducing the gemination of an adjacent consonantal segment or the lengthening of a preceding tautosyllabic vowel, as we shall see takes place in Basaa.

Megan Crowhurst's ‘Diminutives and augmentatives in Mexican Spanish: a prosodic analysis’ (1992, henceforth CRO) takes a fresh look at the thorny problem of explicating the distribution of the allomorphs -itV and -(e)citV of the Spanish diminutive suffix. CRO has conspicuous virtues. Among them is the limitation of data to a single dialect, namely Sonoran, of northern Mexico. This narrow focus is insurance against inadvertently mixing incompatible data, a special risk posed by the considerable dialect variability of productive diminutivisation. At the same time, CRO broadens its data base by taking into account the partial parallelism between diminutive -itV/-(e)citV and augmentative -otV/-(e)sotV.