Introduction

One of the major concerns of laboratory phonology is that of determining the nature of the transition between discrete phonological structure (conventionally, “phonology”) and its expression in terms of nondiscrete physical or psychoacoustic parameters (conventionally, “phonetics”). A considerable amount of research has been devoted to determining where this transition lies, and to what extent the rule types and representational systems needed to characterize the two levels may differ (see Keating 1985 for an overview). For instance, it is an empirical question to what extent the assignment of phonetic parameters to strings of segments (phonemes, tones, etc.) depends upon increasingly rich representational structures of the sort provided by autosegmental and metrical phonology, or upon real-time realization rules – or indeed upon some combination of the two, as many are coming to believe. We are only beginning to assess the types of evidence that can decide questions of this sort, and a complete and fully adequate theory of the phonetics/phonology interface remains to be worked out. A new synthesis of the methodology of phonology and phonetics, integrating results from the physical, biological and cognitive sciences, is required if we are to make significant progress in this area.

The present study examines one question of traditional interest to both phoneticians and phonologists, with roots that go deep into modern linguistic theory. Many linguists have noted the existence of cross-linguistic preferences for certain types of syllable structures and syllable contacts. These have been the subject of descriptive studies and surveys such as that of Greenberg (1978), which have brought to light a number of generalizations suggesting that certain syllable types are less complex or less marked than others across languages.