Introduction

As Beckman and Edwards point out, independent lines of research identify two apparently distinct durational phenomena in speech utterances. Pre-boundary lengthening occurs at the far edge of a domain that, until recently, was identified as syntactic (e.g. Klatt 1975; Cooper and Paccia-Cooper 1980; but see Gee and Grosjean 1983). Stress-timed shortening (as Beckman and Edwards call it) shortens a syllable when there are many of them in a stress foot as compared to when there are few (e.g. Pike 1947/1968). This effect is generally measured by comparing the duration of the stressed syllable in a foot as a function of the total number of syllables in the foot (e.g. Lehiste 1973). Conventional accounts of this phenomenon ascribe it to a rhythmical constraint by which some languages attempt to maintain isochrony between stressed syllables (Abercrombie 1967).

Possibly these two classes of effect are as independent and unrelated as their treatments in the literature suggest. One is described as lengthening, the other as shortening; one occurs at the edge of a domain, the other seems to have to do with prominence peaks in a domain more than with domain edges; by some accounts the domains for the one are syntactic, while those for the other are prosodic or metrical. Alternatively, however, these differences may be more apparent than real. Stressed syllables in monosyllabic stress feet that are identified as unshortened in a stress-timing account can just as well be identified as lengthened; so the absence of stress-timed shortening can instead be described as a lengthening of a stressed syllable at the right edge of a foot.