The English Rhythm Rule as an accent deletion rule

The culminative nature of stress has led to three common assumptions about its representation. These are (i) that it should allow for an unbounded continuum of prominence levels (the INFINITY assumption); (ii) that it should characterise prominence levels as relative to nearby weaker or stronger prominence levels, not as levels that are phonetically interpretable in isolation (the RELATIVITY assumption); and (iii) that all levels of prominence should be represented in terms of the same mechanism (the UNIFORMITY assumption). The infinity assumption is implicit in the Chomsky & Halle (1968) w-ary feature [stress], in Liberman & Prince's (1977: 263) metrical tree, as well as recent interpretations of the grid (Halle & Vergnaud 1987: 35). The relativity assumption was made explicit in Trager & Smith (1957: 35) and Liberman & Prince (1977: 262). The uniformity assumption was commonly held until Liberman & Prince split off the binary feature [stress] (to represent the distinction between reduced and unreduced syllables) from the metrical tree (to represent higher levels of prominence), which formed the basis for Selkirk's (1980) proposal to postulate the foot as a prosodic constituent. The claim defended in this paper is that a representation of stress must be adopted which provides for a finite number of prominence levels (rejecting the infinity assumption); which is locally interpretable (rejecting the relativity assumption); and which is differential in the sense that not all levels of prominence are represented in terms of the same mechanism (rejecting the uniformity assumption).