Beckman and Edwards have presented experimental evidence for the association of syllabic lengthening in speech with two phonological effects. One is the presence of an intonational phrase boundary. Lengthening associated with intonational phrase boundaries occurs consistently across speakers and speech rates. The other effect is the presence of a word boundary. In contrast to phrase-final lengthening, word-final lengthening occurs inconsistently. It is more evident at a slow rate of speech; even then, not all speakers show it with all sentences.

Beckman and Edwards's aim in examining the occurrence of lengthening effects is to make claims about the proper model of English phonological structure. They relate their experiments to a long tradition of phonological interpretation of timing effects in speech, with two separate (and as Beckman and Edwards point out, at least in part incompatible) strands: lengthening which is claimed to accompany the terminal boundary of a phonological unit, and shortening which is claimed to adjust the interval between stressed syllables in the direction of greater regularity. The effects which Beckman and Edwards found in their experiments are of the first sort.

This commentary concerns the assumptions on which undertakings such as Beckman and Edwards's are based. Precisely how does a speaker's performance shed light on the phonology of a language?

Two useful distinctions may be drawn. The first is between what one may call a strong and a weak version of the claim to “psychological reality.”