Ill terminal fall and fade to silence, with a slowing down toward the end." 3 3. That pitch and stress are phonemically independent. It is recognized that changes in stress may affect any level of intonation that happens to be running at the moment, but not, for example, in such a way as to raise a Level 2 pitch to a Level 3 pitch; the changes are phonetically slight and phonemically non-distinctive. My purpose in this article is to deny the third assumption, and to reverse the roles of stress and pitch. I shall offer evidence that far from being a non-distinctive by-product or a completely independent variable, pitch is our main cue to stress. 2. EARLIER NoTIONS OF PITCH AS A CuE TO STRESS The idea that stress may depend on pitch is not new. The experiment of John Muyskens in 1931, using kymographic records, purported to show that the familiar noun-verb pairs like permitpermit are distinguished by higher pitches on their stressed syllables. Kenneth L. Pike' and Daniel Jones 5 demolish this argument by pointing out, in Jones' words, that "it often happens in a language that strong stresses are found on low-pitched syllables and weak stresses on high-pitched syllables." To demonstrate this, all we need to do is turn the permit example into a question: permit? The refutation, however, is based on a persistent fallacy: that in order to serve as a cue to stress, pitch must RISE. Recent discussions and descriptions continue to look for this kind of relationship, 6 and, failing to find it, enter a verdict against pitch in general.-1 The analysis of Spanish juncture and intonation made by Stockwell, Bowen, and Silva-Fuenzalida, Language, XXXII (1956), 641-665, which follows Smith and Trager, throws some doubt on the morphemic status of intonation patterns: "A sequence of pitches up to and including a terminal juncture will be referred to as an INTONATION PATTERN. Whether or not such a sequence of suprasegmental elements is a morph remains to be demonstrated on the morphological level of analysis" (p. 661). It is hard to see on what basis the levels and junctures are contrastive units, if sequences of them are not morphs.
