[dry run for talk at conference on Music, Language and the Mind, Tufts University, July 2008]
Much empirical evidence shows that pitch levels at key points in spoken sentences correspond in highly lawful ways between one speaker and another and also between e.g. normal and raised voice within the same speaker. This makes it possible for listeners to normalize away from differences of pitch range and establish phonological equivalence between one absolute pitch value and another. This ability is obviously crucial in a tone language, but also affects the intonational significance of local peaks, local valleys, endpoints, etc. in languages like English. Though such phonological equivalence is sometimes invoked in discussions of the psychophysics of pitch, it is quantitatively quite distinct from anything in music, in that it involves nothing comparable to chroma or octave equivalence. The normalization is not based on a uniform psychophysical pitch scale (e.g. semitones, ERB, Bark), but rather on proportional equivalence relative to a given pitch span (e.g. mid tone might be defined as 40% of the distance between the bottom and the top of the speaker's pitch span). It is thus conceptually more comparable to establishing musical values like "dominant" and "leading tone" relative to a given key, and arguably cognitive rather than psychophysical in nature.