I have not read this book, just googling my question.
Although most naive native listeners feel that stress has something to do with loudness (as indeed it can, but not necessarily so), the main perceptual dimension in which stress is signalled is pitch (Lehiste 1970: 153). Compare the noun insult with the verb insult, both said as a statement. In both words the stressed syllable has a higher pitch. In English, however, it is not high pitch per se that signals stress. It is a greater pitch deviation, either up or down, from the pitch on surrounding unstressed syllables. To see this, compare the noun insult said as a statement, with the same word said as a question: 'insult?' In the latter, the pitch of the stressed syllable is no longer higher than the adjacent unstressed syllable, but lower.
From Forensic Speaker Identification by Phil Rose, page 155.
I think that 'pitch change' here must mean the change in pitch from the phoneme just before the syllable to the phoneme starting it, coupled with the change in pitch from the phoneme ending the syllable to the phoneme just after that. To illustrate, I think e.g. in the phrase
- Is that an insult?
there is more change from the /n/ of 'an' to /ɪ/ of 'insult' than there is change from, in 'insult', /n/ to /s/, and it's for this that stress appears on the first syllable of 'insult'.
Or am I wrong and what is meant is something else e.g. some total or average pitch of each syllable?