the model speaker, rather than providing a nondistorted motor template for the person who stutters to match, instead provides a pacing or rhythmic one (Johnson & Rosen, 1937); a perspective that would appear consistent with Kalinowski et al’s findings. It is worth noting that all of those mentioned above have also been levelled at the fluency enhancing properties of delayed auditory feedback (DAF), which we discuss below. (The relationship between choral speech and DAF is an important one, and we return to this with regard to therapy in chapter 14.)
This is a type of cued speech which is very closely related to choral and unison versions. Technically, shadowed speech occurs where there is a slight delay between the speech of the model speaker and the person who stutters, as opposed to the simultaneous output produced during unison and choral speech. The difference is that while with choral speech the speaker knows exactly what the model speaker is going to say, shadowed speech can be used to follow the novel speech of the model speaker. Like choral/unison speech, shadowing can produce dramatic results (Cherry & Sayers, 1956; Kelham & McHale, 1966), but like them the gains in fluency tend to be lost once the stimulus of the model speaker has ended. Because of this, the use of choral or shadowed speech is now rare, and usually confined only to moments in therapy or assessment, where it is considered important to have the client experience a moment of fluency, albeit in the knowledge that this method of producing it will not provide any basis for sustainable improvement. What is interesting from our present perspective, however, is the potential relationship between shadowed speech and delayed auditory feedback. As we will see in chapter 16, the fluency enhancing effect of shadowed and choral speech has been put to use in devices which use DAF and frequency auditory feedback (FAF) to approximate the effects of speaking alongside other speakers.
Delayed auditory feedback
It is now over 40 years since Goldiamond and colleagues first stumbled on the potential fluency enhancing effects of delayed auditory feedback (Flanagan, Goldiamond, & Azrin, 1958, 1959; Goldiamond, 1965). Findings from the earliest experiments centred around the vicarious discovery that some people who stuttered experienced improved fluency when they put on headphones and heard their speech played back to them with a slight time delay. (Some readers may already have experienced DAF as an echo effect when speaking on a poor transcontinental telephone line.) Commonly, DAF also results in reduced fluency in nonstuttering speakers (Fukawa, Yoshioka, Ozawa, & Yoshida, 1988; Stuart, Kalinowski, Rastatter, & Lynch, 2002), although
Fukawa et al. observed that people who stutter were significantly more likely to be affected by DAF than nonstutterers, and that male nonstutterers were more susceptible to the effect than females. Most noticeably, Goldiamond (1965) found a tendency for speakers to slow their rate of speech in an effort to counteract the disruptive influences of the delayed feedback. Particularly, at around 250 ms delay (0.25 of a second) a prolonged speech pattern was produced, where vowels became disproportionately more stretched than consonants. The further finding that the extent of the prolonged speech could be controlled by altering the delay times lead to the development of a number of “prolonged speech” programs which used DAF in a systematic way to elicit fluent speech. (See chapter 12 as to how prolonged speech programs have developed.) During the early stages of therapy, DAF was set to encourage excessive prolongation, usually around 250 ms. When clients were able to demonstrate 100 percent fluency in their speech at this delay setting, the next stepwise decrease in DAF (usually in 50 ms increments) was introduced to encourage a slightly faster rate of speech. Again, the client learned to control fluency using decreased prolongation associated with the reduced DAF. The procedure was then repeated at incrementally reduced delay levels, with clients having to demonstrate completely fluent speech at each one before progressing to the next decreased DAF setting. Eventually, the client reached the point where he was able to maintain fluency without any delay in auditory feedback (e.g., Curlee & Perkins, 1969, 1973). At this time it was thought that the fluency enhancing effects of DAF could be explained simply as byproducts of the slower rate speech that it produced. During the mid-1970s and through the 1980s there was a lull in DAF research as clinicians looked to alternative ways of slowing speech for therapy. It was not until the early 1990s when a resurgence of interest occurred, largely driven by findings that increased fluency could indeed result under DAF at normal and even fast rates of speech (e.g., Stuart & Kalinowski, 1996). This finding has led to a new generation of clinicians and researchers becoming interested in DAF as a treatment option for stuttering. We examine the more recent applications in relation to therapy elsewhere (see chapter 14).
Aside from the therapeutic implications, the early findings that DAF could enhance fluency for at least some people who stutter led to a number of theories of stuttering, based on the assumption that timing perception is disturbed.
 Subsequent findings as to the optimal time delay settings have varied widely (see chapter 14).