Experiments such as Savin and Bever's (1970) , in which listeners detect the presence of phonemes or syllables in lists of syllables, are meaningless for determining the perceptual units of speech. We show that the minimum reaction time in such experiments occurs whenever the linguistic level of the target and the search list is the same. Since this level is determined by the experimental design, it is impossible to use the occurrence of minimum reaction time to identify perceptual units. Theoretical questions raised by monitoring experiments include the strategies used in searching lists for targets and the phenomenon of linguistic “transparency,” presumably the basis for the increases in reaction time when the level of the target and search list are different.