A 300-item list of concrete and abstract nouns with varying frequency of occurrence was presented at a 1-sec rate to 144 subjects under imagery or nonimagery instructions. Subjects were then tested on 72 word pairs homogeneous and heterogeneous with respect to concreteness, where the more frequent member of each pair was to be chosen. Frequency discrimination was found to be a function of relative rather than absolute frequency differences between pair members. In addition, a subjective frequency bias for abstract items was found, with the best performance when the more frequent alternative was abstract and less frequent alternative concrete. The worst performance was for the reverse condition, while that for pair types homogeneous with respect to concreteness fell in-between, with better performance when both alternatives were concrete. The results suggest that the role of imagery may be to produce more discriminable subjective relative frequency differences between alternatives and that the imagery effect generally found in verbal discrimination learning may be reconcilable with frequency theory as it currently stands.One of the more reliable fmdings concerning verbal discrimination learning (VOL) over the last few years has been the so-called "imagery effect" when concrete and abstract word materials are employed. Differential performance on word pairs homogeneous with respect to concreteness (i.e., image-arousal capacity) is typically found with faster learning when both members of a pair are concrete rather than abstract in mixed as well as unmixed list designs (Paivio & Rowe, 1970Ullrich & Balogh, 1972).An attempt was initially made (paivio & Rowe, 1970) to integrate this phenomenon within the general theoretical framework of frequency theory (Ekstrand, Wallace, & Underwood, 1966;Wallace, 1972) which has thus far served well to explain VOL. This attempt took the form of assuming implicit imaginal response evocation to correct high-imagery items more than to incorrect high-imagery items and differential frequency accrual analogous to that postulated by frequency theory for implicit associative responses (lARs). Subsequent failures to confirm several predictions based upon this assumption led to the abandonment of frequency theory interpretations in favor of image "tagging" or differential learning strategy interpretations and to the conclusion that imagery and frequency are independent processes in VDL (Rowe, 1972a, b).It now seems that the abandonment of frequency theory may have been somewhat premature. Wallace, Murphy, and Sawyer (1973), in a recent monograph, pointed out that the critical requirement for a frequency theory interpretation of the imagery effect is a demonstration that correct and incorrect alternatives of concrete word pairs are more discriminable in terms of Requests for reprints should be sent to George D. Goede!, Department of Psychology, State University of New York, College at Geneseo, Geneseo, New York 14454. 53 subjective frequency than are alternatives of abstract word pairs. Usi...