This study consisted of two experiments. In the first experiment, 40 college students gave frequency ratings for concrete and abstract words which were equated on normative frequency. From the results it was concluded that abstract (low imagery) words, even though the two sets of words are of equal frequency. In the second experiment, different subjects learned verbal discrimination lists consisting of both abstract and concrete pairs. While the usual concreteness effect was obtained when abstract and concrete items differed widely on phenomenal frequency, it disappeared when these items were equated on perceived frequency. The finding that appropriate frequency manipulations can eliminate the concreteness/imagery effect, coupled with similar findings for other stimulus characteristics, lends strong support to the frequency theory of discrimination learning. (Author/TD)