Four experiments tested whether metaphor increases tenor attribute salience. (Tenor and vehicle refer to the term represented in a metaphor and that representing it, respectively.) Ortony's (1979) salience-imbalance hypothesis predicts that of the possible tenor-vehicle shared attribute pairings (low-low, low-high, high-low, high-high) only low-high pairings are highly metaphoric and that one purpose of metaphor is heightening tenor attribute salience. Together, these concepts imply that low-high tenor-vehicle pairings (only) should cause increases in salience of tenor attributes. Experiment 1 measured subjects' "typicality" ratings of attributes as defined as " ... the degree to which the property associated with the word is rated as being the best or prototypic example of the property ... " (Katz (1982), p. 285). Metaphors are predicted to increase ratings for tenor attributes in low-high pairs. Experiment 2 measured the number of subjects listing a given attribute for a noun. The hypothesis predicts that the attribute shared in low-high pairs will be listed for the tenor more frequently after metaphor treatment. In Experiment 3, naming latencies for attributes task following tenor and vehicle primes were the measure of salience. Metaphors were predicted to reduce naming latencies for tenor attributes in low-high pairs due to increased association. A fourth experiment tested whether relative differences in tenor-vehicle attribute salience are sufficient to effect changes in RTs when tenor and vehicle attribute saliencies are both low or both high. All experiments used randomly constructed Noun-A-is-a-Noun-B metaphors without context. The Pretest/Posttest x Treatment x Tenor/Vehicle interaction for a 24-hour posttest delay was significant in Experiment 1, indicating a stronger effect of metaphor on subjects' likelihood of listing target tenor attributes than on vehicle attributes. In Experiment 2, shared attributes were listed more often after metaphor treatment for all types except high-low. It is concluded that while the evidence fails to support Ortony's hypothesis as stated, a modified version allowing for relative differences could account for the results. The model proposed allows for three salience pairings having metaphoric potential (low-low, low-high, and high-high) as long as the tenor salience is lower than the vehicle salience. (Dissertation Abstracts)