Tested the hypothesis that the rhyming relation restricts the range of response alternatives to the stimulus, practically converting recall into a recognition test in 3 experiments with 30, 12, and 18 undergraduates, respectively. In Exp. I, an assonance (change of last phoneme) rule for pairs, which restricts alternatives about as much as a rhyming rule, facilitated performance as much as did a rhyming rule. In Exp. II, when response alternatives were equated by multiple-choice tests for memory, the advantage for rhyming pairs vanished. Exp. III showed that the presence of some rhyming pairs in a list induced S to generalize this rule inappropriately to other pairs, thereby suffering interference on those pairs composed by re-pairing rhyming units. (PsycINFO Database Record (c) 2006 APA, all rights reserved)