Key points are not available for this paper at this time.
Three experiments evaluated the role of acoustic similarity produced by homophones in interference, each experiment representing different learning tasks. In Experiment 1, a list-discrimination task was used and no evidence for interference was observed. In Experiment 2, a long list of paired associates was presented in which the homophones were paired with different response terms. Interference occurred when the level of learning of the competing pairs was about the same but disappeared entirely when one of the ostensibly competing pairs had a high degree of learning. A retroactive inhibition paradigm was used in the third experiment. The amount of interference observed was essentially as great as that found for the traditional A-B, A-D paradigm. It was not clear why the first experiment failed to produce interference whereas the other two did. Overall, it appears that the acoustic identity produced by homophones is always present, but its effects may be reduced or eliminated by adding discriminatory codes, presumably semantic in nature.
Benton J. Underwood (Sat,) studied this question.