Ms. Seidenberg et al., PSEUDOHOMOPHONE EFFECTS AND MODELS OF WORD RECOGNITION, Journal of experimental psychology. Learning, memory, and cognition, 22(1), 1996, pp. 48-62
Two experiments examined factors that influence the processing of pseu
dohomophones (nonwords such as brane or joak, which sound like words)
and nonpseudohomophones (such as brone and joap, which do not sound li
ke words). In Experiment 1, pseudohomophones yielded faster naming lat
encies and slower lexical-decision latencies than did nonpseudohomopho
nes, replicating results of R. S. McCann and D. Besner (1987) and R. S
. McCann, D. Besner, and E. Davelaar (1988). The magnitude of the effe
ct was related to subjects' speed in lexical decision but not naming.
In Experiment 2, both immediate and delayed naming conditions were use
d. There was again a significant pseudohomophone effect that did not c
hange in magnitude across conditions. These results indicate that pseu
dohomophone effects in the lexical-decision and naming tasks have diff
erent bases. In lexical decision, they reflect the pseudohomophone's a
ctivation of phonological and semantic information associated with wor
ds. In naming, they reflect differences in ease of articulating famili
ar versus unfamiliar pronunciations. Implications of these results con
cerning models of word recognition are discussed, focusing on how pseu
dohomophone effects can arise within models that do not incorporate wo
rd-specific representations, such as the M. S. Seidenberg and J. L. Mc
Clelland (1989) model.