Abstract
Fleming (1976) criticized Rubenstein et al.'s (1971) evidence for phonemic encoding and conducted two studies of his own which he interpreted as supporting an alternative process. His alternative has considerable explanatory power, but Fleming based his evidence for it upon Rubenstein et al.'s RTs for a sample of 69 nonwords and upon being able to prove that the variance between these RTs was not explained by phonemic encoding. The Rubenstein RTs, however, had previously been re-analyzed by Clark (1973) who contended that the reported differences could not be generalized beyond the particular items used in the experiment. Thus, Fleming's evidence for his alternative suffers the same weakness that raised questions about the original experiment. In the process of examining that weakness, we were led to re-analyze the statistics of previous experiments, and we unearthed several serious errors. Our re-analyses strengthen the evidence for phonemic encoding and suggest that Fleming's “alternative” should be considered as another source of variance affecting recognition RTs, not as a mutually exclusive alternative to phonemic encoding.