Effect of Subjective Interletter Similarity, Perceived Word Similarity, and Contextual Variables on the Recognition of Letter Substitutions in a Proofreading Task

Measures of interletter similarity, e.g., template matching, feature analysis, subjective rating, or tachistoscopic confusability, have not yet been used to predict or explain perceptual performance in tasks involving actual reading. The validity of a measure of subjective interletter similarity is investigated by examining the correlational relationship between this index and the ability of subjects to detect letter substitutions systematically introduced into a prose passage. As expected, recognition of errors appears to depend upon subjective interletter similarity, but this effect seems to be mediated by the perceived similarity between words with and without their typographical errors. Further, these effects seem to depend upon various contextual variables such as letter frequency, word frequency, and word length.