Detecting letters and words in prose passages: a test of alternative theoretical accounts of the missing letter effect.

Abstract

College students searched for either h or the in prose passages in which every h occurred in the test word the. In Experiment 1, passage versions differed in that the critical noun phrases were either the alone (i.e., in citation form as a noun referring to itself) or "the definite article." Many more detection errors occurred for letter than word target items, especially with "the definite article." In Experiment 2, passage versions differed in that a given noun phrase containing the test word the occurred as a subject in one version and an object in the other. Again, many more detection errors occurred when the target item was the letter h than when it was the letter sequence the. Also, with letter but not letter sequence targets, more detection errors occurred for object than subject noun phrases. In Experiment 3, passages were presented either in regular format or with all capital letters. Students made more detection errors with the regular than with the capitals format, many more errors occurred when participants searched for letters than for letter sequences, and the effect of target item was larger with regular than with capitals format. These findings suggest that accounts of detection errors in reading must include the influence of unitization and processing time or attentional allocation.