Detection errors on the and and: Evidence for reading units larger than the word

Abstract

In five experiments, subjects read lO0-word passages and circled instances of a given target letter, letter group, or word. In each case subjects made a disproportionate number of detection errors on the common function words the and and. The predominance of errors on these two words was reduced for passages in which the words were placed in an inappropriate syntactic context and for passages in which word-group identification was disturbed by the use of mixed typecases or a list, rather than a paragraph, format. These effects for the word and were not found for the control word ant. These results were taken as evidence that familiar word sequences may be read in units larger than the word, probably short syntactic phrases or word frames. A tentative model of the reading process consistent with these results is proposed. © 1977 Psychonomic Society, Inc.