This project seeks to illuminate the role that implicit speech processes play in silent reading. Although subvocal activity has long been known to occur during reading, its significance and the conditions controlling it are not well understood. There is disagreement about whether this speech activity is functional or merely a by-product of no significance. The central assumption of the theory underlying this project, automatic activation theory, is that identification of printed words routinely causes activation of speech segments associated with letters and with the words themselves. This contrasts with the view ("dual route" theory) that printed words are sometimes "recoded" into speech and at other times identified directly through the "visual route." A second assumption of automatic activation theory is that the speech code, once activated, remains part of the mental representation of a sentence until certain comprehension processes have been completed. The two assumptions of automatic activation theory will be tested in two types of experiments: One type will use backward visual masking, an experimental procedure in which a word is flashed on the screen for a very short period of time, followed by a brief flash of a "mask", a visual display that may or may not share specific characteristics of the word it follows. Recent research has found that the degree of disruption caused by a mask depends both on the degree to which the mask shares speech segments and also on the degree to which the mask shares letters with the word it masks. The masking experiments will vary characteristics of the word being masked that usually affect the speed at which it is identified; this will provide a situation in which the automatic activation and the dual route theories make different predictions. The second kind of experiment will focus on the role of speech in reading by examining the visual tongue-twister effect: The time for silent reading of sentences that repeat speech segments, e.g., "Peter Piper picked a peck of pickled peppers..", is longer than for normal sentences. The experiments will address whether such effects occur in extended text reading and whether they are diminished when the reader has a goal of gist comprehension rather than detailed understanding, situations for which the two theories make different predictions. This research has implications for the nature of reading disability. According to available evidence, it is quite likely that some kinds of reading disability are linked to defective use of speech segments in learning to read and in memory for linguistic input. To the extent that this project sheds some light on normal reading processes, it may help in understanding the nature of the speech-related defect in some cases of disability.
