Automaticity training for dyslexics: An experimental study.

Less-skilled readers may experience a processing bottleneck in reading comprehension produced by a failure to automate word recognition. The automaticity hypothesis predicts that training which increases rapid and automatic word recognition will improve comprehension. To date few studies have tested this hypothesis. Our goal was to test the automaticity hypothesis by training dyslexic readers (n=35) to access the meaning of words more rapidly. Training consisted of speeded word games implemented on a microcomputer that provided feedback concerning subjects' speed and accuracy using sound and graphics. Three experimental tasks were administered both before and after training in which trials with trained and untrained stimuli were randomly intermixed. The measures were latency and accuracy of word vocalization, sentence comprehension, and a dual-task procedure designed to measure automaticity. Although pre/post improvement was larger for trained than untrained stimuli, reaction time on the word vocalization and sentence comprehension tasks improved significantly for both trained and untrained stimuli. In contrast, automaticity and sentence comprehension accuracy improved significantly for trained but not untrained stimuli. The training effect in comprehension remained when controlling for increases in word knowledge. The results support the automaticity hypothesis and further suggest that severely disabled readers may benefit from training in automatic word recognition.