The largest body of evidence relevant to the role of short-term memory as a working memory system comes from studies of reading. A similar conclusion comes from a recent experiment by Coltheart, Avons, and Trollope. The Daneman and Carpenter study used a small group of subjects of high ability but nevertheless obtained correlations in excess of 0.30 between memory span for words and reading comprehension scores. Gathercole and Baddeley argue that their non-word repetition task provides a measure of 'phonological memory' and that this explains its ability to predict vocabulary development. The existence of good general cognitive skills in people with severe short-term memory problems places clear limits on theories about the relationship between short-term memory skills and cognitive development. Cycling information through these two interfaced systems amounts to a simplified model of sub-vocal rehearsal.