Abstract
A general problem in studying children with developmental dyslexia is how to separate inefficiency in learning on the one hand from exposure to written texts on the other. To evaluate dyslexic children's learning abilities with graphemic materials, we tested their improvement in a condition that minimized previous experience with words (i.e., "novel words") and with the standard, horizontal spatial letter array (i.e., a non-canonical "zigzag" format). We selected five pairs of children with dyslexia and (younger) typically developing readers matched for reading speed and accuracy in these specific conditions. Reading performance on novel words in the zigzag format was measured in 23 sessions; learning curves were fitted by power functions. Similar to typically developing readers, children with dyslexia improved their reading of novel words presented in the new format; however, their rate of learning was slower than that of typically developing readers. Furthermore, their learning to read in the new format did not generalize to novel untrained items, whereas significant generalization was present in typically developing readers. As the failure to generalize learning of the spatial format could not be attributed to reduced experience, it indicates a genuine disability and points to impaired perceptual learning as a factor in developmental dyslexia.