Ten children with dyslexia and 11 normal readers performed tasks of phoneme mapping (assigning sounds to letters) and morpheme mapping (relating suffixed words to their roots) during fMRI scanning, before and after 28 hours of comprehensive reading instruction, in a study of the effects of reading instruction on brain activation in children with dyslexia at University of Washington, Seattle, WA. Performance of dyslexic children improved significantly by the end of the 3-week intervention. On the Word Attack subtest of the Woodcock Reading Mastery Test-R (phoneme mapping), the mean standard score for dyslexics increased from 87.0 to 93.7 (p=0.03). On a Comes From task (morpheme mapping), the mean score for dyslexics increased from 70.2 to 74.0 (p=0.04). Before treatment and during the initial phoneme mapping task, dyslexics showed less fMRI brain activation than controls in left middle and inferior frontal gyri, right superior frontal gyrus, left temporal, and bilateral parietal regions; during the morpheme mapping scan, activation was significantly reduced in left middle frontal gyrus, right superior parietal, and fusiform/occipital regions. Following instruction, reading scores and brain activation patterns of dyslexics increased, and closely resembled those of controls. The elimination of group differences at follow-up resulted from increased activation in dyslexics and also from decreased activation in controls, presumably due to a practice effect. 
COMMENT. Reading instruction in children with dyslexia is associated with changes in brain activation patterns during specific language processes that resemble those of control normal readers. Previous studies have demonstrated normalization of brain activation patterns in dyslexics during phonological tasks and following successful remedial training . The above study extends this treatment effect to morpheme mapping tasks, and shows that phoneme and morpheme language processes have different brain activation patterns. Treatment of dyslexia increases brain activation in circuits normally involved in processing language function.
Magnetic source imaging (MSI), a combination of MEG and MRI, has been used to study functional neuroanatomy during reading. Dyslexics failed to activate the left visual and receptive language cortical areas during word presentation, but instead, activated the left inferior frontal lobe (Salmelin R et al, 1996). The activation of the left posterior-temporal lobe during reading aloud or silently has been observed in PET studies of normal readers (Price CJ et al, 1994). The most critical area of dysfunction in dyslexic subjects is in the left posterior temporal lobe. A phonological-linguistic basis of dyslexia is most generally accepted (Denckla MB, 1994), and is preferred to the visual system deficit theory (Lehmkuhle S et al, 1993).