Article Text


Developmental dyslexia
  1. W A Lishman
  1. Emeritus Professor of Neuropsychiatry, Institute of Psychiatry, London, UK
  1. Correspondence to:
 W A Lishman
 9 Elwill Way, Beckenham, Kent BR3 3AB, UK;

Statistics from

The term developmental dyslexia ("specific reading retardation") refers to an unexpected difficulty in reading in children and adults who otherwise possess the intelligence, motivation, and schooling considered necessary for accurate and fluent reading.1 The concept has had a chequered career, with debates about its definition, origin, and causes, and indeed about its validity as a distinct entity. Increasingly, however, psychological and epidemiological investigations have lent it respectability and there are strong indications that genetic influences are at work.2 It is thought to affect between 4 and 7% of children, usually equivalent to a retardation of some 18–24 months in reading relative to expectation.3

A growing consensus highlights problems with phonological processing as a core deficit in the disorder, hence the focus on phonology in the majority of remedial programmes. The phoneme is the smallest identifiable unit of spoken or heard language, individual phonemes being assembled together in the construction of words. An important task in learning to read appears to consist in appreciating the correspondence between such phonemes and their equivalent representations ("graphemes") in written language.

A search is now afoot for possible physiological underpinnings to the condition, and in this, brain imaging has come to make a major contribution, as outlined below.


For obvious reasons, brain imaging was first explored in adults with a history of reading difficulties in childhood, even though by adulthood the problem may have ameliorated. In such subjects, a variety of imaging protocols have shown problems with activation of the language cortex during the performance of reading tasks. Thus, in a ground-breaking positron emission tomography (PET) study, Paulesu et al examined five young men who, although ultimately succeeding academically, had had marked dyslexia in childhood.4 In this sense they were "compensated" dyslexics. Appropriate normal reading controls were used for comparison.

During performance of a visually presented phonological rhyming task, the dyslexics activated severely restricted areas of the language cortex, in fact Broca’s area alone, whereas the controls showed widespread activations from Wernicke’s area posteriorly to Broca’s area anteriorly plus the tissues of the insula in between. On a phonological memory task, by contrast, the dyslexics activated the posterior language areas, including Wernicke’s area and the supramarginal gyrus, but Broca’s area only weakly. In neither task did the dyslexics activate posterior and anterior language areas in concert with one another. This suggested a degree of "disconnection" between components of the language apparatus, with obvious consequences for ease of translation between different language codes.

Shaywitz et al used functional magnetic resonance imaging (fMRI) on a larger group of dyslexics and controls.5 They employed a hierarchical series of tests that made increasing demands on phonological functions, and the scans were analysed to see which brain areas responded with a corresponding progressive increase in activation.

Among the controls, brain activation increased systematically in posterior cortical areas as the phonological demands increased; in Wernicke’s area, the angular gyrus and the visual cortex. The dyslexics failed to show such an effect. This again pointed to an imperfectly functioning system for segmenting words into their phonological constituents. By contrast, in frontal regions around Broca’s area, the dyslexics showed relative overactivation in response to phonological demands, which was thought to reflect the increased effort they expended on attempts at performing the tasks.

Further studies dealing with whole word reading have shown additional differences between dyslexic readers and controls. Brunswick et al monitored the brain activity produced on PET scans when compensated dyslexics and controls read words presented one at a time on a screen.6 During word reading, all subjects activated the visual cortex, the left temporoparietal language areas, and the articulatory cortex of Broca’s area, but the dyslexics differed from the controls in certain respects. Firstly, when reading silently they showed diminished activation of the left temporoparietal region generally. Secondly, they showed increased activity in the region of Broca’s area when reading aloud, but not when reading silently, which perhaps reflected compensatory efforts to decode print successfully for explicit reading. Thirdly, and perhaps most interestingly, they consistently showed decreased activation of area BA37 in the posterior part of the inferior temporal lobe.

Area BA37 appears to play a critical role in the retrieval of the names of words and objects.7 When damaged, for example after a stroke, the patient cannot find the name for an object he is shown even though he is clearly aware of its identity ("word selection anomia"). Any impairment in the functioning of such an area would clearly contribute to the reading difficulties of dyslexic individuals.


Attention has now turned to studies in children to see how far such findings may already be present during the period of literacy acquisition or whether they largely reflect long-term changes as the person matures. The results to date strongly support the former view.

Temple et al examined fMRI scans during a visual phonological rhyming task in 24 dyslexic children aged 8–12 years (mean 10.7) and controls.8 The normal reading controls activated both the left inferior frontal gyrus and the left temporoparietal area. The dyslexics activated the inferior frontal region well (though in a somewhat more anterior location), but temporoparietal activity was virtually absent. Additionally, on a parallel test of orthographic processing (judgements as to whether two visually presented letters were the same) the dyslexic children activated a greatly reduced area of the extrastriate occipital cortex.

Shaywitz et al examined a larger and rather older group of 70 dyslexic children aged 7–18 years (mean 13).9 Rhyming tasks during fMRI were again employed. In this sample, activations were apparent in left hemisphere posterior sites (both temporoparietal and occipitotemporal), although significantly reduced in degree compared with controls. Left frontal activation was also reduced in the dyslexic children. Thus, support was again obtained for the view that disruptions in important language regions of the brain are already present in childhood dyslexics. Significant correlations were observed between the extent of activation in the posterior brain regions and a measure of reading skill across the full cohort of dyslexic and control children.

Shaywitz et al comment on the contrast between the impaired frontal activity in their childhood dyslexics and the increased frontal activity observed in some adult samples. An age analysis on their cohort showed that the older dyslexic children engaged the frontal systems to an increasing extent as the phonological demands of the tasks increased, suggesting therefore that this represents a compensatory process.

Finally, Simos et al have applied magnetoencephalography (MEG) to the problem, examining 10 dyslexic children (mean age 12.6 years) and 8 normal reading controls during performance of a visual word reading task.10

The results were striking. The left basal temporal cortex in the vicinity of the fusiform and lingual gyri was activated first in all subjects (within 200 ms), presumably representing the pre-lexical analysis of print. Normal readers then activated the left temporoparietal language regions of the brain (superior temporal, angular and supramarginal gyri) within some 300 ms, whereas the dyslexics in the main activated the corresponding regions of the right hemisphere. Only one of the dyslexic children showed reliable left temporoparietal activity, and this was delayed and weak in comparison with the right-sided activity.

Simos et al thus concluded that there are marked and consistent differences in the patterns of brain activation between young dyslexic and normal readers during the first half second after reading the printed word. Aberrant patterns of functional connectivity between the basal temporal cortex and the key language areas of the left hemisphere appear to be responsible.


A crucial question concerning the primacy or otherwise of the brain changes observed in dyslexia must lie in whether they are reversible in some degree with remediation. Rather surprisingly, this issue has already been tackled, and both Temple and Simos report encouraging results.

Temple et al have reported a move towards normalisation of fMRI scans in 20 dyslexic children after an intensive 8 week course of therapy.11 This consisted of a computerised battery of exercises (Fast ForWord Language) including, for example, practice in auditory discrimination, phoneme identification, and language comprehension. After treatment, the left temporoparietal cortex showed activation when this had not been present before. The left frontal activation had moved more posteriorly to the area seen in controls. Moreover, significant correlations could be observed between the magnitude of increased left temporoparietal activation and improvements on certain measures of language ability and phonological awareness.

Other brain regions, not active in controls, also showed activation post-treatment, including right hemisphere areas homologous to the language areas on the left. This may have reflected compensatory processes. It was also clear that the new left temporoparietal activation was in a region near to, but not identical with the focus seen in normal controls, indicating that the return towards normality was as yet incomplete. More could scarcely be expected after so brief an intervention.

Simos et al have repeated MEG studies on eight dyslexic children (mean age 11.4 years) after an 8 week period of therapy.12 This consisted of approximately 80 hours of one to one instruction focused on the development of phonological processing and decoding skills. On this occasion, they employed a visual rhyming task in conjunction with the MEG recordings

Before the intervention, all of the dyslexic children had shown little or no activation in the left temporoparietal regions, in sharp contrast to the controls, and the predominant activity had been in the homologous regions of the right hemisphere. After treatment, all showed dramatic changes in regional activation profiles; activation in the left superior temporal gyrus now exceeded that on the right, with non-significant trends in the same direction for the supramarginal and angular gyri. All children showed significant gains in reading skills, and a strong correlation was observed between improvement in response accuracy on the rhyming task and the degree of increased activation in the left superior temporal gyrus.

Again, however, there were indications that the restoration towards normality was incomplete, in that the time to peak development of left superior temporal gyrus activity was longer in the treated dyslexics than among the normal reading controls (at 800 ms and 600 ms respectively).


These relatively recent investigations give strong support for the validity of the concept of developmental dyslexia along with evidence of its neurobiological basis. The results are impressive, not least in revealing dysfunctions in areas where one would expect to find them; in brain regions known to be involved with language generally and with phonological processes in particular. Such dysfunctions appear to be present from an early age, at least from the period of learning to read, yet they have proved to be amenable in some degree to modification with training. This reinforces the importance of identifying vulnerable children at the earliest opportunity and engaging them in appropriate remediation.

Several questions follow on from these findings, such as possible origins for the disturbed brain physiology, its cerebral substrate, and its universality among poor readers generally. In this last regard, it may also be asked how broadly it applies to languages other than English.

With regard to its origins, the obvious contenders lie with genetics and/or adverse intrauterine events, but these may not be the whole answer. Brain development continues through childhood and possibly adolescence, and becomes "fine tuned" in relation to environmental influences. Those connections that are activated appropriately become strengthened and endure, while others are pruned and discarded. When the child becomes involved with language, such modifications no doubt affect the language systems of the brain to a substantial degree. Thus, both genetic and environmental influences may contribute in varying degrees to the final shaping of the dyslexic brain.

The cerebral substrate underlying the dysfunction also remains mysterious. Galaburda and colleagues have reported cortical dysplasias and ectopias in occasional dyslexic subjects at autopsy,13 sometimes in the language cortex itself, but it is impossible to gauge how common these may be. A relative underdevelopment of the left temporal lobe has been suggested by Eliez et al, who found a 12% reduction in volume on magnetic resonance images among dyslexic men, affecting the grey matter predominantly.14 Klingberg et al have found evidence on diffusion tensor imaging (DTI) pointing to microstructural abnormalities of the temporoparietal white matter in adults with reading difficulties, the relevant axons being mainly anteroposterior in orientation.15 DTI is a development of MRI that reflects the integrity of axonal membranes and myelin sheaths, and the coherence of axonal orientation. Significant correlations were observed between reading scores and the severity of the changes in the left white matter tracts that contain the connections between posterior and anterior language areas. However, all such findings rest to date on the investigation of very small numbers of cases.

With regard to the entire spectrum of poor readers, it is noteworthy that brain imaging research has so far been carried out exclusively on subjects with severe and well-diagnosed dyslexia, leaving uncertain the status of the long tail of "other impaired readers". In some, the reading difficulties may appear to have derived from lack of sufficient educational opportunity, or deprivation from an early age of adequate encouragement and stimulation. The question arises whether they too would show abnormal patterns of brain activation when processing written language, or whether this is the prerogative of a small sub-sample alone. Thus, it remains to be determined how far brain imaging will ultimately reveal differences between those who labour with dysfunctional brains from the outset, and those whose reading difficulties have social rather than innate biological causes.

The question also arises whether English, with its so-called "deep orthography", is unique in leading to dyslexic difficulties in association with this particular neurobiological background. English uses 1120 graphemes to represent the 40 phonemes of the language; Italian, by contrast, uses 33 graphemes to suffice for its 25 phonemes. The prevalence of dyslexia across different languages appears to be related to the depth or shallowness of their orthographies. Paulesu et al have nevertheless found that English, Italian, and French dyslexics all show equivalent reductions in activation of the key brain regions known to be affected in English-speaking dyslexics when PET scans are carried out during word reading.16 Whether or not there is a universal brain basis for developmental dyslexia in yet other languages remains to be determined.

Finally, this work on dyslexia may be viewed in the context of other forms of learning disability such as difficulties with numeracy or fine manipulative skills. These too may prove to have distinctive correlates in functional brain changes if carefully examined by modern techniques. It is probable that research in dyslexia has at the moment simply taken the lead because problems with reading are so damaging and disabling in present day society.


View Abstract


  • Adapted and extended from The Marjorie Lishman Memorial Lecture, delivered 8 May 2002

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.