Dyslexia

Abstract

Dyslexia is a specific, persistent difficulty in learning to read that MeSH classifies under language disorders, arising despite adequate instruction, intelligence, and opportunity. Its most robust cognitive cause is a phonological deficit: an impairment in representing and manipulating the sounds of speech that undermines the letter-to-sound mapping on which reading is built. Reading ability is distributed continuously, so dyslexia marks the lower tail of a normal distribution rather than a distinct category with a natural cutpoint. Modern accounts treat it as a multiple-deficit disorder, in which phonological weakness combines with slowed naming speed and other risks, and neuroimaging locates its signature in a left-hemisphere reading network that includes the visual word form area. Early identification and structured phonics instruction measurably improve outcomes.

Keywords: dyslexia, phonological deficit, reading disability

What Dyslexia Is

Dyslexia is a specific and persistent difficulty in learning to read accurately and fluently, present despite conventional instruction, adequate intelligence, and sufficient opportunity to learn. The widely used definition of Lyon and colleagues (2003) frames it as a disorder of neurobiological origin whose core is a difficulty with accurate or fluent word recognition and with spelling and decoding, arising from a deficit in the phonological component of language and often unexpected in relation to the person’s other cognitive abilities. In diagnostic systems it is coded as ICD-11 6A03.0, developmental learning disorder with impairment in reading, and appears in DSM-5 under specific learning disorder with impairment in reading. MeSH files it among the language disorders, reflecting its roots in the spoken-language system rather than in vision.

Two features of the modern concept matter for everything that follows. The first is that dyslexia is dimensional. Reading ability, like height, is distributed continuously across the population and is approximately normally distributed; dyslexia occupies the lower end of that distribution and blends smoothly into the typical range, with no natural break marking where “dyslexic” begins (Shaywitz & Shaywitz, 2005). Any threshold used to diagnose it — a score below the fifth, tenth, or fifteenth percentile — is therefore an administrative convention, and prevalence estimates rise and fall with that cutoff, a point the Worked Example makes quantitative.

The second is that the old discrepancy definition — reading far below the level predicted by IQ — has been abandoned. Vellutino and colleagues (2004) showed that poor readers with and without an IQ–achievement discrepancy do not differ in the phonological skills that actually cause reading difficulty, nor in their response to instruction, so the discrepancy criterion identified no meaningful subgroup. What defines dyslexia is not a gap from IQ but a specific weakness in the language machinery that supports decoding.

Types of Dyslexia

In the Medical Subject Headings hierarchy, Dyslexia is the broad descriptor and carries one narrower descriptor of its own:

Two caveats are needed to read this table correctly. First, MeSH is an indexing classification, not a cognitive theory. It mints a single narrower descriptor — acquired dyslexia — for cataloguing the literature, and it does not enumerate the subtypes that clinicians and researchers use within developmental dyslexia. Those working distinctions are orthogonal to the MeSH child: readers are described as showing a dysphonetic (phonological) profile, with poor nonword decoding; a surface profile, with slow, laborious recognition of whole words; or, on the double-deficit account, a rate profile of slowed naming, and a combined profile impaired on both. These are graded tendencies along the continuous dimensions described below, not the discrete categories the table format suggests.

Second, the one MeSH child — acquired dyslexia — names a fundamentally different disorder from the topic here: it is the breakdown of a reading system that was built and then damaged, whereas developmental dyslexia is a system that never developed typically. They share a name and a symptom but not a cause, and the deep link above leads to the separate article that treats the acquired forms in full.

The Phonological Core

The most replicated finding in half a century of dyslexia research is that its proximal cognitive cause is a phonological deficit — a weakness in the mental representation and manipulation of the sounds of speech (Vellutino et al., 2004). Learning to read an alphabetic script requires discovering that spoken words are built from smaller sound units, phonemes, and that letters and letter groups map onto those units. A child whose phonemic representations are imprecise or hard to access finds this mapping unusually difficult to learn, and the result is slow, effortful, error-prone decoding.

The deficit shows up on three linked behavioural markers that together predict and characterize dyslexia. Phonological awareness — the ability to detect and manipulate the sound structure of words, for instance by deleting a phoneme to turn spin into pin — is reliably weak, and its weakness in preschool is among the strongest predictors of later reading difficulty. Verbal short-term memory, the ability to hold a sequence of sounds briefly in mind, is reduced, reflecting the same degraded phonological code. And rapid automatized naming, the speed of retrieving the spoken names of familiar visual items, is slowed — a marker whose partial independence from awareness motivates the double-deficit account below. The interactive demonstration below presents the phoneme-manipulation task that sits at the centre of this account.

Phoneme deletion is a standard test of phonological awareness — the skill that is weakest in dyslexia. Say the word, remove one sound, and choose what is left. Notice how the task lives entirely in sound, not spelling.

The phonological theory does not claim that dyslexia is a visual problem. Letter reversals such as confusing b and d, the popular image of dyslexia, are common in all beginning readers and are not specific to it; the difficulty lies in connecting the visual symbol to its sound, not in seeing the symbol. This is why dyslexia is classified with the language disorders, and why the interventions that work are built on explicit, structured instruction in the sound–letter code rather than on visual training.

The phonological deficit is universal, but its surface expression varies with the writing system, which is why dyslexia looks different across languages. The psycholinguistic grain-size theory of Ziegler and Goswami (2005) explains this: a transparent orthography such as Italian or Finnish, in which letters map consistently onto sounds, lets readers rely on small, phoneme-sized units, so dyslexia there shows up mainly as slow but accurate reading; a deep orthography such as English, riddled with inconsistent spellings, forces reliance on larger and more variable units and yields the accuracy errors familiar from English-language research. The same underlying phonological weakness thus produces a milder, fluency-dominated profile in a shallow script and a more visible, accuracy-dominated one in a deep script — a caution against treating findings from English as the universal face of the disorder.

Theories and Mechanisms

The phonological deficit is the centre of gravity, but it is not the whole story, and several accounts extend or compete with it. The most influential extension is the double-deficit hypothesis of Wolf and Bowers (1999), which argues that reading fluency depends on two partially independent sources: phonological awareness and the speed of rapid automatized naming. On this view readers fall into four groups — no deficit, a single phonological deficit, a single naming-speed deficit, and a double deficit on both — and the double-deficit group is the most severely and pervasively impaired, because the two weaknesses compound. The demonstration below places a reader in this space and shows how the predicted severity grows as each deficit deepens.

The double-deficit hypothesis places a reader on two axes — phonological awareness and rapid naming speed. Impairment on both compounds, so the double-deficit corner is the most severely affected. Move the sliders and watch the quadrant and the predicted severity change.

The shaded corner is the double-deficit region. There the two weaknesses interact rather than merely add, which is why readers impaired on both fluency sources are the most pervasively affected.

Two further accounts propose that a sensory impairment underlies the phonological one. The magnocellular theory of Stein (2001) holds that dyslexia reflects impaired function in the fast, transient magnocellular pathways of the visual and auditory systems, degrading the timing of the sensory signals from which phonological and orthographic representations are built. A related auditory hypothesis points to difficulty tracking rapid changes in the speech signal. These theories are contested: the multiple-case study of Ramus and colleagues (2003) found that a phonological deficit was present in almost every dyslexic adult tested, whereas auditory, visual, and motor deficits appeared only in a subset and always alongside the phonological one, suggesting that sensory problems are neither necessary nor sufficient and that the phonological deficit is the common core.

The modern synthesis is the multiple-deficit model of Peterson and Pennington (2015). It abandons the search for a single cause and treats dyslexia as the probabilistic outcome of several interacting risk factors — genetic and environmental, cognitive and sensory — no one of which is necessary or sufficient on its own. The phonological deficit is the most consistent contributor, but reading failure emerges when enough risks accumulate, which explains why dyslexia is dimensional, why it is so often comorbid with disorders such as ADHD and dyscalculia, and why individual profiles vary.

The Reading Brain

Skilled reading recruits a distributed left-hemisphere network, and dyslexia is marked by atypical function within it (D’Mello & Gabrieli, 2018). Three regions recur. A posterior ventral occipitotemporal region — the visual word form area that Dehaene and Cohen (2011) describe as tuned to recognize letter strings as visual wholes — supports the fast, automatic word recognition of the fluent reader, and it is characteristically underactivated in dyslexia. A temporoparietal system supports the effortful mapping of letters to sounds that a beginning or struggling reader relies on. And an anterior inferior frontal region is engaged in articulation and phonological processing. Functional imaging by Shaywitz and Shaywitz (2005) found relative underactivation of the two posterior left-hemisphere systems in dyslexic readers, with compensatory reliance on frontal and right-hemisphere regions.

The neural differences have a developmental and genetic basis. Galaburda and colleagues (2006) traced a path from candidate dyslexia-susceptibility genes to their role in neuronal migration during fetal cortical development, linking the disorder’s heritability to subtle anomalies in how the reading network is wired. This does not mean the brain differences are immutable: because the reading network is shaped by learning, its function shifts with effective instruction, and the underactivation is in part a consequence of less successful reading experience as well as a cause — part of why early, intensive intervention changes both behaviour and brain response.

Worked Example

Because reading ability is continuous and approximately normal, the number of people counted as dyslexic depends entirely on where the diagnostic cutoff is placed — a direct consequence of the dimensional view. This can be made exact. Suppose a reading composite is standardized to a normal distribution with mean 100 and standard deviation 15, as most reading tests are. A diagnostic threshold set at k standard deviations below the mean corresponds to a standard score of 100 + 15 × z, and the fraction of the population below it is the standard-normal cumulative probability Φ(z).

Moving the cutoff by half a standard deviation at a time takes the estimated prevalence from about 16% to about 7% to about 2% — a nearly sevenfold range produced by nothing but the choice of threshold, with no change in any child. This is why published prevalence figures for dyslexia span roughly 5–17% depending on the criterion used, and why arguing about the “true” rate is misconceived: on a dimensional trait the rate is defined by the cutoff, not discovered. The lesson is practical as well as conceptual — a child just above a strict cutoff is not meaningfully different from one just below it, so services keyed to a rigid threshold will miss children who need help. The demonstration below recomputes Φ(z) as the cutoff is moved, so the prevalence and the shaded tail respond live.

Reading ability is continuous and roughly normal (mean 100, SD 15). Because dyslexia is the lower tail of that curve, its estimated prevalence is set by the diagnostic cutoff, not discovered. Slide the threshold and watch the shaded tail — and the prevalence — move.

At z = −1.0 the tail holds about 16% of readers; at z = −2.0, about 2%. Nothing about any child changes — only the line drawn through a continuous distribution.

Discussion

Dyslexia is the most studied of the specific learning disorders, and its arc illustrates how a cognitive science matures. It began as a puzzle of “word blindness” framed in visual terms, moved through the discrepancy era in which it was defined by a gap from IQ, and settled on a language-based account in which a phonological deficit disturbs the letter-to-sound learning at the heart of reading (Vellutino et al., 2004). Each step discarded an intuitive but wrong idea — that the problem is in the eyes, or that it is defined by intelligence — in favour of a mechanism that the evidence actually supported.

The construct binds outward across cognitive psychology. It is a disorder of a written-language skill that draws on the visual perception of letter strings and on the same sound system studied in speech perception; its verbal short-term memory marker ties it to working memory; it sits within the broader category of learning disorders; and it is frequently comorbid with dyscalculia, the parallel disorder of number. Its separate acquired form, the loss of reading after brain injury, is the natural experiment that maps the same reading system by damaging it.

Two cautions temper the account. The subtypes and theories are idealizations over a continuous, multiply determined trait, so individual readers rarely fit one label cleanly (Peterson & Pennington, 2015). And the neural differences are both cause and consequence of reading experience, so they are best read as the signature of a developmental process, not as a fixed lesion.

Current Directions

The most active frontier is early identification, and it turns on what Ozernov-Palchik and Gaab (2016) call the dyslexia paradox: the disorder is most treatable early, yet is usually not diagnosed until a child has already failed to read for two or three years. Their work uses pre-literacy behavioural markers and brain measures — taken before formal reading instruction begins — to flag risk in kindergarten, so that intervention can start before failure compounds into the secondary anxiety and disengagement that a late diagnosis brings.

A second line pushes the sensory account into precise neuroscience. Goswami’s temporal-sampling framework links dyslexia to atypical neural tracking of the slow amplitude modulations of the speech envelope, proposing that imprecise oscillatory sampling of speech degrades the phonological representations downstream — an auditory-neuroscience extension of the phonological theory rather than a rival to it.

The third front is intervention science. Hulme and Snowling (2016) synthesize the evidence that structured, phonologically grounded instruction — explicit training in phoneme awareness and letter–sound mapping, embedded in reading and spelling practice — produces reliable gains, and that reading and oral-language difficulties often need to be targeted together. The convergence of early-risk screening with well-specified instruction is moving dyslexia from a disorder identified by failure toward one identified by risk and addressed before failure occurs (Snowling et al., 2020).

Common Misconceptions

“Dyslexia means seeing letters backwards.”

Letter reversals such as b for d are common in all beginning readers and are not specific to dyslexia. The disorder is a language-based difficulty in connecting letters to sounds, not a visual problem of reversed or scrambled perception (Vellutino et al., 2004).

“Dyslexia is a sign of low intelligence.”

Dyslexia is defined as an unexpected difficulty with reading, occurring across the full range of intelligence. The discarded IQ-discrepancy criterion identified no meaningful subgroup, and reading difficulty is caused by a specific phonological weakness, not by general ability (Lyon et al., 2003).

“There is a sharp line between dyslexic and normal readers.”

Reading ability is continuous and approximately normally distributed; dyslexia is its lower tail. Any diagnostic cutoff is a convention, and prevalence estimates swing widely with the threshold chosen (Shaywitz & Shaywitz, 2005).

“Children grow out of dyslexia.”

Dyslexia is persistent, but its outcomes are highly modifiable: early, structured, phonologically grounded instruction produces measurable gains and can change brain response, even though the underlying vulnerability remains (Hulme & Snowling, 2016).

Glossary

Acquired dyslexia.

The loss of a previously mastered ability to read, caused by brain injury; a different disorder from the developmental dyslexia of this article, though filed beneath it in MeSH.

Comorbidity.

The co-occurrence of two disorders in the same person more often than chance; dyslexia is frequently comorbid with ADHD and with dyscalculia.

Decoding.

The process of converting printed letters and letter groups into their corresponding speech sounds to recognize a written word; the skill most directly impaired in dyslexia.

Dimensional.

Distributed on a continuum rather than as a present-or-absent category; reading ability is dimensional, so dyslexia is its lower tail with no natural boundary.

Discrepancy criterion.

The abandoned definition of dyslexia as reading achievement far below the level predicted by IQ; shown to identify no distinct or meaningful subgroup of poor readers.

Double-deficit hypothesis.

The proposal that reading fluency depends on two partly independent skills, phonological awareness and naming speed, so that impairment on both is the most severe profile.

Magnocellular theory.

The account that dyslexia stems from impaired function in the fast, transient magnocellular sensory pathways, degrading the timing of the signals that build reading representations.

Multiple-deficit model.

The view that dyslexia is the probabilistic outcome of several interacting risk factors, none necessary or sufficient alone, with the phonological deficit the most consistent contributor.

Orthography.

The writing system of a language — its letters and the conventions mapping them to sounds; the depth of an orthography shapes how readily its decoding is learned.

Phoneme.

The smallest unit of speech sound that distinguishes one word from another; the unit that alphabetic letters map onto and that dyslexic readers represent imprecisely.

Phonological awareness.

The ability to detect and manipulate the sound structure of spoken words; its weakness is the core marker of dyslexia and a strong early predictor of reading difficulty.

Phonological deficit.

The most robust proximal cause of dyslexia: an impairment in representing and accessing the sounds of speech that undermines the letter-to-sound mapping of reading.

Rapid automatized naming.

The speed of retrieving the spoken names of familiar visual items such as colours or digits; slowed in dyslexia and the second deficit of the double-deficit account.

Visual word form area.

A region of left ventral occipitotemporal cortex tuned to recognize letter strings as visual wholes; characteristically underactivated in dyslexic readers.

Key Researchers

Stanislas Dehaene (Collège de France; NeuroSpin, Paris). Advanced the neuronal-recycling account of reading and identified the visual word form area, framing dyslexia as a disorder of the reading network’s ventral stream. ORCID - Wikipedia - Wikidata - Google Scholar - Faculty

Albert M. Galaburda (Harvard Medical School; Beth Israel Deaconess Medical Center, Emeritus). Documented the cortical neuroanatomy of dyslexia and, with Norman Geschwind, proposed a developmental account of its biological origins. Wikipedia - Wikidata - Google Scholar - Faculty

Usha Goswami (University of Cambridge, Centre for Neuroscience in Education). Developed the temporal-sampling framework linking dyslexia to atypical neural tracking of the speech amplitude envelope, extending the phonological account into auditory neuroscience. ORCID - Wikipedia - Wikidata - Google Scholar - Faculty

Charles Hulme (University of Oxford, Department of Education). Established, with Snowling, the evidence base for phonologically grounded reading interventions and the multiple-deficit view of reading disorders. ORCID - Wikidata

Sally E. Shaywitz (Yale University, Yale Center for Dyslexia & Creativity). Co-led the Connecticut Longitudinal Study establishing dyslexia’s dimensional nature and used functional imaging to localize its neural signature. ORCID - Wikipedia - Wikidata - Faculty

Margaret J. Snowling (University of Oxford, President of St John’s College, Emerita). Formulated the phonological-deficit account of dyslexia and, through prospective family-risk studies, the multiple-deficit and comorbidity model. ORCID - Wikipedia - Wikidata

Frequently Asked Questions

What is dyslexia?
Dyslexia is a specific, persistent difficulty in learning to read accurately and fluently, present despite adequate instruction, intelligence, and opportunity. Its core cause is a phonological deficit — a weakness in processing the sounds of speech that disrupts the mapping between letters and sounds (Lyon et al., 2003).

What causes dyslexia?
Its most robust proximal cause is a phonological deficit, but modern accounts treat it as a multiple-deficit disorder in which phonological weakness, slowed naming speed, and other genetic and environmental risks combine. No single factor is necessary or sufficient on its own (Peterson & Pennington, 2015).

Is dyslexia a visual problem?
No. Letter reversals occur in all beginning readers and are not specific to dyslexia. The difficulty is in connecting a letter to its sound, a language-based problem, which is why dyslexia is classified with the language disorders (Vellutino et al., 2004).

Does dyslexia mean a person is not intelligent?
No. Dyslexia occurs across the whole range of intelligence and is defined as reading difficulty that is unexpected given a person’s other abilities. The old idea of defining it by a gap from IQ has been abandoned as unsupported (Lyon et al., 2003).

How common is dyslexia?
Estimates range from about 5% to 17%, and the range is largely an artefact of where the diagnostic cutoff is set. Because reading ability is continuous, moving the threshold by half a standard deviation can change the estimated prevalence severalfold (Shaywitz & Shaywitz, 2005).

What is the double-deficit hypothesis?
It proposes that reading fluency rests on two partly independent skills — phonological awareness and rapid naming speed — so readers impaired on both show the most severe and pervasive difficulty, worse than those with either single deficit (Wolf & Bowers, 1999).

What happens in the brain in dyslexia?
Dyslexic readers show relative underactivation of two posterior left-hemisphere systems — a ventral occipitotemporal region including the visual word form area, and a temporoparietal region — with compensatory reliance on frontal and right-hemisphere areas (Shaywitz & Shaywitz, 2005; Dehaene & Cohen, 2011).

Can dyslexia be treated?
Outcomes are highly modifiable. Structured, explicit instruction in phoneme awareness and letter–sound mapping produces reliable gains, and the earlier it starts the better, because intervention is most effective before reading failure compounds (Hulme & Snowling, 2016).

Support Organizations

Organizations that provide information, assessment guidance, and advocacy for dyslexia and reading disability.

References