Language Disorders

Abstract

A language disorder is a persistent difficulty acquiring or using spoken, written, or signed language that is out of keeping with a person's age or general ability. In the MeSH classification it is a communication disorder, and it divides into two broad families: acquired disorders, in which established language is lost after brain injury such as stroke, producing the aphasias; and developmental disorders, in which language fails to develop typically in childhood, the most common being developmental language disorder. Cognitive psychology treats language as a set of separable processes for comprehension and production, mapped onto a left-hemisphere network, so that different lesions and developmental profiles impair different components while sparing others. This article surveys how these disorders are classified, localized in the brain, measured, and explained by models of the language system.

Keywords: language disorders, aphasia, developmental language disorder, comprehension and production, dual-stream model

A language disorder is defined by a gap: between the language a person can understand or produce and what their age, schooling, and non-verbal ability would predict. It is a disorder of the language system itself rather than of the sensory or motor machinery that carries language, which is why a child with normal hearing and a clear articulatory apparatus can still fail to acquire grammar, and why an adult with intact vocal cords can lose the ability to retrieve words after a stroke. MeSH places language disorders within the broader category of communication disorders, alongside speech disorders and hearing disorders, and treats them as conditions of comprehension or expression of written and spoken language (Geschwind, 1970). The cognitive interest of the category is that language is not a single faculty but a layered system of sound, word, and sentence processing, so damage or disruption at different layers yields recognisably different disorders rather than a uniform loss.

Types of Language Disorders

MeSH files language disorders as a child of the broader category Communication Disorders and, in turn, gives them their own narrower descriptors. These direct subtypes carve the category along two orthogonal lines at once: by the language process affected (writing in agraphia, word retrieval in anomia, reading in dyslexia) and by the developmental-versus-acquired distinction (language development disorders on one side, the acquired speech and language disorders on the other). The classification is an indexing hierarchy for the medical literature, not a theory of the language system, so the subtypes overlap and a single patient may carry more than one: an aphasia after stroke, for instance, routinely produces both anomia and agraphia. Table 2 lists the direct MeSH children of the descriptor.

Note. Subtypes are listed as MeSH classifies them and are not mutually exclusive; a single patient may present with several at once. None currently has its own article on this site, so the terms are listed without links.

Acquired and Developmental Disorders

The single most important division within language disorders is between the acquired and the developmental. An acquired language disorder is a loss: language developed normally and was then damaged, almost always by focal injury to the left hemisphere from stroke, tumour, trauma, or a neurodegenerative disease. The prototypical acquired disorder is aphasia, and its study has been the main road to understanding how language is organised in the brain, because the pattern of loss reveals which components were served by the damaged tissue (Hillis, 2007). A developmental language disorder is instead a failure to acquire: the system never reaches typical function in the first place, in a child with adequate hearing, non-verbal intelligence in the normal range, and no other condition that would explain the difficulty (Bishop, 2006). The two families differ not just in timing but in mechanism, course, and treatment, which is why the contrast in Table 1 anchors the rest of this article.

The Language Network in the Brain

The modern understanding of language disorders begins with the nineteenth-century observation that damage to different parts of the left hemisphere produces different language deficits. Damage to the posterior inferior frontal lobe, Broca's area, leaves comprehension relatively intact but makes speech effortful, halting, and agrammatic; damage to the posterior superior temporal lobe, Wernicke's area, leaves speech fluent but empty and comprehension impaired. Norman Geschwind revived and formalised this into a connectionist account in which Wernicke's area stores the sound patterns of words, Broca's area programs their articulation, and the arcuate fasciculus — a white-matter tract — carries information between them, so that severing the tract produces conduction aphasia, a selective failure of repetition with fluent speech and good comprehension (Geschwind, 1970). This Wernicke-Geschwind model, mapped onto speech perception and production, dominated teaching for a century.

Two developments revised it. First, diffusion imaging of white-matter tracts showed that the perisylvian language network is richer than a single arcuate cable: multiple dorsal and ventral pathways connect frontal, temporal, and parietal language regions, and individual differences in these tracts predict language ability (Catani, Jones, & ffytche, 2005). Second, careful lesion-symptom mapping in large patient samples showed that the classical localisations were too neat: comprehension deficits in particular depend on a wider swathe of middle and anterior temporal cortex than Wernicke's area alone, so the classical map misattributed comprehension to the wrong region (Dronkers et al., 2004). Functional neuroimaging across hundreds of studies has since replaced the two-box picture with a distributed left-hemisphere system whose nodes handle heard speech, spoken language, and reading in overlapping but dissociable ways (Price, 2012).

Figure 1

The Left-Hemisphere Language Network and Its Two Streams

Mapping Aphasia to Lesions

The classical value of the connectionist model is that it predicts a small set of aphasia syndromes from the location of damage, and three behaviours are enough to tell them apart: whether speech is fluent, whether comprehension is preserved, and whether repetition is preserved. Broca's aphasia is non-fluent with good comprehension and poor repetition; Wernicke's aphasia is fluent with poor comprehension and poor repetition; conduction aphasia spares fluency and comprehension but selectively ruins repetition, the signature of a disconnection between the posterior and anterior language regions; and global aphasia impairs all three. This three-question triage is the working logic behind the classical bedside classification of aphasia, even though modern imaging shows the underlying anatomy is less discrete than the syndromes suggest (Hillis, 2007). The first demonstration places a lesion on a schematic language network and reads off the resulting profile.

The Dual-Stream Model

The influential modern replacement for the single-cable model is the dual-stream model of speech processing, which borrows the ventral/dorsal logic of the visual system and applies it to language (Hickok & Poeppel, 2007). Early cortical stages turn the acoustic signal into a sound-based representation in the superior temporal lobe of both hemispheres. From there the pathway divides. A ventral stream, running down into the middle and inferior temporal lobe, maps sound onto meaning and so supports comprehension; it is largely bilateral, which is why comprehension is more robust to unilateral damage than the classical model implies. A dorsal stream, running up toward parietal and frontal cortex, maps sound onto articulatory representations and so supports repetition, speech production, and the learning of new word forms; it is strongly left-dominant. The model reinterprets the classical syndromes: conduction aphasia is dorsal-stream damage, comprehension deficits reflect ventral-stream and wider temporal involvement, and the resilience of comprehension after left-hemisphere stroke follows from ventral bilaterality.

The dual-stream framework has held up as lesion and imaging methods have improved. A large re-analysis of aphasia after stroke found that the deficits map better onto damage to these streams and their cortical hubs than onto the classical Broca-Wernicke dichotomy, and that fluency, comprehension, and repetition load onto separable anatomical factors (Fridriksson et al., 2018). The second demonstration sends a spoken word through the two streams under different task demands and shows which stream carries the load.

Developmental Language Disorder

The developmental side of the category is dominated by developmental language disorder (DLD), the condition long studied under the name specific language impairment (SLI). A child with DLD has marked, persistent difficulty understanding or producing language — typically most visible in grammar and in the retrieval and learning of words — that is not explained by hearing loss, low non-verbal intelligence, autism, or a known neurological condition (Leonard, 2014). The change of name matters. The older term specific implied that language was impaired in isolation and that a gap between language and non-verbal IQ was diagnostic; both claims proved untenable. Deficits in DLD often extend to procedural learning and processing speed, and requiring a language-versus-IQ discrepancy excluded many children with genuine, disabling language difficulties while predicting neither prognosis nor response to treatment (Bishop, 2006).

The current terminology was fixed by the CATALISE consensus, a structured multinational Delphi study among specialists. Its first phase agreed the criteria for identifying language problems in children (Bishop et al., 2016), and its second phase settled on developmental language disorder as the label, dropping the non-verbal IQ cut-off and reserving separate terms for language difficulties that accompany a known biomedical condition (Bishop et al., 2017). Why DLD arises is still debated. The procedural deficit hypothesis proposes that the core problem is an impairment of the brain's procedural memory system, which normally supports the implicit learning of grammatical rules and sequences, so that the grammatical difficulties of DLD are one expression of a broader learning deficit rather than a language-specific fault (Ullman & Pierpont, 2005). On this view DLD is the developmental mirror of the acquired disorders: not a lesion in a language box, but an atypically built learning system that leaves language most visibly affected.

How Common Are Language Disorders?

Language disorders are common, and developmental disorders far outnumber acquired ones in the population. The first rigorous epidemiological estimate came from a large kindergarten screening that found specific language impairment in about 7.4% of five-year-olds (Tomblin et al., 1997). Two decades later the population-based SCALES study, using a two-stage design and a broader, IQ-independent definition, estimated that about 7.6% of children starting school have a developmental language disorder of unknown origin, with roughly another 2.3% showing language disorder associated with a known condition (Norbury et al., 2016). A crucial finding of that study was that prevalence depends heavily on where the non-verbal-ability line is drawn: relaxing the old discrepancy requirement, as CATALISE later recommended, identifies more children as having a treatable disorder. The third demonstration varies the diagnostic cut-off and population size to show how the expected number of affected children changes.

Assessment

Because language disorders are defined behaviourally, they are diagnosed by structured testing of the language system rather than by a single biological marker. Standardised, norm-referenced instruments sample both receptive language (comprehension) and expressive language (production) across levels — sounds, words, grammar, and connected discourse — and compare a person's scores against age expectations; a range of these language tests exist for both children and adults. In acquired aphasia, assessment additionally profiles the three localising behaviours — fluency, comprehension, and repetition — and adds naming and reading, so that the aphasia can be classified and change tracked over recovery (Hillis, 2007). In children, assessment must separate a true disorder from the wide normal variation in the pace of language development and from the effects of limited exposure to the test language, which is why population screening uses two stages and why the diagnostic threshold is a matter of judgement rather than a natural boundary (Norbury et al., 2016). Language testing also draws on general cognitive capacities such as working memory, whose verbal component is frequently limited in DLD.

Worked Example

Consider what a prevalence figure means on the ground. Take the SCALES estimate that about 7.6% of children entering school have a developmental language disorder (Norbury et al., 2016). In a single classroom of 30 children, the expected number affected is 0.076 x 30 = 2.28, so on average between two and three children per class. Across a small primary school of 300 children the expectation is 0.076 x 300 = 22.8, roughly 23 children. Set this against the prevalence of autism, commonly estimated near 1%: at 0.076 versus 0.010, DLD is about 7.6 times as common, yet it attracts a small fraction of the recognition and resources — the paradox that motivated the CATALISE effort to standardise the diagnosis (Bishop et al., 2017).

The cut-off dependence sharpens the point. Suppose a screening programme covers a birth cohort of 600,000 children. Using a strict definition that requires a language score more than 1.5 standard deviations below the mean identifies about the bottom 6.7% of a normal distribution, or roughly 40,000 children. Relaxing the threshold to 1.25 standard deviations captures about the bottom 10.6%, or roughly 63,600 children. The same underlying reality yields estimates differing by more than 20,000 children according to a single methodological choice, which is why epidemiological prevalence for language disorder is always reported together with the definition used, and why the field treats the diagnostic threshold as a decision rather than a discovery (Tomblin et al., 1997; Norbury et al., 2016).

Discussion

Language disorders occupy a revealing place in cognitive science because they are the natural experiments that made a componential theory of language testable. The aphasias showed that comprehension, production, and repetition can be dissociated by damage, which is only possible if they are served by partly separate machinery; the dual-stream model turned those dissociations into an anatomy (Hickok & Poeppel, 2007; Fridriksson et al., 2018). The developmental disorders make the complementary point from the other direction: if language can fail to develop while non-verbal intelligence, hearing, and articulation are intact, then language acquisition draws on mechanisms that are at least partly specialised, even if, as the procedural-deficit account argues, those mechanisms are not exclusively linguistic (Ullman & Pierpont, 2005).

The category also carries a cautionary lesson about classification. The classical Broca-Wernicke map was elegant, teachable, and, in its strong localisationist form, partly wrong: comprehension turned out to depend on tissue the model assigned elsewhere, and the syndromes blur at their edges (Dronkers et al., 2004). The developmental side made the same error in a different key, with the discarded discrepancy criterion that excluded children who needed help (Bishop, 2006). In both cases the correction came from large samples and better measurement rather than from a new intuition, which is the recurring shape of progress in this field: the disorders are more distributed, more graded, and more overlapping than the first clean models proposed, and the useful theory is the one that survives that messiness.

Current Directions

Two lines of current work stand out. The first is the continued refinement of lesion-symptom mapping in acquired aphasia. Large, imaging-rich patient databases now let researchers relate specific language behaviours to precise patterns of grey- and white-matter damage, testing and extending the dual-stream account and beginning to predict which patients will recover which functions (Fridriksson et al., 2018). This matters clinically because a prediction of recovery, tied to a lesion, is the basis for individualised rehabilitation. The second is the consolidation of developmental language disorder as a recognised, defined condition following the CATALISE consensus, which has shifted research from arguing about terminology toward epidemiology, genetics, and intervention on an agreed target (Bishop et al., 2017). Population studies with IQ-independent criteria are producing defensible prevalence figures and, with them, the case for services scaled to a disorder that affects roughly one child in every classroom (Norbury et al., 2016). Both lines share a direction: from describing and naming language disorders toward predicting their course and changing it.

Common Misconceptions

A language disorder is the same as a speech disorder.

They are different. A speech disorder affects the motor production of sounds and fluency, such as a lisp or a stammer, while a language disorder affects the underlying system of meaning, words, and grammar. A person can have clear articulation and still have a severe language disorder, and MeSH lists the two as separate children of communication disorders (Geschwind, 1970).

Developmental language disorder means low intelligence.

By definition DLD occurs in children with non-verbal intelligence in the normal range; the difficulty is specific to language and its supporting learning systems. Indeed the consensus that renamed the condition deliberately dropped the requirement of a gap between language and IQ, because it excluded affected children without improving prediction (Bishop et al., 2017).

Comprehension lives only in Wernicke's area.

The classical map is too neat. Careful lesion analysis shows that comprehension depends on a wider region of middle and anterior temporal cortex than Wernicke's area alone, and the dual-stream model attributes it to a largely bilateral ventral pathway, which is why comprehension is relatively robust to unilateral damage (Dronkers et al., 2004; Hickok & Poeppel, 2007).

Glossary

Agraphia.

An acquired loss of the ability to write despite adequate hand function, often accompanying aphasia.

Anomia.

A difficulty retrieving words, especially names, despite preserved knowledge of their meaning; the most common residual sign of aphasia.

Aphasia.

An acquired language disorder caused by brain injury, impairing some combination of production, comprehension, naming, and repetition.

Arcuate fasciculus.

A white-matter tract connecting posterior and anterior language regions; its disruption is the classical basis of conduction aphasia.

Broca's area.

A region of the posterior inferior frontal lobe associated with effortful, agrammatic speech production when damaged.

Developmental language disorder.

A persistent childhood difficulty acquiring language, not explained by hearing loss, low non-verbal IQ, autism, or a known medical condition; formerly specific language impairment.

Dorsal stream.

The left-dominant pathway mapping speech sound onto articulation, supporting repetition and speech production.

Dual-stream model.

Hickok and Poeppel's account splitting speech processing into a ventral sound-to-meaning stream and a dorsal sound-to-articulation stream.

Dyslexia.

A specific difficulty with accurate and fluent reading; developmental in childhood or acquired after brain injury.

Expressive language.

The production side of language: formulating and articulating words and sentences to convey meaning.

Perisylvian language network.

The interconnected frontal, temporal, and parietal regions around the Sylvian fissure of the left hemisphere that process language.

Receptive language.

The comprehension side of language: recovering meaning from heard, read, or signed input.

Specific language impairment.

The former name for developmental language disorder, implying a language-specific deficit and a language-versus-IQ discrepancy now abandoned.

Ventral stream.

The largely bilateral pathway mapping speech sound onto meaning, supporting comprehension.

Wernicke's area.

A region of the posterior superior temporal lobe classically associated with fluent but empty speech and impaired comprehension when damaged.

Key Researchers

Dorothy V. M. Bishop (b. 1952). Emeritus professor of developmental neuropsychology at the University of Oxford; she led the CATALISE consensus that established the diagnosis of developmental language disorder and has shaped the study of its causes and epidemiology. Wikipedia - Faculty Page - ORCID - Google Scholar

Nina F. Dronkers (contemporary). Neuropsychologist at the University of California; her lesion-symptom mapping revised the classical map of language comprehension and helped move the field toward a distributed account. Wikipedia - Faculty Page - ORCID - Google Scholar

Norman Geschwind (1926-1984). American behavioral neurologist at Harvard Medical School; he revived and formalised the connectionist Wernicke-Geschwind model of language and disconnection syndromes that dominated twentieth-century teaching. Wikipedia

Gregory Hickok (contemporary). Cognitive scientist at the University of California, Irvine; with David Poeppel he proposed the dual-stream model of speech processing that now frames the anatomy of language disorders. Faculty Page - Google Scholar

Argye E. Hillis (contemporary). Professor of neurology at Johns Hopkins University; she has synthesised a quarter-century of progress in aphasia and advanced the understanding of acquired language disorders after stroke. Faculty Page - Google Scholar

Courtenay Frazier Norbury (contemporary). Professor of developmental language and communication disorders at University College London; she led the SCALES population study that produced a defensible, IQ-independent prevalence estimate for language disorder. Faculty Page - ORCID - Google Scholar

David Poeppel (b. 1964). Neuroscientist at New York University and the Ernst Strüngmann Institute; with Gregory Hickok he developed the dual-stream model of the cortical organisation of speech processing. Wikipedia - Faculty Page - ORCID - Google Scholar

Frequently Asked Questions

What is a language disorder?
A language disorder is a persistent difficulty understanding or using spoken, written, or signed language that is out of keeping with a person's age and general ability. It is a disorder of the language system itself, not of hearing or of the motor production of speech (Geschwind, 1970).

How is a language disorder different from a speech disorder?
A speech disorder affects the motor production of sounds and fluency, such as a stammer, whereas a language disorder affects the underlying system of words, grammar, and meaning. Someone can articulate clearly and still have a severe language disorder, and the two are classified separately in MeSH (Geschwind, 1970).

What is the difference between acquired and developmental language disorders?
Acquired disorders, the aphasias, are a loss of language that developed normally and was then damaged by brain injury such as stroke. Developmental disorders are a failure to acquire language typically in childhood, with no single lesion and a persistent course (Hillis, 2007).

What is aphasia?
Aphasia is an acquired language disorder caused by brain injury, usually to the left hemisphere. Different lesions produce different profiles of fluency, comprehension, and repetition, which is how the classical aphasia syndromes are told apart (Hillis, 2007).

What is developmental language disorder?
Developmental language disorder, formerly called specific language impairment, is a lasting childhood difficulty acquiring language that is not explained by hearing loss, low non-verbal intelligence, autism, or a known medical condition (Bishop et al., 2017).

How common are language disorders?
Developmental language disorder affects roughly 7 to 8 percent of children starting school, making it far more common than autism. Population studies show the exact figure depends on where the diagnostic threshold is set (Norbury et al., 2016).

Which parts of the brain support language?
Language depends on a network in the left hemisphere linking frontal, temporal, and parietal regions. Modern models describe a ventral stream mapping sound to meaning and a dorsal stream mapping sound to articulation, refining the older Broca and Wernicke picture (Hickok & Poeppel, 2007).

Can language disorders be treated?
Acquired aphasia is treated with speech and language therapy, aided by some spontaneous recovery in the months after injury, and developmental language disorder is managed with targeted language intervention. Accurate assessment guides both (Hillis, 2007).

References