Aphasia

Abstract

Aphasia is an acquired disorder of language caused by injury to the brain — most often a left-hemisphere stroke — in which the capacity to produce or understand words is degraded while general intellect is largely preserved. Since Paul Broca and Carl Wernicke localized speech in the nineteenth century, the syndromes have been organized around three bedside axes — fluency, comprehension, and repetition — whose combinations define the classical types, from the effortful nonfluent output of Broca aphasia to the empty fluency of Wernicke aphasia. Voxel-based lesion-symptom mapping and the dual-stream model have since loosened the tidy region-to-syndrome links of the classical account without discarding the perisylvian network it described. This article surveys the taxonomy, its classical and revised anatomy, and the course of recovery, with a worked example and three interactive demonstrations.

Keywords: aphasia, language, fluency

Aphasia sits at the historical origin of cognitive neuroscience: the first firm evidence that a mental faculty could be tied to a piece of cortex came from a patient who had lost speech. It remains a working laboratory for the architecture of language, because the ways language breaks are not arbitrary — they fall into recurring patterns that any theory of the normal system must explain (Damasio, 1992).

The account below moves from what aphasia is, through the family of syndromes MeSH recognizes and the classical model that named them, to the three clinical axes that classify a patient at the bedside, a worked example of how those axes generate the syndrome list, the anatomy as lesion mapping has revised it, and the course of recovery. Throughout, the theme is that aphasia is not one disorder but a set of dissociable failures of a shared network.

Figure 1

The Left Perisylvian Language Network

What Aphasia Is

Aphasia is the loss or impairment of language caused by damage to the brain regions that support it. The damage is *acquired* — the person had language and lost it — which distinguishes aphasia from the developmental language disorders of childhood. It is a disorder of language itself: of retrieving words, assembling them into grammatical strings, and decoding the words of others. It is not a disorder of the vocal apparatus. A patient whose speech is slurred because the muscles of articulation are weak has dysarthria, not aphasia; a patient who cannot find the word “comb” while holding one, or who produces fluent nonsense, has aphasia even if every muscle works.

The overwhelmingly common cause is stroke in the territory of the left middle cerebral artery, which supplies the perisylvian cortex of the language-dominant hemisphere. Roughly a third of acute stroke patients are aphasic on admission (Pedersen et al., 1995). Aphasia also follows traumatic injury, tumour, and — in a slow, degenerative form — neurodegeneration, where it is called primary progressive aphasia (Mesulam, 2001) and is now classified into three variants by a consensus scheme (Gorno-Tempini et al., 2011). Because language is lateralized to the left hemisphere in almost all right-handers and most left-handers, aphasia is overwhelmingly a consequence of left-sided damage.

What aphasia is *not* is a loss of thought. Aphasic patients recognize their families, navigate their homes, play cards, and grasp situations they cannot describe. The dissociation between spared cognition and lost language is exactly what made aphasia the founding evidence of cognitive neuroscience (Damasio, 1992): it showed that a faculty of mind could be selectively removed by removing a piece of cortex, and therefore that the faculty was, in some sense, *located* there.

Types of Aphasia

Aphasia is not a single condition but a family of syndromes. The Medical Subject Headings (MeSH) vocabulary, which the United States National Library of Medicine uses to index the biomedical literature, files aphasia under speech and language disorders and records four descriptor-level subtypes. MeSH is an *indexing classification*, not a clinical theory: its categories are chosen to organize the literature, and they overlap and cross-cut the way clinicians actually group patients. The four indexed subtypes are largely orthogonal — a given patient’s syndrome is defined by *where* the lesion falls and *which* language functions survive, so the same anatomy can be described under more than one heading.

The classical bedside taxonomy adds several syndromes MeSH does not index at the descriptor level — global aphasia (all functions lost), the transcortical aphasias (repetition spared while other functions fail), and anomic aphasia (a residual word-finding deficit). The next two sections show how the whole list arises from a small number of dissociable functions.

The Classical Model

The modern study of aphasia begins in 1861, when Paul Broca examined a patient, Leborgne, who could utter only the syllable “tan” yet understood much of what was said to him. At autopsy Broca found a lesion in the left inferior frontal gyrus — the region now bearing his name — and concluded that articulate speech was localized there. Thirteen years later Carl Wernicke described the mirror-image case: patients who spoke fluently but produced meaningless strings and could not understand speech, whose lesions lay in the posterior superior temporal gyrus. Wernicke went further than localization. He proposed a *connectionist* model: a sensory word-image centre (his area) and a motor word-image centre (Broca’s), joined by a fibre pathway, so that language depended not only on the centres but on the connection between them.

Wernicke’s model predicted a syndrome no one had yet described — a fluent aphasia with intact comprehension but failed repetition, produced by cutting the connection while sparing both centres. That syndrome, conduction aphasia, was later confirmed. A century after Wernicke, Norman Geschwind revived and systematized the connectionist framework in his account of the disconnexion syndromes (Geschwind, 1965), organizing the aphasias around named cortical centres joined by white-matter tracts. The resulting Wernicke–Geschwind model — Broca’s area for output, Wernicke’s for comprehension, the arcuate fasciculus between them — was taught to a generation of students and remains the scaffold on which the syndromes are introduced, including here.

The Three Clinical Axes

The classical syndromes are not memorized as a list; they are *derived* from a small number of features assessed at the bedside. Three carry most of the diagnostic weight.

Fluency is the ease and flow of spontaneous output. Fluent aphasics produce speech at a normal rate with normal melody and grammatical scaffolding, even when the content is empty or error-laden; nonfluent aphasics produce halting, effortful, telegraphic speech. Fluency roughly separates anterior lesions (nonfluent) from posterior ones (fluent).

Auditory comprehension is the ability to understand spoken language, tested by following commands and answering yes/no questions. It separates the syndromes with posterior temporal involvement (impaired) from those without (spared).

Repetition is the ability to repeat words and sentences spoken by the examiner. Because repetition requires the whole loop — hearing, holding, and reproducing — it is disproportionately sensitive to damage of the connecting pathway, and it is the axis that isolates conduction aphasia (spared comprehension, failed repetition) and, by its *sparing*, the transcortical aphasias.

Two further features — naming (the near-universal deficit in aphasia, retrieving words on demand; its isolated form is anomia) and grammar — refine the picture, but the three axes above generate the primary syndrome list, as the next section shows.

Worked Example: Three Axes, Eight Syndromes

Treat each of the three axes as roughly binary — intact or impaired — and the classical taxonomy becomes a truth table. With three binary features the number of distinct cells is

$$N = 2^k = 2^3 = 8$$

and each of the eight cells corresponds to a named classical syndrome. This is why the syndrome list has the length it does: it is the enumeration of the combinations, not an arbitrary catalogue.

The pattern is legible. Holding repetition *failed* and reading down the first four rows recovers the four perisylvian aphasias — Broca, Wernicke, conduction, and global — the ones whose lesions fall inside the repetition loop. Switching repetition to *spared* recovers the transcortical aphasias, whose lesions spare the loop but disconnect it from the wider cortex, which is precisely why the patient can still repeat what is said even when spontaneous speech or comprehension has failed. Anomic aphasia, all three axes nominally intact, is the residual category: fluent, comprehending, able to repeat, yet unable to find words — the mildest and most common end-state, and the form many other aphasias resolve into. The demonstration above lets the reader set the three switches and read off the cell.

The Anatomy Revisited

The clean region-to-syndrome mapping of the classical model was built from a modest number of autopsied cases, and it has not survived contact with quantitative lesion analysis. When Nina Dronkers and colleagues re-examined the brains of Broca’s original patients with modern imaging, the lesions extended well beyond the eponymous area. More systematically, *voxel-based lesion-symptom mapping* — a method that tests, at every voxel, whether damage there predicts a given deficit across a large patient sample (Bates et al., 2003) — re-derived the anatomy of the aphasias from data rather than from named centres. The results (Dronkers et al., 2004) showed that comprehension depends on a distributed set of temporal regions, not on Wernicke’s area alone, and that the classical mappings were approximations.

Two influential reframings followed. The *dual-stream model* of Gregory Hickok and David Poeppel (Hickok & Poeppel, 2007) reorganized the neurobiology of language around two pathways: a ventral stream mapping sound to meaning, and a dorsal stream mapping sound to articulation. On this view the aphasia syndromes are breakdowns at the level of streams, not boxes. A large lesion study by Julius Fridriksson and colleagues (Fridriksson et al., 2018), pointedly titled *Anatomy of aphasia revisited*, found that the same cortical regions contribute to multiple syndromes and argued that Broca’s and Wernicke’s areas, as classically defined, are not the discrete language centres the model assumed — a conclusion foreshadowed by Tremblay and Dick’s blunt review title, “Broca and Wernicke are dead” (Tremblay & Dick, 2016). The dynamics of the network after injury have also been traced directly: Saur and colleagues (Saur et al., 2006) followed language reorganization from the acute phase through recovery, and modern surgical and imaging frameworks (Middlebrooks et al., 2017) now map language before neurosurgery precisely because the textbook anatomy is too coarse to trust in an individual brain.

None of this discards the network Wernicke drew. It refines it: the perisylvian language system is real, its dorsal and ventral pathways are real, and lesions along it still produce recognizable syndromes — but the boundaries are graded, the regions multifunctional, and the individual variation large.

Assessment and Recovery

Aphasia is assessed with standardized batteries that sample the axes above — spontaneous speech, comprehension, repetition, naming, reading, and writing — and yield both a syndrome classification and a severity score. The same profile that names the syndrome sets the therapy target: a nonfluent patient works on output, a patient with comprehension loss on decoding.

Most aphasia after stroke improves. Recovery is fastest in the first weeks and months, driven initially by the resolution of oedema and the return of blood flow to tissue that was hypoperfused but not destroyed, and later by reorganization of surviving networks (Saur et al., 2006). The single largest determinant of outcome is initial severity, followed by lesion size and location; smaller, more posterior lesions and milder initial deficits recover more fully (Pedersen et al., 1995). Speech-and-language therapy works: a large systematic review of randomized trials (Brady et al., 2016) found that therapy improves functional communication, reading, writing, and expressive language relative to no treatment, with more intensive schedules yielding larger gains. Contemporary accounts frame recovery as neuroplasticity of the language network (Kiran & Thompson, 2019) and are beginning to model both the deficits and the recovery computationally (Stefaniak et al., 2020), moving the field from box-and-arrow syndromes toward distributed, degradable, and re-trainable representations.

Discussion

The enduring lesson of aphasia is dissociation. That comprehension can fail while fluency is spared, or repetition fail while both are intact, tells us that language is not a single faculty but a set of separable operations, each with its own neural substrate and its own way of breaking. This is the logic on which cognitive neuropsychology was built: the pattern of what survives and what is lost constrains the architecture of the normal system more sharply than intact performance ever could.

The tension running through the field is between the *localizationist* impulse — this function lives here — and the *network* reality that language is distributed, redundant, and individually variable. The classical model erred toward the first; lesion mapping and the dual-stream account have pulled toward the second (Fridriksson et al., 2018). The productive synthesis is that both are partly right: there is real regional specialization within a network whose boundaries are graded and whose components cooperate. Aphasia is what happens when specific parts of that cooperative system are removed.

Cognitive Implications

For cognitive psychology, aphasia is a natural experiment in the componential structure of language. The repetition axis, in particular, connects the aphasia literature to models of working memory: the disproportionate repetition failure of conduction aphasia has been read as damage to a phonological store, linking a clinical syndrome to a component of the normal memory architecture. The near-universal naming deficit speaks to the retrieval stage of lexical access, and the agrammatism of nonfluent aphasia to the separability of syntax from the lexicon.

Aphasia also disciplines theory. Any account of how words are stored, retrieved, sequenced, and understood must predict the syndromes that actually occur and forbid the ones that do not — and the eight-cell truth table above is a compact statement of what a theory must explain. That the observed syndromes cluster where they do, and that certain combinations are common while others are vanishingly rare, is data no purely behavioural study of intact speakers could supply.

Current Directions

The active research front has moved decisively from localization to network dynamics and quantitative prediction. Large-sample lesion-symptom mapping, pooling hundreds of patients, continues to redraw the anatomy of each syndrome and to expose the limits of the classical categories (Fridriksson et al., 2018). The open goal is prediction at the level of the individual: given a specific lesion, forecast the deficit and the likely trajectory of recovery well enough to plan therapy — a goal that neurocomputational models of the language network are beginning to make tractable (Stefaniak et al., 2020). In parallel, the neuroplasticity of the recovering network is being mapped directly, with an eye to timing and targeting therapy to the periods and regions where the brain is most able to reorganize (Kiran & Thompson, 2019). The unresolved questions are whether the classical syndrome labels should be retired in favour of deficit-and-lesion descriptions, and how far therapy can be personalized from an individual’s lesion and network profile.

Common Misconceptions

Aphasia means the person has lost their intelligence.

No. Aphasia is a loss of language, not of thought. Aphasic people reason, remember, and understand situations they cannot put into words; the dissociation between spared cognition and lost language is the founding observation of the field (Damasio, 1992), and conflating the two is both wrong and demeaning.

Aphasia is a problem with the voice or the speech muscles.

That is dysarthria (or apraxia of speech), a disorder of articulation. Aphasia is a disorder of language — word retrieval, grammar, comprehension — and can occur with perfectly intact articulation, and equally in reading and writing (Damasio, 1992).

Broca’s area is the speech centre and Wernicke’s is the comprehension centre.

A useful first approximation, but lesion mapping shows both functions are distributed across networks, that these regions are multifunctional, and that damage confined to the classical areas does not reliably produce the classical syndrome (Tremblay & Dick, 2016).

Aphasia is permanent.

Most post-stroke aphasia improves, substantially so in the first months, and speech-and-language therapy produces measurable functional gains (Brady et al., 2016). Primary progressive aphasia is the exception, worsening over time because its cause is neurodegeneration.

Glossary

Agrammatism.

The omission of grammatical words and inflections, producing telegraphic speech; characteristic of nonfluent aphasia.

Anomia.

Difficulty retrieving words, especially names of objects; the most common and most persistent aphasic symptom.

Aphasia.

An acquired impairment of language caused by brain damage, sparing general cognition.

Arcuate fasciculus.

The white-matter fibre tract classically held to connect Wernicke’s and Broca’s areas; its disruption is the traditional explanation of conduction aphasia.

Comprehension.

The ability to understand spoken or written language; one of the three axes classifying aphasia.

Dual-stream model.

The account of language as a ventral sound-to-meaning stream and a dorsal sound-to-articulation stream, reframing the aphasia syndromes as stream-level breakdowns.

Dysarthria.

A disorder of speech articulation caused by weakness or incoordination of the speech muscles; distinct from aphasia.

Fluency.

The ease, rate, and melodic flow of spontaneous speech; the axis separating anterior (nonfluent) from posterior (fluent) aphasia.

Global aphasia.

The most severe syndrome, in which fluency, comprehension, and repetition are all lost; it follows large perisylvian lesions.

Naming.

The retrieval of words on demand, tested by asking a patient to name objects; impaired in nearly every aphasia.

Perisylvian.

The cortex surrounding the Sylvian fissure of the left hemisphere, where the core language network lies.

Primary progressive aphasia.

A neurodegenerative aphasia that begins insidiously and worsens over years, rather than following an acute lesion.

Repetition.

The ability to repeat words and sentences on demand; the axis that isolates conduction and, by its sparing, the transcortical aphasias.

Transcortical aphasia.

A family of syndromes in which repetition is spared while spontaneous speech or comprehension fails, because the lesion disconnects the intact perisylvian loop from wider cortex.

Voxel-based lesion-symptom mapping.

A statistical method that tests, at each voxel, whether damage predicts a deficit across a patient sample, re-deriving functional anatomy from data.

Key Researchers

Paul Broca

(1824–1880). French physician and anthropologist whose 1861 report on the patient Leborgne localized articulate speech to the left inferior frontal gyrus, founding the study of aphasia and giving nonfluent aphasia its name. Wikipedia - Wikidata

Nina F. Dronkers

(living). Neuroscientist at the University of California, Berkeley whose voxel-based lesion-symptom mapping re-derived the anatomy of the aphasias from data, showing the classical syndrome-to-lesion mappings were only approximate. ORCID - Wikidata

Norman Geschwind

(1926–1984). American behavioural neurologist whose disconnexion-syndrome framework revived the classical connectionist model and organized the aphasia syndromes around named cortical centres joined by white-matter tracts. Wikipedia - Wikidata

Gregory Hickok

(living). Cognitive scientist at the University of California, Irvine and co-architect of the dual-stream model of speech processing, which reorganizes the neurobiology of language around dorsal and ventral streams and reframes the aphasia syndromes as stream-level breakdowns. ORCID

Argye E. Hillis

(living). Neurologist at the Johns Hopkins University School of Medicine whose work on acute stroke aphasia showed that hypoperfused but structurally intact cortex can produce aphasic deficits that resolve when blood flow is restored, refining lesion-symptom inference. ORCID

Marsel Mesulam

(living). Behavioural neurologist at Northwestern University who defined primary progressive aphasia as a neurodegenerative, non-stroke route to aphasia, extending the syndrome beyond acute focal lesions. Wikidata

Matthew A. Lambon Ralph

(living). Neuroscientist at the University of Cambridge whose neurocomputational account of language and its breakdown models both the deficits of aphasia and the mechanisms of recovery, moving the field from box-and-arrow syndromes to distributed representations. ORCID

Carl Wernicke

(1848–1905). German neurologist who in 1874 described the fluent, comprehension-impaired aphasia that bears his name and proposed the first connectionist model of language in the brain, distinguishing sensory from motor speech centres. Wikipedia - Wikidata

Frequently Asked Questions

What is the difference between aphasia and dysarthria?

Aphasia is a disorder of language (retrieving words, building grammatical sentences, understanding others), while dysarthria is a disorder of the physical act of speaking, caused by weakness or incoordination of the speech muscles. A person can have one without the other: an aphasic patient may articulate clearly but produce the wrong words, and a dysarthric patient may know exactly what to say but slur it.

What causes aphasia?

The most common cause is stroke affecting the left middle cerebral artery, which supplies the language cortex. Aphasia also results from head trauma, brain tumours, infections, and, in its primary progressive form, neurodegenerative disease.

Can people recover from aphasia?

Most people with post-stroke aphasia improve, especially in the first weeks and months, and speech-and-language therapy produces measurable gains in communication. The degree of recovery depends mainly on the initial severity and on the size and location of the lesion. Primary progressive aphasia, by contrast, worsens over time.

Does aphasia mean a loss of intelligence?

No. Aphasia impairs language, not thought. People with aphasia typically retain their memories, reasoning, and knowledge of the world; the difficulty is in encoding and decoding language, not in the ideas behind it.

How many types of aphasia are there?

The classical bedside taxonomy recognizes eight syndromes, generated by combining three roughly binary axes: fluency, comprehension, and repetition. MeSH indexes four subtypes at the descriptor level: Broca, Wernicke, conduction, and primary progressive aphasia.

Why is repetition tested so carefully?

Because repetition requires hearing, briefly holding, and reproducing a spoken string, it uses the whole language loop and is uniquely sensitive to damage of the connecting pathway. A selective failure of repetition with intact comprehension defines conduction aphasia, and preserved repetition despite other deficits defines the transcortical aphasias.

Is aphasia the same as Alzheimer’s disease?

No, though they can overlap. Aphasia is a specific language impairment; Alzheimer’s is a broad dementia affecting memory and many cognitive domains. However, one neurodegenerative syndrome, primary progressive aphasia, presents first and predominantly as a worsening aphasia, and can be an early presentation of an underlying degenerative disease.

Which hemisphere is affected in aphasia?

Almost always the left. Language is lateralized to the left hemisphere in nearly all right-handers and the large majority of left-handers, so aphasia is overwhelmingly a consequence of left-hemisphere damage; right-hemisphere language dominance is rare.

Support Organizations

References