Conduction Aphasia

Abstract

Conduction aphasia is a type of aphasia in which a person understands speech well and speaks fluently, yet cannot repeat what was just heard — a deficit in repetition out of all proportion to comprehension or spontaneous output. Since Carl Wernicke first predicted it in 1874, it has been the clearest clinical evidence that repeating a word is a distinct operation from understanding or producing one, marked by self-corrected phonemic near-misses clinicians call *conduites d’approche*. This article treats the syndrome as a window onto the architecture that links hearing to speaking: the classical account of a severed connection, the competing account of an overloaded phonological short-term store, and the modern dual-stream reframing that unites them. Three interactive demonstrations trace the deficit from the language loop to the length effect in repetition.

Keywords: conduction aphasia, repetition, arcuate fasciculus

Repeating a just-heard word feels like the simplest of language acts — so automatic that a child does it before understanding, and a parrot does it without understanding at all. Yet it is not simple. To repeat a word, the brain must hold its sound in mind long enough to reassemble it as an articulatory plan, and that brief bridge from perception to production is a component of language in its own right, separable from both ends it connects. Conduction aphasia is the syndrome that separates it. The patient comprehends normally and speaks fluently, but the bridge is out, and so the one thing they cannot reliably do is carry a sound across it intact (Geschwind, 1965).

The syndrome is defined by a triad: fluent spontaneous speech, good auditory comprehension, and disproportionately impaired repetition. That third feature, present against the spared background of the first two, is the diagnostic signature. When a patient who converses easily and follows instructions without difficulty is nonetheless unable to repeat a phrase back, the classical inference is that the machinery of comprehension and the machinery of production are each intact but have been disconnected from one another. The history of conduction aphasia is the history of arguments about what that disconnection really is — a cut cable, a small memory store, or a failure of sensory-motor translation — and each argument has sharpened the cognitive model of how the brain turns hearing into speech.

What Conduction Aphasia Is

Conduction aphasia is an acquired language disorder, most often the result of a stroke in the left perisylvian region, whose defining feature is a selective impairment of repetition in a patient whose comprehension and fluency are largely preserved (Bernal & Ardila, 2009). Asked to converse, the patient does so easily and to the point; asked what a sentence means, they answer correctly; but asked simply to repeat a word or phrase verbatim, they falter, and the longer or less familiar the target, the worse they do. It is this dissociation — repetition failing while the abilities it seems to sit between are intact — that gives the syndrome its theoretical importance.

The speech that patients do produce is fluent but marred by phonemic paraphasias: errors in which the target word’s sounds are substituted, transposed, added, or dropped, so that *hippopotamus* might come out as *hipponamus* or *hikkopotamus*. Critically, these are errors of *sound*, not of *meaning* — the patient is plainly reaching for the right word and mangling its phonology, not selecting the wrong word. They are usually aware of the error and try again, producing a series of successive approximations that edge toward the target, the behaviour Frenchspeaking neurologists named *conduites d’approche* (“approach behaviours”). This self-monitoring is itself diagnostic: it shows that comprehension and error-detection are working, and that the failure lies specifically in assembling and holding the phonological form for output.

It is essential to separate conduction aphasia from the two syndromes it sits between. In Wernicke’s aphasia, comprehension is impaired and speech, though fluent, is full of errors the patient does not notice; in Broca’s aphasia, speech is effortful and non-fluent. Conduction aphasia is neither: comprehension is good and output is fluent, and the patient knows when they have erred. The name itself encodes the classical theory — Wernicke called it *Leitungsaphasie*, a disorder of *conduction* between the speech centres — and although that theory has been repeatedly revised, the clinical picture it names has proved remarkably stable for a century and a half (Ardila, 2010).

The Classical Disconnection Account

The syndrome began as a prediction, not an observation. In 1874 Carl Wernicke, reasoning from his new model of the two speech centres — a posterior sensory centre for understanding words and an anterior motor centre for producing them — argued that a lesion of the *connection* between them, sparing both centres, should produce a distinctive picture: preserved comprehension, preserved fluent speech, but disrupted repetition, because repeating a heard word requires passing its sound-based representation from the sensory centre to the motor one. He called this predicted syndrome *Leitungsaphasie*, conduction aphasia, and cases matching it were duly found.

Nearly a century later, Norman Geschwind revived and formalised the account within his broader theory of the disconnexion syndromes — disorders produced not by damage to a processing centre but by the severing of the white-matter tracts connecting centres (Geschwind, 1965). In this framework, conduction aphasia is the textbook disconnection: the lesion interrupts the arcuate fasciculus, the great dorsal fibre bundle arching from the temporal comprehension region forward to the frontal production region, so that the two intact cortical areas can no longer communicate. Comprehension (a temporal function) and fluent output (a frontal function) survive; repetition, which uniquely requires the connection, does not. For a generation this arcuate-disconnection story was the standard explanation, and it remains the account taught first because it captures the core logic of the syndrome so cleanly.

The lesion evidence at first seemed to confirm it. Hanna Damasio and Antonio Damasio, in a landmark 1980 study, mapped the lesions of conduction aphasics and found damage clustered in the region the arcuate fasciculus traverses — the left supramarginal gyrus and the underlying white matter of the insula and temporoparietal junction — consistent with an interruption of the dorsal pathway (Damasio & Damasio, 1980). The account was anatomically grounded, mechanistically simple, and predictively successful. Its difficulties came later, and from two directions: doubts about whether the arcuate fasciculus is really the critical structure, and a rival cognitive account of what repetition actually requires.

The Repetition Deficit and Phonological Short-Term Memory

The rival account came from cognitive psychology rather than neurology. In 1977 Tim Shallice and Elizabeth Warrington argued that at least some conduction aphasics do not have a *disconnection* at all, but a defective auditory-verbal short-term store — a shrunken phonological memory that can no longer hold a sound-based representation long enough to reproduce it (Shallice & Warrington, 1977). On this view repetition fails not because the signal cannot be *routed* from comprehension to production, but because it cannot be *held* during the brief interval that routing requires. The prediction is specific and testable: if the deficit is a limited store, then repetition should fail as a function of *length* — a single word might be repeated, two or three words not, because the store overflows.

This length effect is exactly what such patients show, and it distinguishes the short-term-memory account from the disconnection one. A pure disconnection would be expected to disrupt repetition roughly uniformly; a limited buffer predicts a graded collapse, fine for short material and failing progressively as the string lengthens, because each additional item is another thing the overloaded store must retain. The distinction between the two variants — a *reproduction* conduction aphasia tied to the sensorimotor conversion and a *repetition* conduction aphasia tied to the short-term store — became a standard refinement of the syndrome, and the phonological-store account tied conduction aphasia directly to the wider psychology of working memory and the phonological loop.

A particularly clean demonstration of what the store does and does not preserve came from Juliana Baldo, Nina Dronkers, and colleagues, who showed that conduction aphasics “get the gist but lose the trace”: presented with a sentence they cannot repeat verbatim, patients nonetheless retain and can report its *meaning*, so the failure is specific to the surface phonological form, not to comprehension or memory in general (Baldo, Klostermann, & Dronkers, 2008). The verbatim trace decays while the gist survives — precisely what one expects if the damaged component is a short-term store for sound rather than a store for meaning. This dissociation between retaining a sentence’s form and retaining its content is one of the sharpest results in the literature and a direct link between the aphasia and the architecture of verbal memory.

Worked Example

The most diagnostic quantitative signature of the short-term-memory account is the length effect in repetition, and it follows directly from the idea of a limited phonological store. Suppose repeating a sequence requires holding each of its phonological units — roughly, each word or syllable — in the store until it can be produced, and that a damaged store retains any one unit, intact and in order, with some fixed independent probability $p$. Because the sequence is repeated correctly only if *every* unit survives, the probability of correctly repeating an $n$-unit sequence is

$$P(\text{correct}) = p^{\,n},$$

and the expected error rate is $1 - p^{\,n}$. The error rate therefore climbs steeply and with acceleration as the target lengthens: each extra unit multiplies in another chance for the store to drop something.

Take a plausible per-unit retention of $p = 0.75$ for a moderately impaired store — one that holds any given unit three times in four. The model then predicts the repetition error rates in Table 1. A single word is missed about a quarter of the time; a four-word phrase more than two-thirds of the time; a six-word sentence over four-fifths. This steep, curved rise with length — nearly usable for the shortest targets, collapsing for the longest — is the fingerprint of a buffer disorder, and it is what separates a short-term-memory conduction aphasia from a comprehension deficit, where difficulty tracks a sentence’s meaning and structure rather than its raw length (Shallice & Warrington, 1977).

Table 1

Predicted Repetition Error Rate by Sequence Length, Phonological-Store Model (per-unit retention p = 0.75)

The model is deliberately simple — it treats every unit as equally vulnerable and independent — but it captures the essential clinical fact that repetition in conduction aphasia degrades with length in a curved, accelerating way rather than failing all at once. The interactive demonstration below lets the reader vary both the per-unit retention and the sequence length, so the length effect emerges as a surface the two parameters trace out together.

Conduites d’approche and Phonemic Paraphasia

The most vivid clinical feature of conduction aphasia is the way patients *pursue* a word they cannot quite reproduce. Asked to repeat or name a target, the patient produces an approximation, hears that it is wrong, and tries again — and again — each attempt typically closer to the target than the last: *“caterpillar … pattercillar … catapiller … caterpillar.”* These successive self-corrections are the *conduites d’approche*, and they are among the most theoretically loaded behaviours in aphasiology, because they reveal several intact systems working around a single broken one.

For a patient to approach a target this way, three things must be true. The auditory representation of the target must be intact, or there would be nothing to aim at; the patient plainly knows the word they want. The error-monitoring system must be intact, or the patient could not tell that an attempt had missed — and conduction aphasics reliably do detect their own phonemic errors, in sharp contrast to Wernicke’s aphasics, who do not. What is broken is the step between: the assembly of a stable phonological plan for output. The approach behaviour is thus a real-time trace of a system trying to converge on a target it can represent and evaluate but cannot cleanly generate, and it fits both the classical and the short-term-memory accounts — a failing sensorimotor translation or a decaying buffer would each produce attempts that fall short and must be retried.

The errors themselves, the phonemic paraphasias, carry the same message. They are substitutions (*teligision* for *television*), transpositions (*aminal* for *animal*), omissions, and additions of *phonemes* — the sounds are wrong but they are the target’s neighbours, clustering around the intended word rather than replacing it with a different one. This is the phonological signature that distinguishes conduction aphasia from the *semantic* paraphasias of deeper language disorders, where the wrong *word* is produced. In conduction aphasia the word is right and its sounds are scrambled — the mark of a fault in phonological encoding, not in lexical selection (Ardila, 2010).

The Anatomy Revisited

The clean arcuate-disconnection story ran into trouble as the anatomy was examined more closely. The first difficulty was empirical: careful lesion studies found that damage restricted to the arcuate fasciculus does not reliably produce conduction aphasia, and that the syndrome can arise from cortical lesions of the supramarginal gyrus or superior temporal region with the arcuate relatively spared. James Anderson and colleagues reported a re-examination of the Wernicke–Geschwind model in which the classical arcuate account did not fit the data as neatly as the textbooks implied, pointing instead to a larger role for cortical damage in the temporoparietal junction (Anderson et al., 1999). The tract, it seemed, was neither necessary nor sufficient in the simple way the model required.

The second difficulty was that the arcuate fasciculus turned out to be more complicated than a single cable. Using diffusion tractography, Marco Catani, Derek Jones, and Dominic ffytche resolved the perisylvian language network into *multiple* segments: a long direct pathway connecting temporal and frontal regions, and a pair of shorter, indirect pathways relaying through the inferior parietal lobe (Catani, Jones, & ffytche, 2005). A syndrome once attributed to cutting “the” arcuate now had to be re-described against a branching architecture in which different segments might carry different aspects of the sound-to-speech mapping, and in which a lesion’s effect would depend on exactly which branch it caught.

Reviews weighing all of this reached a nuanced verdict. Byron Bernal and Alfredo Ardila, surveying the role of the arcuate fasciculus, concluded that its damage is *associated* with conduction aphasia but is neither the whole story nor invariably present, so the syndrome is better understood as a disturbance of a perisylvian *network* than as the cutting of one tract (Bernal & Ardila, 2009). The most direct challenge came from large-sample lesion work: Julius Fridriksson and colleagues, re-deriving the aphasia syndromes from data in *Anatomy of aphasia revisited*, questioned whether conduction aphasia maps cleanly onto arcuate-fasciculus damage at all, finding its lesion correlates more distributed and more cortical than the classical model predicts (Fridriksson et al., 2018). The arcuate account survives as a first approximation and a teaching tool, but not as the final anatomy.

The Dual-Stream Reframing

The account that now does the most theoretical work reframes the whole problem. Gregory Hickok and David Poeppel proposed that speech processing is organised into two streams: a *ventral* stream, running into the temporal lobe, that maps sound onto meaning, and a *dorsal* stream, running up through the parietal lobe to frontal motor regions, that maps sound onto articulation — an auditory-motor integration system (Hickok & Poeppel, 2007). Comprehension is a ventral-stream function; repetition and the on-line guidance of speech by its sound are dorsal-stream functions. On this view conduction aphasia is precisely a dorsal-stream disorder: the sensorimotor interface that lets an auditory representation constrain a motor one is damaged, so the patient understands (ventral stream intact) but cannot faithfully reproduce (dorsal stream broken).

This reframing has a decisive virtue: it *unifies* the two older accounts. The dorsal stream is at once a *connection* between auditory and motor cortex — so damaging it disconnects them, as the classical model said — and the substrate of a *phonological short-term store*, because holding a sound in mind to reproduce it is exactly what an auditory-motor buffer does, as the short-term-memory account said. Hickok’s later computational account of speech production makes this explicit, casting the dorsal stream as a state feedback control system in which auditory targets guide motor commands, and conduction aphasia as the loss of that feedback loop — a single mechanism from which both the disconnection picture and the memory picture fall out (Hickok, 2012).

The evidence for the sensorimotor-integration view is convergent. Bradley Buchsbaum and colleagues, pooling lesion and functional-imaging data, showed that a region deep in the left temporoparietal junction — area *Spt*, the sylvian parietal-temporal region — is active during both the perception and the covert production of speech, exactly the property required of an auditory-motor interface, and that conduction aphasia, phonological short-term memory, and sensorimotor integration all converge on this territory (Buchsbaum et al., 2011). The syndrome that began as a severed cable and became an overloaded buffer is now best understood as the failure of a computation — the on-line alignment of what is heard with what is said — that the dorsal stream exists to perform.

Assessment and Treatment

Because the defining feature of conduction aphasia is a *dissociation*, assessment is organised around demonstrating it: showing that repetition is impaired against a background of spared comprehension and fluent output. The clinician contrasts performance across tasks — conversational speech, auditory comprehension, naming, reading, and repetition — and looks for the signature profile: fluent connected speech with phonemic paraphasias, good comprehension, and repetition that is disproportionately poor and that worsens with the length and unfamiliarity of the target. The presence of *conduites d’approche* and of self-monitored phonemic errors supports the diagnosis and helps separate it from Wernicke’s aphasia, where errors go unnoticed, and from Broca’s, where speech is non-fluent (Middlebrooks et al., 2017).

Treatment follows the cognitive analysis. Where the deficit is best understood as a weakened phonological short-term store, therapy targets the maintenance and manipulation of sound-based representations, using repetition and rehearsal tasks graded by length to rebuild span. Where the emphasis is on phonological encoding for output, cueing hierarchies and self-monitoring strategies help the patient exploit their intact comprehension and error-detection to converge on targets more efficiently — in effect, to make the *conduites d’approche* shorter and more reliable. Conduction aphasia also has a comparatively favourable natural history: many patients recover substantial repetition ability over months, and treatment is set against a background of frequently good spontaneous improvement (Stefaniak, Halai, & Lambon Ralph, 2020). Matching the therapy to the patient’s specific profile — store versus encoding, and the severity of each — is the clinical payoff of taking the cognitive model seriously.

Discussion

Conduction aphasia occupies a special place in the cognitive neuroscience of language because it isolates a component that intuition would not have separated: the act of holding a heard sound and reproducing it, distinct from understanding it and from speaking spontaneously. That this act can be selectively lost is the strongest evidence that repetition is not a trivial pass-through but a computation with its own machinery, and the syndrome has served as a proving ground for successive models of what that machinery is. Each model — disconnection, short-term store, dual-stream integration — has had to honour the same stubborn clinical facts: spared comprehension, fluent output, length-dependent repetition failure, phonemic paraphasia, and self-monitored approach behaviour.

The trajectory of these models is itself instructive about how cognitive neuroscience works. The disconnection account was anatomically motivated and mechanistically clean, but too simple for the anatomy (Fridriksson et al., 2018). The short-term-memory account was cognitively precise and made the right quantitative prediction — the length effect — but was agnostic about mechanism (Shallice & Warrington, 1977). The dual-stream account subsumes both by identifying a single neural computation, auditory-motor integration, whose failure produces at once a disconnection and a memory deficit (Hickok, 2012). The progression is not a series of refutations but a deepening: each account preserved the true observations of its predecessor and reinterpreted them at a more fundamental level.

Cognitive Implications

Beyond the clinic, conduction aphasia has shaped how cognitive science thinks about the interface between perception and action. The dorsal auditory-motor stream is a concrete instance of a sensorimotor transformation — a mapping from a perceptual representation onto the motor commands that would reproduce it — and the same computational idea recurs wherever perception must guide skilled action, from reaching toward a seen object to imitating an observed gesture. That speech has such a stream, and that its damage produces a specific and recognisable aphasia, is evidence that the brain treats *saying what one hears* as a species of the general problem of *doing what one perceives*.

The syndrome also anchors the psychology of verbal short-term memory to a neural substrate. The “gist but not the trace” dissociation shows that the momentary, verbatim store for speech sounds is separable from the retention of meaning, giving a lesion-based reality to the phonological loop of working memory models (Baldo et al., 2008). And the syndrome’s history is a case study in the value of *double dissociation* and *convergent evidence*: it was cognitive dissociations that first challenged the pure disconnection account, and the pooling of lesion with functional-imaging data that finally localised the sensorimotor interface (Buchsbaum et al., 2011). Conduction aphasia, in this light, is less a narrow clinical entity than a recurring lesson in how the mind maps hearing onto speech.

Current Directions

Current research on conduction aphasia is moving with the broader reappraisal of the classical model of language in the brain. A first front is the dismantling of the Broca–Wernicke–arcuate picture itself: Pascale Tremblay and Anthony Dick, arguing that “Broca and Wernicke are dead,” document how thoroughly modern data have outgrown the nineteenth-century centres-and-cables framework, and conduction aphasia — the syndrome that framework was built to explain — is squarely part of that re-examination (Tremblay & Dick, 2016). The question is no longer *which tract is cut* but *which distributed network computation has failed*, and how the classical syndromes map onto it.

A second front is data-driven lesion mapping on large patient samples, which is redrawing the anatomical correlates of every aphasia type, conduction included, and finding them more cortical and more distributed than the arcuate model held (Fridriksson et al., 2018). A third is intraoperative and preoperative language mapping, where the practical need to spare repetition and naming during neurosurgery has driven a contemporary, network-based framework of language organisation that must account for conduction-type deficits when a dorsal-stream structure is stimulated or resected (Middlebrooks et al., 2017). Running through all three is the dual-stream model as the organising theory and the growing use of tractography and functional imaging to test, patient by patient, exactly which part of the auditory-motor interface a given lesion has damaged — and, increasingly, to predict and support recovery (Stefaniak et al., 2020).

Common Misconceptions

Conduction aphasia means the person cannot understand speech.

No — comprehension is a defining spared feature. Patients follow conversation and grasp meaning normally; what they cannot do is repeat verbatim. The deficit sits between understanding and producing, not in understanding itself (Bernal & Ardila, 2009).

The errors are word-choice mistakes.

Not usually. The characteristic errors are phonemic paraphasias — the right word with its sounds substituted or rearranged (television → teligision) — not semantic errors in which a different word is chosen. The word is right; its phonology is scrambled (Ardila, 2010).

It is always caused by cutting the arcuate fasciculus.

No. The arcuate-disconnection account is the classical teaching model, but careful and large-sample lesion studies show the syndrome can arise from cortical temporoparietal damage with the arcuate spared, and that arcuate damage alone does not reliably produce it. It is better seen as a perisylvian network disorder (Fridriksson et al., 2018).

Patients do not notice their errors.

The opposite is true and diagnostic. Conduction aphasics reliably detect their own phonemic errors and try again, producing the successive approximations of conduites d’approche — unlike Wernicke’s aphasics, whose comprehension and self-monitoring are impaired (Baldo et al., 2008).

Glossary

Aphasia.

An acquired disorder of language from brain damage; conduction aphasia is one of its classical fluent subtypes.

Arcuate fasciculus.

The dorsal white-matter tract arching from temporal to frontal language cortex; its interruption was the classical explanation of conduction aphasia.

Area Spt.

The sylvian parietal-temporal region active in both perceiving and covertly producing speech, proposed as the auditory-motor interface.

Conduction aphasia.

An aphasia marked by fluent speech and good comprehension but disproportionately impaired repetition, with phonemic paraphasias.

Conduites d’approche.

The successive self-corrected approximations a patient makes toward a target word, edging closer with each attempt; a hallmark of conduction aphasia.

Disconnexion syndrome.

A disorder produced by severing the white-matter connections between intact cortical centres rather than by damaging a centre itself.

Dorsal stream.

The auditory-motor pathway that maps speech sounds onto articulation; its damage is the modern account of conduction aphasia.

Dual-stream model.

The theory that speech is processed by a ventral sound-to-meaning stream and a dorsal sound-to-articulation stream.

Gist-versus-trace.

The dissociation whereby conduction aphasics retain a sentence’s meaning while losing its verbatim phonological form.

Length effect.

The steep, accelerating rise in repetition error rate with the length of the target, the signature of a limited phonological store.

Phonemic paraphasia.

A sound-level error — substitution, transposition, omission, or addition of phonemes — that leaves the target word recognisable.

Phonological short-term store.

A brief memory for speech sounds; on one account its impairment is the core of conduction aphasia.

Repetition.

Reproducing aloud a word or phrase just heard; the ability selectively impaired in conduction aphasia.

Ventral stream.

The sound-to-meaning pathway of the dual-stream model; spared in conduction aphasia, which is why comprehension is intact.

Key Researchers

Marco Catani

Neuroanatomist at the Institute of Psychiatry, Psychology and Neuroscience, King’s College London, whose diffusion tractography resolved the perisylvian language network into direct and indirect segments, reframing the arcuate anatomy on which conduction aphasia was built. ORCID

Antonio Damasio

Neuroscientist and director of the Brain and Creativity Institute at the University of Southern California who, with Hanna Damasio, produced the modern lesion-anatomical account of conduction aphasia, localising it to the supramarginal and insular region the arcuate traverses. ORCID - Wikipedia - Wikidata

Nina F. Dronkers

Neuroscientist at the University of California, Berkeley and the VA Northern California Health Care System whose voxel-based lesion-symptom mapping redefined the anatomy of the aphasias, and whose work on conduction aphasia showed that patients retain the gist of a sentence while losing its verbatim trace. ORCID - Wikidata

Julius Fridriksson

Professor of communication sciences and disorders at the University of South Carolina, whose large-sample lesion study *Anatomy of aphasia revisited* re-derived the aphasia syndromes from data and questioned whether conduction aphasia maps cleanly onto arcuate-fasciculus damage. ORCID

Norman Geschwind

(1926–1984). American behavioural neurologist whose theory of the disconnexion syndromes revived and formalised Wernicke’s account, making interruption of the arcuate fasciculus the textbook explanation of conduction aphasia for a generation. Wikipedia - Wikidata

Gregory Hickok

Professor of cognitive sciences at the University of California, Irvine and co-architect of the dual-stream model of speech processing, which recasts conduction aphasia as a breakdown of dorsal-stream auditory-motor integration rather than a simple fibre disconnection. ORCID

Carl Wernicke

(1848–1905). German neurologist who in 1874 predicted conduction aphasia — *Leitungsaphasie* — as a disconnection between the sensory and motor speech centres before it had been clinically documented, the founding hypothesis of the syndrome. Wikipedia - Wikidata

Frequently Asked Questions

What exactly is conduction aphasia?

Conduction aphasia is an acquired language disorder in which a person understands speech well and speaks fluently but cannot reliably repeat what they have just heard. The impairment of repetition is out of proportion to any difficulty in comprehension or spontaneous speech, and that dissociation is what defines the syndrome.

Why can the person speak and understand but not repeat?

Because repeating a word is a distinct operation from understanding or producing one: it requires holding the word’s sound in mind and passing it to the speech system to be reproduced. In conduction aphasia the comprehension and production systems are each intact, but the process that links them for repetition — whether a connecting pathway or a short-term sound store — is damaged.

What kind of errors do patients make?

They make phonemic paraphasias: the sounds of the target word are substituted, transposed, dropped, or added, so the word comes out distorted but recognisable (for example, *television* as *teligision*). These are errors of sound, not of meaning, and patients usually notice them and try again.

What are conduites d’approche?

They are the successive self-corrected attempts a patient makes when reaching for a word — each try typically closer to the target than the last, as in *“caterpillar … pattercillar … catapiller … caterpillar.”* They show that the patient knows the target and can hear their own errors, and that the fault lies in assembling the phonological form for output.

What part of the brain is affected?

Classically the arcuate fasciculus, the dorsal tract connecting temporal and frontal language areas, and the overlying supramarginal gyrus. Modern lesion studies place the damage more broadly across the left temporoparietal junction, and the dual-stream model attributes the syndrome to the dorsal auditory-motor stream rather than to any single tract.

How is it different from Wernicke’s and Broca’s aphasia?

In Wernicke’s aphasia comprehension is impaired and errors go unnoticed; in Broca’s aphasia speech is effortful and non-fluent. Conduction aphasia is neither: comprehension is good, speech is fluent, and the patient detects their own errors — only repetition is disproportionately impaired.

Is repetition worse for longer material?

Yes. Repetition typically degrades with the length of the target, failing progressively as words or syllables are added. This length effect is a central piece of evidence for the view that a limited phonological short-term store underlies the deficit.

Can conduction aphasia improve?

Often, yes. Conduction aphasia has a comparatively favourable outlook, with many patients regaining substantial repetition ability over months. Speech-language therapy targeted at the specific profile — rebuilding phonological span or supporting self-monitored phonological encoding — is set against this generally favourable natural recovery.

Support Organizations

References