VIEWS & REVIEWS

Aphasia
Progress in the last quarter of a century

Argye E. Hillis, MD ABSTRACT In the last 25 years, characterization of aphasia has shifted from descriptions of the
language tasks that are impaired by brain damage to identification of the disrupted cognitive pro-
Address correspondence and
reprint requests to Dr. Argye E. cesses underlying language. At the same time advances in technology, including functional imaging,
Hillis, Department of Neurology, electrophysiologic studies, perfusion imaging, diffusion tensor imaging, and transcranial magnetic
Phipps 126, Johns Hopkins
Hospital, 600 North Wolfe
stimulation, have led to new insights regarding the relationships between language and the brain.
Street, Baltimore, MD 21287 These insights, together with computational models of language processes, converge on the view that
[email protected]
a given language task relies on a complex set of cognitive processes and representations carried out
by an intricate network of neural regions working together. Recovery from aphasia depends on resto-
ration of tissue function or reorganization of the cognitive/neural network underlying language, which
can be facilitated by a number of diverse interventions. The original research by the author reported in
this article was supported by NIH R01 DC05375. NEUROLOGY 2007;69:200–213

In the nineteenth century, Dax,1 Broca,2 Wernicke,3 and their contemporaries made a number
of important discoveries about the locations of brain lesions that cause disruption of lan-
guage (aphasia) by studying the brains of individuals who had been aphasic. The most reli-
able finding was that damage to the left hemisphere was discovered in patients who had had
language impairment. Another novel and important observation was that damage to more
anterior parts of the brain, particularly the left posterior inferior frontal cortex, was usually
found in patients whose spoken output was limited or poorly articulated,4 while damage to
more posterior regions in the temporal lobe was found in patients whose spoken output was
well articulated but meaningless.3
Although localization of functions in the brain lost favor in the early 20th century, these
early observations provided some of the groundwork for Norman Geschwind’s5 seminal
writings on aphasia classification and associated sites of lesions, briefly reviewed in this
article. Geschwind’s students and colleagues in Boston developed a well known clinical test,
the Boston Diagnostic Aphasia Examination (BDAE6) to assist clinicians in identifying pa-
tients with each type of aphasia. Although the earliest descriptions of aphasia may have
confounded impairments of speech (the motor processes involved in production of verbal
language) and language (the abstract symbols or representations and syntactic processes that
underlie verbal and written communication), Geschwind and his followers used the term to
describe disorders of language. When CT scans became available, other investigators con-
firmed that the various types of aphasia identified by Geschwind and classified with the
BDAE are generally associated with lesions in relatively distinct areas of the brain7,8 (but see
contradictory evidence9).
However, in the 1980s three developments revolutionized our thinking about aphasia.
First, PET and later functional MRI (fMRI) and magnetoencephalography showed that areas
in both hemispheres of the brain are activated specifically during language tasks, although the
left hemisphere reliably shows more activation in the majority of neurologically normal
adults.10-13 Furthermore, these functional imaging studies have revealed not only coordinated
activations in the previously described language centers, but also activation of more distant
From the Departments of Neurology and Physical Medicine and Rehabilitation, Johns Hopkins University School of Medicine, and Department of
Cognitive Science, Johns Hopkins University, Baltimore, MD.
Supported by NIH (NIDCD) through R01 DC 05375.
Disclosure: The author reports no conflicts of interest.

200 Copyright © 2007 by AAN Enterprises, Inc.

areas of cortex, such as inferior and anterior grids.23,24 If this area is essential for some
temporal cortex14 and in the basal ganglia component of the reading or naming tasks,
and thalamus.15 Secondly, there was mount- these functions should be disrupted during
ing appreciation for the complexity of the the inhibitory stimulation. In fact, cortical
most basic language tasks, such as naming a stimulation of nearby posterior basal tempo-
picture or understanding a word. Linguists, ral cortex disrupted reading of kanji (which
psychologists, neuroscientists, and mathe- represent meanings, not sounds) and picture
maticians began to dissect out the cognitive naming, which both require conversion of se-
processes and representations that are com- mantic to phonologic representations, but
putationally necessary to accomplish such did not affect comprehension of spoken
seemingly simple tasks. Advances in technol- words, reading of kana (which represent
ogy, such as parallel processing, allowed de- sounds of words), copying, or tool use.25 Al-
velopment of computational models that ternatively, one might observe whether tran-
simulated tasks of reading, naming, or word sient hypoperfusion in this area (seen on
recognition.16,17 The complexity of the tasks perfusion MRI or PET) in cerebral ischemia
seemed to fit well with the complexity of the disrupts reading or naming. In fact, hypoper-
network of neural regions that are engaged fusion or infarct of this area is strongly asso-
during the tasks as revealed by functional im- ciated with impaired oral reading and
aging. Third, the recognition that focal neu- naming in acute stroke.26 Another way to test
rodegenerative disease could cause primary if an area is essential for a particular task is
progressive aphasia18 allowed exploration of to use inhibitory repetitive transcortical
patterns of language deficits that are caused magnetic stimulation (rTMS) in normal sub-
by atrophy of regions of the brain that are jects during that task.27 More detailed exam-
not typically damaged by stroke. For exam- ples of these methods of exploring brain/
ple, progressive atrophy of left more than language relationships are described later,
right anterior and inferior temporal cortex after first reviewing recent insights regarding
(which are rarely affected by stroke) is asso- the aphasias described by Geschwind.5
ciated with progressive dissolution of
“CLASSIC APHASIA CATEGORIZATION”: VAS-
modality-independent semantics, or mean- CULAR SYNDROMES Broca’s aphasia is character-
ings of words and objects in a syndrome ized by nonfluent spontaneous speech and sentence
known as semantic dementia.19,20 repetition with relatively spared comprehension.
Furthermore, new technologies that per- “Nonfluency” often includes reduced phrase length,
mit the investigator to cause (or image) tem- impaired melody and articulatory agility, dimin-
porary dysfunction of focal regions of brain ished words per minute, or agrammatic sentence
have allowed direct tests of hypotheses about production. Because the concept is multidimen-
sional, it is often difficult to judge. A patient might
structure/function relationships derived from
have poor melody and/or disrupted articulation, but
functional imaging. To illustrate, functional produce complete, grammatic sentences or have flu-
imaging studies reliably show activation in ently articulated agrammatic speech. Fluency in
midfusiform gyrus during reading tasks,21 in- each of these dimensions is also a continuum rather
dicating that this area is consistently engaged than a dichotomy. Furthermore, many patients have
in reading. Other studies show reliable acti- predominantly nonfluent speech along all or most
vation in the left midfusiform gyrus during dimensions, but have islands of fluent production,
naming and other lexical tasks.22 To test the particularly of “overlearned” or social phrases, such
hypothesis that this area is necessary for a as, “You know what I mean.” For these reasons,
many patients cannot be reliably classified.6,28 Nev-
particular component of reading and/or
ertheless, poor fluency in each of these dimensions
naming, one might cause temporary dysfunc-
tends to co-occur, so that some patients have perva-
tion of the left midfusiform gyrus by inhibit- sively nonfluent speech, that is reliably perceived as
ing this region with cortical stimulation in such. These individuals tend to have good compre-
patients undergoing surgery during awake hension of conversations, single words, and gram-
craniotomy23 or through implanted subdural matically simple sentences, but have trouble

Neurology 69 July 10, 2007 201

 understanding sentences with grammatically com- drome, Broca’s aphasia is generally associated with
plex or “non-canonical” sentence structure, such as damage or dysfunction in posterior, inferior frontal
passive voice sentences and sentences with object gyrus, which includes Broca’s area (Brodmann area
relative clauses (e.g., “The girl that the boy kissed 44 and 45). However, when damage is isolated to
ran away”29) and those requiring maintenance and Broca’s area, or part of Broca’s area, then the only
processing of word order or comparisons of word deficit may be impaired motor planning and pro-
meanings (e.g., “Is a horse larger than a dog?”30). gramming of motor speech (“apraxia of speech”),
Such sentences are generally understood in conver- which frequently is observed in Broca’s aphasia but
sation because the context limits the possible inter- can occur in isolation.37 The entire syndrome of
pretations. Broca’s aphasia is often associated with Broca’s aphasia may require damage or neural com-
impaired spelling of familiar words, as well as diffi- promise of a larger area surrounding Broca’s area
culty using letter-sound associations (phonics) to supplied by the superior division of the left MCA.
read or spell unfamiliar words. The motor deficit of apraxia of speech that is usu-
Whether the various language deficits observed ally associated with the language deficits in Broca’s
in Broca’s aphasia can be accounted for by a single aphasia can account for several aspects of impaired
underlying deficit has been a matter of energetic de- fluency, including disrupted articulatory agility and
bate. One proposal was that the agrammatic speech prosody.
and impaired comprehension of grammatically There is one caveat in attempting to relate lan-
complex sentences could be explained by an under- guage syndromes to particular areas of the brain,
lying syntactic processing deficit.31 However, this whether one identifies areas by Brodmann areas,
hypothesis was undermined by the reports of pa- gyri, or groups of voxels on scans registered to a
tients with agrammatic speech without difficulty particular atlas, spatial coordinates, or an “ideal-
understanding grammatically complex sentences.32 ized” brain image. There is a great deal of individual
Reduced lexical activation33 or impairment in main- variability in the boundaries of the cytoarchitectural
taining and processing the sequence of linguistic ele- fields identified by Brodmann, as well as in the pat-
ments might account for many of the frequently tern of sulci and gyri and general shape of individual
observed deficits, including those in reading, spell- brains.38,39 Therefore, “localization” of functions
ing, and repetition,30 but like all “unifying” impair- can only be approximate.
ments would have difficulty explaining cases that It is also important to note that dysfunctional
show selective impairment of just one of the tasks. neural tissue surrounding an infarct can also con-
Furthermore, individuals with Broca’s aphasia often tribute to the observed deficits. Thus, patients with
have difficulty naming verbs relative to nouns34,35 or hypoperfusion but no infarct in and around Broca’s
producing particular verb forms, in addition to area often have the typical vascular syndrome de-
comprehension of grammatically complex struc- scribed above.30,40 Deficits have been attributed to
tures and other deficits described above. This array hypoperfusion, rather than the small, usually
of deficits may be difficult to explain by proposing a deeper, infarcts in the same patients, because restor-
single underlying impairment. The various dissocia- ing blood flow to the hypoperfused regions resulted
tions between frequently impaired language func- in immediate recovery of language functions.30,41-44
tions have led some authors to argue that Broca’s Wernicke’s aphasia is characterized by fluent but
aphasia is not a theoretically coherent syndrome relatively meaningless spontaneous speech and rep-
with a single underlying impairment.36 Neverthe- etition and relatively poor comprehension of words,
less, it may well be a vascular syndrome—impair- sentences, and conversation. Spoken language may
ment of a collection of speech and language be limited to jargon comprised of either real words
functions that depend on nearby brain regions all or neologisms (nonwords such as “klimorata”) or a
supplied by the superior division of the left middle combination of the two. In contrast to those with
cerebral artery (MCA). This hypothesis would ex- Broca’s aphasia, the individual with Wernicke’s
plain the frequent association between agrammatic aphasia is typically unaware of the errors. The ap-
speech, impaired articulation, verb production, and propriate melody or intonation may give the im-
asyntactic comprehension in the same way as the pression that the person is speaking another
frequent association between agrammatic speech language. Particularly in the acute stage, there is of-
and right hemiplegia—they are both caused by ten a profound impairment of comprehension, such
damage or dysfunctional tissue in posterior frontal that the patient may listen to others and respond
cortex, but are not caused by a single underlying fluidly with language-like, meaningless utterances
functional deficit. for hours, with no apparent inkling that he or she
Consistent with the concept of a vascular syn- has neither understood anything others have said

202 Neurology 69 July 10, 2007

nor said anything that could be understood by oth- TCM, motor speech is spared, as evidenced by
ers. Often the person will intermittently include a spared repetition, while other functions typically af-
coherent “social” phrase, such as, “yes, that’s fected in Broca’s aphasia are impaired. Likewise,
right.” Written output is typically similar to spoken transcortical sensory (TCS) aphasia is similar to
output—written words with little or no content, of- Wernicke’s aphasia, except that repetition is accu-
ten including nonword letter strings. Reading com- rate. It is caused by lesions involving areas sur-
prehension is typically no better than spoken rounding Wernicke’s area, in the watershed
comprehension. Repetition is generally similar to territories between the MCA and posterior cerebral
spontaneous speech—fluent jargon. These deficits artery (PCA) or the PCA territory.50 Finally, mixed
have been attributed to impaired inhibition of lexi- transcortical aphasia (MTA) is comparable to
cal activation, so that the person cannot select the global aphasia, except that sentence repetition is
appropriate word, sound, or meaning from compet- spared. These individuals appear to be echolalic and
ing linguistic units that are also activated.33 Al- have lesions surrounding Broca’s and Wernicke’s
though such an underlying impairment would area, but sparing language cortex itself.51 Because of
account for many of the observed language deficits, this localization, the syndrome is sometimes known
it could not easily account for cases with relatively as “isolation of the speech area,” as it appears to
preserved or relatively impaired categories of disconnect speech and language from broadly dis-
words, such as animals or tools,45 or impaired nouns tributed meanings of words.
relative to verbs.35 This collection of deficits is usu- Conduction aphasia is characterized by relatively
ally caused by neural dysfunction in regions sup- fluent, accurate spontaneous speech with phonemic
plied by the inferior division of the left MCA, paraphasias (well articulated responses that are
including Wernicke’s area (most of Brodmann area phonetically similar to the target word) sometimes
22, in the posterior, superior temporal gyrus). This produced in a series of increasingly closer approxi-
hypothesis can account for occasional dissociations mations of the target (e.g., “splant, plant, plants,
between the typical deficits or anomalous lesion pants” for pants—termed “conduit d’approche”)
sites in patients with Wernicke’s aphasia by the fact and disproportionately impaired repetition. Ge-
that there is individual variability in the cerebral schwind5 hypothesized that the repetition deficit
vasculature and the areas supplied by particular was caused by a lesion in the arcuate fasciculus, a
arteries. white matter tract that runs between Broca’s and
Global aphasia refers to the combination of defi- Wernicke’s areas. This hypothesis has been chal-
cits described in Broca’s and Wernicke’s aphasia. lenged on the basis that patients with conduction
Typically, the person has severely impaired compre- aphasia more often have lesions in the supramar-
hension of single words, sentences, and conversa- ginal gyrus or deep parietal white matter,52 and on
tion, and very limited, if any, spoken output. the basis that lesions of the arcuate fasciculus do not
Spontaneous speech, naming, and repetition may be reliably cause conduction aphasia.53 For these rea-
limited to a single perseverative word (e.g., no, no, sons, the author believes that lesions in the arcuate
no) or nonword utterance. Reading and writing are fasciculus have no causative role in the clinical syn-
also profoundly impaired. In most cases, both Bro- drome of conduction aphasia.
ca’s area and Wernicke’s area are damaged46 or Pure alexia (alexia with agraphia) refers to im-
functionally compromised.40 paired reading in the presence of spared writing and
The transcortical aphasias refer to language syn- relatively spared recognition of words spelled
dromes similar to those described above, but with aloud. This syndrome often results from a combina-
relatively normal sentence repetition. Transcortical tion of two lesions, both caused by occlusion or ste-
motor (TCM) aphasia has many of the characteris- nosis of the left posterior cerebral artery. A lesion in
tics of Broca’s aphasia, except that repetition is flu- the left occipital cortex results in right homony-
ent and grammatic. This vascular syndrome is mous hemianopsia, such that all visual information
caused by lesions just anterior or superior to (sur- is initially processed in the right occipital cortex. A
rounding) Broca’s area,47 often caused by occlusion second lesion in the splenium of the corpus callosum
of the anterior cerebral artery (ACA)48,49 or “water- prevents visual information in the right hemisphere
shed” areas between the ACA and the MCA. This from being transferred to the left hemisphere lan-
observation may account for the overlap in syn- guage cortex. Therefore, the person cannot read
dromes, consistent with Mohr’s proposal that dam- printed words but can recognize words spelled
age to Broca’s area alone causes a motor speech aloud (because the latter can initially be processed
deficit, while damage to the surrounding areas in the left hemisphere of the same patients54-58). In
causes the other symptoms of Broca’s aphasia. In one case the patient had a left occipital infarct and

Neurology 69 July 10, 2007 203

 Table 1 Vascular aphasic syndromes

MTA and
Broca Wernicke Conduction TCM TCS global Anomic Optic

Fluency* Poor Good Fair–good Poor Good Poor Good Good

Content Good Poor Good Good Poor Poor Good Good

Comprehension Intact words Poor Intact words Intact words Poor Poor Good Good
and simple and simple and simple
sentences sentences sentences

Repetition Poor, nonfluent Poor, fluent Poor Good Good MTA good; Good Good
jargon global poor

Naming Worse for verbs Worse for Fair–good Fair–good Poor Poor Worse for Poor with
nouns nouns visual stimuli

Spelling Poor Poor May be spared May be Poor Poor May be Good
spared spared

Reading Poor Poor May be spared May be Poor Poor May be Poor
spared spared

Associated Right arm Superior visual Poor working Abulia Right Right Right
signs weakness; field cut memory field cut hemiplegia hemianopia
apraxia of
speech

*Fluency includes grammaticality, prosody, melody, articulatory agility, and rate of speech, which can be differentially affected.
TCM ⫽ transcortical motor; TCS ⫽ transcortical sensory; MTA ⫽ mixed transcortical aphasia.

hypoperfusion of the splenium; his reading recov- of visual information to the left hemisphere lan-
ered when the splenium was reperfused.59 Consis- guage cortex, caused by disrupted white matter con-
tent with this hypothesis of a “disconnection” nections in the splenium. Similarly, patients with
between right occipital cortex and left hemisphere pure alexia can often read very slowly, letter-by-
language areas as an account of pure alexia, many letter, perhaps for the same reason. The naming er-
pure alexic patients also are unable to name visual ror pattern observed in optic aphasia was simulated
stimuli, although they can name the same items by Plaut and Shallice62 with a parallel distributed
from tactile exploration or in response to a verbal processing model of naming, by causing disrupted
description. This pattern of performance, known as interactions between visual input and semantics.
optic aphasia,60,61 can also result from the combina- Pure alexia and optic aphasia are not strictly apha-
tion of left occipital and splenium lesions, and can sic syndromes, since the representations and pro-
be seen as a disconnection between visual process- cesses that constitute language are intact, but there
ing in the right hemisphere and language output in is impaired access to them. However, they have
the left hemisphere.58,59
been discussed here, because they have relatively re-
Patients with optic aphasia can often produce
liable associated lesions and because they often are
gestures for visual objects that they cannot name,
mistaken for aphasia. However, there is inadequate
indicating that they can access at least some mean-
space to discuss other related vascular syndromes,
ing. However, tests of picture association demon-
such as alexia with agraphia, pure agraphia,63 and
strate that they cannot reliably distinguish between
pure word deafness.64
semantically related pictures.58 Their errors in nam-
In summary, the classic aphasia classifications
ing and word/picture matching likewise show access
to some meaning, but not sufficient meaning to se- are vascular syndromes consisting of frequently as-
lect between semantically related items. That is, sociated deficits that reflect damage or dysfunction
they often produce semantic paraphasias, names of regions of neural tissue (essential for particular
that are related in meaning to the target (e.g., mitten language functions) supplied by a particular artery.
named as “sock”). Interestingly, given sufficient For example, occlusion or critical stenosis of the su-
time, they often slowly hone in on the precise mean- perior division of the left MCA causes tissue dys-
ing. For example, in an attempt to name a shoe, a function in left posterior inferior frontal cortex and
patient said, “I tie them . . . wear them . . . not for insula, which apparently includes areas necessary
the hand but for the foot.” In contrast, the same for grammatic sentence production, planning and
patient was able to immediately name a shoe from programming of speech articulation, and other
tactile exploration. This pattern of errors seems to functions that are impaired in Broca’s aphasia. The
reflect slow, bit by bit access to semantic informa- same branch supplies the lateral motor strip, ac-
tion from vision, perhaps due to slow transmission counting for the frequent association between Bro-

204 Neurology 69 July 10, 2007

 predicting recovery67 and for selecting patients for
Figure 1 Imaging in a patient with Broca’s aphasia
particular language therapies.68

APHASIA CHARACTERIZED BY DISRUPTION OF

SPECIFIC COGNITIVE PROCESSES UNDERLY-
ING LANGUAGE TASKS Although the classic
aphasia syndromes have been useful clinically, they
have been less useful for developing theories of how
language is represented and processed. Most recent
theories of sentence comprehension, sentence pro-
duction, naming, reading, and so on, assume that
these tasks require a number of distinct cognitive
representations and processes that might rely on dif-
ferent brain regions. The proposal that each under-
lying component can be individually disrupted by
brain damage can account for the various patterns
of performance on language tasks that have been
described. Not only are language tasks decompos-
Diffusion-weighted image ca’s aphasia and right arm weakness. These able into discrete representations and processes, but
(left) and perfusion-weighted vascular syndromes are summarized in table 1. many of the cognitive representations might be
image (right) in a patient with
Broca’s aphasia associated
The vascular syndromes are clinically useful, in composites of features distributed across regions of
with a tiny diffusion that they allow strong predictions about what parts the brain. For example, the semantic representation
abnormality, but much larger of the brain are ischemic. Although recent structural of “horse” might include features of how it moves,
perfusion abnormality imaging, particularly diffusion-weighted imaging represented in areas of the brain critical for motion
involving Broca’s area.
(DWI), is highly sensitive to dense ischemia or in- recognition (middle temporal visual area—area
farct, it does not reveal all areas that are dysfunc- MT—and middle superior temporal area—area
tional due to hypoperfusion. Dynamic contrast MST),69,70 features specifying what it eats in another
(bolus-tracking) perfusion-weighted imaging (PWI) area, and features of how it is used by humans in
or CT perfusion scans can be used to reveal areas of another area of the brain.71 Activation of all of the
poor blood flow that that often better account for features simultaneously in different areas of cortex
the patient’s clinical deficits.43,44,65 However, perfu- might constitute a “semantic representation” (the
sion imaging is not always obtainable, for example meaning, for example, what makes a horse a horse).
when a patient has no vascular access for contrast Such distributed semantic representations might ac-
injection. The clinical syndrome and severity of def- count for patients with category-specific language
icits can predict the site and volume of perfusion deficits. For example, damage to area MT/MST,
abnormality, and can be used with the DWI to iden- disrupting activation of features of movement,
tify patients with a large diffusion– clinical mis- might affect access to the meaning of animals and
match who might benefit from aggressive treatment modes of transportation, but not buildings. In fact,
to restore blood flow.66 For example, the patient patients with selective deficits in naming and com-
whose scans are shown in figure 1 had a classic Bro- prehending the names of animals, as well as patients
ca’s aphasia, but just tiny areas of infarct on DWI a with selective preservation of animals (and to a
few hours after onset of symptoms. The clinical syn- lesser extent, transportation), have been de-
drome indicated there was likely to be a perfusion scribed.45,70 Many cases of selective impairments in-
abnormality that included posterior inferior frontal volving “living things” or “biologic kinds” relative
cortex. In this case, the perfusion abnormality was to artifacts, or the opposite, have been reported.72-76
confirmed with PWI. However, even if it had it not Several computational models of semantics have
been confirmed, this patient would be considered simulated category-specific semantic deficits by dis-
someone who needs urgent treatment to reperfuse rupting access to a particular type of feature.77 Oth-
the cortex and prevent infarct in the critical lan- ers have accounted for selective impairments
guage cortex. Each of the classic aphasia types has involving living things by assuming that living
been observed with hypoperfusion of the areas typi- things have strongly correlated features (e.g., the
cally associated with the syndrome in a study of pa- presence of eyes, ears, and limbs are highly corre-
tients with small subcortical stroke, and patients lated) or that their distinctive features are selectively
with similar subcortical strokes who did not have lost because they are weakly correlated to other fea-
hypoperfusion of the cortex did not have aphasia.40 tures.78 Others have proposed that category-specific
The aphasia syndromes have also been useful for deficits have an evolutionary basis, such that certain

Neurology 69 July 10, 2007 205

 2); access to the meaning of horse (the semantic rep-
Figure 2 Cognitive processes and representations underlying picture naming
resentation); access to the learned pronunciation of
horse (the phonologic representation); and motor
planning/programming of the muscles of the lips,
tongue, palate, vocal folds, and muscles of respira-
tion required to articulate the word. Many investi-
gators also include a modality-independent level of
lexical (word) access (e.g., a “lemma” level, which
specifies the syntactic role of the word, but not the
meaning or phonology84,85). A schematic representa-
tion of these cognitive processes underlying naming
is shown in figure 2. Although this depiction sug-
gests serial processing from one level to the next,
there is evidence that these sorts of cognitive repre-
sentations may be activated in parallel or in cascade,
and there are likely to be feedback as well as feed-
forward interactions, at least between some levels of
processing.86
Evidence for proposing each of these compo-
nents of the naming process comes from patterns of
impaired performance after brain damage or degen-
erative disease. Evidence that naming requires ac-
cess to a structural description for visual
A schematic representation
of the cognitive processes areas of the brain became devoted to processing recognition comes from patients with apperceptive
and representations
particular categories (e.g., animals, which were visual agnosia, who cannot distinguish drawings of
underlying picture naming real items from unreal items (e.g., a dog with ant-
both enemies and food79). Still others have proposed
that names of tools and action verbs are represented lers) or name visual stimuli, but can name the same
in areas critical for producing associated limb mo- items in response to a verbal description. Patients
tions (e.g., posterior frontal cortex70) or that manip- with optic aphasia, described earlier, can distin-
ulable items activate frontoparietal regions guish real from unreal objects but are impaired in
subserving hand motion representations.80 Al- activating a complete semantic representation from
though controversies remain, the category-specific the structural description. They can, however, ac-
language deficits have provided essential data for cess full semantics from tactile exploration or verbal
developing models of particular language functions. description. Optic aphasia can be accounted for by
This new approach to conceptualizing aphasia assuming partially impaired access to semantics
will be illustrated by considering one language task from an intact structural description.
that is most frequently impaired by left hemisphere In contrast, patients with semantic dementia can-
damage (and frequently tested at the bedside)— not fully access semantics, particularly of less famil-
picture naming. Impaired naming may be the resid- iar items, from any modality. They often use objects
ual deficit after partial recovery from almost any inappropriately (e.g., try to eat soup with a knife)
aphasia type, and thus may result from damage to a and cannot identify associations between pictures.
variety of brain regions. These observations can be Evidence for proposing a distinction between the
explained by assuming that picture naming requires complete semantic representation and a subset of
a number of different cognitive representations and “defining” or “identification” features is found in
processes that each depend on distinct, but perhaps the contrasting patterns of performance by patients
overlapping, brain regions. Damage to any one of with semantic dementia and patients who fail to un-
these areas might disrupt picture naming, but per- derstand words (e.g., point to a picture of a knife
haps in different ways. It would not be possible to when asked to point to spoon) but do not use ob-
review all of the many theories and simulations of jects inappropriately. The latter pattern, which is
cognitive processes underlying picture naming.62,81-85 common after stroke, indicates that access to some
However, it is widely agreed that naming a picture general knowledge (e.g., about how to use an ob-
of a horse would require, at the least, processes for ject) and personal knowledge (e.g., which brush is
abstracting from the visual stimulus the features my brush) is intact, but there is inadequate semantic
that allow recognition of the item as something fa- information to identify which related items are
miliar (access to the structural description in figure linked to a particular name. A person with this sort

206 Neurology 69 July 10, 2007

of lexical semantic deficit might call a horse a cow tongue, jaw, palate, respiratory muscles, or vocal
or a deer, and might point to a picture of a cow or folds), but nevertheless make numerous and vari-
deer when asked to point to a horse, but would not able errors in articulation of words, particularly
try to saddle a cow or deer. This distinction between longer words. Such a deficit, sometimes called
semantic errors after stroke vs the semantic errors in apraxia of speech, results in disruption of prosody
patients with semantic dementia has been elo- as well as distortions, transpositions, omissions, ad-
quently described by Jeffries and Lambon Ralph,87 ditions, and substitutions of phonemes. The patient
although their characterization of the cause of se- is quite aware of the errors and tries to correct them,
mantic errors after stroke is somewhat different. but often shows articulatory struggle and frustra-
Evidence for proposing a level of modality- tion. Although apraxia of speech is a motor speech
independent lexical access is that damage to such a disorder rather than a true language disorder, it fre-
level of processing would account for the common quently co-occurs with aphasia. In isolation, it is
deficit of “anomia,” in which the individual can re- rare and usually transient.
trieve neither the phonologic representation The pattern of errors across tasks, as well as er-
(learned pronunciation) nor the orthographic repre- ror types, provide indications regarding the compo-
sentation (learned spelling) of a particular item, de- nent of naming that is impaired in a particular
spite intact semantics. Such a deficit is experienced aphasic patient, as illustrated in table 2.
as the tip of the tongue phenomenon. Patients with It is important to note that while selective deficits
anomia have full access to the meaning and may re- in particular cognitive processes underlying a task
trieve semantically related names or phonologically such as naming have been very informative for de-
related names, but are aware that such names are veloping theories, such discrete deficits are rela-
not what they are searching for. They know the tively uncommon. The majority of aphasic
grammatic word class, and sometimes partial pho- individuals have impairments involving more than
nologic information (e.g., the first letter or the ap- one cognitive process.
proximate length of the word).
Separate modality-specific representations of the NEURAL CORRELATES OF THE COGNITIVE
name must also be postulated to explain perfor- PROCESSES UNDERLYING NAMING Functional
mance of patients who are able to retrieve the pho- imaging studies of naming show widespread activa-
nologic representation to say the name but not the tion of perisylvian language cortex, including Bro-
orthographic representation to spell the name, or ca’s area and Wernicke’s area, as well as more
the opposite pattern, despite the absence of motor inferior and anterior temporal and occipital cortex
output deficits.88-93 Sometimes these deficits are spe- during picture naming.11,14 Some studies have re-
cific to a particular grammatic category. For exam- ported more localized areas of activation associated
ple, one reported patient was selectively unable to with particular components of naming.95 Such re-
orally name or read aloud verbs but could write the sults are consistent with the hypothesis that com-
same verbs and could orally name or read aloud plex neural networks, as well as complex cognitive
(and write) nouns, while another patient had the op- systems, are recruited for apparently simple lan-
posite pattern with respect to output modality—she guage tasks such as picture naming.11,14,94-96 Various
was selectively unable to retrieve the written name language tasks, including naming and word com-
of verbs, but could retrieve the oral names of verbs prehension, recruit overlapping networks of brain
and oral and written names of nouns.89 Yet another regions.11,97 Such results suggest that language tasks
patient was selectively impaired in oral naming of such as naming or reading emerge from distinct dis-
nouns relative to verbs using the same stimuli.93 Al- tributions of activation across various brain re-
though such patients sometimes produce semanti- gions98 and are consistent with the hypothesis that
cally related words in the affected output modality, various language tasks depend on overlapping neu-
they do not have deficits in accessing meanings (se- ral as well as cognitive processes. Most studies also
mantics) or linking meanings to words (lexical- show different patterns of activation depending on
semantic deficits), as demonstrated by their the semantic category of stimuli70 consistent with
accurate naming in the other modality and by excel- the hypothesis that semantic representations are dis-
lent word comprehension. tributed across various regions in temporal and per-
Finally, evidence for proposing a distinct system haps parietal cortex. Difficulty of task and rate of
for motor planning and programming of speech ar- presentation also influence the pattern of activation
ticulation comes from patients who have no dysar- in both the left and right hemispheres,99 indicating
thria (i.e., no impairment in the rate, range, that some activation in functional imaging may also
strength, or timing of movements of the lips, reflect nonlinguistic aspects of the task, such as the

Neurology 69 July 10, 2007 207

 Table 2 Patterns of errors in naming associated with selective impairments of distinct cognitive processes

Spoken word
Cognitive process Use of objects comprehension Oral naming Written naming Repetition

Semantics Errors on low Semantic errors Semantic errors Semantic errors Normal
familiarity items (e.g., dog/cat) (e.g., dog¡cat) (e.g., dog¡cat)

Lexical–semantics† Normal* Semantic errors Semantic errors Semantic errors Normal

(e.g., dog/cat) (dog¡cat; dog¡bone) (e.g., dog¡cat; dog¡bone)

Modality–independent Normal Normal Semantic errors Semantic errors Normal

lexical access (dog¡cat; dog¡bone) (e.g., dog¡cat; dog¡bone)

Phonologic Normal Normal Semantic errors Normal Normal

representations (dog¡cat; dog¡bone)
or phonologically
similar words
or nonwords

Orthographic Normal Normal Normal Semantic errors Normal

representations (e.g., dog¡cat; dog¡bone)
or orthographically
similar words
or nonwords

Articulatory Normal Normal Variable off-target Normal Variable off-target

planning and articulations articulations
programming

*Processes that are normal when there is selective impairment of a component may be impaired when the patient has other
deficits.
†Lexical-semantics refers to mechanisms for linking a subset of semantic features that define the word to a particular lexical repre-
sentation (e.g., what makes a dog a dog).

difficulty of response selection100 or other executive sue dysfunction across seven left hemisphere Brod-
processes. Likewise, some activated regions during mann areas: BA22/Wernicke’s area, BA44 and 45/
picture naming may reflect automatic access to per- part of Broca’s area, BA 38/anterior temporal,
sonal memories or emotions connected with the pic- BA37/posterior middle–inferior temporal and fusi-
tured item. As has been frequently discussed, form gyri, BA 39/angular gyrus, and BA21/inferior
functional imaging studies show areas of the brain temporal.106a Other studies have highlighted the im-
that are engaged in a particular task, but cannot de- portance of a particular region for a given compo-
termine which of these areas are essential for the nent on naming. For instance, lesions in left BA 37
tasks.101 Nevertheless, functional imaging studies are associated with modality-independent lexical
have been important in showing that different re- access,106,107 and restoration of tissue function in BA
gions within Wernicke’s area102 or Broca’s area103 37 is associated with recovery of modality-
subserve different functions, and for showing that independent lexical access.108 Diffusion tensor imag-
the distinct patterns of activation across a number ing and tractography are promising methods of
of brain regions support a given language identifying white matter tracks that connect various
function.12,95,104 components of the neural network underlying nam-
Lesion studies show that overlapping networks ing and other language tasks.53,109 To complement
of brain regions are also essential for various lan- the imaging of structural connections, magnetoen-
guage tasks, such as naming or reading. Recent cephalography can reveal neural networks through
studies indicate that damage to the same regions can the functional connectivity between areas revealed
disrupt naming, written word comprehension, and by simultaneous activation.110
spoken word comprehension, but different areas Some levels of representation computed in nam-
have greater weight in predicting error rates on each ing are likely to be more distributed than others. For
of these tasks in regression models.105 Each underly- example, semantic representations might be widely
ing component process shared by these tasks might distributed in the temporal (particular anterior and
depend on a particular “cog” in the system (a partic- lateral inferior temporal cortex14), frontal, and pari-
ular brain region) or all components could all de- etal cortex, with features of some items such as ac-
pend on the same or overlapping brain regions, but tions represented closer to motor systems, and
with some areas being more critical to one compo- features of other items, such as colors, represented
nent or the other. A recent study using discriminant closer to visual areas. Lesions that affect just part of
function analysis indicated that disruption of differ- the distributed semantic representations might
ent components of the naming process (depicted in cause category-specific semantic impairments.
figure 1) could be distinguished by six discriminant Some areas, such as Wernicke’s area, might be criti-
functions that reflected distinct distributions of tis- cal for linking these widespread semantic represen-

208 Neurology 69 July 10, 2007

tations— or just the subset of defining features—to treatment must be individualized to the aphasic per-
specific words for word comprehension and nam- son’s deficits, needs, and goals.
ing.11,111 Other areas, such as Broca’s area or other Some current trends in rehabilitation research
parts of lateral prefrontal cortex, might be critical are to evaluate the effects of pharmacologic aug-
for selecting or activating certain types of semantic mentation of therapy, with stimulants,126 cholinest-
information to accomplish specific tasks.103 erase inhibitors, dopamine agonists, and other
medications that influence availability of particular
RECOVERY A wealth of literature on aphasia re- neurotransmitters.127,128 There is a paucity of ran-
covery demonstrates that most patients make at domized, placebo-controlled trials in this area, but
least some recovery and the majority make substan- several are under way. There is no evidence that
tial or complete recovery. Early recovery is likely to medications are useful in the absence of language
be due to restoration of blood flow and other mech- therapy, however. Transcranial magnetic stimula-
anisms of tissue recovery, while later stages of re- tion is also being investigated as a method of en-
covery are likely to depend on reorganization of hancing aphasia recovery, with some small case
structure/function relationships, as well as reorgani- series showing some benefit.129,130
zation of cognitive functions and compensatory
CONCLUSION The study of aphasia and its associ-
mechanisms.67,112 Many functional lesion studies of
ated lesions in the late 19th century led to many in-
language recovery have shown increased activation
sights about the neural organization of language
of right hemisphere homologues of language areas,
functions. Many of these insights have been con-
while others show increased activation of perile-
firmed and elaborated in more recent studies using
sional areas of the left hemisphere during recov- advanced imaging to localize areas of dysfunctional
ery.104,113,114 Several recent studies have indicated brain tissue associated with particular language def-
that the areas recruited for a particular language icits or using functional imaging to identify areas of
task change over the course of recovery, with mini- the brain that activated during a particular language
mal elicited activation (or hemodynamic response) task in normal controls14,103 or in recovering aphasic
during the language task in the acute stage of stroke, individuals.104,131 More detailed theories and compu-
predominantly right hemisphere activation in the tational models of particular language tasks have
subacute stage, and a return to predominant left contributed also to the understanding of how each
hemisphere activation in the more chronic stage in task might be carried out in the brain.
patients who show good recovery of the task.115
Received November 29, 2006. Accepted in final form February
TREATMENT This review cannot do justice to re- 7, 2007.
cent findings in the domain of aphasia treatment.
However, a few words are important. First, aphasia REFERENCES
therapy can be very useful in improving language 1. Dax M. Lesions de la moitie gauche de l’encephale coince-
dent avec trouble des signes de la pensie. Montpelier 1936.
and communication in a variety of settings.116,117
2. Broca P. Perte de la parole. Romollisement chronique et
Several recent investigations have shown that in- destruction partielle du lobe anterieur gauche du cerveau.
tense treatment (e.g., at least 4 days per week, at Bull Soc Anthropol 1861;2.
least 2 hours per day) for a short time is more effec- 3. Wernicke C. Lehrbruch der Gehirnkrankheiten. Berlin:
tive than a similar number of sessions spread out Theodore Fisher; 1881.
4. Broca P. Sur la faculte du langage articule. Paris Bull Soc
over a longer period.118 Some treatments focus on
Anthr 1865;6:337–393.
treating the underlying impairment, such as improv- 5. Geschwind N. Disconnexion syndromes in animals and
ing the component of the naming process that is dis- man. Brain 1965;88:237–294, 585–644.
rupted or teaching the patient to rely on other 6. Goodglass H, Kaplan E. The assessment of aphasia and
components to compensate for the damage related disorders. Philadelphia, PA: Lea & Febiger; 1972.
7. Naeser MA, Hayward RW. Lesion localization in aphasia
process.119-124 Most of the direct treatments rely on
with cranial computed tomography and Boston Diagnos-
the principle that the more often a patient produces tic Aphasia Examination. Neurology 1978;28:545–551.
a particular correct response (with as much facilita- 8. Alexander MP.. Aphasia: clinical and anatomical aspects.
tion as needed), the more often that person will be In: Feinberg TE, Farah MJ, eds. Behavioral neurology and
able to produce that same correct response indepen- neuropsychology. New York: McGraw Hill; 1997:133–
150.
dently in the future. Other interventions focus on
9. Willmes K, Poeck K. To what extent can aphasia syn-
communicative function— how to be successful in a dromes be localized? Brain 1993;116:1527–1540.
particular communicative task, rather than treating 10. Peterson SE, Fox PT, Posner MI, Mintun M, Raichle ME.
a particular impairment.125 In both approaches, the Positron emission tomography studies of the cortical

Neurology 69 July 10, 2007 209

 anatomy of single word processing. Nature 1988;331:585– 29. Grodzinsky Y. The neurology of syntax: language use
589. without Broca’s area. Behav Brain Sci 2000;23:1–71.
11. Fridriksson J, Morrow L. Cortical activation associated 30. Davis C, Kleinman JT, Newhart M, Heidler-Gary J, Hil-
with language task difficulty in aphasia. Aphasiology lis AE. Speech and language functions that depend on Bro-
2005;19:239–250. ca’s area. Brain Lang 2006;99:142–143.
12. Binder JR. Neuroanatomy of language processing studied 31. Caramazza A, Zurif E. Dissociation of algorithmic and
with functional MRI. Clin Neurosci 1997;4:87–94. heuristic processes in language comprehension. Brain
13. Crinion JT, Lambon-Ralph MA, Warburton EA, Lang 1976;3:572–582.
Howard D, Wise RJ. Temporal lobe regions engaged dur- 32. Miceli G, Mazzucchi A, Menn L, Goodglass H. Contrast-
ing normal speech comprehension. Brain 2003 126(Pt 5): ing cases of Italian agrammatic aphasia without compre-
1193–1201. hension disorder. Brain Lang 1983;33:273–295.
14. Wise RJS. Language systems in normal and aphasic hu- 33. Blumstein SE, Milberg WP. Language deficits in Broca’s
man subjects: functional imaging studies and inferences and Wernicke’s aphasia: a singular impairment. In:
from animal studies. Br Med Bull 2003;65:95–119. Grodzinsky Y, Shapiro L, Swinney D, eds. Language and
15. Kraut MA, Kremen S, Moo LR, Segal JB, Calhoun V, the brain. New York: Academic Press; 2000.
Hart JJ. Object activation in semantic memory from vi- 34. Miceli G, Silveri MC, Nocetini U, Caramazza A. Patterns
sual multimodal feature input. J Cogn Neurosci 2002;14: of dissociation in comprehension and production of
37–47. nouns and verbs. Aphasiology 1988;2:351–358.
16. McClelland J, Rumelhart DM. An interactive-activation 35. Zingeser LB, Berndt RS. Retrieval of nouns and verbs in
model of context effects in letter perception. Psychol Rev agrammatism and anomia. Brain Lang 1990;39:14–32.
1981;88:375–407. 36. Badecker W, Caramazza A. On considerations of method
17. Weems SA, Reggia JA. Stimulating single word process- and theory governing the use of clinical categories in neu-
ing in the classic aphasia syndromes based on the rolinguistics and cognitive neuropsychology: the case
Wernicke-Lichtheim-Geschwind theory. Brain Lang 2006; against agrammatism. Cognition 1985;20:97–125.
98:291–309. 37. Mohr J, Pessin M, Finkelstein S, Funkenstein H, Duncan
18. Mesulam MM. Slowly progressive aphasia without gener- G, Davis K. Broca aphasia: pathological and clinical.
alized dementia. Ann Neurol 1982;11:592–598. Neurology 1978;28:311–324.
19. Mummery CJ, Patterson K, Price CJ, Ashburner J, Frack- 38. Rademacher J, Caviness VS, Steinmetz H, Galaburda
owiak RS, Hodges JR. A voxel-based morphometry study AM. Topographical variation in the human primary cor-
of semantic dementia: relationship between temporal lobe tices: implications for neuroimaging, brain mapping, and
atrophy and semantic memory. Ann Neurol 2000;47:36– neurobiology. Cereb Cortex 1993;3:313–329.
45. 39. Amunts K, Schleicher A, Burgel U, Mohlberg H, Uylings
20. Gorno-Tempini ML, Dronkers NF, Rankin KP, et al. HBM, Zilles K. Broca’s region revisited: cytoarchitecture
Cognition and anatomy in three variants of primary pro- and intersubject variability. J Comp Neurol 1999;412:
gressive aphasia. Ann Neurol 2004;55:335–346. 319–341.
21. Cohen L, Lehericy S, Chochon F, Lemer C, Rivaud S, De- 40. Hillis AE, Barker PB, Wityk RJ, et al. Variability in sub-
haene S. Language-specific tuning of visual cortex? Func- cortical aphasia is due to variable sites of cortical hypo-
tional properties of the visual word form area. Brain 2002; perfusion. Brain Lang 2004;89:524–530.
125:1054–1069. 41. Hillis AE, Kane A, Tuffiash E, et al. Reperfusion of spe-
22. Price CJ, Winterburn D, Giraud AL, Moore CJ, Nop- cific brain regions by raising blood pressure restores selec-
peney U. Cortical localisation of visual and auditory word tive language functions in subacute stroke. Brain Lang
form areas: a reconsideration of the evidence. Brain Lang 2002;79:495–510.
2003;86:272–286. 42. Hillis AE, Wityk R, Barker PB, Caramazza A. Neural re-
23. Ojemann GA. Cortical stimulation and recording in lan- gions essential for writing verbs. Nat Neurosci 2003;6:19–
guage. London: Academic Press; 1994. 20.
24. Hart J, Lesser R, Gordon B. Selective interference with the 43. Hillis AE, Wityk RJ, Barker PB, et al. Subcortical aphasia
representation of size in the human by direct cortical stim- and neglect in acute stroke: the role of cortical hypoperfu-
ulation. J Cogn Neurosci 1992;4:337–344. sion. Brain 2002;125:1094–1104.
25. Usui K, Ikeda A, Takayama M, et al. Conversion of se- 44. Croquelois A, Wintermark M, Reichhart M, Meul R, Bo-
mantic information into phonological representation: a gousslavsky J. Aphasia in hyperacute stroke: language fol-
function in left posterior basal temporal area. Brain 2003; lows brain penumbra dynamics. Ann Neurol 2003;54:
126:632–641. 321–329.
26. Philipose LE, Newhart M, Kleinman J, Heidler-Gary J, 45. Hillis AE, Caramazza A. Category-specific naming and
Hillis A. Network of neural regions essential for reading comprehension impairment: A double dissociation. Brain
and spelling of words and pseudowords. Neurology 2006; 1991;114:2081–2094.
66:A53. Abstract. 46. Mazzocchi F, Vignolo LA. Localization of lesions in apha-
27. Pascual-Leone A, Walsh V, Rothwell J. Transcranial sia: clinical CT scan correlations in stroke patients. Cor-
magnetic stimulation in cognitive neuroscience: virtual le- tex 1979;15:627–653.
sion, chronometry, and functional connectivity. Curr 47. Freedman M, Alexander MP, Naeser MA. Anatomic ba-
Opin Neurobiol 2000;10:232–237. sis of transcortical motor aphasia. Neurology 1984;34:
28. Hillis-Trupe AE. Reliability of rating spontaneous speech 409–417.
in the Western Aphasia Battery: Implications for classifi- 48. Masdeu JC, Schoene WC, Funkenstein HH. Aphasia fol-
cation. In: Brookshire RH, ed. Clinical aphasiology. Min- lowing infarction of the left supplementary motor area.
neapolis: BRK Publishers; 1984:55–69. Neurology 1979;15:627–653.

210 Neurology 69 July 10, 2007

49. Rubens AB. Transcortical motor aphasia. In: Whitaker 68. Chapey R. Language intervention strategies in aphasia
H, Whitaker H, eds. Studies in neurolinguistics. New and related neurogenic communication disorders. 4th ed.
York: Academic; 1976:293–306. Baltimore, MD: Lippincott Williams & Wilkins; 2001.
50. Alexander MP, Hiltbrunner B, Fischer RS. Distributed 69. Kable JW, Lease-Spellmeyer J, Chatterjee A. Neural sub-
anatomy of transcortical sensory aphasia. Arch Neurol strates of action event knowledge. J Cogn Neurosci 2002;
1989;46:885–892. 14:795–805.
51. Rapcsak SZ, Krupp LB, Rubens AB, Reim J. Mixed 70. Damasio H, Tranel D, Grabowski TJ, Adolphs R,
transcortical aphasia without anatomic isolation of the Damasio A. Neural systems behind word and concept re-
speech area. Stroke 1990;21:953–956. trieval. Cognition 2004;92:179–229.
52. Palumbo CL, Alexander MP, Naeser MA. CT scan lesion 71. Martin A, Chao LL. Semantic memory and the brain:
sites associated with conduction aphasia. In: Kohn SE, ed. structure and processes. Curr Opin Neurobiol 2001;11:
Conduction aphasia. Hillsdale, NJ: Lawrence Erlbaum; 194–201.
1992:51–75. 72. McKenna P, Warrington EK. Category-specific naming
53. Selnes OA, van Zijl P, Barker PB, Hillis AE, Mori S. MR preservation: a single case study. J Neurol Neurosurg Psy-
diffusion tensor imaging documented arcuate fasciculus chiatry 1978;41:571–574.
lesion in a patient with normal repetition performance. 73. Warrington EK, McCarthy RA. Categories of knowledge:
Aphasiology 2002;16:897–902. further fractionations and an attempted explanation.
54. Dejerine J. Sur un cas de cécité verbale avec agraphie, Brain 1987;110:1273–1296.
suivi d’autopsie. Comptes Rendus Hebdomadaires des Sé- 74. Warrington EK, Shallice T. Category specific semantic
ances et Mémoires de la Société de Biologie 1891; Ninth impairments. Brain 1984;107:829–853.
series 3:197–201. 75. Sachett C, Humphreys GW. Calling a squirrel a squirrel
55. Damasio A, Damasio H. Hemianopia, hemiachromatop- but a canoe a wigwam: a category-specific deficit for arte-
sia and the mechanisms of alexia. Cortex 1986;22:161– factual objects and body parts. Cogn Neuropsychol 1992;
169. 9:73–86.
76. Hart J, Berndt RS, Caramazza A. Category-specific nam-
56. Chialant D, Caramazza A. Perceptual and lexical factors
ing deficit following cerebral infarction. Nature 1995;316:
in a case of letter-by-letter reading. Cogn Neuropsychol
338.
1998;15:167–201.
77. Farah M, McClelland J. A computational model of se-
57. Cohen L, Martinaud O, Lemer C, et al. Visual word rec-
mantic memory impairment: modality specificity and
ognition in the left and right hemispheres: anatomical and
emergent category specificity. J Exp Psychol Gen 1991;
functional correlates of peripheral alexias. Cereb Cortex
120:339–357.
2003;13:1313–1333.
78. Randall B, Moss HE, Rodd JM, Greer M, Tyler LK. Dis-
58. Hillis AE, Caramazza A. Cognitive and neural mecha-
tinctiveness and correlation in conceptual structure: be-
nisms underlying visual and semantic processing. J Cogn
havioral and computational studies. J Exp Psychol Learn
Neurosci 1995;:457–478.
Mem Cogn 2004;30:393–406.
59. Marsh EB, Hillis AE. Cognitive and neural mechanisms
79. Caramazza A, Shelton J. Domain specific knowledge sys-
underlying reading and naming: evidence from letter-by-
tems in the brain: the animate-inanimate distinction. J
letter reading and optic aphasia. Neurocase 2005;11:325–
Cogn Neuropsychol 1998;10:1–34.
328.
80. Saccuman MC, Cappa SF, Bates EA, et al. The impact of
60. Freund CS. Uber optische aphasia and seelenblindheit.
semantic reference on word class: an fMRI study of action
Arch Psychiatr Nervenkrank 1889;20:276–297.
and object naming. Neuroimage 2006;32:1865–1878.
61. Lhermitte E, Beauvois MF. A visual-speech disconnexion
81. Humphreys GW, Riddoch MJ, Quinlan PT. Cascade pro-
syndrome: report of a case with optic aphasia, agnosic cesses in picture identification. Cogn Neuropsychol 1988;
alexia and colour agnosia. Brain 1973;96:695–714. 5:67–104.
62. Plaut D, Shallice T. Perseverative and semantic influences 82. Hillis AE, Caramazza A. The compositionality of lexical-
on visual object naming errors in optic aphasia: a connec- semantic representations: clues from semantic errors in
tionist account. J Cogn Neurosci 1993;5:89–117. object naming. Memory 1995;3:333–358.
63. Hillis AE. Alexia and agraphia. In: Godefroy O, Bogous- 83. Butterworth B. Lexical access in speech production. In:
slavsky J, eds. The behavioural and cognitive neurology Marslen-Wilson W, ed. Lexical representation and pro-
of stroke. Cambridge: Cambridge University Press (in cess. Cambridge, MA: MIT Press; 1989.
press). 84. Dell G, O’Sheaghdha P. Stages of lexical access in lan-
64. Buchman AS. Pure word deafness:100 years later. J Neu- guage production. Cognition 1992;42:287–314.
rol Neurosurg Psychiatry 1986;49:489–499. 85. Levelt WJM, Schriefers H, Vorberg D, Meyer AS, Pech-
65. Beaulieu C, de Crespingy A, Tong DC, Mosely ME, Al- mann T, Havinga J. The time course of lexical access in
bers GW, Marks MP. Longitudinal magnetic resonance speech production: A study of picture naming. Psychol
imaging study of perfusion and diffusion in stroke: evolu- Rev 1991;98:122–142.
tion of volume and correlation with clinical outcome. Ann 86. Rapp BC, Goldrick M. Interactivity and discreteness in
Neurol 1999;46:568–578. spoken word production. Psychol Rev 2000;107:460–499.
66. Reineck LA, Hillis AE. “Diffusion-Clinical Mismatch” 87. Jefferies E, Lambon Ralph MA. Semantic impairment in
predicts potential for early recovery of aphasia. Stroke stroke aphasia versus semantic dementia: a case-series
2004;35:287. comparison. Brain 2006;129:2132–2147.
67. Kertesz A. Recovery of aphasia. In: Feinberg TE, Farah 88. Ellis AW, Miller D, Sin G. Wernicke’s aphasia and normal
MJ, eds. Behavioral neurology and neuropsychology. language processing: a case study in cognitive neuropsy-
New York: McGraw Hill; 1997:167–182. chology. Cognition 1983;15:111–114.

Neurology 69 July 10, 2007 211

 89. Caramazza A, Hillis AE. Lexical organization of nouns 111. Hillis AE, Wityk RJ, Tuffiash E, et al. Hypoperfusion of
and verbs in the brain. Nature 1991;349:788–790. Wernicke’s area predicts severity of semantic deficit in
90. Caramazza A, Hillis AE. Where do semantic errors come acute stroke. Ann Neurol 2001;50:561–566.
from? Cortex 1990;1:95–122. 112. Hillis AE, Heidler J. Mechanisms of early aphasia recov-
91. Rapp BC, Benzing L, Caramazza A. The autonomy of lex- ery: evidence from MR perfusion imaging. Aphasiology
ical orthographic representations. Cogn Neuropsychol 2002;16:885–896.
1997;14:71–104. 113. Warburton E, Swinburn K, Price CJ. Mechanisms of re-
92. Hillis AE, Rapp BC, Caramazza A. When a rose is a rose covery from aphasia: evidence from positron emission to-
in speaking but a tulip in writing. Cortex 1999;35:337– mographic studies. J Neurol Neurosurg Psychiatry 1999;
356. 66:155–161.
93. Hillis AE, Caramazza A. The representation of grammat- 114. Marsh EB, Hillis AE. Recovery from aphasia following
ical categories of words in the brain. J Cogn Neurosci brain injury: the role of reorganization. Prog Brain Res
1995;7:396–407. 2006;157:143–156.
94. Mesulam M-M. From sensation to cognition. Brain 1998; 115. Saur D, Lange R, Baumgaertner A, et al. Dynamics of
121:1013–1052. language reorganization after stroke. Brain
95. Howard D, Patterson K, Wise R, et al. The cortical local- 2006;129:1371–1384.
ization of the lexicons. Brain 1992;115:1769–1782. 116. Robey RR. The efficacy of treatment for aphasia persons:
96. Fridriksson J, Morrow KL, Moser D, Fridriksson A, Bay- a meta-analysis. Brain Lang 1994;47:585–608.
lis GC. Neural recruitment associated with anomia treat- 117. Robey RR. A meta-analysis of clinical outcomes in the
ment. NeuroImage 2006;32:1403–1412. treatment of aphasia. J Speech Lang Hear Res 1998;41:
97. Wise R, Chollet F, Hadar U, Friston K, Hoffner E, Frack- 172–187.
owiak R. Distribution of cortical neural networks in- 118. Bhogal SK, Teasell R, Speechley M. Intensity of aphasia
volved in word comprehension and word retrieval. Brain therapy, impact on recovery. Stroke 2003;34:987–993.
1991;114:1803–1817. 119. Humphreys G, Riddoch MJ. Cognitive neuropsychology
98. Price CJ, Thierry G, Griffiths T. Speech-specific auditory and cognitive rehabilitation. Hillsdale, NJ: Lawrence Erl-
processing: where is it? Trends Cogn Sci 2005;9:271–276. baum Associates; 1994.
99. Grabowski TJ, Damasio AR. Investigating language with 120. Raymer AM, Rothi LG. Cognitive approaches to im-
functional imaging. In: Toga AW, Mazziotta JC, eds. pairments of word comprehension and production. In:
Brain mapping: the systems. San Diego: Academic Press; Chapey R, ed. Language intervention strategies in
2000:425–461. aphasia and related neurogenic communication disor-
100. Thompson-Schill SL, Gabrieli JDE. Priming of visual and ders. 4th ed. Baltimore, MD: Lippincott Williams &
functional knowledge on a semantic classification task. J Wilkins; 2001.
Exp Psychol Learn Mem Cogn 1999;25:41–53. 121. Raymer AM, Thompson CK, Jacobs B, Le Grand HR.
101. Chatterjee A. A madness to the methods in cognitive neu- Phonological treatment of naming deficits in aphasia:
roscience? J Cogn Neurosci 2005;17:847–849. model-based generalization analysis. Aphasiology 1993;7:
102. Wise RJS, Scott SK, Blank SC, Mummery CJ, Murphy K, 27–53.
Warburton EA. Separate neural subsystems within “Wer- 122. Mitchum CC, Berndt RS. Cognitive neuropsychological
nicke’s area”. Brain 2001;124:83–95. approaches to diagnosing and treating language disor-
103. Bookheimer S. Functional MRI of language: new ap- ders: production and comprehension of sentences. In:
proaches to understanding the cortical organization of se- Chapey R, ed. Language intervention strategies in aphasia
mantic processing. Annu Rev Neurosci 2002;25:151–188. and related neurogenic communication disorders. Balti-
104. Price CJ, Crinion J. The latest on functional imaging stud- more, MD: Lippincott Williams & Wilkins; 2001.
ies of aphasic stroke. Curr Opin Neurol 2005;18:429–434. 123. Beeson P, Hillis AE. Comprehension and production of
105. Newhart M, Ken L, Kleinman JT, Heidler-Gary J, Hillis written words. In: Chapey, ed. Language intervention
AE. Neural networks essential for naming and word com- strategies in aphasia and related neurogenic communica-
prehension. Cogn Behav Neurol (in press). tion disorders. 4th ed. Baltimore, MD: Lippincott Wil-
106. Raymer A, Foundas AL, Maher LM, et al. Cognitive neu- liams & Wilkins; 2001:572–604.
ropsychological analysis and neuroanatomical correlates 124. Wambaugh JL, Linebaugh CW, Doyle PJ, Martinez
in a case of acute anomia. Brain Lang 1997;58:137–156. AL, Kalinyak-Fliszar M, Spencer KA. Effects of two
106a.DeLeon J, Gottesman RF, Kleinman JT, et al. Neural re- cueing treatments on lexical retrieval in aphasic speak-
gions essential for distinct cognitive processes underlying ers with different levels of deficit. Aphasiology 2001;15:
picture naming. Brain 2007;130:1408 –1422. 933–950.
107. Hillis AE, Newhart M, Heidler J, et al. The roles of the 125. Holland AL. Assessment and treatment of pragmatic as-
“visual word form area” in reading. Neuroimage 2005;24: pects of communication in aphasia. In: Hillis AE, ed. The
548–559. handbook of language disorders. New York, NY: Psy-
108. Hillis AE, Kleinman KT, Newhart M, et al. Restoring ce- chology Press; 2002.
rebral blood flow reveals neural regions critical for nam- 126. Walker-Batson D, Curtis S, Natarajan R, et al. A double-
ing. J Neurosci 2006;26:8069–8073. blind, placebo-controlled study of the use of amphet-
109. Catani M, Jones DK, ffytche DH. Perisylvian language amine in the treatment of aphasia. Stroke 2001;32:2093–
networks of the human brain. Ann Neurol 2005;57:8–16. 2098.
110. Salmelin R, Kujala J. Neural representation of language: 127. Klein RB, Albert ML. Can drug therapies improve lan-
activation versus long-range connectivity. Trends Cogn guage functions of individuals with aphasia? A review of
Sci 2006;10:519–525. the evidence. Semin Speech Lang 2004;25:193–204.

212 Neurology 69 July 10, 2007

128. Small SL. Biological approaches to the treatment of aphasia. 130. Naeser MA, Martin PI, Nicholas M, et al. Improved pic-
In: Hillis AE, ed. The handbook of adult language disorders: ture naming in chronic aphasia after TMS to part of right
integrating cognitive neuropsychology, neurology and reha- Broca’s area: an open-protocol study. Brain Lang 2005;93:
bilitation. New York: Psychology Press; 2002:397–412. 95–105.
129. Mottaghy FM, Sparing R, Topper R. Enhancing picture 131. Cappa SF, Perani D, Grassi F, et al. A PET follow-up
naming with transcranial magnetic stimulation. Behav study of recovery after stroke in acute aphasics. Brain
Neurol 2006;17:177–189. Lang 1997;56:55–67.

Neurology 69 July 10, 2007 213

 Aphasia: Progress in the last quarter of a century
Argye E. Hillis
Neurology 2007;69;200-213
DOI 10.1212/01.wnl.0000265600.69385.6f

This information is current as of July 9, 2007

Updated Information & including high resolution figures, can be found at:
Services http://n.neurology.org/content/69/2/200.full

References This article cites 107 articles, 10 of which you can access for free at:
http://n.neurology.org/content/69/2/200.full#ref-list-1
Citations This article has been cited by 6 HighWire-hosted articles:
http://n.neurology.org/content/69/2/200.full##otherarticles
Subspecialty Collections This article, along with others on similar topics, appears in the
following collection(s):
All Cerebrovascular disease/Stroke
http://n.neurology.org/cgi/collection/all_cerebrovascular_disease_strok
e
All Cognitive Disorders/Dementia
http://n.neurology.org/cgi/collection/all_cognitive_disorders_dementia
All Rehabilitation
http://n.neurology.org/cgi/collection/all_rehabilitation
Permissions & Licensing Information about reproducing this article in parts (figures,tables) or in
its entirety can be found online at:
http://www.neurology.org/about/about_the_journal#permissions
Reprints Information about ordering reprints can be found online:
http://n.neurology.org/subscribers/advertise

Neurology ® is the official journal of the American Academy of Neurology. Published continuously since
1951, it is now a weekly with 48 issues per year. Copyright . All rights reserved. Print ISSN: 0028-3878.
Online ISSN: 1526-632X.