doi:10.1093/brain/awm011
Brain (2007), 130, 1408 ^1422
Neural regions essential for distinct cognitive
processes underlying picture naming
Jessica DeLeon,3 Rebecca F. Gottesman,1 Jonathan T. Kleinman,1 Melissa Newhart,1 Cameron Davis,1
Jennifer Heidler-Gary,1 Andrew Lee1 and Argye E. Hillis1,2,3
1
Departments of Neurology, 2Physical Medicine and Rehabilitation, Johns Hopkins University School of Medicine and
Department of Cognitive Science, Johns Hopkins University, Baltimore, MD, USA
3
Correspondence to: Argye E. Hillis, MD, Department of Neurology, Phipps 126, Johns Hopkins Hospital, 600 North Wolfe
Street, Baltimore, MD 21287, USA.
E-mail: [email protected]
We hypothesized that distinct cognitive processes underlying oral and written picture naming depend on intact
function of different, but overlapping, regions of the left hemisphere cortex, such that the distribution of tissue dysfunction in various areas can predict the component of the naming process that is disrupted.To test this hypothesis,
we evaluated 116 individuals within 24 h of acute ischaemic stroke using a battery of oral and written naming and
other lexical tests, and with magnetic resonance diffusion and perfusion imaging to identify the areas of tissue dysfunction. Discriminant function analysis, using the degree of hypoperfusion in various Brodmann’s areasçBA 22
(including Wernicke’s area), BA 44 (part of Broca’s area), BA 45 (part of Broca’s area), BA 21 (inferior temporal
cortex), BA 37 (posterior, inferior temporal/fusiform gyrus), BA 38 (anterior temporal cortex) and BA 39 (angular
gyrus)ças discriminant variables, classified patients on the basis of the primary component of the naming process
that was impaired (defined as visual, semantics, modality-independent lexical access, phonological word form,
orthographic word form and motor speech by the pattern of performance and types of errors across lexical
tasks). Additionally, linear regression analysis demonstrated that the areas contributing the most information to
the identification ofpatientswith particularlevels of impairmentin the naming processwerelargelyconsistent with
evidence for the roles of these regions from functionalimaging.This study provides evidence that the levelof impairment in the naming process reflects the distribution of tissue dysfunction in particular regions of the left anterior,
inferior and posterior middle/superior temporal cortex, posterior inferior frontal and inferior parietal cortex.
While occipital cortex is also critical for picture naming, it is likely that bilateral occipital damage is necessary to
disrupt visual recognition.These findings provide new evidence that a network of brain regions supports naming,
but separate components of this network are differentially required for distinct cognitive processes or representations underlying the complex task of naming pictures.
Keywords: aphasia; magnetic resonance perfusion weighted imaging; naming; anomia; language
Abbreviations: BA ¼ Brodmann’s area; DWI ¼ diffusion-weighted imaging; PWI ¼ perfusion-weighted imaging;
TTP ¼ time to peak
Received June 08, 2006. Revised October 23, 2006. Accepted January 12, 2007. Advance Access publication March 2, 2007
Introduction
Impaired ability to name objects or actions is nearly
ubiquitous among individuals with aphasia. However,
aphasic individuals may have trouble naming pictures for
different reasons. Naming a pictured object is a complex
process. It involves a number of relatively distinct (but
interacting) mental representations and cognitive processes.
To illustrate, naming a picture of a particular horse
as ‘horse’ demands at least the following processes:
(1) recognition of the visual stimulus as an instance of
a familiar concept; (2) access to the meaning of horse
(or what makes a horse a horse); (3) access to the
phonological word form (the learned pronunciation of the
word) and (4) motor programming and planning of
articulation (as well as implementation of the movement
sequences) to say the word. Although these functions may
not be completely segregated anatomically, they can be
individually impaired by brain damage. Therefore, they are
depicted as distinct levels of processing in a schematic
ß 2007 The Author(s)
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial use License (http://creativecommons.org/lisences/by-nc/2.0/uk/)
which permits unrestricted non-commercial use, distributed, and reproduction in medium, provided the original work is properly cited.
Neural regions essential for naming
representation of the representations and processes underlying naming shown in Fig. 1. This ‘model’ can be criticized
on a variety of grounds, as discussed below, but serves as
a basis for distinguishing the various impairments in
picture naming that are observed consequent to brain
lesions.
The meaning, or ‘semantic’ component is complex, and
probably includes (at least) an amodal representation of
everything we know about an item, including general
knowledge and personal knowledge about the item, as well
as those features that make it what it is and distinguishes it
from related items (what we will refer to, for simplicity, as
‘the defining features’ or ‘lexical-semantics’). To link the
complex semantic representation to a word form representation it is necessary to select the defining features from the
entire semantic representation. That is, to name a picture of
a horse, we need to select information about what makes a
horse a horse and not a deer (e.g. mane, can be
domesticated). We do not need to access information
about how they are used in a particular region of the world,
or about horses we have known. This linking mechanism
might take the form of executive processes for selecting a
lexical representation from semantics (Jeffries and LambonRalph, 2006) or identification mechanisms (Miller and
Johnson-Liard, 1976; Hillis et al., 1990). The defining
features (‘lexical-semantics’) selected by an executive
function or identification mechanism can be thought of
as a subset of the distributed semantic representation. That
is, what we mean by ‘lexical-semantics’ is restricted to the
type of semantic information needed to name pictures or to
determine whether or not a picture depicts an instance of
the referent of a particular word.
The distinction between the entire semantic representation and lexical-semantics is best captured by distinct
patterns of performance when each is disrupted. Disruption
or degradation of the semantic representation, including
amodal general and personal knowledge, is seen in patients
with semantic dementia, who late in the disease use all but
the most familiar items inappropriately (e.g. spread butter
with a can opener). This behaviour is rarely observed in
patients with aphasia caused by unilateral stroke, probably
because disruption of the amodal general and personal
knowledge requires left (and perhaps to a lesser degree,
right) anterior inferior and lateral temporal damage as
shown in patients with semantic dementia (Mummery
et al., 2000; Gorno-Tempini et al., 2004). Stroke rarely
causes damage to anterior, inferior temporal cortex (Huther
et al., 1998; Caviness et al., 2002; Wise, 2003) or bilateral
temporal cortex. Disruption of mechanisms for linking this
semantic representation (including amodal general and
personal knowledge) to a particular word form representation (for both word comprehension and naming), via
lexical-semantics, is frequently observed after stroke,
in patients with transcortical sensory aphasia, Wernicke’s
aphasia or global aphasia. These patients typically have
damage to superior temporal gyrus, inferior parietal cortex
Brain (2007), 130, 1408 ^1422
1409
or prefrontal cortex (Goodglass and Wingfield, 1997; Hart
and Gordon, 1990; Hillis et al., 2001). However; damage to
these areas might result in reduced activation of more
rostral temporal cortex (Wise, 2003).
Both patients with damage to lexical-semantics after
stroke and patients with semantic dementia make semantic
errors in naming and comprehension, but show distinctive
patterns across tests and stimuli, as elegantly demonstrated
by Jeffries and Lambin-Ralph (2006). In their study,
patients with semantic dementia were sensitive to familiarity of the item, showed item consistency across tests and
made mostly ‘coordinate’ semantic errors (e.g. deer for
horse). Patients with post-stroke aphasia were not sensitive
to familiarity, did not show item consistency, made
associative semantic errors (e.g. hay for horse) as well as
coordinate semantic errors and often named items correctly
in response to phonemic cues. The authors interpreted the
latter pattern as evidence that the amodal semantic core
is intact, and accounted for the pattern of errors as
a reflection of impaired executive processes. However,
the pattern of performance might also be explained by
underspecified lexical-semantic information that is consistent with a number of different word-form representations
(e.g. 5large, hooved, herbivore4 would be consistent with
‘horse’, ‘cow’, ‘donkey’, ‘deer’ etc. and might lead to
inconsistent responses to the same item across tests and
trials of the same test (see Hillis et al., 1990 or Hillis and
Caramazza, 1995a for further details of this account).
The phonemic cue would provide additional information
that would allow the patient to select the correct word.
The fact that such patients can be ‘mis-cued’ to produce the
wrong word with an incorrect phonemic cue (Howard
and Orchard-Lisle, 1984; Hillis and Caramazza, 1995b;
Howard and Gatehouse, 2006) is consistent with the view
that a correct response to a phonemic cue does not imply
that the meaning of the item is intact. Furthermore,
since features of the semantic representation that are not
part of lexical-semantics would be spared in stroke patients
with impaired lexical-semantics, they might access some of
these spared features in naming and produce associative
semantic errors, as well as coordinate semantic errors, as
reported by Jeffries and Lambon-Ralph (see also Hillis
et al., 1990). Additionally, patients with impaired amodal
general knowledge (SD) might try to saddle a deer, while
a patient with impaired lexical-semantics would not.
Thus, impairments of lexical-semantics versus impairments
of the more inclusive semantic representation would
explain the distinct patterns of performance in naming
and behaviour after stroke versus semantic dementia.
Many theories of the cognitive processes underlying
naming also specify an intermediate level of representation
between the semantic representation and the phonological
word form—a modality-independent lemma. Although the
original use of the term ‘lemma’ referred to a representation with semantic content (Kempen and Huijbers, 1983),
more recent papers have assumed that the lemma has
1410
Brain (2007), 130, 1408 ^1422
J. DeLeon et al.
Fig. 1 Schematic representation of the cognitive processes underlying picture naming. The representations that receive the most
activation, and are therefore selected for output, are shown in bold. Solid lines indicate processes involved in oral naming; dashed
lines indicate processes involved in written naming. More peripheral mechanisms for output (e.g. output buffers, mechanisms for motor
programming of output) are not shown.
no semantic content (Dell, 1986; Levelt, 1989, 1999;
Levelt et al., 1991; Dell et al., 1997). Instead, the lemma
mediates access to the phonological word form or
orthographic word form (the learned spelling of the
word) from semantics, and specifies the syntactic role of
the word (e.g. grammatical word class and gender; Garrett,
1988; Roelofs, 1992; Badecker et al., 1995). Whether or not
this level of representation is essential has been hotly
debated (Roelofs, 1992; Caramazza and Miozzo, 1997, 1998;
Roelofs et al., 1998). There is evidence that syntactic
information is accessed independently from phonological
information in naming (Caramazza and Miozzo, 1997;
Vigliocco et al., 1999), but one does not require completed
access to the other (inconsistent with the two-stage model
of Levelt (1989); but consistent with models that assume
cascaded or parallel activation of different representations,
such as Dell, 1986). As the role of a syntactic level of
representation remains controversial, we will not further
explore the neural correlates of this level of processing.
Nevertheless, we will assume that there is a level or site of
processing that allows access to both the phonological and
orthographic word forms, but is independent of semantics
(both amodal general and personal knowledge and lexicalsemantics). This assumption is based on (1) the tip-of-thetongue state, in which neither the orthographic nor the
phonological word form is accessible and (2) the observation that a common deficit after focal brain damage is
the inability to retrieve either the phonological or the
orthographic word form, in the face of intact knowledge
of the meaning of the object and word (pure anomia).
A much less common, but reported, pattern is the selective
impairment in accessing either the phonological or
orthographic word form, but not both, despite intact
motor output in both modalities (Ellis et al., 1983;
Caramazza and Hillis, 1990; Rapp et al., 1997; Hillis
et al., 1999). The more common pattern of anomia,
reflected in the association between impaired oral and
written naming with intact word comprehension, could be
explained by adjacent areas of the brain that are responsible
for access to phonological and orthographic word forms, or
could be explained by a common element needed for both.
The latter can more easily account for a high concordance
between oral and written naming with respect to the items
affected. This level of processing, which we will call
‘modality independent lexical access’, might be best
represented as the ‘hidden links’ in computational models
of lexical access (e.g. Plaut and Shallice, 1993) or the ‘Llevel’ (which avoids claims about whether this level is best
represented as a ‘lexeme’ or a ‘lemma’ level) in computational models of Rapp and Goldrick (2000). The theoretical
distinction between ‘modality-independent lexical access’
and ‘lexical-semantics’ is that modality-independent lexical
access is required specifically for word production (not
word comprehension), while ‘lexical-semantics’ is necessary
for linking words to their meanings in both production and
comprehension. Both components are assumed to be
processing mechanisms or access procedures that are
important for linking more distributed representations, as
schematically represented in Fig. 1. In this figure, modalityindependent lexical access is shown as a single node for a
single item, but this level might be a distributed
representation as well, as in parallel distributed processing
(PDP) models such as those in Seidenberg and McClelland
(1989), Plaut and Shallice (1993) and others. In addition,
Neural regions essential for naming
arrows indicate that there might be not only feedforward
activation between levels, but also feedback interaction,
although the degree of feedback interaction is a matter of
controversy (Rapp and Goldrick, 2000).
These various representations and processes underlying
naming are revealed by patients with distinct patterns of
performance across tasks. Patients who have trouble at
the visual recognition level of the naming process may have
trouble accessing a ‘structural description’ of horse, so that
they cannot distinguish a drawing of a horse from an
unreal but visually similar drawing (e.g. a drawing of horse
with a dog’s head) or may not be able to match unusual
views of horses to one another. This pattern of performance
is sometimes known as ‘aperceptive visual agnosia’. Other
patients may access a structural description (so they can
accomplish the above tasks), but may not be able to
name the horse due to impaired access to lexical-semantics
(or to the entire semantic representation) of horse from
the structural description. Such a deficit is often known
as ‘optic aphasia’ (Lhermitte and Beauvois, 1973; Riddoch
and Humphreys, 1987; Manning and Campbell, 1992,
1995c; Marsh and Hillis, 2005) (or ‘associative agnosia’
DeRenzi and Saetti, 1997; Chanoine et al., 1998). A patient
with this deficit would be able to identify which horse
is different from three identical drawings plus one distinct
drawing of horses, but might not be able to sort
photographs or drawings of horses versus deer (Hillis and
Caramazza, 1995c). Nevertheless, as the semantic representation is intact (but inaccessible from vision) in these
patients, they would be able to verbalize the difference
between horses and deer and correctly name ‘horse’ in
response to a verbal description (Hillis and Caramazza,
1995c). In contrast, patients with impairment of the
semantic representation (including amodal knowledge and
lexical-semantics, such as patients with SD) might perform
poorly on any semantic tasks, including naming (of pictures
or to definitions), sorting pictures of horses versus deer
or pointing to pictures associated with horses.
Still other patients would fail to correctly name a horse
from a picture or from a description of a horse, because
of a deficit at the level of lexical-semantics. Such a patient
may have an unstable or underspecified lexical-semantic
representation of horse. For example, patients with
impaired lexical-semantics might access sufficient information to identify a horse as an animal, but may not access
information about what makes a horse different from
a deer. For example, as shown in Fig. 1, they might access
many components of lexical-semantics, but if they failed
to activate those that distinguish a horse from a deer
(e.g. mane) or if they inadvertently activated spurious
components (e.g. non-domesticated), such patients would
make semantic paraphasias (e.g. calling a horse a ‘deer’)
in oral and written naming and in word/picture matching.
However, they would have no difficulty saying whether or
not some horses can be ridden, pointing to pictures
associated with horses, or other tasks that depend on
Brain (2007), 130, 1408 ^1422
1411
amodal general or personal knowledge or the intact
components of lexical-semantics.
Other patients accurately access complete lexicalsemantics, and so they perform well in word/picture
matching and other comprehension tasks, but cannot
retrieve the phonological or orthographic word form.
These patients may have disruption of the naming process
at the level of modality-independent lexical access, or their
problem may occur in directly accessing the phonological
and orthographic word form representations from semantics. In this paper, we will refer to the pattern of equivalent
impairments in oral naming and written naming,
with spared comprehension, as disruption at the level of
the modality-independent lexical access, recognizing that
the pattern can be alternatively explained by two separate
‘downstream’ impairments. Patients with impaired modality-independent lexical access make semantic paraphasias
in both speech and writing, but are aware that the
paraphasias are errors. Other errors include circumlocutions (e.g. horse ! ‘big animal, races, cowboys ride them’),
associative semantic errors (horse ! ‘hay’), or omissions.
Similar errors are made by patients whose naming problem
arises only in accessing the phonological word form.
However, such patients may be able to spell the word,
indicating that access to modality-independent access and
access to the orthographic word form are spared.
For example, such a patient might say ‘deer’ in response
to a picture of a horse, but may simultaneously write horse
correctly (e.g. Caramazza and Hillis, 1990; Rapp et al.,
1997). Other patients show the opposite pattern; they might
incorrectly write deer but correctly say ‘horse’ in response
to the picture of the horse, indicating intact modalityindependent lexical access and access to phonological word
forms, but impaired access to orthographic word forms
(Caramazza and Hillis, 1991; Hillis et al., 1999). Again,
these patients are typically aware of their errors, because
lexical-semantic representations are intact. In both cases,
the errors cannot be due to a problem in motor planning
or execution of speech or writing, because they perform the
movements necessary for articulating or writing the word,
but in producing the wrong words. In other patients,
however, the naming impairment is primarily due to motor
programming or planning of speech articulation (apraxia of
speech). These patients make various off-target attempts
to produce the words, make errors that increase with
articulatory complexity (e.g. polysyllabic words, consonant
blends), and have halting, effortful prosody. Still other
patients may have impaired rate, range or strength
of muscles involved in speech (dysarthria) or writing
(hemiparesis), but these output problems are not generally
considered naming deficits.
The various components of the naming process depicted
in Fig. 1 may depend on different brain regions or may
depend on a network of brain regions that underlie all
of the components. Consistent with either hypothesis,
functional imaging studies of naming show widespread
1412
Brain (2007), 130, 1408 ^1422
activation of left perisylvian and extrasylvian cortex during
picture naming (Howard et al., 1992; Hirsch et al., 2001;
Abrahams et al., 2003; Grabowski et al., 2003; Martin et al.,
2005; Harrington et al., 2006; Price et al., 2005, 2006;
Kemeny et al., 2006; Saccuman et al., 2006). Likewise,
damage to many different brain regions can impair picture
naming (Goodglass and Wingfield, 1997). It is plausible
that lesions in different brain regions disrupt different
components of the naming process, but all resulting in
impaired naming. There is some evidence for this
possibility. For example, impaired naming of both pictures
and tactile stimuli in the presence of intact word comprehension (a pattern indicative of disruption at the level of
modality-independent lexical access or phonological word
form) is associated with tissue dysfunction (reduced blood
flow and/or infarct) in left Brodmann’s area (BA) 37,
part of the posterior middle/inferior temporal and fusiform
gyri (Raymer et al., 1997; Foundas et al., 1998; Hillis et al.,
2002a; Hillis et al., 2005). Furthermore, restored neural
function (by restoring blood flow) in left BA 37 is
associated with improvement in naming but not comprehension (Hillis et al., 2002b; Hillis et al., 2006). Additional
evidence that this area is important for naming comes from
functional imaging studies of normal subjects that show
activation of left middle temporal gyrus (Howard et al.,
1992) and fusiform gyrus (Grabowski et al., 2003; Kemeny
et al., 2006; Price et al., 2006; Saccuman et al., 2006) during
naming tasks. In contrast, impaired word comprehension,
attributable to impairment at the level of lexical-semantics
or linking word forms and lexical-semantic representations,
is associated with tissue dysfunction in left BA 22, (Lesser
et al., 1986; Hart and Gordon, 1990; Hillis et al., 2001) and
restored tissue function in this area results in recovery of
both comprehension and naming, consistent with restored
lexical-semantics (Hillis et al., 2001; Hillis et al., 2002; Hillis
and Heidler, 2002). Converging evidence from functional
imaging studies shows activation in left BA 22, in addition
to inferior frontal gyrus (BA 44/45) and inferior temporal
(BA 21) during word comprehension tasks (Papathanassiou
et al., 2000; Booth et al., 2002a; Fridriksson and Morrow,
2005). However, other studies only show more anterior or
inferior temporal (BA 20/21/38), parietal (BA 7/40) or
frontal activation tasks in semantic decisions about words,
especially concrete words (e.g.. Zahn et al., 2000; Sabsevitz
et al., 2005). For example, Mummery et al. (1998) found
activation in more anterior/medial temporal cortex during
perceptual decisions about words (e.g. same colour?) and
activation in temporo-parietal-occipital junction during
associative or locative decisions about words. Some of the
more widespread activation seen during these semantic
tasks may reflect activation of the complex semantic
representation (general and personal knowledge about the
item, in addition to lexical-semantics). It is also plausible
that the various cognitive processes underlying naming
emerge from distinct patterns of connectivity across the
same, or overlapping, brain regions specialized for
J. DeLeon et al.
particular inputs or outputs (Price et al., 2005; see also
Mesulam, 1998, for a similar view). In this case, although
all component processes might rely on the same regions,
one component process might rely more heavily on the
functioning of some regions (and connections between
them) than on others.
Based on these preliminary data, we hypothesized that
the relative location(s) of damage or dysfunction in acute
stroke patients might discriminate between patients with
damage to separate components of the naming process.
As an initial test of this hypothesis, we identified the
‘earliest’ or primary component of the naming process that
was impaired in each of 116 patients with acute stroke,
based on the pattern of performance across lexical tasks and
the types of errors made in naming. By ‘earliest’ we mean
the first component in the cognitive architecture depicted
in Fig. 1. It is often difficult to detect disruption in more
distal components of the process, when there is a deficit in
an earlier, or more proximal, component. That is, patients
who make semantic errors (e.g. ‘dog’ for cat) in all naming
and comprehension tasks have a disrupted semantic
representation; they may or may not also have impairment
in accessing phonological and orthographic word forms
from the semantic representation. Diffusion-weighted
imaging (DWI; which shows areas of acute infarct
or dense ischaemia) and perfusion-weighted imaging
(PWI, which shows areas of hypoperfused, dysfunctional
tissue) of the patients were analysed to identify the entire
area of tissue dysfunction in each patient. We used
discriminant function analysis to identify distributions of
dysfunction across areas that discriminated between groups
of patients defined by their earliest disrupted component of
the naming process.
Material and methods
Participants
We studied a consecutive series of right-handed individuals who
were admitted to the Johns Hopkins Hospital or Johns Hopkins
Bayview Medical Center within 24 h of onset of acute, left
hemisphere ischaemic stroke, and had none of the following
exclusion criteria: (i) contraindication for MRI or contrast;
(ii) below an education level of tenth grade; (iii) dementia or
other premorbid neurological condition; (iv) uncorrected hearing
or visual loss; and (v) lack of proficiency in English prior to
stroke; (vi) diminished level of consciousness; or (vii) unable
to give consent or indicate a family member to give consent.
Informed consent was attained from each patient or a designated
family member, using procedures and forms approved by the
Johns Hopkins Institutional Review Board. A total of 116 patients
were enrolled in this study.
Test battery
Patients were administered a series of language tasks within 24 h
of presentation, which included the following tests: (i) auditory
and written word comprehension (n ¼ 51 trials each); (ii) tactile
naming (n ¼ 17 objects); (iii) oral and written naming of pictured
Neural regions essential for naming
objects (n ¼ 34 items); (iv) oral reading (n ¼ 34 words and
24 pseudowords); (v) spelling to dictation (n ¼ 34 words and 24
pseudowords); (vi) repetition (n ¼ 34 words and 24 pseudowords);
(vii) visual and auditory lexical decision (n ¼ 34 words and 24
pseudowords); and (viii) spoken yes/no questions (n ¼ 10). For
word comprehension, three trials of word/picture verification with
each stimulus were presented. That is, the patient was shown
a picture of an object three times: once with a semantic foil,
once with a phonological foil and once with the target word.
These three presentations of the picture were presented in a
random sequence, intermixed with other picture/word pairs.
The patient had to accurately identify the correspondence between
the picture and the target word and reject the correspondence
between the picture with both the semantic and the phonological
foil in order for the response to be scored as correct.
Imaging
Each patient had an MRI scan, including DWI trace images
(bmax ¼ 1000 s/mm2; TR/TE of 10 000 ms/120 ms) and PWI
(TR/TE of 2000/60 ms) sequences on a GE Signa 1.5 tesla, echo
planar imaging capable scanner, with 5 mm slices. Apparent
diffusion coefficient (ADC) maps were generated from the
b ¼ 1000 and b ¼ 0 images to confirm that DWI lesions were
acute. PWI images were generated after injection of 20 cc
GdDTPA (gadolinium) at 5 cc/s. PWI scans were post-processed
to maps of time to peak (TTP) arrival of the gadolinium in each
voxel. The DWI and TTP maps were co-registered to T2-epi
images (generated with the same slices). For each image,
a technician blinded to language testing selected the Damasio
and Damasio (1989) BA template that most closely corresponded
to the anatomical slice was selected, and manually drew the
following BAs onto each slice 10, 11, 18, 19, 20, 21, 22, 37, 38, 39,
40, 44 and 45. The severity of hypoperfusion in each left
hemisphere BA was defined as the mean number of seconds delay
in TTP arrival of contrast in that BA, relative to the TTP delay in
the homologous region of in the normal (right) hemisphere.
Areas with a mean delay of 47 s or DWI abnormality were
assigned a value of 7.0 s delay, since this severity of hypoperfusion
or infarct represents complete dysfunction (Neumann-Haefelin
et al., 1999; Hillis, Wityk et al., 2001; Thijs et al., 2001; Sobesky
et al., 2004). Infarct was defined as bright on DWI and dark on
ADC maps. Although some areas that are bright on DWI and dark
on ADC are potentially salvageable if blood flow is immediately
restored, such areas are densely ischaemic and not functional.
Data analyses
Patients were classified into groups with disruption of different
components of the naming process, as identified by the pattern of
performance across tasks and the types of errors made. See Table 1
for the criteria for identifying disruption of each component of
the naming process. These criteria are based on single case studies
of patients who have been reported to have deficits at each level of
the naming task, on the basis of detailed analysis of performance
across a wide range of tasks (e.g. Butterworth et al., 1984; Howard
and Orchard-Lisle, 1984; Kay and Ellis, 1987; Caramazza and
Hillis, 1990; Hillis et al., 1990; Howard and Gatehouse, 2006).
Although many authors have used more detailed criteria involving
many more tests (e.g. Howard and Gatehouse, 2006), our criteria
are shared by many studies and serve as a ‘working definition’
based on the schematic architecture of the naming process
Brain (2007), 130, 1408 ^1422
1413
shown in Fig. 1. For example, impairment at the level of
‘lexical-semantics’ (the subset of semantic information required to
be able to name an object and determine whether or not a picture
depicts an instance of items referred to by that name) would
be reflected in impairments in both spoken and written word
comprehension (measured in our study with word/picture
verification), as well as oral and written naming of both pictures
and objects (see Hillis et al., 2000 for discussion). It is widely
agreed that patients with semantic impairments make semantic
errors in naming (Howard and Gatehouse, 2006; Jeffries and
Lambon-Ralph, 2006), although patients with deficits at the level
of visual perception (e.g. accessing semantics from a visual/
structural description) or in accessing the phonological word form
can also make such errors (Hillis and Caramazza, 1995a; Howard
and Gatehouse, 2006). However, patients with visual/perceptual
deficits will make such errors only in response to visual stimuli,
and patients with impaired access to the phonological word form
make semantic errors in oral but not written naming. In contrast,
patients with impaired lexical-semantics make semantic errors in
all of these tasks (Hillis et al., 1990). We did not attempt to
distinguish between impairments in access versus representation,
because such distinctions may require extensive testing (Shallice,
1987) or may be impossible (Caramazza et al., 1990).
A score of 410% errors on a particular task was selected as the
criterion for impairment for that lexical task, because (i) this cutoff was 42 standard deviations below the mean for normal
subjects on our original normative data from non-brain damaged
subjects and (ii) no normal subject made more than 10% errors
on any of these tasks. Regrettably, we did not have sufficient time
in the acute stage of stroke to administer tasks that would
distinguish between deficits in lexical-semantics or more broadly
in semantics, including amodal general and personal knowledge
(although the latter is uncommon after stroke).
Discriminant function analysis utilizing SPSS (SPSS for
windows, Rel. 12.0.1 2003: SPSS Inc.) was used to identify the
areas of hypoperfusion and/or infarct on MRI scans that could
discriminate between patients with disruption in different
components of the naming process. In this analysis, the input
consists of the independent (‘discriminant’) variables—the areas of
tissue dysfunction (severity of hypoperfusion in each Brodmann
area, defined by mean TTP) in each patient and the group
membership defined by level of impairment (see Table 1) of that
patient. Data are entered in a stepwise fashion (to identify the
least number of areas that could distinguish between groups),
and a series of discriminant functions (based on the status of each
of the independent variables) are identified that each represent
a pattern in the independent variables that distinguished a
particular group from the other groups. Discriminant functions
that could distinguish between groups with an alpha level of
P50.05 were considered significant.
Subsequent analyses
As we cannot rule out the possibility that patients with
comparably impaired oral and written naming have deficits
in access to both phonological and orthographic word forms
(rather than impaired modality-independent lexical access), and as
there were small numbers of patients in some of the groups, we
carried out two additional analyses. First, we combined patients
with deficits in modality-independent lexical access with patients
with impaired access to phonological and orthographic word
1414
Brain (2007), 130, 1408 ^1422
J. DeLeon et al.
Table 1 Criteria for impairment
Level of impairment
Impairment criteria
Group size
Visual/perceptual
Lexical-semantics
Tactile naming4(20%) Picture naming
410% error in:
Spoken Word/Picture Verification AND
Written Word/Picture Verification AND
Tactile Naming AND
Oral Naming of Pictures AND
Written Naming of Pictures AND
Semantic errors in at least one of the above tasks
410% error in:
Oral Naming of Pictures AND
Tactile Naming AND
Written Naming AND
510% error in:
Spoken Word/Picture Verification
410% error in:
Tactile Naming AND
Oral Naming of Pictures AND
510% errors in Written Naming of Pictures
Semantically and/or phonologically related errors
and/or circumlocutions in at least one of the above tasks
410% error in:
Written Naming of Pictures AND
Spelling to Dictation of Irregular Words AND
Semantically and/or visually related errors and/or omissions
or phonologically plausible errors (e.g. ‘phone’ spelled foan) in spelling
410% error in:
Tactile Naming AND
Oral Naming of Pictures AND
Oral Reading AND
Repetition AND
Production of articulatory errors
1
46
Modality-independent lexical access
Phonological word form
Orthographic word form
Motor speech
forms into one group (into a single ‘lexical access’ impairment
group), and retained the other groups. Second, we created two
deficit groups, semantic (with 410% errors in naming and word/
picture verification) and post-semantic (410% errors in naming,
but 510% errors in word/picture verification) in addition to the
‘no deficit’group. We carried out discriminant function analysis in
the same way as described above with these classifications.
To demonstrate that the original classification can generalize
to a new set, which would be essential from a clinical perspective,
we carried out a ‘leave one out’ cross-validation procedure to
determine if group assignments using the identified discriminant
functions were reliable. This procedure uses a random subset of
the initial subjects.
Linear regression analyses were performed to determine
the contribution of degree of hypoperfusion in each of the BAs
identified in the primary discriminant function analysis to error
rate on auditory word/picture verification and to error rate in oral
naming.
Results
Using these criteria, one patient was classified as having
impairment at a visual/perceptual (structural disruption)
level; 46 patients were classified as having impairment
17
2
8
2
at the level of the lexical-semantic system; 17 patients were
classified as having impairment at the level of modalityindependent lexical access; two patients were classified
as having impairment at the level of the phonological word
form; eight patients were classified as having impairment of
the orthographic word form level, and two patients
were classified as having impairment of motor speech.
Another 40 patients were classified as having no naming
deficits. Table 2 shows the range of scores for each task by
patients in each group based on the ‘earliest’ level of
processing that was found to be impaired.
Initial analysis
Multiple (five) discriminant functions were found that
could classify cases into groups based on the level of
disruption of the naming process (as defined by Table 2),
using the status of each of seven BAs (of the 13 BAs
entered; see imaging section). Each discriminant function
has different coefficients, thus weighting each of these BAs
differentially. Together, these five discriminant functions
successfully distinguished the behaviourally-defined groups
(Wilk’s; lambda ¼ 0.255; I2 ¼ 93.5; df35, P50.0001).
Neural regions essential for naming
Brain (2007), 130, 1408 ^1422
The seven independent variables, mean severity of hypoperfusion within each BA, that were incorporated
into the discriminant functions were: BA 22 (including
Wernicke’s area), BA 44 (part of Broca’s area), BA 45 (part
of Broca’s area), BA 21 (inferior temporal cortex), BA 37
(posterior, inferior temporal/fusiform gyrus), BA 38 (anterior temporal cortex) and BA 39 (angular gyrus). Table 3
displays the correlations between each independent variable
and each discriminant function. The mean TTP in the other
six analysed BAs (10, 11, 18, 19, 20 and 40) did not
contribute to classifying the patients.
Function 1 (most highly correlated with the degree
of hypoperfusion in BA 22 (r ¼ 0.899) had the highest value
in the group of patients with disruption of naming
at the level of semantics. Function 2 was most correlated
with the severity of hypoperfusion in BA 37 (r ¼ 0.995)
and had the highest value in cases with disruption at the
level of modality-independent lexical access. Function 3,
with highest correlations with degree of hypoperfusion
in BA 44 (r ¼ 0.870) and BA 45 (r ¼ 0.862), had the
highest value in cases with motor speech impairment.
1415
Function 4 was most correlated with the degree of
hypoperfusion in BA 39 (r ¼ 0.401); this function had the
highest value in the two patients with disruption at the level
of phonological word form. Finally, the addition of
Function 5, which correlated most highly with severity
of hypoperfusion in BA 38 (r ¼ 0.786), accounted for only
an additional 0.1% of the variance; it had the highest
absolute (negative) value in the one patient with disruption
at the level of visual perception/structural description.
In summary, for patients with impaired lexical-semantics,
Function 1 (correlated with dysfunction in left BA 22,
which includes all or part of Wernicke’s area) was the
highest discriminating function. Among patients with
disruption at the level of modality-independent lexical
access, Function 2 (correlated with degree of tissue dysfunction in left BA 37, posterior middle temporal/fusiform
gyrus) was highest. For patients with impairment at the
level of the phonological word form, Functions 2 and 4
(correlated with degree of tissue dysfunction in left BA 39,
angular gyrus) had the highest values. Among patients with
disruption at the level of the orthographic word form,
Table 2 Range of error rates on each test for each group (defined by initial deficit, as defined in Table 1)
Number of patients
Oral picture naming
Oral tactile naming
Written picture naming
Oral reading
Repetition
Spelling to dictation
Spoken word comprehension
Written word comprehension
Visual
Semantic
Modalityindependent
lexical access
Phonological
word form
Orthographic
word form
Motor speech
1
24
0
47
90
8
0
12
29
46
12^100
12^100
18 ^100
18 ^100
0 ^100
9^100
12^100
18 ^100
17
12^100
12^100
18 ^100
2^94
0 ^16
12^54
0^6
0^6
2
18 ^24
24
0^6
27
0 ^10
0
0^6
0^6
8
0^6
0^6
12^71
2^28
0^8
14 ^50
0^6
0^6
2
18 ^71
18 ^71
0 ^ 47
86
18 ^ 68
0 ^ 85
0^6
0
Table 3 Structure matrix for the initial analysis
BA 22
BA 21
BA 37
BA 39
BA 44
BA 45
BA 38
Cumulative percentage
of variance
Function 1
(highest in
semantic group)
Function 2
(highest in
modalityindependent
lexical access
and phonological
group)
Function 3
(highest in
motor speech
group)
Function 4
(highest in
orthographic
word form
group)
Function 5
0.899
0.347
0.092
0.162
0.032
0.090
0.194
80.3%
0.271
0.256
0.823
0.704
0.089
0.052
0.031
89.8%
0.232
0.116
0.038
0.009
0.870
0.862
0.093
97.7%
0.021
0.042
0.182
0.401
0.061
0.354
0.299
99.9%
0.246
0.081
0.034
0.086
0.409
0.301
0.786
100.0%
Pooled within-groups correlations between discriminating variables and standardized canonical discriminant functions. The highest absolute
correlations between each variable (degree of hypoperfusion within each BA) and each function are displayed in bold print. Functions are
identified by the analysis (not a priori). We have added in parentheses the group that has the highest value for each function.
1416
Brain (2007), 130, 1408 ^1422
J. DeLeon et al.
Function 4 (most correlated with severity of hypoperfusion
in left BA 39, angular gyrus) was highest. Figure 2 shows
magnetic resonance DWI and PWI scans of two patients,
with distinct regions of hypoperfusion associated with
a deficit at the level of modality-independent lexical
access (Panel A) versus a deficit at the level of accessing
orthographic word forms (Panel B). There was only one
patient with a deficit in visual recognition and only two
patients whose ‘earliest’ (only) disruption in naming was
in motor speech, so results are likely to be unreliable
in these cases.
Fig. 2 (A) DWI (left) and PWI (right) scans of a patient with
impairment at the level of modality-independent lexical access
(with deficits in oral and written picture naming, but normal comprehension). (B) DWI (left) and PWI (right) scans of a patient with
impairment at the level of accessing orthographic word forms. On
these PWI scans, dark areas represent regions of hypoperfusion.
Subsequent analyses
As there were too few cases with deficits in access to
visual/structural descriptions, modality-specific deficits in
accessing phonological word forms, or isolated motor
speech impairment to draw conclusions regarding these
groups, we carried out two post hoc analyses with fewer
classes of patients. Similar results were found. When we
combined patients with impaired modality-independent
lexical access (Group 3) with impaired access to phonological word forms (Group 4) or orthographic word forms
(Group 5), two discriminant functions based on severity
of hypoperfusion of the same seven BAs successfully
distinguished the behaviourally-defined groups (Wilk’s
lambda ¼ 0.274; I2 ¼ 137; df14, P50.0001). Function 1
was most correlated with severity of hypoperfusion in left
BA 22 (0.965), but also with hypoperfusion of left BA 21
(r ¼ 0.420) and BA 38 (0.253) and was highest in patients
with semantic deficits. Function 2, which was highest
in patients with deficits in lexical access (modalityindependent lexical access or access to orthographic
and/or phonological word forms) was more strongly
correlated with the degree of hypoperfusion in left BA 37
(r ¼ 0.626) and BA 44 (r ¼ 0.542).
When patients were then reclassified as having no
deficits, semantic deficits or post-semantic deficits, very
similar results were obtained. Two functions, based on the
severity of hypoperfusion in the same seven BAs were
identified that distinguished between groups (Wilk’s
lambda ¼ 0.238; I2 ¼ 142; df14, P50.0001). Table 4 displays
the correlation coefficients for each independent variable
for each function in the two post hoc analyses. Figure 3
displays the values of discriminant functions 1 (x axis) and
2 (y axis) for each case of semantic or post-semantic deficit,
as behaviourally defined in this last analysis.
Cross-Validation. We ran a cross-validation using a
leave-one-out procedure. The cross-validation for each
analysis yielded identical or almost identical results for
Table 4 Structure matrix for the subsequent analyses
Secondary analysis 1
BA 22
BA 21
BA 38
BA 37
BA 44
BA 45
BA 39
Cumulative percentage
of variance
Secondary analysis 2
Function 1
(highest in
semantic group)
Function 2
(highest in lexical
access group)
Function 1
(highest in
semantic group)
Function 2
(highest in
post-semantic
group)
0.965
0.420
0.253
0.305
0.312
0.339
0.263
82.5%
0.034
0.171
0.099
0.626
0.542
0.397
0.334
100%
0.942
0.385
0.269
0.248
0.254
0.291
0.226
76.3%
0.193
0.214
0.045
0.602
0.548
0.404
0.334
100.0%
Pooled within-groups correlations between discriminating variables and standardized canonical discriminant functions. The highest absolute
correlations between each variable (degree of hypoperfusion within each BA) and each function are displayed in bold print. Functions are
identified by the analysis (not a priori). We have added in parentheses the group that has the highest value for each function
Neural regions essential for naming
Brain (2007), 130, 1408 ^1422
1417
of lexical-semantics). This model indicates that hypofusion
of left BA 22 (most of the superior temporal gyrus,
extending to the superior temporal sulcus) was the
strongest predictor of the severity of word comprehension
impairment acutely after stroke. It was the only area where
the severity of hypoperfusion independently predicted error
rate (P50.0001) after adjustment for the severity of
hypoperfusion of the other areas.
Error rate in oral naming of pictures was best predicted
by the following model derived from linear regression,
where NE stands for naming error rate and HP stands
for ‘severity of hypoperfusion of’ as defined earlier:
NE ¼ HP BA22 8:7 þ HP BA 45 3:5 þ HP BA 37
3:3 þ HP 21 1:7 HP 38 1:6 HP 39
Fig. 3 Scatterplot showing the values of function 1 (x axis) and
function 2 (y axis) in cases of semantic impairment (open squares)
and post-semantic impairment (solid diamonds).
the new population as for the original population. For
example, in the last analysis 80.0% of the original cases and
77.1% of the cross-validated cases were correctly classified.
In the earlier analysis, 77.7% of the original group were
correctly classified, and 76.8% of the cross-validated cases
were correctly classified.
Linear regression analysis. The discriminant function
analyses demonstrated that the distribution of tissue
dysfunction across seven BAs distinguished between
groups of patients behaviourally identified as having
different levels of impairment in picture naming.
However, these analyses do not directly reveal the
independent contributions of each area to each type of
impairment. To evaluate these contributions, we performed
two separate linear regression analyses: one to determine
the contributions of dysfunction in each of the seven BAs
(as identified in the discriminant function analysis) to
the severity of the lexical-semantic deficit (indicated by
the error rate on auditory word/picture verification), and
another to determine contribution of these BAs to severity
of the naming deficit (indicated by error rate in oral
naming of pictures).
The error rate in auditory word/picture verification was
best predicted by the following model derived from linear
regression, where VerifErr stands for verification error rate
and HP stands for ‘severity of hypoperfusion of’ as defined
earlier:
VerifErr ¼ HP BA22 8:6 þ HP BA 45 5:0
þ HP BA 21 1:9 þ HP 37 0:16
HP 38 1:2 HP 39 0:71
þ 3:9 ðr ¼ 0:81; r2 ¼ 66; P50:0001Þ
The coefficients indicate the ‘weight’ of the contribution
to error rate in word/picture verification (a measure
3:9 þ 13 ðr ¼ 0:78; r2 ¼ 61; P50:0001Þ
The areas where the degree of hypoperfusion independently contributed to the error rate in oral naming
of pictures, after adjusting for the degree of hypoperfusion
in the other areas, were: left BA 22 (P50.0001), BA 37
(posterior middle and inferior temporal/fusiform gyrus;
P ¼ 0.019); and BA 39 (angular gyrus; P ¼ 0.002).
Discussion
The results of this study provide evidence that distinct
distributions of tissue dysfunction in seven cortical areas
of the left hemisphere are associated with disruption of
different components of the naming task. We did not find
a one-to-one relationship between areas of tissue dysfunction and impaired components of naming. Instead, our
results indicate that each component depends on the
function of a number of areas in a network. Although
certain areas were more heavily weighted than others
in each discriminant function, all of the discriminant
functions depended on the status of several regions.
This finding is consistent with functional imaging studies
that generally show several areas where activation is
correlated with any component of the naming task (see
introduction). It is also consistent with the proposal of
‘comibinatorics’ by Price and colleagues (2005) that distinct
cognitive processes emerge from the interactions between
several regions that are specialized for particular inputs or
outputs. Damage to a single part of the network would
likely disrupt picture naming performance, but would
differentially affect the various cognitive processes required
for accurate naming.
The data from this study largely converge with evidence
from other modalities investigating brain–language relationships. For instance, the severity of tissue dysfunction in left
BA 22 had the greatest contribution in predicting the error
rate in word/picture verification, a measure of the semantic
deficit (lexical-semantics or possibly more broad deficits
in semantics). Left BA 22 is sometimes referred to as
Wernicke’s area (e.g. Kober et al., 2001), although others
1418
Brain (2007), 130, 1408 ^1422
define Wernicke’s area as a more limited area in posterior
BA 22, and others define it as a broader area that includes
part of angular gyrus. The role of BA 22 is consistent with
results of functional imaging studies that demonstrate
activation in left BA 22 during word comprehension tasks
(Papathanassiou et al., 2000; Booth et al., 2002a;
Whatmough and Chertkow, 2002; Fridriksson and
Morrow, 2005; Saur et al., 2006) and detecting semantic
anomalies in sentences (Friederici et al., 2003; see also
Maess et al., 2006). However, other functional imaging
studies provide evidence for specialization within BA 22
(Wise et al., 2001; Scott et al., 2003) as well as other BAs
(Poldrack et al., 1999). For example, there is converging
evidence from several studies that more rostral areas of left
lateral temporal cortex, linked via the uncinate fasciculus to
ventrolateral prefrontal cortex, are engaged in processing
word meaning; while caudal temporal regions, linked via
the arcuate fasciculus to the dorsolateral prefrontal cortex,
are engaged in processing phonetic and phonological
structure (Wise, 2003; see also Hickok and Poeppel,
2000). We did not attempt to determine the areas within
each BA that may have such specialized roles. Nevertheless,
we did find evidence that at least part of left BA 22 is
critical for accessing the type of semantic information
needed to name pictures and to determine whether a word
matches a picture, consistent with data from chronic stroke
and direct cortical stimulation reviewed in the introduction
showing that damage or temporary dysfunction of left BA
22 is associated with deficits in accessing word meaning.
The finding that severity of tissue dysfunction in left BA
37 correlated most strongly with the discriminant function
that most contributed to classifying patients with impairment at the level of modality-independent lexical access
also converges with evidence from functional imaging
and chronic lesion studies. Left BA 37 includes part of
posterior/lateral and ventral inferior temporal gyrus, which
shows activation during word retrieval relative to later
stages of picture naming (Kemeny et al., 2006) or during
naming of tools in response to pictures or sounds (Tranel
et al., 2005), or naming pictures relative to face orientation,
more strongly in men than women (Grabowski et al., 2003).
Left BA 37 also includes the midfusiform gyrus, which
reliably shows activation in reading words (Cohen et al.,
1997, 2000, 2002), spelling words (Rapcsak et al., 2004),
and naming objects (Abrahams et al., 2003; Price et al.,
2005, Price et al., 2006). Price and Devlin (2003, 2004) have
proposed that the left midfusiform gyrus is important in
modality-independent lexical processing, rather than specific to visual word forms. Cohen and colleagues
(Cohen et al., 2004a, b) have argued that there are two
distinct areas of the left fusiform gyrus, both in BA 37: one
involved specifically in reading and one involved in
modality-independent lexical processing. Our data cannot
evaluate this hypothesis, since we did not look for distinct
areas of tissue dysfunction within left BA 37, but our data
J. DeLeon et al.
do support the hypothesis that part of left BA 37 is
important for modality-independent lexical processing.
In patients with disruption at the level of the orthographic word form, the discriminant function that was
most correlated with severity of hypoperfusion in left BA 39
(angular gyrus) had the highest value. This finding is
consistent with many lesion studies that show a strong
association between chronic lesions involving left angular
gyrus and impairment of both reading and writing
(see Benson, 1979 or Hillis and Rapp, 2004, for review).
Functional imaging studies show activation in left angular
gyrus during spelling, at least in some patients (Beeson
et al., 2003; Hseih and Rapp, 2004). Together, these results
indicate some role of the left angular gyrus in spelling.
Although this area is plausibly involved in accessing
orthographic representations, we did not attempt to distinguish this deficit from damage to sublexical phonology to
orthography, which might affect spelling of words also.
Functional imaging studies inconsistently show activation
of left BA 39 during oral picture naming tasks (Farias et al.,
2005; Fridriksson et al., in press; Harrington et al., 2006;
Kemeny et al., 2006). Other functional imaging studies have
revealed greater activation in BA 39 in response to semantic
relative to phonological judgements (e.g. Mummery et al.,
1998), semantically congruent sentences relative to incongruent or pseudoword sentences (Humphries et al., 2006);
or cross-modal tasks (e.g. spelling judgements with auditory
input or rhyming judgements with visual input; Booth
et al., 2002b).
There were not enough patients with deficits in the other
components of the naming process to comment on their
relationship to dysfunction in particular BAs. Nevertheless,
tissue dysfunction in left BA 38, anterior temporal cortex,
BA 44 and 45, posterior inferior frontal gyrus and left BA
21, inferior temporal cortex, also contributed to severity of
naming and comprehension error rates, and to distinguishing the level of impairment in picture naming. The severity
of hypoperfusion in BA 38 and 21 were most correlated
with the discriminant function that had a high value in
patients with semantic deficits (see Table 4), while BA 44
and 45 correlated more strongly with the discriminant
function that had a high value in patients with postsemantic deficits (and those with motor speech deficits in
the initial analysis; Table 3). These findings are consistent
with studies of progressive aphasic patients with atrophy in
anterior temporal gyrus (BA 38) and inferior temporal
cortex (BA 21) who initially present with spoken word
retrieval deficits, but eventually develop semantic dementia
or ‘semantic’ variety of primary progressive aphasia
(Mummery et al., 1999; Gorno-Tempini et al., 2004).
Patients with lesions in left BA 38 are impaired in naming
famous landmarks, relative to patients with lesion in right
BA 38, consistent with the hypothesis that this area may
be critical in naming unique entities (Damasio et al., 2004;
Tranel et al., 2006). Lesion studies also show chronic infarct
or progressive atrophy involving BA 44 and 45 in patients
Neural regions essential for naming
with motor speech impairments (Mohr et al., 1987;
Amici et al., 2006).
Likewise, several functional imaging studies reveal
activation of left BA 38 (Grabowski et al., 2003), left BA
21 (Papathanassiou et al., 2000), and left BA 44/45
(Papathanassiou et al., 2000; Hirsch et al., 2001;
Abrahams et al., 2003; Martin et al., 2005) in overt
naming. Activation in left BA 21 and 38 (as well as
posterior superior temporal sulcus) likely reflects semantic
processing, compatible with our findings, because these
areas are active in comprehension of normal discourse,
apart from metalinguistic task demands (Crinion et al.,
2003, 2006). Activation in left BA 44/45 was not seen in
these normal comprehension tasks, but is seen during
motor speech relative to non-speech articulatory movements (Bonilha et al., 2006).
Many of the cited functional imaging studies show
bilateral activation in many of the regions we have discussed
as well as in bilateral cingulate cortex. Damage to these right
cortical regions does not normally interfere with naming
common objects of the type used in this study. However, it
is widely recognized that functional imaging studies reveal
areas of the brain that are reliably engaged in a particular
function, even if they are not essential for the function. For
example, picture naming often elicits recollection of personal
knowledge or emotions linked to the concept that might
activate more widespread areas. Therefore, lesion studies like
the current one can complement functional neuroimaging in
exploring the neural regions necessary for a particular
cognitive function or task (Chatterjee, 2005).
We should note that the network of regions we identified
as important for naming may not include all of the areas
that can support naming. Patients with damage to any
of these regions may well recover naming over time, as the
cognitive processes underlying naming begin to emerge
from other ‘degenerate’ systems that can support naming
(Price et al., 1999). This study only identified areas where
tissue dysfunction was associated with acutely impaired
picture naming, indicating that the area was necessary
for the task before reorganization of structure/function
relationships. Furthermore, this study revealed the areas
where tissue dysfunction affects one cognitive process
underlying naming more than it affects other components
of naming. If tissue dysfunction in some other region were
to affect all components equally, it might interfere with
naming, but would not be identified in this analysis.
We should also note that analysis of regions of
tissue dysfunction by Brodmann’s areas has limitations
(e.g. known individual variability in cytoarchitectural
fields). However, these limitations are not ameliorated or
solved by using a voxel-based approach or analysis of the
involved gyri, since areas with distinct functions likely
correspond to regions with a particular cytoarchitecture
within each individual brain. Use of BA templates
(which estimate the location of distinct cytoarchitectural
fields) has high interjudge reliability, some theoretical
Brain (2007), 130, 1408 ^1422
1419
rationale, as well as some empirical support (because
lesions involving specific BAs have been associated with
various language deficits, and activation in particular BAs
has been associated with particular language processes).
Nevertheless, our references to particular BAs should be
considered approximate, since we have no data on the
cytoarchitecture of these areas. Furthermore, as discussed
earlier, there may be specialization of distinct areas within
each BA.
Another limitation of this study is that at least some
of the patients had impairment involving more than one
level of processing in the naming task. For example, many
patients had mild impairments of motor speech in addition
to a ‘primary’ deficit at the level of lexical-semantics,
modality-independent lexical access, or orthographic/phonological word form. Such patients made recognizable
semantic paraphasias, despite misarticulations (often selfcorrected). Only two patients had a pure motor speech
disorder, so we were unable to identify areas of tissue
dysfunction that reliably discriminated motor speech
deficits from other deficits. Nevertheless, this analysis was
quite successful in discriminating patients with functional
damage to different components of the naming task. Future
studies are needed to confirm that these discriminant
functions (using tissue dysfunction in the seven identified
regions as the discriminant variables) applied sequentially
can successfully classify an altogether new set of patients
into the same groups defined by the primary impaired
component of the naming process. Our cross-validation
used a subset of the same patients, which provides
a measure of reliability, but does not assure that these
functions will be as successful in distinguishing the level
of impairment in naming in other patients.
In summary, we have shown that each patient’s level
of impairment in the naming process reflected the
distribution of tissue dysfunction in particular regions of
the left anterior, inferior, and posterior middle/superior
temporal cortex, posterior inferior frontal, and inferior
parietal cortex. These same regions show activation during
propositional and non-propositional speech in functional
imaging studies (Blank et al., 2002). While occipital cortex
is also critical for picture naming, it is likely that bilateral
occipital damage is necessary to disrupt visual recognition.
Our results complement functional imaging studies by
demonstrating which of the neural regions engaged in
naming are also critical for naming.
Acknowledgements
The research reported in this paper was supported by NIH
R01 DC05375 and P41 RR15241. Funding to pay the Open
Access publication charges for this article was provided by
an NIH grant, DC05375 to AH.
References
Abrahams S, Goldstein LH, Simmons A, Brammer MJ, Williams SC,
Giampietro VP, et al. Functional magnetic resonance imaging of verbal
1420
Brain (2007), 130, 1408 ^1422
fluency and confrontation naming using compressed image acquisition
to permit overt responses. Hum Brain Mapp 2003; 20: 29–40.
Amici S, Gorno-Tempini ML, Ogar JM, Dronkers NF, Miller BL. An
overview on Primary Progressive Aphasia and its variants. Behav Neurol
2006; 17: 77–87.
Badecker W. The two-stage model of lexical retrieval: evidence from a case
with anomia with selective preservation of grammatical gender.
Cognition 1995; 57: 193–216.
Beeson PM, Rapcsak SZ, Plante E, Chargualaf J, Chung A, Johnson SC,
et al. The neural substrates of writing: a functional magnetic resonance
imaging study. Aphasiology 2003; 17: 647–55.
Benson DF. Aphasia, alexia, and agraphia. New York: Churchill
Livingstone; 1979.
Blank SC, Scott SK, Murphy K, Warburton E, Wise RJ. Speech production:
Wernicke, Broca and beyond. Brain 2002; 125: 1829–38.
Bonilha L, Moser D, Rorden C, Baylis GC, Fridriksson J. Speech apraxia
without oral apraxia: can normal brain function explain the physiopathology? Neuroreport 2006; 17: 1027–31.
Booth JR, Burman DD, Meyer JR, Gitelman DR, Parrish TB,
Mesulam MM. Functional anatomy of intra- and cross-modal lexical
tasks. Neuroimage 2002a; 16: 7–22.
Booth JR, Burman DD, Meyer JR, Gitelman DR, Parrish TB,
Mesulam MM. Modality independence of word comprehension. Hum
Brain Mapp 2002b; 16: 251–61.
Butterworth B, Howard D, McLoughlin P. The semantic deficit in aphasia:
the relationship between semantic errors in auditory comprehension and
picture naming. Neuropsychologia 1984; 22: 409–26.
Caramazza A, Hillis AE. Where do semantic errors come from? Cortex
1990; 1: 95–122.
Caramazza A, Hillis AE. Lexical organization of nouns and verbs in the
brain. Nature 1991; 349: 788–90.
Caramazza A, Miozzo M. The relation between syntactic and phonological
knowledge in lexical access: evidence from the ‘tip-of-the-tongue’
phenomenon. Cognition 1997; 64: 309–43.
Caramazza A, Miozzo M. More is not always better: a response to Roelofs,
Meyer, and Levelt. Cognition 1998; 69: 231–41.
Caramazza A, Hillis AE, Rapp BC, Romani C. Multiple semantic or
multiple confusions? Cogn Neuropsychol 1990; 7: 161–8.
Caviness V, Makris N, Montinaro E, Sahin N, Bates J, Schwamm L, et al.
Anatomy of stroke, Part I: An MRI-based topographic and volumetric
system of analysis. Stroke 2002; 33: 2549–56.
Chanoine V, Teixeira-Ferreira C, Demonet JF, Nespoulous JL, Poncet M.
Optic aphasia with pure alexia: a mild form of visual associative
agnosia? A case study. Cortex 1998; 34: 437–48.
Chatterjee A. A madness to the methods in cognitive neuroscience? J Cogn
Neurosci 2005; 17: 847–9.
Cohen L, Dehaene S, Naccache L, Lehéricy S, Dehaene-Lambertz G,
Hénaff MA, et al. The visual word form area: spatial and temporal
characterization of an initial stage of reading in normal subjects and
posterior split brain patients. Brain 2000; 123: 291–307.
Cohen L, Lehericy S, Chochon F, Lemer C, Rivaud S, Dehaene S.
Language-specific tuning of visual cortex? Functional properties of the
visual word form area. Brain 2002; 125: 1054–69.
Cohen L, Henry C, Dehaene S, Molko N, Lehéricy S, Martinaud O, et al.
The pathophysiology of letter-by-letter reading. Neuropsychologia
2004a; 42: 1768–80.
Cohen L, Jobert A, Le Bihan D, Dehaene S. Distinct unimodal and
multimodal regions for word processing in the left temporal cortex.
Neuroimage 2004b; 23: 1256–70.
Crinion JT, Lambon-Ralph MA, Warburton EA, Howard D, Wise RJ.
Temporal lobe regions engaged during normal speech comprehension.
Brain 2003; 126: 1193–201.
Crinion JT, Warburton EA, Lambon-Ralph MA, Howard D, Wise RJ.
Listening to narrative speech after aphasic stroke: the role of the left
anterior temporal lobe. Cereb Cortex 2006; 16: 1116–25.
Damasio H, Damasio A. Lesion analysis in neuropsychology. New York:
Oxford University Press; 1989.
J. DeLeon et al.
Damasio H, Tranel D, Grabowski T, Adolphs R, Damasio A. Neural
systems behind word and concept retrieval. Cognition 2004; 92:
179–229.
Dell GS. A spreading activation theory of retrieval in sentence production.
Psychol Rev 1986; 93: 283–321.
Dell GS, Schwartz ME, Martin N, Saffran EM, Gagnon DA. Lexical access
in aphasic and nonaphasic speakers. Psychol Rev 1997; 104: 801–38.
DeRenzi E, Saetti MC. Associative agnosia and optic aphasia: qualitative or
quantitative difference? Cortex, 1997; 33: 115–30.
Ellis AW, Miller D, Sin G. Wernicke’s aphasia and normal language
processing: a case study in cognitive neuropsychology. Cognition 1983;
15: 111–4.
Farias ST, Harrington G, Broomand C, Seyal M. Differences in functional
MR imaging activation patterns associated with confrontation naming
and responsive naming. AJNR Am J Neuroradiol 2005; 26: 2492–9.
Foundas A, Daniels SK, Vasterling JJ. Anomia: case studies with lesion
localization. Neurocase 1998; 4: 35–43.
Fridriksson J, Morrow L. Cortical activation and language task difficulty in
aphasia. Aphasiology 2005; 19: 239–50.
Fridriksson J, Morrow KL, Moser D, Baylis GC. Neural recruitment
associated with anomia recovery in aphasia. Neuroimage: In press.
Fridriksson J, Rorden C, Morgan PS, Murrow KL, Bayliss GC. Measuring
the hemodynamic response in chronic hypoperfusion. Neurocase: In
press.
Friederici AD, Ruschemeyer SA, Hahne A, Fiebach CJ. The role of left
inferior frontal and superior temporal cortex in sentence comprehension: localizing syntactic and semantic processes. Cereb Cortex 2003; 13:
170–7.
Garrett MF. The Cambridge survey of lingusistics. Biological and
psychological aspects of language. Vol. 3. Cambridge, MA: Harvard
University Press; 1988.
Goodglass H, Wingfield A. Word-finding deficits in aphasia: brainbehavior relations and symptomatology. Anomia. London: Academic
Press; 1997.
Gorno-Tempini ML, Dronkers NF, Rankin KP, Ogar JM, Phengrasamy L,
Rosen HJ, et al. Cognition and anatomy in three variants of primary
progressive aphasia. Ann Neurol 2004; 55: 335–46.
Grabowski TJ, Damasio H, Eichhorn GR, Tranel D. Effects of gender on
blood flow correlates of naming concrete entities. Neuroimage 2003; 20:
940–54.
Harrington GH, Buonocore MH, Farias ST. Intrasubject reproducibility of
functional MR imaging activation in language tasks. AJNR Am J
Neuroradiol 2006; 27: 939–44.
Hart J, Gordon B. Delineation of single-word semantic comprehension
deficits in aphasia, with anatomical correlation. Ann Neurol 1990; 27:
226–31.
Hickok G, Poeppel D. Towards a functional neuroanatomy of speech
perception. Trends Cogn Sci 2000; 4: 131–8.
Hillis AE, Barker P, Beauchamp N, Gordon B, Wityk R. MR perfusion
imaging reveals regions of hypoperfusion associated with aphasia and
neglect. Neurology 2000; 55: 782–8.
Hillis A, Barker P, Beauchamp N, Winters B, Mirski M, Wityk R.
Restoring blood pressure reperfused Wernicke’s area and improved
language. Neurology 2001; 56: 670–2.
Hillis AE, Caramazza A. The compositionality of lexical-semantic
representations: clues from semantic errors in object naming. Memory
1995a; 3: 333–58.
Hillis A, Caramazza A. Cognitive and neural mechanisms underlying visual
and semantic processing. J Cogn Neurosci 1995b; 7: 457–78.
Hillis AE, Caramazza A. Converging evidence for the interaction of
semantic and phonological information in accessing lexical information
for spoken output. Cogn Neuropsych 1995c; 12: 187–227.
Hillis AE, Heidler J. Mechanisms of early aphasia recovery: evidence from
MR perfusion imaging. Aphasiology 2002; 16: 885–96.
Hillis AE, Kane A, Tuffiash E, Ulatowski JA, Barker P, Beauchamp N, et al.
Reperfusion of specific brain regions by raising blood pressure restores
Neural regions essential for naming
selective language functions in subacute stroke. Brain Lang 2002b; 79:
495–510.
Hillis AE, Kleinman KT, Newhart M, Heidler-Gary J, Gottesman R,
Barker PB, et al. Restoring cerebral blood flow reveals neural regions
critical for naming. J Neurosci 2006; 26: 8069–73.
Hillis AE, Newhart M, Heidler J, Barker PB, Herskovits E, Degaonkar M.
The roles of the ‘‘visual word form area’’ in reading. Neuroimage 2005;
24: 548–59.
Hillis AE, Rapp BC. Cognitive and neural substrates of written language
comprehension and production. The new cognitive neurosciences.
3rd edn. Cambridge? MA: MIT Press; 2004. p. 755–88.
Hillis AE, Rapp BC, Caramazza A. When a rose is a rose in speaking but a
tulip in writing. Cortex 1999; 35: 337–788.
Hillis AE, Rapp BC, Romani C, Caramazza A. Selective impairment of
semantics in lexical processing. Cogn Neuropsychol 1990; 7: 191–224.
Hillis AE, Tuffiash E, Wityk RJ, Barker PB. Regions of neural dysfunction
associated with impaired naming of actions and objects in acute stroke.
Cogn Neuropsychol 2002a; 19: 523–34.
Hillis AE, Wityk R, Tuffiash E, Beauchamp NJ, Jacobs MA, Barker PB,
et al. Hypoperfusion of Wernicke’s area predicts severity of semantic
deficit in acute stroke. Ann Neurol 2001; 50: 561–6.
Hirsch J, Moreno DR, Kim KH. Interconnected large-scale systems for
three fundamental cognitive tasks revealed by functional MRI. J Cogn
Neurosci 2001; 13: 389–405.
Howard D, Gatehouse C. Distinguishing semantic and lexical word retrieval
deficits in people with aphasia. Aphasiology 2006; 20: 921–50.
Howard D, Orchard-Lisle V. On the origin of semantic errors in naming:
evidence from a case of a global aphasic. Cogn Neuropsych 1984; 1:
163–90.
Howard D, Patterson KE, Wise R, Brown WD, Friston K, Weiller C, et al.
The cortical localization of the lexicons. Brain 1992; 115: 1769–82.
Hsieh L, Rapp B. Functional magnetic resonance imaging of the cognitive
components of the spelling process. Brain Lang 2004; 91: 40–1.
Humphries C, Binder JR, Medler DA, Liebenthal E. Syntactic and semantic
modulation of neural activity during auditory sentence comprehension.
J Cogn Neurosci 2006; 18: 665–79.
Huther G, Dorfl J, Van der Loos H, Jeanmonod D. Microanatomic and
vascular aspects of the temporomesial region. Neurosurgery 1998; 43:
1118–36.
Jefferies E, Lambon Ralph MA. Semantic impairment in stroke aphasia
versus semantic dementia: a case-series comparison. Brain 2006; 129:
2132–47.
Kay J, Ellis AW. A cognitive neuropsychological case study of anomia –
implications for psychological models of word retrieval. Brain 1987; 110:
613–29.
Kemeny S, Xu J, Park GH, Hosey LA, Wettig CM, Braun AR. Temporal
dissociation of early lexical access and articulation using a delayed
naming task—an FMRI study. Cereb Cortex 2006; 16: 587–95.
Kempen G, Huijbers P. The lexicalization process in sentence
production and naming: indirect election of words. Cognition 1983;
14: 185–209.
Kober H, Moller M, Nimsky C, Vieth J, Fahlbusch R, Ganslandt O. New
approach to localize speech relevant brain areas and hemispheric
dominance using spatially filtered magnetoencephalography. Hum Brain
Mapp 2001; 14: 236–50.
Lesser R, Luders M, Dinner N, Dinner DS, Klem G, Hahn J, et al.
Electrical stimulation of Wernicke’s area interferes with comprehension.
Neurology 1986; 36: 658–63.
Levelt WJM. Speaking: from intention to articulation. Cambridge, MA:
MIT Press; 1989.
Levelt WJM. Producing spoken language: a blueprint of the speaker. In:
Brown CM, Hagoort P, editors. The neurocognition of language.
New York: Oxford University Press; 1999.
Levelt WJM, Schriefers H, Vorberg D, Meyer AS, Pechmann T, Havinga J.
The time course of lexical access in speech production: a study of
picture naming. Psychol Rev 1991; 98: 122–42.
Brain (2007), 130, 1408 ^1422
1421
Lhermitte E, Beauvois MF. A visual-speech disconnection syndrome
syndrome: report of a case with optic aphasia, agnosic alexia and colour
agnosia. Brain 1973; 96.
Maess B, Herrmann CS, Hahne A, Nakamura A, Friederici AD. Localizing
the distributed language network responsible for the N400 measured by
MEG during auditory sentence processing. Brain Res 2006; 1096:
163–72.
Manning L, Campbell R. Optic aphasia with spared action naming: a
description and possible loci of impairment. Neuropsychologia 1992; 30:
587–92.
Martin PI, Naeser MA, Doron KW, Bogdan A, Baker EH, Kurland J, et al.
Overt naming in aphasia studied with a functional MRI hemodynamic
delay design. Neuroimage 2005; 28: 194–204.
Marsh EB, Hillis AE. Cognitive and neural mechanisms underlying reading
and naming: evidence from letter-by-letter reading and optic aphasia.
Neurocase 2005; 11: 318–25.
Mesulam M-M. From sensation to cognition. Brain 1998; 121: 1013–52.
Miller GA, Johnson-Laird PN. Language and perception. Cambridge, MA:
Harvard University Press; 1976.
Mummery CJ, Patterson K, Hodges JR, Price CJ. Functional neuroanatomy
of the semantic system: divisible by what? J Cogn Neurosci 1998; 10:
766–77.
Mummery CJ, Patterson K, Wise RJ, Vandenbergh R, Price CJ, Hodges JR.
Disrupted temporal lobe connections in semantic dementia. Brain 1999;
122: 61–73.
Mummery CJ, Patterson K, Price CJ, Ashburner J, Frackowiak RS,
Hodges JR. A voxel-based morphometry study of semantic dementia:
relationship between temporal lobe atrophy and semantic memory. Ann
Neurol 2000; 47: 36–45.
Neumann-Haefelin T, Wittsack HJ, Wenserski F, Siebler M, Seitz RJ,
Modder U, et al. Diffusion and perfusion-weighted MRI: the DWI/PWI
mismatch region in acute stroke. Stroke 1999; 30: 1591–7.
Papathanassiou D, Etard O, Mellet E, Zago L, Mazoyer B,
Tzourio-Mazoyer N. A common language network for comprehension
and production: a contribution to the definition of language epicenters
with PET. Neuroimage 2000; 11: 347–57.
Plaut D, Shallice T. Perseverative and semantic influences on visual object
naming errors in optic aphasia: a connectionist account. J Cogn
Neurosci 1993; 5: 89–117.
Poldrack RA, Wagner AD, Prull MW, Desmond JE, Glover GH,
Gabrieli JDE. Functional specialization for semantic and phonological
processing in the left inferior prefrontal cortex. Neuroimage 1999; 10:
15–35.
Price CJ, Devlin JT. The myth of the visual word form area. Neuroimage
2003; 19: 473–81.
Price CJ, Devlin JT. The pros and cons of labeling a left occipitotemporal
region: ‘‘The visual word form area’’. Neuroimage 2004; 22: 477–9.
Price CJ, Mummery CJ, Moore CJ, Frakowiak RS, Friston KJ. Delineating
necessary and sufficient neural systems with functional imaging studies
of neuropsychological patients. J Cogn Neurosci 1999; 11: 371–82.
Price CJ, Devlin JT, Moore CJ, Morton C, Laird AR. Meta-analyses of
object naming: effect of baseline. Hum Brain Mapp 2005a; 25: 70–82.
Price CJ, Thierry G, Griffiths T. Speech-specific auditory processing: where
is it? Trends Cogn Sci 2005b; 9: 271–6.
Price CJ, McCrory E, Noppeney U, Mechelli A, Moore CJ, Biggio N, et al.
How reading differs from object naming at the neuronal level.
Neuroimage 2006; 29: 643–8.
Rapcsak SZ, Beeson PM. The role of left posterior inferior temporal cortex
in spelling. Neurology 2004; 62: 2221–9.
Rapp BC, Goldrick M. Interactivity and discreteness in spoken word
production. Psychol Rev 2000; 107: 460–99.
Rapp BC, Benzing L, Caramazza A. The autonomy of lexical orthographic
representations. Cogn Neuropsychol 1997; 14: 71–104.
Raymer A, Foundas AL, Maher LM, Greenwald ML, Morris M, Rothi LG,
et al. Cognitive neuropsychological analysis and neuroanatomical
correlates in a case of acute anomia. Brain Lang 1997; 58: 137–56.
1422
Brain (2007), 130, 1408 ^1422
Riddoch MJ, Humphreys GW. Visual optic processing in optic aphasia:
A case of semantic access agnosia. Cogn Neuropsych 1987; 4: 131–85.
Roelofs A. A spreading activation theory of lemma retrieval in speaking.
Cognition 1992; 42: 107–42.
Roelofs A. A case for the lemma/lexeme distinction in models of speaking:
comment on Caramazza and Miozzo. Cognition 1998; 69: 219–30.
Sabsevitz DS, Medler DA, Seidenberg M, Binder JR. Modulation of
the semantic system by word imageability. Neuroimage 2005; 27:
188–200.
Saccuman MC, Cappa SF, Bates EA, Arevalo A, Della Rosa P, Danna M,
et al. The impact of semantic reference on word class: an fMRI study of
action and object naming. Neuroimage 2006; 32: 1865–78.
Saur D, Lange R, Baumgaertner A, Schraknepper V, Willmes K, Rijntjes M,
et al. Dynamics of language reorganization after stroke. Brain 2006; 129:
1371–84.
Scott SK, Leff AP, Wise RJS. Going beyond the information given:
a neural system supporting semantic interpretation. Neuroimage 2003;
19: 870–6.
Seidenberg M, McClelland JL. A distributed, developmental model of
visual word recognition and naming. Psychol Rev 1989; 96: 523–68.
Shallice T. Impairments of semantic processing: multiple dissociations. In:
Coltheart M, Sugihara G, Kaminaga T, Sugishita M, Interindividual
uniformity and variety of the ‘‘Writing center’’: a functional MRI study.
Neuroimage 2006; 32: 1837–49.
Sobesky J, Zaro Weber O, Lehnhardt FG, Hesselmann V, Thiel A,
Dohmen C, et al. Which time-to-peak threshold best identifies
penumbral flow? A comparsion of perfusion-weighted magnetic
J. DeLeon et al.
resonance imaging and position emission tomograpghy in acute
ischemic stroke. Stroke 2004; 35: 2843–7.
Thijs V, Adami A, Neumann-Haelfelin T, Moseley M, Marks M, Albers G.
Relationship between severity of MR perfusion deficit and DWI lesion
evolution. Neurology 2001; 57: 1205–11.
Tranel D. Impaired naming of unique landmarks is associated with left
temporal polar damage. Neuropsychology 2006; 20: 1–10.
Tranel D, Grabowski TJ, Lyon J, Damasio H. Naming the same entities
from visual or from auditory stimulation engages similar regions of left
inferotemporal cortices. J Cogn Neurosci 2005; 17: 1293–305.
Vigliocco G, Antonini T, Garett MF. Is ‘‘count’’ and ‘‘mass’’ information
available when the noun is not? An investigation of the tip of the tongue
states and anomia. J Mem Lang 1999; 40: 534–58.
Whatmough C, Cherkow H. Neuroanatomical aspects of naming. In: Hillis
AE, editor. Handbook of adult language disorders: integrating cognitive
neuropsychology, neurology, and rehabilitation. Philadelphia:
Psychology Press; 2002.
Wise RJS. Language systems in normal and aphasic human subjects:
functional imaging studies and inferences from animal studies. Br Med
Bull 2003; 65: 95–119.
Wise RJS, Scott SK, Blank C, Mummery CJ, Murphy K, Warbuton EA.
Separate neural subsystems within ‘‘Wernicke’s area’’. Brain 2001; 124:
83–95.
Zahn R, Huber W, Drews E, Erberich S, Krings T, Willmes K, Schwarz M.
Hemispheric lateralization at different levels of human auditory word
processing: a functional magnetic resonance imaging study. Neurosci
Lett 2000; 287: 195–8.