doi:10.1093/brain/awh659
Brain (2005), 128, 2858–2871
Right anterior superior temporal activation
predicts auditory sentence comprehension
following aphasic stroke
Jenny Crinion and Cathy J. Price
Wellcome Department of Imaging Neuroscience, Institute of Neurology, London, UK
Correspondence to: Jenny Crinion, Wellcome Department of Imaging Neuroscience, Institute of Neurology,
12 Queen Square, London WC1N 3BG, UK
E-mail: [email protected]
Previous studies have suggested that recovery of speech comprehension after left hemisphere infarction may
depend on a mechanism in the right hemisphere. However, the role that distinct right hemisphere regions play
in speech comprehension following left hemisphere stroke has not been established. Here, we used functional
magnetic resonance imaging (fMRI) to investigate narrative speech activation in 18 neurologically normal
subjects and 17 patients with left hemisphere stroke and a history of aphasia. Activation for listening to
meaningful stories relative to meaningless reversed speech was identified in the normal subjects and in each
patient. Second level analyses were then used to investigate how story activation changed with the patients’
auditory sentence comprehension skills and surprise story recognition memory tests post-scanning. Irrespective of lesion site, performance on tests of auditory sentence comprehension was positively correlated with
activation in the right lateral superior temporal region, anterior to primary auditory cortex. In addition, when
the stroke spared the left temporal cortex, good performance on tests of auditory sentence comprehension was
also correlated with the left posterior superior temporal cortex (Wernicke’s area). In distinct contrast to this,
good story recognition memory predicted left inferior frontal and right cerebellar activation. The implication of
this double dissociation in the effects of auditory sentence comprehension and story recognition memory is that
left frontal and left temporal activations are dissociable. Our findings strongly support the role of the right
temporal lobe in processing narrative speech and, in particular, auditory sentence comprehension following left
hemisphere aphasic stroke. In addition, they highlight the importance of the right anterior superior temporal
cortex where the response was dissociated from that in the left posterior temporal lobe.
Keywords: aphasia; auditory sentence comprehension; stroke
Abbreviations: CAT = comprehensive aphasia test; fMRI = functional MRI; SPM = statistical parametric mapping
Received March 17, 2005. Revised August 5, 2005. Accepted September 12, 2005. Advance Access publication October 18, 2005
Introduction
The neural basis of speech recovery following aphasic stroke
remains unknown. It is generally accepted that there will be
some degree of functional reorganization following cerebral
damage. This will either involve changes within the
pre-existing language network (e.g. Leff et al., 2002) or the
recruitment of supplementary brain areas in the left or right
hemisphere. Clinical observations combined with structural
neuroimaging have suggested that if the lesion is large the
prognosis is poor; particularly, it has been proposed, if there is
ventral extension of the infarct into the middle and inferior
temporal gyri (Selnes et al., 1983; Naeser et al., 1987; Hart
#
and Gordon, 1990). The inverse correlation between the size
of the left hemisphere lesion and the degree of language
recovery (Demeurisse et al., 1985; Heiss et al., 1993) has
led some to suggest that it is the left rather than right
hemisphere that plays an important role in recovery. However, there are also examples of patients with left hemisphere
strokes, whose language impairments were impaired further
when a second stroke damaged the right hemisphere (Basso
et al., 1989). This suggests that, at least in these patients,
the right hemisphere was also contributing to language
processing.
The Author (2005). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: [email protected]
Right activity predicts aphasics’ comprehension
Functional imaging studies of aphasic speech recovery have
differed in their conclusions concerning the contribution of
left and right hemisphere activation. This inconsistency may
partly reflect the tasks tested. During auditory repetition tasks,
for example, the majority of functional imaging studies have
emphasized that good language recovery depends on left
hemisphere activation, although right hemisphere activation
may be observed when left hemisphere regions are no longer
available (Heiss et al., 1997, 1999; Karbe et al., 1998; Kessler
et al., 2000). In contrast, during auditory comprehension
tasks, a greater emphasis has been placed on the role of
the right hemisphere. For example, when patients with left
temporal lobe damage listened to words presented at varying
rates, right posterior temporal activation was more correlated
with increasing word rate than in control subjects (Leff
et al., 2002). Right hemisphere activation in patients with
left temporal damage has also been shown to correlate with
response accuracy when subjects respond to simple and repetitive auditory instructions (Musso et al., 1999) or perform
semantic decisions on auditory words (Sharp et al., 2004).
Critically, however, these results do not necessarily reflect
a ‘laterality shift’ because the same right hemisphere areas
are also engaged by neurologically normal subjects
(Warburton et al., 1999; Zahn et al., 2002, 2004). This is
clearly illustrated in the study by Sharp et al. (2004) who
found that right anterior fusiform activation correlated
with performance accuracy in neurologically normal controls
as well as the patients.
An outstanding question that we wish to address is whether
right hemisphere activation contributes to long-term language functioning after aphasic stroke. Previous studies
have addressed this issue by investigating how activation
changes over the course of recovery (Cardebat et al., 2003;
Fernandez et al., 2004); by correlating activation with behavioural responses on the task used during scanning (Musso
et al., 1999; Blasi et al., 2002; Sharp et al., 2004) or by correlating activation with language assessments conducted
outside the scanner (Rosen et al., 2000; Blank et al., 2003;
Noppeney et al., 2005). Although significant activation
changes over time or within scanning session have been
observed, correlations between activation and language performance outside the scanner have not revealed significant
effects in aphasic patients. For example, neither Blank et al.
(2003) nor Rosen et al., (2000) observed a relationship
between right hemisphere activation and speech production
ability measured outside the scanner. These authors, therefore, concluded that right hemisphere activation may represent behavioural strategies (Rosen et al., 2000) or transcallosal
disinhibition (Blank et al., 2003) that do not necessarily
contribute to the recovery of speech production processes.
The role of the right hemisphere, however, is likely to be
critically dependent on the task. For example, according to
the model of right hemisphere language proposed by Zaidel
(1976), the right hemisphere should selectively contribute
to the recovery of language comprehension but not to the
recovery of language production.
Brain (2005), 128, 2858–2871
2859
In the present study, we investigated the brain regions that
contribute to speech comprehension processes by regressing
activation evoked when patients listened to short stories with
(i) auditory sentence comprehension and (ii) subsequent
memory abilities, measured outside the scanner. The functional MRI (fMRI) paradigm engaged subjects in automatic
‘on-line’ narrative speech comprehension without any
requirement to make a response. The intention was to reduce
activation related to working memory and other executive
functions that are not directly related to normal, on-line
comprehension of narrative speech. We predicted that the
activation pattern would vary, across subjects, according to
their auditory comprehension abilities. In addition, we assessed how activation changed with auditory recognition memory as measured by a post-scanning surprise story recognition
test. This allowed us to distinguish linguistic (auditory comprehension) from metalinguistic executive (auditory recognition memory) processes that are not directly related to
normal, on-line narrative speech comprehension. Although
speech comprehension proficiency may well depend on working memory capacity (Just and Carpenter, 1992; MacDonald
et al., 1992) and the speech comprehension problems of some
patients with aphasia have been attributed to verbal working
memory deficits (Hermann et al., 1992; Wilson and Baddeley,
1993; Aziz et al., 2002) previous functional imaging studies
have not dissociated activation that is related to understanding speech (primarily at a linguistic level) from executive
(metalinguistic) aspects of speech processing.
In summary, previous functional imaging studies have
reached divergent conclusions concerning the neural mechanisms of recovery and this may reflect the very variable
stimuli, tasks and patient groups that have been investigated.
It, therefore, remains unknown as to whether right hemisphere activation reflects a maladaptive strategy or a beneficial
one. In non-fluent patients, it has been proposed that the
pattern of overactivation may limit, rather than account
for, aphasia recovery (Belin et al., 1996; Rosen et al., 2000).
Each hemisphere may be important, depending on the type of
language behaviour and when it was examined (Weiller et al.,
1995; Basso et al., 1996; Belin et al., 1996; Mimura et al., 1998;
Cao et al., 1999; Ansaldo et al., 2002; Hund-Georgiadis et al.,
2002; Ansaldo et al., 2004). In this study we aimed to differentiate auditory sentence comprehension (linguistic) processes
from auditory recognition memory processes (executive,
metalinguistic) in a group of chronic aphasic patients by
regressing activation evoked when patients listened to short
stories with (i) auditory sentence comprehension and (ii)
subsequent memory abilities, measured outside the scanner.
Methods
Participants
Seventeen patients (12 males, mean 62 years, SE 2.7) and 18
neurologically normal subjects (8 males, mean 58 years, SE 2.8)
were studied. Each had English as their first language, was righthanded and gave informed consent to participate in the study.
2860
Brain (2005), 128, 2858–2871
J. Crinion and C. J. Price
Fig. 1 Infarct location in aphasic patients. The lesion overlap maps show the extent of damage across patients. Sagittal, transaxial and
coronal projections are displayed rendered onto a single-subject brain T1 template. The range of the colour scale relates to the absolute
number of patients in each group. In the patients with temporal lobe damage (Temporal patients), lesion overlap is highest in the left
superior temporal lobe. In the Control patients, lesion overlap is highest in the left motor cortices.
All patients acquired their aphasia following a single left hemisphere stroke. They had a volumetric MRI to aid co-registration
with their fMRI scan and to define their allocation to the appropriate
patient group based on the site of their lesions (Fig. 1). One group
of eight patients had left temporal lobe damage (Temporal patients);
mean age 58 years, mean time since infarct 41 months. The remaining nine patients who had left hemisphere lesions sparing the temporal lobes acted as an aphasia control group (Control patients);
mean age 66 years, mean time since infarct 48 months. There was no
significant difference between the groups in terms of age (P = 0.09,
SE = 4.3) or time of study after infarct (P = 0.69, SE = 17.9).
Patients’ language abilities were assessed using the comprehensive
aphasia test (CAT) (Swinburn et al., 2005). For single words, subsets
of items manipulate variables such as word frequency, imageability,
animacy and length, with other confounding variables controlled.
Imageability values were drawn from the MRC Psycholinguistic
Database (Coltheart, 1981) and word frequency from Francis and
Kucera (1982). In the sentence level subtests, clausal and syntactic
complexities have been manipulated (varying factors like sentence
‘reversibility’ and active/passive structures). With all spoken output
tasks in the CAT the focus of interest is in the linguistic rather than
motor aspects of speech production. Therefore, articulatory errors
not affecting the perceptual identity of the target are scored correct
so that patients with dysarthria would not be penalized.
Scanning stimuli
Subjects listened to two conditions during their fMRI scans: stories
(St) and reversed version of the same stories (RSt). The length of each
story was 64 words (25 s duration). All stories were recorded by a
single female narrator. A typical example is:
‘Yesterday one person was killed. There was a big storm. The
storm blew down many trees. John was driving home after work.
The storm was very bad. A tree fell on the road. John did not see the
tree. John’s car hit the tree. The police found John. The police called
the doctor. It was too late. John died on the way to the hospital.’
Using SoundEdit16 computer software the waveform for each
story was reversed (RSt) and used as the baseline condition. Reversing the speech signal destroys intelligibility while retaining the overall
spectrotemporal complexity of the original speech signal. The
reversed versions of the narratives were expected to control for
early acoustic processing of the speech signal in both left and
right superior temporal cortex. However, as reversing speech destroys normal segmental stress patterns, suprasegmental prosody and
voice identity, it was expected that the activation patterns in the
contrast of St with RSt would not reflect linguistic processing alone.
During each subject’s study they heard 32 stories (St) and 8
reversed stories (RSt), with each story and its reversed version
used only once per subject. The order of presentation was randomized, both within and between subjects. Presentation was binaural
with the volume set at a comfortable level for each subject. The
subjects were asked to simply listen and try to understand the
story. They were aware that the reversed stories were unintelligible
but were asked to pay attention to the sounds.
To ensure that the subjects had attended to the stimuli, an
eye-tracker was used to check that the subjects were awake throughout scanning acquisition; and a surprise story recognition test was
presented to each subject after scanning. In this post-scanning test,
Right activity predicts aphasics’ comprehension
subjects were presented with 38 individual phrases and asked to
indicate after each one whether they recognized it as being part of
a story played to them in the scanner or not.
fMRI scanning
A Siemens 1.5 T scanner was used to acquire T 2*-weighted echoplanar images with BOLD contrast. Each echoplanar image comprised 35 axial slices of 2.0 mm thickness with 1 mm inter-slice
interval and 3 · 3 mm in-plane resolution. Volumes were acquired
with an effective repetition time (TR) of 3.15 s/volume and the first
six (dummy) volumes of each run were discarded to allow for
T1 equilibration effects. A total of 460 volume images were taken
in two separate runs, each of 10 min duration. After the two functional runs, a T1-weighted anatomical volume image was acquired
for all subjects.
Behavioural analyses
All patients’ individual performance on the comprehensive battery of
language tests and the recognition memory test post-scan are tabulated in Table 1. To investigate group differences, the data were
entered into SPSS 10.0 (SPSS, Inc., Chicago, IL) and group means
for each behavioural variable were compared using independent
sample t-tests. Bivariate correlations estimated significance of
Pearson correlations between the behavioural variables. Significance
was set at P < 0.05 (two-tailed) for all analyses.
Statistical parametric mapping
Statistical parametric mapping was performed using SPM2 software
(Wellcome Department of Imaging Neuroscience, London, UK),
running under Matlab 6.5 (Mathworks Inc., Sherborn, MA, USA).
All volumes from each subject were realigned using the first as reference and resliced with sinc interpolation. Within SPM, using Imcalc
an 8 mm smoothed study specific template was created using the
images from both the normals and patients to aid the accuracy of the
normalization of this subject group rather than using the standard
SPM template based on normal healthy young volunteers. The functional images were then spatially normalized to the study specific
template (Friston et al., 1995a) with SPM2 normalization software,
using non-linear-basis functions with high regularization that limits
the local distortions within the image. Functional data were spatially
smoothed, with a 12 mm full width at half maximum isotropic
Gaussian kernel, to compensate for residual variability after spatial
normalization and to permit application of Gaussian random field
theory for corrected statistical inference.
First, the statistical analysis was performed in a subject-specific
fashion. To remove low-frequency drifts, the data were high-pass
filtered using a set of discrete cosine basis functions with a cut-off
period of 128 s. Each experimental condition was modelled independently by convolving each block with a synthetic haemodynamic
response function. The parameter estimates were calculated for all
brain voxels using the general linear model, and contrast images
comparing each story against reversed stories (i.e. the baseline)
were computed (Friston et al., 1995b). These subject-specific contrast
images were then entered into three separate second level ANOVAs
that identified:
(i) Group differences and commonalities in story activation.
Differences were identified by comparing story activation in
neurologically normal subjects, patients with left temporal
lobe damage and patients without left temporal lobe damage.
Commonalities were identified with the conjunction of patient
Brain (2005), 128, 2858–2871
2861
activation and the normal activation inclusively masked (P <
0.001 uncorrected) with the same contrasts.
(ii) The impact of auditory sentence comprehension on story
activation. Within a single analysis, each patient group was
modelled independently and story activation was regressed
(in each group separately) on the auditory sentence comprehension score from the CAT. This multiple regression within
ANOVA analysis enabled us to find common effects of auditory
sentence comprehension and differences between patient
groups. For both common and differential effects, we identified
the effect of sentence comprehension in conjunction with the
main effect of story activation, in order to limit the search to
areas that responded positively to intelligible speech. As neurologically normal control subjects are all at ceiling on the CAT
auditory sentence comprehension test, they were excluded from
the analysis.
Unless otherwise indicated, we report and discuss regions
that showed significant effects at P < 0.05 (corrected for multiple comparisons across the whole brain) with an extent
threshold, for each cluster, of 5 voxels.
(iii) Activation associated with good story recognition memory.
Within a single analysis, each patient group and the neurologically normal controls were modelled independently and story
activation (for each subject within each group) was regressed on
subsequent story recognition memory established from the
surprise recognition test. This multiple regression within
ANOVA analysis enabled us to find common effects of subsequent recognition memory as well as group differences. For
both common and differential effects, we identified the effect of
subsequent story recognition in conjunction with the main
effect of story activation, in order to limit the search to
areas that responded positively to the narrative speech.
Results
Behavioural data
The effect of left hemisphere lesions on
language function
Although many of the patients continued to have significant
difficulties on tests of speech production and verbal fluency,
all patients performed above chance on tests of auditory sentence comprehension (minimal score 73% accuracy) (see
Table 1).
A direct comparison of the patient groups indicated
lower performance for patients with left temporal lobe
damage on both auditory (P = 0 .009, SE: 1.9) and written
sentence comprehension (P = 0.002, SE: 2.7) tasks (see
Table 2). However, there was considerable variability within
groups and no clear relationship between performance on the
language tests, and the site and extent of damage (see Table 1).
For example, Patients 7 and 5 (WC and GH) had extensive left
temporal lobe damage but still scored 28/30 and 30/30 on
single word comprehension (chance = 7.5/30) and 25/32 and
23/32 on auditory sentence comprehension (chance = 8/32).
Analysis of the extent and site of the lesion was, therefore,
not a good predictor of language comprehension. There
were no significant correlations between age of the patients
or time post-stroke, and performance on any of the
language tests.
66
63
75
69
71
59
59
66
66
M
F
F
M
M
M
F
M
M
Frontal
Frontal
Frontal
Parietal
Parietal
Parietal
Subcortical
Subcortical
Subcortical
86
42
37
78
95
38
20
13
24
4
7
6
125
80
17
36
51
27
25
16
33
31
33
27
33
19
Mean = 32.29
SD = 3.26
Range = 24–37
(max = 38)
(n = 18)
Mean = 31.73
SD = 0.67
Range = 30–32
Cut off=29
(max = 32)
(n = 26)
Mean = 29.78
SD = 2.5
Range = 24–32
Cut off= 23
(max = 32)
(n = 23)
Mean = 29.63
SD = 0.79
Range = 27–30
Cut off= 27
(max = 30)
(n = 27)
Mean = 46.37
SD = 1.6
Range = 42–48
Cut off=43
(max = 48)
(n = 27)
Mean = 30.17
SD = 1.85
Range = 26–32
Cut off= 27
(max = 32)
(n = 23)
Mean = 29.15
SD = 1.35
Range = 25–30
Cut off = 25
(max = 30)
(n = 27)
18
n/c
21
20
27
12
18
30
32
32
15*
24*
29*
32
32
32
32
32
6*
28*
32
30
0*
32
25*
29
32
30
28
29
32
32
31
31
7*
7*
30
24
24
9*
29
12*
28
30
27*
28
30
30
30
29
29
6*
28
30
30
30
27*
24*
22*
48
48
31*
45
48
48
48
48
48
4*
16*
48
46
47
2*
37*
28*
Recognition
memory
post-scan
26*
32
30
24*
30
32
32
29
28
21*
27*
29
29
23*
18*
25*
16*
Written word Written
Repetition
comprehension sentence
single word
comprehension
28
30
29
30
26
30
30
28
28
20*
30
27
30
30
16*
28
26
Months
Auditory word Auditory
Naming
post-stroke comprehension sentence
comprehension
Brain (2005), 128, 2858–2871
The individual scores for each patient on the language subtests of the comprehensive aphasia test battery (CAT) and the surprise story recognition memory test post-fMRI-scanning. The
cut-off where mentioned is the score that at least 95% of normal subjects exceed; as a result the scores that are below the cut-off represent aphasic performance. Individual test scores
highlighted in bold italic typeface with an asterisk (*) represent aphasic performance. aST = anterolateral superior temporal lobe; pST = posterior superior temporal lobe; a + pST =
damage to both anterior and posterior superior temporal cortices; Recognition memory post-scan = surprise recognition memory test post-scanning; Temp. = patients with left temporal
lobe damage; n/c= not collected.
Non-aphasic
performance
AK
BC
DR
DC
PB
RJ
SB
TS
AM
pST
pST
pST
aST
a + pST
a + pST
a + pST
a + pST
9
10
11
12
13
14
15
16
17
M
M
M
F
F
M
M
M
Control
Control
Control
Control
Control
Control
Control
Control
Control
64
69
48
34
41
67
75
64
1
2
3
4
5
6
7
8
Temp.
Temp.
Temp.
Temp.
Temp.
Temp.
Temp.
Temp.
GE
GB
VW
LH
GH
MG
WC
DD
Patient Patient Age
Sex Lesion
number initials (years)
location
Patient
group
Table 1 Patients’ behavioural profiles
2862
J. Crinion and C. J. Price
Right activity predicts aphasics’ comprehension
Brain (2005), 128, 2858–2871
2863
Table 2 Group language comprehension data
Patient group
Temporal (n = 8)
Control (n = 9)
Maximum score
Auditory word
comprehension
Auditory sentence
comprehension
Written word
comprehension
Written sentence
comprehension
Mean
SE
Mean
SE
Mean
SE
Mean
SE
25.9
28.8
30
1.8
0.5
23.5*
29.2*
32
1.7*
0.9*
24.6
29
30
2.9
1.1
17.8*
30.4*
32
3.5*
0.5*
The table displays mean scores and standard error of the patient groups with left superior temporal lobe damage (Temporal patients)
and those without damage to the left temporal lobe (Control patients) on auditory and written single word and sentence comprehension
testing. We performed independent t-tests for each variable, and significance was set at P < 0.05 (two-tailed). The Control patients
were as a group significantly better on tests of auditory and written sentence comprehension than the Temporal patients (P = 0.009, SE: 1.9;
P = 0.002, SE: 2.7, respectively), these scores are highlighted in bold italic typeface with an asterisk (*). There was no difference between
the patient groups on tests of single word comprehension, with both groups’ mean performance >80% accuracy on these tests.
Recognition memory performance on the surprise
story recognition test
The Normal group’s mean performance on the surprise recognition test post-scan (mean = 32.4/38; SE 0.8) was significantly higher (P < 0.0001) than the group of 16 patients (mean =
24.4/38; SE 1.7). Note: Patient 2, GB from the group of
patients with temporal lobe damage did not participate in
the surprise recognition test post-scan. The mean performance of the seven patients with temporal lobe damage was
lower (20.9/38; SE 2.3) than the nine patients without temporal lobe damage (27.1/38; SE 2.1) but this mean difference
did not quite reach significance (P = 0.06) due to the wide
variability within each group.
There were no significant correlations between recognition
memory scores and
(i) the age of the patients and normal subjects, (P = 0.45,
n = 34),
(ii) the time post-infarction in the patients (P = 0.27, n = 16) or
(iii) sentence comprehension (P = 0.41, n = 16).
However, patient performance on the post-scan recognition
test was significantly correlated with auditory single word
comprehension (P = 0.04, n = 16).
Functional imaging data
Activation for listening to stories relative to baseline
In the Normal group, listening to stories relative to baseline,
(reversed stories) activated lateral left temporal cortex, with
five main peaks extending from posterolateral to anterolateral
temporal lobe centred on the superior temporal sulcus (STS),
and at the junction of the superior temporal gyrus (STG) with
the inferior parietal lobe (IPL). Activation in the right
superior temporal cortex did not extend into the STG/IPL
junction, with four main peaks in the lateral temporal lobe.
Additional activations were observed bilaterally in the ventral
temporal lobe, within the occipito-temporal sulcus separating
the inferior temporal gyrus (ITG) and fusiform gyrus (FG),
and in five regions of the left frontal and bilateral motor
cortices (Fig. 2, a; Table 3, a).
The Control patients, i.e. those without temporal lobe damage, activated the same system as the Normal group, apart
from two regions in bilateral motor cortices that the Normal
group activated significantly more (Fig. 2, b1 and b2; Table 3, b
and b2).
The patients with left superior temporal lobe damage activated the same right temporal lobe system as the Normal group
(Fig. 2, c1; Table 3, c). However, as expected they had significantly less activation within the left superior temporal cortices, within three main peaks extending from the STG to
posterolateral and anterolateral temporal lobe centred on
the STS (Fig. 2, c2, c3; Table 3, c2 and c3). Analyses of individual patients revealed that left anterior temporal activation
was significantly reduced even when this region was spared in
patients with posterior temporal damage, see Fig. 3: Patient
GE (c). Likewise, left posterior temporal activation was
significantly reduced even when this region was spared in
patients with left anterior temporal damage, see Fig. 3:
Patient 4, LH (c).
Neither the Control patients nor the patients with temporal
lobe damage activated any region significantly more than
the Normal group. In particular, there was no evidence
of a difference in any right temporal region to suggest a
‘laterality shift’ of function in the patients with left temporal
lobe damage.
Auditory sentence comprehension
For both patient groups, there was a positive correlation
between auditory sentence comprehension and activation
in the right lateral anterior temporal lobe (x = +62, y =
8, z = 16; Z = 6.0) with no significant effect of group
in this area (see Fig. 5). For the Control patients, there was
also a positive correlation between auditory sentence comprehension and activation in the left lateral posterior temporal
cortex (x = 46, y = 56, z = +6; Z = 4.7). This effect was not
observed in the patients with temporal lobe damage and the
difference between the patient groups was significant at
P < 0.001 uncorrected (Z = 3.9). There were no significant
negative correlations.
2864
Brain (2005), 128, 2858–2871
J. Crinion and C. J. Price
Fig. 2 Regional activation for stories > baseline. fMRI activation for the contrast Speech (St) > baseline (RSt) in (a) Normal subjects, (b1)
Control patients and (c1) patients with temporal lobe damage (Temporal patients); (b2) and (c2) illustrate where Normal subjects activated
significantly more than Control and Temporal patients, respectively; (c3) illustrates where Control patients activated significantly more
than Temporal patients. For each group, results are shown in colour rendered onto the SPM2 single subject brain template. Details
regarding the location and significance of activations are given in Table 3.
Subsequent recognition memory from surprise story
recognition test post-scan
Positive correlations between activation and performance on
the surprise story recognition test post-scan, common to controls and patients, were observed in the left prefrontal cortex
(x = 42, y = +8, z = +20; Z = 5.32) and right cerebellum (x =
+16, y = 80, z = 40; Z = 5.17) (Fig. 4). There were no
significant differences between subject groups. There were no
significant negative correlations.
of patients had left temporal lobe damage (Temporal
patients); another had left hemisphere damage that spared
the left temporal lobe (Control patients). We found that auditory sentence comprehension ability was positively correlated
with (i) bilateral temporal activation when the temporal lobes
were spared; but (ii) only a right anterior temporal region
when the left temporal lobe was damaged. Below, we discuss
the implications of our results for understanding speech comprehension following aphasic stroke.
Discussion
Lesion site and speech comprehension
deficits
Clinical observations over many decades have demonstrated
that verbal comprehension, both lexical and sentential
semantics, is largely lateralized to the left perisylvian cortices
in most individuals. Nevertheless, speech comprehension
abilities can be very well preserved following left hemisphere
lesions. We investigated whether good speech comprehension
was related to left or right hemisphere activation. One group
Auditory sentence comprehension was significantly more
impaired in the group of patients with temporal damage
compared with the group of patients without temporal
damage. However, within the group of patients with temporal
damage, there was variability in both the lesion site and
the speech comprehension scores. On auditory single word
comprehension tasks, all patients scored significantly above
Z
6
42
42
8
20
30
26
28
46
56
56
56
42
36
32
2
10
12
50
60
54
56
z
28
46
18
16
6
2
14
10
24
28
24
8
2
14
18
20
ns
1.7
ns
ns
4.3
4.8
3.7
5.5
3.7
2.7
5.8
6.6
5.3
5.9
6.1
7.5
7.0
7.6
7.0
Z
x
46
56
42
52
56
54
54
58
6
–
–
–
–
0
–
–
–
26
28
32
0
10
10
–
–
–
y
z
50
2
14
10
22
28
28
18
(c) Temporal pts
ns
ns
ns
ns
2.7
ns
ns
ns
2.2
2.6
4.7
5.1
4.4
4.8
ns
ns
ns
2.3
ns
Z
x
38
30
46
46
y
6
38
–
36
18
–
–
–
–
–
–
–
–
–
–
–
–
–
–
z
28
22
64
52
2.2
ns
1.7
4.8
5.0
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
ns
Z
(b2) Normal > Control pts
x
38
36
30
56
46
46
56
56
56
54
60
54
60
58
54
y
38
42
36
10
0
10
20
28
26
10
48
38
10
6
8
z
28
28
22
42
50
18
16
4
2
28
8
2
14
18
26
2.7
3.8
3.1
3.6
2.3
2.6
3.8
2.2
1.7
ns
ns
ns
3.1
ns
7.4
6.0
6.5
5.7
4.8
Z
(c2) Normal > Temporal pts
x
36
48
56
56
48
58
60
50
62
62
58
y
–
42
–
–
–
8
20
30
–
–
36
4
–
–
48
40
6
6
10
z
28
18
16
6
8
22
8
0
16
18
18
ns
1.9
ns
ns
ns
3.8
2.1
4.0
ns
ns
1.9
1.7
ns
ns
4.6
5.8
5.0
5.0
5.1
Z
(c3) Control pts > Temporal
pts
The first three columns show the coordinates and Z-scores for the condition effect (St > RSt) in (a) the Normal group (visualized in Fig. 2a), (b), the patient Control group (visualized in
Fig. 2b), and (c) the patients with temporal lobe damage (visualized in Fig. 2c). The next three columns (b2, c2 and c3) show the coordinates and Z-scores for the condition effect (St > RSt)
where the Normal group activated significantly more than Control patients (b2) and Temporal patents (c2), and where the Control patients activated significantly more than the Temporal
patients ( c3), visualized in Fig. 2 (b2), (c2) and (c3), respectively. Coordinates are given as x, y and z according to the atlas of Talairach and Tournoux (1988). Normal = normal subject
group; Control pts = patients whose lesions spared the left temporal cortices; Temporal pts = patients whose lesions involved the left superior temporal cortices. Significant effects at
P < 0.05 (corrected for multiple comparisons across the whole brain) are shown in bold typeface.
–
–
–
48
40
8
6
8
60
50
60
58
58
y
x
z
x
y
(b) Control pts
(a) Normal
Left superior temporal cortices
1
60
48
8 8.3
2
54
38
2 8.1
3
60
8
14 8.3
4
58
6
18 8.2
5
58
8
20 7.4
Right superior temporal cortices
1
50
34
8 7.8
2
58
4
22 8.2
3
54
10
28 8.0
4
54
10
24 7.6
Left frontal cortices
1
46
10
18 4.6
2
56
20
16 7.4
3
56
28
4 4.6
4
56
26
2 5.7
5
42
28
14 5.9
Left motor cortices
1
46
0
50 7.1
Right motor cortices
1
56
10
42 5.3
Left ventral temporal cortex
1
36
42
28 5.3
2
30
36
22 5.4
Right ventral temporal cortex
1
38
38
28 4.6
Region
Table 3 Regional activation for stories > baseline
Right activity predicts aphasics’ comprehension
Brain (2005), 128, 2858–2871
2865
2866
Brain (2005), 128, 2858–2871
J. Crinion and C. J. Price
Fig. 3 Location of temporal damage and abnormal activation in patients GE and LH. Sagittal (x = 58) and coronal (y = 14 and 50) slices
are from the MRI scans of each patient after spatial normalization and indicate the extent of the lesions and the validity of the normalization.
Arrows highlight the areas of temporal damage. Reduced activation (Stories > baseline) compared with neurologically Normal subjects
are shown below in yellow on the same MRI slices (Normal > GE/LH).
chance (8/32 = 25%) with the lowest score (16/32 = 50%)
coming from Patient 6 (MG) who had both anterior and
posterior superior temporal lobe damage. On the auditory
sentence comprehension task, again, all patients scored above
chance with the lowest score (16/32) coming from Patient 8
(DD) following extensive left superior temporal damage.
The remainder of the patients made relatively few errors
on our word and sentence comprehension tests even when
left temporal lobe damage was extensive. For example,
sentence comprehension remained excellent following
damage to either posterior (Patient 1: GE) or anterior
(Patient 4: LH) temporal cortices as well as in Patient 7
(WC) who had lost virtually all his anterior and posterior
superior temporal cortices. These observations do not
reveal any appreciable relationship between site of
damage and speech comprehension ability. Instead, they
suggest that, when the left temporal lobe is damaged, areas
outside the left temporal lobe may be sustaining speech
comprehension.
Narrative speech activation in
control subjects
When listening to narrative speech compared with reversed
speech, our neurologically Normal subjects and Control
patients activated bilateral posterior and anterior temporal
regions and the left inferior frontal and premotor cortices.
The role of the right temporal activation in speech comprehension remains unclear. As the reversal of speech destroys
normal stress patterns, intonation, prosody and many
features that signal unique voice identity, the bilateral
anterior temporal signals may not be attributed exclusively
to linguistic processing of speech. Many authors have associated right temporal lobe activation in normal subjects
predominantly with processing of the non-verbal components (intonation, prosody, voice identity, etc.) of speech
(Buchanan et al., 2000; Belin et al., 2002; Mitchell et al.,
2003; Wildgruber et al., 2004). This lateralization of different functions is the usual clinical view. However, there is
Right activity predicts aphasics’ comprehension
Brain (2005), 128, 2858–2871
2867
Fig. 4 Activation increases with story comprehension. Regional activation (stories > baseline) increases with performance on the auditory
sentence comprehension test, from the CAT, are illustrated (above) in colour on models of the brain from SPM2 (P < 0.05 corrected).
Two main peaks were observed in the right (x = 64, y = 10, z = 14) and left (x = 50, y = 44, z = 0) superior temporal lobes.
The x-axes on the graphs (below) indicate the individual subjects’ scores on the auditory sentence comprehension test from the CAT; the
y-axes indicate the mean centred activation for stories more than baseline. In the right temporal region there is a common effect for
Control patients (black diamonds) and patients with temporal lobe damage (red circles), with no significant difference in the
regression slopes (black slope, Rsq = 0.47; red slope, Rsq = 0.73). In the left temporal region, there was a positive correlation for the
Control patients but this effect was not observed in the patients with temporal lobe damage. The regression slope was stronger for
Control patients (black slope, Rsq = 0.66) than patients with temporal lobe damage (red slope, Rsq = 0.0009) and the difference between
the patient groups was significant at P < 0.001 uncorrected (Z = 3.9).
good evidence for participation of the right hemisphere
in lexical semantic functions in normal subjects (Ellis and
Shepherd, 1974; Hines, 1975; Day, 1977, 1979; Young and
Ellis, 1985) and in patients with semantic dementia who
are more impaired in lexical semantic processing when
anterior and ventral temporal atrophy is bilateral (Chan
et al., 2001; Galton et al., 2001; Lambon Ralph et al., 2001).
The effect of left temporal lobe damage
on narrative speech activation
As expected, left temporal lobe damage significantly reduced
left superior temporal activation compared with the neurologically Normal subjects and the Control patients. More
remarkably, analyses of individual patients revealed that
left anterior temporal activation was significantly reduced
even when this region was spared in patients with posterior
temporal damage (see Patient GE in Fig. 3). Likewise, left
posterior temporal activation was significantly reduced
even when this region was spared in patients with anterior
temporal damage (see Patient LH in Fig. 3). The results from
these two patients whose lesions involved either posterior
or anterior superior temporal cortices suggest a mutual
dependence between left anterior and posterior superior
temporal lobe activation when subjects are listening to narrative speech.
The consequence of left temporal lobe damage was also
reflected in the observation that activation for narrative
speech was strongly right lateralized (see Fig. 2, c1). This
apparent laterality shift (left dominant activation in Normals
2868
Brain (2005), 128, 2858–2871
and Control patients right lateralized activation following
left temporal lobe damage), did not reflect compensatory
right hemisphere activation because examination of activation in right hemisphere regions revealed that the patients
with left temporal lobe damage activated within the normal
range. This cannot be attributed to abnormalities in the
BOLD response, because the structural integrity of the
right hemisphere was not compromised in any of the patients.
The results, therefore, suggest that right superior temporal
activation was dissociated from that in the left superior
temporal lobe. This contrasts with previous observations
that have shown how damage to the left frontal operculum
can enhance activation in the right frontal operculum
during speech production tasks (Rosen et al., 2000; Blank
et al., 2003).
The effect of auditory sentence
comprehension score on narrative
speech activation
Right lateralized activation following left temporal lobe damage suggests that narrative speech comprehension might be
sustained by the right hemisphere. Moreover, in some patients
(e.g. Patient 7, WC), who had little remaining left temporal
cortex, right hemisphere activation appeared to be sufficient
for good auditory single word and sentence comprehension.
However, there are well-recognized difficulties when drawing
such inferences. First, the absence of left hemisphere activation may be due to a lack of sensitivity in damaged cortex
(i.e. it could be a null result). Second, there is always a tradeoff between engaging patients in an executive paradigm that
requires a motor response versus engaging patients in more
naturalistic speech comprehension paradigms that do not
involve a motor response. The disadvantage of the former
is that performance on the decision-making aspects of the
task cannot be controlled across subjects. The disadvantage
of the latter is that we do not have a measure of how much of
our narratives were understood while the subjects were in the
scanner.
Our conclusion that the right hemisphere contributes
to speech comprehension is, therefore, dependent on our
observation that right anterior temporal activation could
be predicted not by lesion site but on the basis of auditory
sentence comprehension ability measured outside the scanner
(see Fig. 4). This correlational approach is becoming increasingly popular in functional imaging studies of patient populations. For example, Shaywitz et al. (2003) have shown that
left occipito-temporal activation increases with reading ability
in neurologically normal and dyslexic children; while
Noppeney et al. (2005) have shown that right hemisphere
activation correlates with reading ability following left
anterior temporal lobe resection. Perhaps surprisingly, the
aphasia literature has not previously reported correlations
between right hemisphere activation and speech comprehension and production abilities measured outside the scanner.
J. Crinion and C. J. Price
For example, neither Rosen et al. (2000) nor Blank et al.
(2003) found any relationship between right frontal activation
and speech production measures. Our findings, in the context
of speech comprehension, therefore, support previous suggestions that the right hemisphere might be able to sustain
speech comprehension better than speech production (Zaidel,
1976; Zaidel et al., 2000).
In patients with left temporal lobe damage, auditory sentence comprehension ability was only predicted by activation
in a right lateral superior temporal region, anterior to primary
auditory cortex. Activation in the same right anterior temporal area also predicted auditory sentence comprehension
ability in patients with lesions outside the left temporal lobe,
but in addition, these patients also showed a correlation
between left posterior superior temporal activation and auditory sentence comprehension performance. One interpretation of these findings could be that auditory sentence
comprehension relies on a bilaterally-distributed system.
This is demonstrated in terms of bilateral activations in Normals and Control patients. The reduced comprehension performance of the group of patients with left temporal lobe
damage would, therefore, be reflected in reduced, total temporal lobe activation—in this case the remaining activation of
the right temporal region, with sentence comprehension performance predicted by the sum of left and right temporal
activation.
However, this does not explain our data. In the patients
with left temporal lobe damage there was no negative correlation between the remaining activation in the left temporal
lobe with the degree of right temporal activation. Patients
with left temporal lobe damage and good sentence comprehension did not have higher right temporal responses than the
Control patients (who had both left and right activation), and
patients with bilateral temporal activation (e.g. Patients 16
and 17) did not have better sentence comprehension than
patients with only right temporal activation (e.g. Patients 3
and 4). An alternative interpretation is that, in this group of
chronic patients with left temporal damage, the response in
the right temporal lobe is independent of that in the damaged
left temporal lobe. In other words, the results suggest that in
the chronic phase post-stroke the effect in the right anterior
temporal lobe is independent of that in the left posterior
temporal lobe: the Control patients showed both effects
while patients with left temporal lobe damage only showed
the effect in the right hemisphere.
The anterior location of the right hemisphere correlation is
also interesting. The literature on aphasic stroke patients
emphasizes the importance of the posterior superior temporal
and inferior parietal cortices in speech comprehension
(Alexander et al., 1989) as do functional imaging studies of
aphasia following left hemisphere lesions (Leff et al., 2002;
Musso et al., 1999). However, these studies were not able to
segregate speech perception from auditory comprehension
processes. In contrast, we focused on auditory sentence comprehension in the regression analysis, which highlighted the
role of the right anterior superior temporal cortex (and the left
Right activity predicts aphasics’ comprehension
Brain (2005), 128, 2858–2871
2869
Fig. 5 Activation increases with story recognition. Regional activation (stories > baseline) increases with performance on the surprise story
recognition memory test post-scan for all subject groups are shown in colour on models of the brain from SPM2 (P < 0.05 corrected). Two
main peaks were observed, in the right cerebellum (x = 16, y = 80, z = 40) and in the left prefrontal cortex (x = 42, y = 8, z = 20). The
x-axes in the graphs indicate the individual subjects’ scores on the surprise story recognition memory test post-scan; while the y-axes
indicate the mean centred activation for stories more than baseline. No significant difference in the regression slopes for Normals (green
stars), Control patients (black tirangles) or patients with temporal lobe damage (red circles) was observed in either region. Green slopes
illustrate Normal regression slopes (right Rsq = 0.32; left Rsq = 0.28). Black slopes illustrate patient regression slopes, Control and
Temporal lobe damaged patients combined as one group (right Rsq = 0.26; left Rsq = 0.26).
posterior superior temporal cortex when it was spared in
Control patients). The anterior location of the right
hemisphere effect is consistent with studies of semantic
dementia that show progressive and ultimately profound
loss of speech comprehension following bilateral anterior
and ventral temporal lobe atrophy (Mummery et al., 1999;
Chan et al., 2001; Galton et al., 2001).
The effect of story recognition memory
on narrative speech activation
In distinct contrast to the correlation of auditory sentence comprehension with narrative speech activation,
performance on the surprise story recognition test postfMRI-scanning predicted left inferior frontal and right cerebellar activation (see Fig. 5). This was observed across all
subject groups, Normal and both patients groups, with no
significant differences between groups, even though error
rates were higher in the patients than the normal controls.
The implication of this double dissociation in the effects
of auditory sentence comprehension and story recognition
memory is that left frontal and left temporal activations
are dissociable. For example, prefrontal and motor activation
may be associated with working memory and other executive
functions not directly related to normal, on-line speech
processing (Crinion et al., 2003). Indeed, left frontal lesions
do not typically impair simple narrative speech comprehension; and the Control patients included in the present
study, three of whom had left frontal lesions, performed
as a group significantly better on language comprehension
tests than the patients with left temporal lobe lesions.
Together the dissociation in the effects of speech comprehension and story recognition memory suggest that left
inferior frontal activation during speech processing may
be associated with ‘top-down’ rather than ‘bottom-up’
processing.
2870
Brain (2005), 128, 2858–2871
Implications for long-term language
functioning after aphasic stroke
Our results indicate that the right superior temporal cortex
does play a role in auditory sentence comprehension following aphasic stroke, at least, in the chronic phase. The fact that
the patients’ deficits with speech production were far more
enduring (see Table 1) is consistent with the model of right
hemisphere language proposed by Zaidel (1976) in which the
right hemisphere should selectively contribute to the recovery
of language comprehension but not to the recovery of language production. Future studies can, therefore, investigate
whether right anterior temporal activation changes longitudinally over time during the acute or chronic stage of recovery.
For example, it may be the case that recovery of speech comprehension depends on the pre-morbid function of the right
anterior temporal lobe. Alternatively, the function of the
right anterior temporal lobe may depend on experience/
therapy.
It has been proposed that the left posterior STS and supratemporal plane provide a system whereby speech sequences
are briefly held in memory and can be repeated by projection
forwards of the encoded speech sequences to the caudal frontal lobe via the superior longitudinal fasciculus (Hickok et al.,
2000; Wise et al., 2001). Thus, the normal left lateralization of
speech comprehension systems may be more a reflection of
left superior temporal connections with speech production
faculties within the left frontal lobe rather than purely auditory speech processing systems per se. To fully dissociate these
possibilities will require additional behavioural data from
patients with right superior temporal lesions. It will also be
important to investigate whether successful speech comprehension rehabilitative techniques are associated with
increased responsiveness of the right anterolateral superior
temporal lobe.
Conclusions
In this study, we have shown that:
(i) Auditory speech comprehension both at the single word
and sentence level can be relatively well preserved following left anterior and posterior superior temporal cortical damage. This suggests that alternative neuronal
mechanisms are available to support speech comprehension following aphasic stroke.
(ii) Lesions involving the left superior temporal cortices result in significantly less activation not only in the area of
lesioned cortex but also ipsilaterally along the length of
the superior temporal lobe even in structurally intact
cortices. This suggests that, for processing narrative
speech, left posterior and anterior superior temporal
cortices respond together.
(iii) Auditory sentence comprehension abilities, as measured
outside the scanner, predicted the level of activation in
the left posterior and right anterior superior temporal
cortices. When the left superior temporal region was
J. Crinion and C. J. Price
damaged, the correlation with sentence comprehension
was only observed in the right anterior superior temporal
cortex. This suggests that right superior temporal
responses are dissociated from those in the left superior
temporal lobe.
These results have implications for the development of rehabilitative and pharmacological approaches to the treatment of
aphasic stroke. Once cerebral infarction is established, rehabilitative techniques, whether behavioural or pharmacological,
must be directed at intact cortices. Following extensive left
temporal lobe infarction, we are proposing that right anterolateral, superior temporal cortices, may be a target for such
therapy.
Acknowledgements
This work was funded by the Wellcome Trust. We thank
John Ashburner, Uta Noppeney and Andrea Mechelli for
their advice and assistance.
References
Alexander MP, Hiltbrunner B, Fischer RS. Distributed anatomy of transcortical sensory aphasia. Arch Neurol 1989; 46: 885–92.
Ansaldo AI, Arguin M, Roch LA. The contribution of the right cerebral
hemisphere to the recovery from aphasia: a single longitudinal case
study. Brain Lang 2002; 82: 206–22.
Ansaldo AI, Arguin M, Lecours AR. Recovery from aphasia: a longitudinal
study on language recovery, lateralization patterns, and attentional
resources. J Clin Exp Neuropsychol 2004; 26: 621–7.
Aziz MA, McKenzie JC, Wilson JS, Cowie RJ, Ayeni SA, Dunn BK. The human
cadaver in the age of biomedical informatics. Anat Rec 2002; 269:
20–32.
Basso A, Gardelli M, Grassi MP, Mariotti M. The role of the right hemisphere
in recovery from aphasia. Two case studies. Cortex 1989; 25:
555–66.
Basso A, Corno M, Marangolo P. Evolution of oral and written confrontation
naming errors in aphasia. A retrospective study on vascular patients. J Clin
Exp Neuropsychol 1996; 18: 77–87.
Belin P, van Eeckhout P, Zilbovicius M, Remy P, Francois C, Guillaume S,
et al. Recovery from nonfluent aphasia after melodic intonation therapy:
a PET study. Neurology 1996; 47: 1504–11.
Belin P, Zatorre RJ, Ahad P. Human temporal-lobe response to vocal sounds.
Brain Res Cogn Brain Res 2002; 13: 17–26.
Blank SC, Bird H, Turkheimer F, Wise RJ. Speech production after stroke: the
role of the right pars opercularis. Ann Neurol 2003; 54: 310–20.
Blasi V, Young AC, Tansy AP, Petersen SE, Snyder AZ, Corbetta M. Word
retrieval learning modulates right frontal cortex in patients with left frontal
damage. Neuron 2002; 36: 159–70.
Buchanan TW, Lutz K, Mirzazade S et al. Recognition of emotional prosody
and verbal components of spoken language: an fMRI study. Brain Res Cogn
Brain Res 2000; 9: 227–38.
Cao Y, Vikingstad EM, George KP, Johnson AF, Welch KM. Cortical language
activation in stroke patients recovering from aphasia with functional MRI.
Stroke 1999; 30: 2331–40.
Cardebat D, Demonet JF, De Boissezon X, Marie N, Marie RM, Lambert J,
et al. Behavioral and neurofunctional changes over time in healthy and
aphasic subjects: a PET Language Activation Study. Stroke 2003; 34:
2900–6.
Chan D, Fox NC, Scahill RI, Crum WR, Whitwell JL, Leschziner G, et al.
Patterns of temporal lobe atrophy in semantic dementia and Alzheimer’s
disease. Ann Neurol 2001; 49: 433–42.
Right activity predicts aphasics’ comprehension
Coltheart M. The MRC psycholinguistic database. Q J Exp Psychol 1981; 33A:
497–505.
Crinion JT, Lambon-Ralph MA, Warburton EA, Howard D, Wise RJ.
Temporal lobe regions engaged during normal speech comprehension.
Brain 2003; 126: 1193–201.
Day J. Right-hemisphere language processing in normal right-handers. J Exp
Psychol Hum Percept Perform 1977; 3: 518–28.
Day J. Visual half-field word recognition as a function of syntactic class and
imageability. Neuropsychologia 1979; 17: 515–9.
Demeurisse G, Capon A, Verhas M. Prognostic value of computed tomography in aphasic stroke patients. Eur Neurol 1985; 24: 134–9.
Ellis HD, Shepherd JW. Recognition of abstract and concrete words presented
in left and right visual fields. J Exp Psychol 1974; 103: 1035–6.
Fernandez B, Cardebat D, Demonet JF, Joseph PA, Mazuax JM, Barat M, et al.
Functional MRI follow-up study of language processes in healthy subjects
and during recovery in a case of aphasia. Stroke 2004; 35: 2171–6.
Francis, WN, Kucera, H. Frequency analysis of English usage: Lexicon and
grammar. Boston: Houghton Mifflin; 1982.
Friston KJ, Ashburner J, Frith CD, Poline J-B, Heather JD, Frackowiak RSJ.
Spatial registration and normalization of images. Hum Brain Mapp 1995a;
2: 165–88.
Friston KJ, Holmes AP, Worsley KJ, Poline J-B, Frith CD, Frackowiak RSJ.
Statistical parametric maps in functional imaging: a general linear
approach. Hum Brain Mapp 1995b; 2: 189–210.
Galton CJ, Patterson K, Graham K, Lambon-Ralph MA, Williams G,
Antoun N, et al. Differing patterns of temporal atrophy in Alzheimer’s
disease and semantic dementia. Neurology 2001; 57: 216–25.
Hart J, Gordon B. Delineation of single-word semantic comprehension
deficits in aphasia, with anatomical correlation. Ann Neurol 1990; 27:
226–31.
Heiss WD, Kessler J, Karbe H, Fink GR, Pawlik G. Cerebral glucose metabolism as a predictor of recovery from aphasia in ischemic stroke. Arch
Neurol 1993; 50: 958–64.
Heiss WD, Karbe H, Weber-Luxenburger G, Herholz K, Kessler J, Pietrzyk U,
et al. Speech-induced cerebral metabolic activation reflects recovery from
aphasia. J Neurol Sci 1997; 145: 213–7.
Heiss WD, Kessler J, Thiel A, Ghaemi M, Karbe H. Differential capacity of left
and right hemispheric areas for compensation of poststroke aphasia. Ann
Neurol 1999; 45: 430–8.
Hermann BP, Seidenberg M, Haltiner A, Wyler AR. Adequacy of language
function and verbal memory performance in unilateral temporal lobe
epilepsy. Cortex 1992; 28: 423–33.
Hickok G, Erhard P, Kassubek J, Helms-Tillery AK, Naeve-Velguth S, Strupp
JP, et al. A functional magnetic resonance imaging study of the role of left
posterior superior temporal gyrus in speech production: implications for
the explanation of conduction aphasia. Neurosci Lett 2000; 287: 156–60.
Hines D. Independent functioning of the two cerebral hemispheres for recognizing bilaterally presented tachistoscopic visual-half-field stimuli. Cortex
1975; 11: 132–43.
Hund-Georgiadis M, Lex U, Friederici AD, von Cramon DY. Non-invasive
regime for language lateralization in right- and left-handers by
means of functional MRI and dichotic listening. Exp Brain Res 2002;
145: 166–76.
Just MA, Carpenter PA. A capacity theory of comprehension: individual
differences in working memory. Psychol Rev 1992; 99: 122–49.
Karbe H, Thiel A, Weber-Luxenburger G, Herholz K, Kessler J, Heiss WD.
Brain plasticity in poststroke aphasia: what is the contribution of the right
hemisphere? Brain Lang 1998; 64: 215–30.
Kessler J, Thiel A, Karbe H, Heiss WD. Piracetam improves activated blood
flow and facilitates rehabilitation of poststroke aphasic patients. Stroke
2000; 31: 2112–6.
Lambon Ralph MA, McClelland JL, Patterson K, Galton CJ, Hodges JR. No
right to speak? The relationship between object naming and semantic
impairment: neuropsychological evidence and a computational model.
J Cogn Neurosci 2001; 13: 341–56.
Brain (2005), 128, 2858–2871
2871
Leff A, Crinion J, Scott S, Turkheimer F, Howard D, Wise R. A physiological
change in the homotopic cortex following left posterior temporal lobe
infarction. Ann Neurol 2002; 51: 553–8.
MacDonald MC, Just MA, Carpenter PA. Working memory constraints on
the processing of syntactic ambiguity. Cognit Psychol 1992; 24: 56–98.
Mimura M, Kato M, Kato M, Sano Y, Kojima T, Naeser M, et al. Prospective
and retrospective studies of recovery in aphasia. Changes in cerebral blood
flow and language functions. Brain 1998; 121: 2083–94.
Mitchell RL, Elliott R, Barry M, Cruttenden A, Woodruff PW. The neural
response to emotional prosody, as revealed by functional magnetic resonance imaging. Neuropsychologia 2003; 41: 1410–21.
Mummery CJ, Patterson K, Wise RJ, Vandenbergh R, Price CJ, Hodges JR.
Disrupted temporal lobe connections in semantic dementia. Brain 1999;
122: 61–73.
Musso M, Weiller C, Kiebel S, Muller SP, Bulau P, Rijntjes M. Traininginduced brain plasticity in aphasia. Brain 1999; 122: 1781–90.
Naeser MA, Helm-Estabrooks N, Haas G, Auerbach S, Srinivasan M. Relationship between lesion extent in ‘Wernicke’s area’ on computed tomographic scan and predicting recovery of comprehension in Wernicke’s
aphasia. Arch Neurol 1987; 44: 73–82.
Noppeney U, Price CJ, Duncan JS, Koepp MJ. Reading skills after left
anterior temporal lobe resection: an fMRI study. Brain 2005;
Jun;128(Pt 6): 1377–85.
Rosen HJ, Petersen SE, Linenweber MR, Synder AZ, White DA, Chapman L,
et al. Neural correlates of recovery from aphasia after damage to left inferior
frontal cortex. Neurology 2000; 55: 1883–94.
Selnes OA, Knopman DS, Niccum N, Rubens AB, Larson D. Computed
tomographic scan correlates of auditory comprehension deficits in aphasia:
a prospective recovery study. Ann Neurol 1983; 13: 558–66.
Sharp DJ, Scott SK, Wise RJ. Retrieving meaning after temporal lobe infarction: the role of the basal language area. Ann Neurol 2004; 56: 836–46.
ShaywitzSE,ShaywitzBA,FulbrightRK,SkudlarskiP,MenclWE,ConstableRT,
et al. Neural systems for compensation and persistence: young adult outcome of childhood reading disability. Biol Psychiatry 2003; 54: 25–33.
Swinburn, K, Baker, G, and Howard, D. CAT: The Comprehensive Aphasia
Test. New York: Psychology Press; 2005.
Talairach J, Tornoux M. Co-planar stereotaxic atlas of the human brain. New
York: Thieme Medical Publishers, 1988.
Warburton E, Swinburn K, Price CJ, Wise RJS. Mechanisms of recovery from
aphasia: evidence from positron emission tomographic studies. J Neurol
Neurosurg Psychiatr 1999; 66: 155–61.
Weiller C, Isensee C, Rijntjes M, Huber W, Muller S, Bier D, et al. Recovery
from Wernicke’s aphasia: a positron emission tomographic study. Ann
Neurol 1995; 37: 723–32.
Wildgruber D, Hertrich I, Riecker A, Erb M, Anders S, Grodd W, et al.
Distinct frontal regions subserve evaluation of linguistic and emotional
aspects of speech intonation. Cereb Cortex 2004; 14: 1384–9.
Wilson BA, Baddeley A. Spontaneous recovery of impaired memory span:
does comprehension recover? Cortex 1993; 29: 153–9.
Wise RJ, Scott SK, Blank SC, Mummery CJ, Murphy K, Warburton EA.
Separate neural subsystems within ‘Wernicke’s area’. Brain 2001; 124:
83–95.
Young AW, Ellis AW. Different methods of lexical access for words
presented in the left and right visual hemifields. Brain Lang 1985; 24:
326–58.
Zahn R, Huber W, Drews E, Specht K, Kemeny S, Reith W, et al. Recovery of
semantic word processing in transcortical sensory aphasia: a functional
magnetic resonance imaging study. Neurocase 2002; 8: 376–86.
Zahn R, Drews E, Specht K, Kemeny S, Reith W, Willmes K, et al. Recovery of
semantic word processing in global aphasia: a functional MRI study. Brain
Res Cogn Brain Res 2004; 18: 322–36.
Zaidel E. Auditory vocabulary of the right hemisphere following brain bisection or hemidecortication. Cortex 1976; 12: 191–211.
Zaidel E, Kasher A, Soroker N, Batori G, Giora R, Graves D. Hemispheric
contributions to pragmatics. Brain Cogn 2000; 43: 438–43.