ARTICLE IN PRESS
Brain and Language xxx (2003) xxx–xxx
www.elsevier.com/locate/b&l
Activations of ‘‘motor’’ and other non-language structures
during sentence comprehension
Laurie A. Stowe,a,* Anne M.J. Paans,b Albertus A. Wijers,c and Frans Zwartsd
b
d
a
Department of Linguistics, Graduate School of Behavioral and Cognitive Neurosciences, University of Groningen, Groningen, The Netherlands
PET Center University Hospital, Graduate School of Behavioral and Cognitive Neurosciences, University of Groningen, Groningen, The Netherlands
c
Department of Experimental Psychology, Graduate School of Behavioral and Cognitive Neurosciences, University of Groningen, Groningen,
The Netherlands
Department of Dutch Linguistics, Graduate School of Behavioral and Cognitive Neurosciences, University of Groningen, Groningen, The Netherlands
Accepted 26 August 2003
Abstract
In this paper we report the results of an experiment in which subjects read syntactically unambiguous and ambiguous sentences
which were disambiguated after several words to the less likely possibility. Understanding such sentences involves building an initial
structure, inhibiting the non-preferred structure, detecting that later input is incompatible with the initial structure, and reactivating
the alternative structure. The ambiguous sentences activated four areas more than the unambiguous sentences. These areas are the
left inferior frontal gyrus (IFG), the right basal ganglia (BG), the right posterior dorsal cerebellum (CB) and the left median superior
frontal gyrus (SFG). The left IFG is normally activated when syntactic processing complexity is increased and probably supports
that function in the current study as well. We discuss four hypotheses concerning how these areas may support comprehension of
syntactically ambiguous sentences. (1) The left IFG, right CB and BG could support articulatory rehearsal used to support the
processing of ambiguous sentences. This seems unlikely since the activation pattern associated with articulatory rehearsal in other
studies is not similar to that seen here. (2) The CB acts as an error detector in motor processing. Error detection is important for
recognizing that the wrong sentence structure has been chosen initially. (3) The BG acts to select and sequence movements in the
motor domain and in cognitive domains may serve to inhibit competing and completed plans which is not unlike inhibiting the
initially non-preferred structure or ‘‘unchoosing’’ the initial choice when incompatible syntactic input is received. (4) The left median
SFG is relevant for the evaluation of plausibility. Evaluating the plausibility of the two possibilities provides an important basis for
choosing between them. The notion of the use of domain general cognitive processes to support a linguistic process is in line with
recent suggestions that the a given area may subserve a specific cognitive task because it carries out an appropriate sort of computation rather than because it supports a specific cognitive domain.
! 2003 Elsevier Inc. All rights reserved.
Keywords: Ambiguity resolution; Sentence processing; Cerebellum; Basal ganglia; Superior frontal gyrus
1. Introduction
Sentence comprehension is a very complex task,
which can vary in its cognitive demands according to the
nature of the sentences and contexts involved. It is traditional to see language processing as involving a specialized brain system, consisting of a relatively small
assemblage of specialized processing modules handling
*
Corresponding author. Fax: +31- 50-363-6855.
E-mail address: [email protected] (L.A. Stowe).
0093-934X/$ - see front matter ! 2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0093-934X(03)00359-6
specific parts of the process. The involvement of anterior
and posterior language areas has been confirmed by
many neuroimaging studies (e.g., Caplan, Alpert, &
Waters, 1998; Stowe et al., 1998). However other evidence from neuroimaging shows that language and
other cognitive domains such as music perception and
motor imagery (Binkofsky et al., 2000; Koelsch et al.,
2002) activate common regions. This can be explained if
tasks which share subcomponents (in this example, sequencing) are processed more according to the specific
type of computation than to the cognitive nature of the
task as a whole (Doya, 2000).
ARTICLE IN PRESS
2
L.A. Stowe et al. / Brain and Language xxx (2003) xxx–xxx
In this article we will discuss the results of an experiment investigating the processing of syntactically ambiguous sentences. Most neuroimaging studies of
sentence processing have focused on violations or
complexity; few studies of ambiguity resolution have
been published (but see Cooke et al., 2001). We have
good cognitive models of the processing of these sentences based on behavioral and event-related potential
experiments, but relatively little understanding of those
areas in the brain which support specific aspects of this
process; the experiment was designed to provide an
initial localization of those areas of the brain which are
activated during ambiguity resolution. Under the classical model, it would be predicted that these activations
would most likely be in language areas such as Broca!s
and Wernicke!s areas; this is not the result which we
found. The results of this study thus provide a good
example of activations during language processing of
areas whose involvement is better explained in terms of
more general processes.
First let us consider the cognitive operations which
are involved in processing syntactically ambiguous sentences. In a sentence such as The red drops. . ., drops can
be either a noun or a verb. If the sentence continues
. . .fell onto the floor, then drops is a noun; if it continues
. . .onto the floor, it!s a verb. Syntactic disambiguation
may not occur for several words. However, in many
cases, one of the potential syntactic structures is simpler
or more frequent than the other, allowing an immediate
selection on the basis of structural preference (e.g.,
Frazier, 1978; Spivey & Tanenhaus, 1998). Semantic
plausibility may also be sufficient for a choice between
the structures (e.g., The wood drops. . .). Choice between
the two structures on the basis of syntactic and semantic
factors is thus one aspect of processing these sentences
(Stowe, 1991; Tanenhaus, Spivey-Knowlton, & Hanna,
2000). Secondly, later context may make it clear that the
initial choice was wrong. In The red drops from the dye
bottle onto the floor, it turns out that drops must be a
verb, even though on the basis of the frequency of red as
an adjective, it may have initially appeared that drops
was probably a noun. When an error of this type is
detected, reanalysis must occur, involving reactivation
or reconstruction of the alternative structure.
The goal of the current study was to identify candidate areas which might be involved in these processes
of choice, error detection and reanalysis. Subjects read
syntactically ambiguous sentences and unambiguous
controls. By disambiguating to the less preferred
structure, we ensured that subjects would carry out all
three processes. To determine which of these candidate
area carries out which component, further experiments
are necessary, but comparison with what is known
about the functions of the activated areas allows us to
formulate hypotheses about the contribution of each
area.
2. Study
2.1. Subjects
16 Dutch native speakers with no known neurological
or perceptual deficits (aged 19–47; mean ¼ 23; 8F, 8M)
volunteered for the study after giving informed consent
under a procedure approved by the University Hospital
Medical Ethics committee.
2.2. Materials
Each subject read lists of sentences while four PET
scans were made. Two lists consisted of eight ambiguous
Dutch sentences followed by several unambiguous fillers. Early in the sentence a word occurred which was
ambiguous with regard to syntactic category. The immediately following context was consistent with either
possibility so that the ambiguity remained unresolved
for at least four words (similar to The red drops from the
dye bottle onto. . . : in which the ambiguity is maintained
through the phrase in bold and only then resolves, to the
less preferred reading). A single word late in the sentence
could only form a grammatically correct sequence
within the less preferred alternative, defined operationally as the structure which was less frequently used to
complete the ambiguous context in a pretest (e.g.,
Complete: The red drops from the dye bottle _________).
A variety of category ambiguities were employed. In two
control lists, unambiguous sentences were presented.
These lists were matched with the ambiguous sentences
in pretest plausibility, and in word length and logarithmic word frequency. The syntactic structures of the
unambiguous sentences were the same as those of the
ambiguous sentences after ambiguity resolution (e.g., a
sentence like The girl came from the upstairs flat to the
party).
2.3. Procedure
Subjects were placed in a Siemens CTI 951/31 positron emission tomograph camera. At the beginning of
each scan, 1.86 GBq H15
2 O was injected as a bolus in
saline via a cannula placed in the right brachial vein.
Measurement began 23 s. after injection and continued
for 60 s. Subjects read one of the four lists of sentences
described above, presented word by word in the center
of a computer screen. Subjects were instructed to ‘‘read
the sentences attentively for comprehension’’; no further
task was carried out. The order of presentation was
varied so the four sentence lists occurred equally frequently as first, second, third or fourth scan. The two
ambiguous lists never occurred consecutively. Sentences
began seven seconds before measurements were started.
Each word remained on the screen for 750 ms. and was
immediately replaced by the following word; this speed
ARTICLE IN PRESS
3
L.A. Stowe et al. / Brain and Language xxx (2003) xxx–xxx
voxel fairly dorsal (Fig. 1, above left), right hemisphere
caudate nucleus (Fig. 1, above right), right hemisphere
posterior dorsal cerebellum (CB) (Fig. 1, bottom left)
and left hemisphere superior frontal gyrus (SFG)(Fig. 1,
bottom right). No increases were found to be more activated during reading of unambiguous sentences relative to ambiguous sentences.
was found to ensure comprehension in a pretest. Between sentences an asterisk appeared in the center of the
screen for 1800 ms.
2.4. Data analysis
Regional cerebral blood flow was estimated for each
voxel within the field of view of the camera and rescaled
to voxel size 2.2 " 2.2 " 2.4 mm. Using the Statistical
Parametric Mapping programs (Wellcome Institute for
Neuroimaging), data were corrected for movement between scans and the data were normalized into the stereotactic coordinate system of Talairach and Tournoux
(1988) in order to align across subjects (Friston et al.,
1995a). Since this alignment is unlikely to be entirely
successful, a Gaussian filter was also applied (20 mm in
the x and y dimensions and 12 in the z dimension); in
addition to reducing noise, this filter spreads the effects
of an activation spatially so that nearly aligned activations are more likely to coincide across subjects.
Regional cerebral blood flow was then compared
across the two conditions. Since multiple voxel-by-voxel
comparisons were made, a correction comparable to the
Bonferroni correction was used (Friston et al., 1995b).
Additionally, the number of contiguous activated voxels
at the threshold Z > 3:0 was also calculated. Since it is
not likely that a large cluster of voxels will be activated
by chance, a calculation was also made of the significance level of the size of the activation (Friston, Worsley, Frackowiak, Mazziotta, & Evans, 1994). In this
article, activations will only be regarded as significant if:
(1) the maximal voxel within an activation is significant
at a corrected P < :05 or (2) has a significant extent of
activation at P < :05.
3. Discussion
While processing syntactically ambiguous sentences,
readers must evaluate which possibility is more likely
and inhibit the less probable analysis; if incoming material is inconsistent with the initial choice, the error has
to be recognized and the alternative analysis will have to
be reactivated. In the current study, we have shown that
four areas of the brain (the left IFG, the right caudate
nucleus, the right CB and the left median SFG) are more
activated when these processes must be carried out than
when unambiguous sentences are read.
Only the left IFG has been classically associated with
language processing. This area has been shown to be
activated in a number of studies in which the complexity
of syntactic processing is manipulated (e.g., Caplan et
al., 1998; Stowe et al., 1998); it seems likely that this area
is involved in linguistic processing of the ambiguous
structures, although the area activated in this study extends somewhat higher than the area usually activated
by syntactic manipulations. The left IFG is also involved in motor planning and articulatory rehearsal,
which we will return to below.
The role of the other three regions is more problematic and we will concentrate on them in this discussion.
In the following discussion, we will first briefly review
evidence from lesion and patient studies that the CB and
basal ganglia (BG) may play some role in sentence
processing. Second, we will consider what is known of
the functions of these latter three areas and what these
functions might contribute to comprehension of syntactically ambiguous sentences.
2.5. Results
The areas in Table 1 showed significant increased
blood flow during processing of ambiguous sentences:
left hemisphere inferior frontal gyrus (IFG), primarily
BA 45 extending into BA 44 and 9, with the maximal
Table 1
Areas more activated during the reading of ambiguous sentences than of unambiguous sentences
Area
L inferior frontal gyrus (BA 45)
R caudate
R posterior dorsal cerebellum
L superior frontal (BA 8/9)
Max voxel
x
y
z
)48
18
38
)14
20
6
)80
36
28
16
)24
44
Z
Corrected P
Extent (in voxels)
P extent
4.70
4.60
4.28
4.30
.005
.007
.026
.024
854
296
631
709
.059
.433
.128
.097
Location of the maximal voxel is given in Talairach and Tournoux coordinates, in which X ¼ left ())/right(+) of midline; Y ¼ anterior (+)/
posterior()) to the anterior commissure; Z ¼ dorsal(+)/ventral()) to the plane of the anterior and posterior commissure; statistics are reported for the
maximal voxel (Z-value and P-value corrected for multiple comparisons and for the probability of a cluster of voxels of this extent reaching the
threshold of Z ¼ 3:0).
ARTICLE IN PRESS
4
L.A. Stowe et al. / Brain and Language xxx (2003) xxx–xxx
Fig. 1. Areas activated by syntactically ambiguous sentences Areas of activation (marked with an arrow) are shown projected onto an MRI template
in radiological convention (left ¼ right): Upper left, left inferior frontal gyrus; upper right, right basal ganglia; lower left, right posterior dorsal
cerebellum; and lower right, left superior frontal gyrus.
3.1. Evidence for a role for the cerebellum and basal
ganglia in sentence processing: A review of the literature
3.1.1. Cerebellum
There have been a number of reports of productive
and receptive agrammatism following cerebellar damage
(see Mari€en, Engelborghs, Fabbro, & De Deyn, 2001,
for a review). Cases showing syntactic symptoms typically involve right cerebellar lesions (Riva & Giorgi,
2000; although see Fabro, Moretti, & Bava, 2000, for
some exceptions) and frequently show relatively good
recovery. Morphosyntactic production errors and decreased mean length of utterance characterize these
cases; comprehension also tends to fail on sentences
which depend on syntactic structure.
One common explanation for sentence level deficits
associated with cerebellar lesions is that they affect left
frontal lobe function via diaschisis (damage in one
area leading to hypoactivity in connected areas).
However, in at least one case, it was explicitly shown
that linguistic deficits accompanying cerebellar dysfunction were not coupled with frontal hypoperfusion,
ruling out diaschisis (Gasparini et al., 1999). The
current results suggest that for during comprehension
of some sorts of ambiguous sentences, the CB plays a
relatively ‘‘active’’ role itself, which also argues against
ARTICLE IN PRESS
L.A. Stowe et al. / Brain and Language xxx (2003) xxx–xxx
a classic diaschisis interpretation of sentence comprehension deficits.
3.1.2. Basal ganglia
A number of cases of language deficits following lesions ‘‘more or less limited’’ to the BG have also been
reported in the literature (see Nadeau & Crosson, 1997; ,
for review). A number of reports include difficulties with
sentence production and comprehension (Alexander,
Naeser, & Palumbo, 1987;Fabbro, Clarici, & Bava,
1996; Hochsteinbach, Spaendonck, van Cools, Horstinck, & Mulder, 1998; Pickett, Kuniholm, Protopapas,
Friedman, & Lieberman, 1998). Additionally, a number
of studies have shown that patients with Parkinson!s,
degenerative disease primarily affecting the BG, have
difficulties with syntactic structure (Grossman, 1999;
Lieberman et al., 1992). Although most attention has
been directed to left BG lesions, Hochstein et al. (1998)
found that damage on either side was equally likely to
cause deficits in sentence comprehension and verbal
fluency.
3.1.3. Why are the basal ganglia and cerebellum not
consistently associated with language dysfunction?
It is clear that lesions of the BG and CB do not
consistently have consequences for sentence processing,
which suggests that possibly damage must occur to a
particular area within either structure. The CB in humans is a large structure with massive interconnectivity
with the cortex. It can be expected that different parts of
the CB participate in different cortical functions due to
the exact interconnectivity of the region involved; this is
concistent the fact that by far the majority of the lesions
associated with language production and comprehension problems are to the right cerebellar hemisphere,
which is interconnected to the left cerebral hemisphere.
The BG likewise receives input from—and sends
output via the thalamus to—multiple areas of the cortex.
Several attempts have been made to define the effects of
damage to specific regions in and around the BG, cf
Alexander et al. (1987). However, it should be noted
that Nadeau and Crosson (1997), limiting their inquiries
to cases of left striatocapsular infarction, found that
only about half of their patients showed any linguistic
disturbance and that the nature of the disturbance was
extremely variable in those who did, which is not consistent with this view.
3.2. A motor pathway?
On hearing that the IFG, the CB and the BG are
activated together, the first idea which comes to most
people!s minds is that they form parts of a motor
pathway and that their function is motor-related. If the
function is not obviously motor in nature, as in this
study, one option is that this motor pathway supports a
5
form of working memory which is necessary for the
task. Articulatory rehearsal is the most likely on the
assumption that this pathway primarily supports motor
processing.
It is known that articulatory rehearsal supports some
aspects of sentence processing, since articulatory suppression interacts with propositional complexity and/or
center-embedding (Caplan & Waters, 2000; Withaar,
2002). In the current task, with word by word presentation at a relatively slow rate, working memory demands are likely to be particularly high and retrieval of
preceding words in some format is necessary for revision. If tasks which involve articulatory rehearsal activate the same areas which are activated in the current
study, it would provide evidence that articulatory rehearsal is indeed involved in syntactic disambiguation,
probably for use in reanalysis.
There are, however, several problems with this account of both the experimental and the lesion data.
First, the right BG is not expected to be involved in a
circuit with either left frontal lobe or right CB.1 Second,
as can be seen in Table 2, a selection of articulatory
rehearsal tasks typically activate both frontal and cerebellar areas somewhat similar to those seen in the current study. However, they also activate premotor 6
bilaterally, do not activate the BG and the activation of
the CB is bilateral rather than unilateral, so that the
global pattern is quite different from that found in the
current study. Third, it has also been demonstrated that
short term memory deficits can be seen following right
cerebellar lesions without any apparent agrammatism
(Silveri, di Betta, Filippini, Leggio, & Molinari, 1998),
while Riva and Giorgi (2000) report sentence level
comprehension problems in patients who were able to
repeat complex sentences verbatim, also suggesting a
dissociation between articulatory rehearsal and sentence
processing deficits in cerebellar lesions. Taken together
these facts suggest first, that if a motor-like pathway
supporting working memory is involved here, it unlikely
to include the BG and second, that the representation
which is involved is not articulatory, but some other
higher-order cognitive form of representation. If this
hypothesis is to be interesting and explanatory, it is
necessary to determine what form of representation is
relevant.
As we noted earlier, both the CB and the BG exhibit
massive interconnectivity with various areas of the cortex. Their interaction with these areas is probably not
necessarily motor-based, but rather dependent on the
nature of the process in each cortical area. The strategy
of considering the kind of computations which are car1
This lack of connectivity between left frontal and right basal
ganglia also argues against Ullman et al.!s (1997) account of basal
ganglia involvement in agrammatism as resulting from a frontostriatal
network. For this reason we will not discuss this theory further here.
ARTICLE IN PRESS
6
L.A. Stowe et al. / Brain and Language xxx (2003) xxx–xxx
Table 2
Overview of activations reported for articulatory rehearsal studies
Study
Left
X
Inferior frontal gyrus
Paulesu, Frith, and Frackowiak (1993)
Awh et al. Exp1 (1996)
Smith, Jonides, and Koeppe, 1 (1996)
Awh et al. Exp1 (1996)
Smith et al. Exp1 (1996)
Awh et al. Exp2 (1996)
Smith et al. Exp 3 (1996)
Fiez, Raife, Balota, Schwarz, and Raichle,
1 (1996)
Smith et al. Exp 2 (1996)
Celsis, Agniel, D"emonet, and Marc-Vergnes
(1991)
Right
Y
Z
)46
)55
)55
)44
)44
)42
)42
)29
2
3
3
12
12
17
17
25
16
20
20
22
22
22
22
4
)37
n.a.
44
20
Mean
)44
15
19
Premotor 6
Awh et al. Exp 1 (1996)
Fiez et al. Exp 1 (1996)
Awh et al. Exp 2 (1996)
Awh et al. Exp 2 (1996)
Smith et al. Exp 3 (1996)
Awh et al. Exp 2 (1996)
)48
)48
)28
)28
)28
)24
)6
)6
1
1
1
3
40
40
52
50
52
52
Mean
)33
0
)28
)26
)62
)67
Cerebellum
Smith et al. Exp 3 (1996)
Smith et al. Exp 3 (1996)
Smith et al. Exp 1 (1996)
Awh et al. Exp 1 (1996)
Awh et al. Exp 2 (1996)
Awh et al. Exp 2 (1996)
Smith et al. Exp 3 (1996)
Fiez et al. (1996)
Fiez et al. (1996)
Paulesu et al., 1993
Mean
X
48
Y
Z
4
12
Comparison
Maintain letter lists–visual recognition
Letter recognition–match
Letter memory–match
Letter recognition–match
Letter memory–match
N-back–plain match
Verbal 2-back–match
Maintain word list– fix
Verbal n-back–spatial
List learning–listening
26
24
3
3
50
52
48
24
2
50
)52
)50
24
)62
)45
33
33
33
28
)60
)60
)60
)60
)25
)25
)25
)38
)26
)67
)50
)30
)17
)9
)18
)49
)77
)47
)54
)45
)24
)22
)16
27
14
)63
)60
)14
)16
)22
)60
)37
27
)61
)27
Letter recognition–match
Word list maintain–fix
N-back–plain match
N-back–rehearsal
Verbal 2-back–match
N-back–plain match
Verbal maintenance–verbal control
Verbal maintenance–verbal control
Verbal–verbal control
Letter recognition–match
N-back–search
N-back–rehearsal
Verbal maintenance–verbal control
Maintenance–fixation
Maintenance–fixation
Letter list–visual control
Locations are organized into left and right hemisphere activations of two areas within the frontal lobe and the cerebellum. The location of the
maximal voxel is given in Talairach and Tournoux coordinates (cf. Table 1) and the nature of the comparison which elicited the activation is shown in
the righthand column.
ried out by the CB and BG in motor processing may
reveal the nature of their contribution to other cognitive
functions. Recent evidence suggests that their contribution in these domains can be dissociated into error
monitoring in the CB and sequential processing in the
BG (Jueptner & Weiller, 1998).
3.3. Cerebellum as error detector
In motor tasks, the CB uses perceptual information
to monitor plans generated by the cortex for errors
(Molinari, Filippini, & Leggio, 2002). This function
underlies the role of the CB in motor learning (e.g.,
activation decreases as a motor task is learned, Friston,
Frith, Passingham, Liddle, & Frackowiak, 1992). One
possibility is that the CB is an error detector with regard
to many sorts of cortical processes, that is, that the CB is
well-suited for this sort of computation (Doya, 2000).
Recently evidence for this monitoring function has
been extended to another verbal domain. Fiez, Petersen,
Cheney, and Raichle (1992) suggested that the activation of the right lateral CB in word generation tasks
(e.g., verbal fluency, verb generation, stem completion,
translation, synonym generation) are all due to the need
to detect outputs which do not fit the criteria. Fiez et al.
(1992) showed that a right cerebellar patient showed
many errors in word generation tasks with no learning
or automatization of the task over time. Desmond,
Gabrieli, and Glover (1998) showed that the CB is
particularly activated during stem completion when
ARTICLE IN PRESS
L.A. Stowe et al. / Brain and Language xxx (2003) xxx–xxx
there are few compatible completions, so that most
options generated by the cortex will be errors. Crucially
Fiez et al. (1992) assumes that there is an input to the
CB from the cortex which can be compared to the
output in order to detect errors. For the word generation tasks, the criteria is more abstract than a motor
representation.
An error detection mechanism is also necessary for
various aspects of sentence processing. In production,
failure of monitoring for errors according to a syntactic
criterion would lead to morphosyntactic errors, which
are frequently seen in agrammatic cerebellar patients.
Failures in comprehension of ambiguous sentences may
also occur if errors are not detected efficiently. There is
little definite evidence for this possibility, but it makes
some clear, testable predictions about when cerebellar
activation similar to that found in the current study will
occur. For example, the posterior dorsal CB should not
be activated when there is no evidence against the initially preferred structure of a grammatically ambiguous
sentence. Further, any sentence with a similar ungrammaticality would be expected to activate this area. A
number of experiments have been carried out in which
ungrammatical sentences were presented to subjects.
The CB has not typically been activated in these studies;
however, the CB has not typically been scanned in these
studies either. (Moro et al., 2001) who scanned the
whole brain, report a right lateralized cerebellar activation for detection of morphosyntactic errors. Both
these predictions should be tested further in future research.
3.4. Basal ganglia as selection device
In the motor domain, the BG appear to play an important role in action selection (Bergman et al., 1998).
The final output of the striatum via the globus pallidus is
tonically inhibitory, diminishing the likelihood of any
action occurring; an action is chosen by ‘‘inhibiting inhibition.’’ In normal function, there seems to be a winner takes all result in this disinhibition, so that only one
action takes place (Bergman et al., 1998). In motor sequence planning, it is proposed that this process focuses
effort on the current action (‘‘choosing’’) and that loops
within the BG are responsible for the smooth change to
the following action involving an element of ‘‘unchoosing’’ or re-inhibiting the first movement (Onlo-or &
Winstein, 2001). From this viewpoint it is not surprising
that BG lesions and Parkinson!s disease frequently lead
to speech production errors (Lieberman et al., 1992;
Pickett et al., 1998).
More broadly, this sort of procedure would allow the
BG to adjudicate between competing actions which are
planned in various cortical systems or between competing plans within the same domain. ‘‘Action’’ selection
certainly applies in cognitive domains other than motor
7
output planning (Vakil, Kahan, Huberman, & Osimani,
2000). Godbout and Doyon (2000) demonstrated that
patients have more problems with irrelevant intrusions
when describing a familiar sequence of actions (e.g.,
getting up in the morning) than either normal controls
or frontal patients. ‘‘Action’’ selection is not even limited to sequencing. Copeland, Chenery, and Murdoch
(2001) reported that both Parkinson!s and BG lesion
patients had problems in selecting an appropriate
meaning for a semantically ambiguous word in sentence
context (e.g., He dug with the spade), showing priming
from the irrelevant meaning much later than normals.
Moretti et al. (2001) noted that Parkinson!s patients!
verbal fluency errors are frequently irrelevant intrusions
and that subthalamic nucleus stimulation, which improves motor symptoms, also leads to a decrease in intrusions.
Syntactically ambiguous sentences are clearly a phenomenon in which a selection has to be made. First, a
choice is made when the ambiguity is recognized and
there is sufficient evidence to make a decision. Inhibiting
the irrelevant possibility is important to prevent confusion. Secondly, when it turns out that there is an error, it
is necessary to ‘‘unchoose’’ the initially selected structure. The activation in the current study is consistent
with a role for the right BG in one or both of these
procedures. Recently Frisch, Kotz, von Cramon, and
Friederici (2003) have shown that left BG lesion patients
show no P600 to ungrammaticalities. The P600 is an
ERP response which is normally associated with recognizing an error and attempting to reanalyze the sentence to ‘‘fix’’ it. This suggests that the activation seen in
the current study may reflect some aspect of the reanalysis procedure (see also evidence from Friederici,
R€
uschemeyer, Hahne, & Fiebach, 2003 that the left BG
is activated by ungrammatical sentences). The evidence
that making an initial choice for semantically ambiguous words is also slowed in Parkinson!s patients and
right BG lesion patients (Copeland et al., 2001) suggests
that the initial choice may also be supported by this
area. The degree to which there are lateralization of
these processes in the BG is an interesting one for future
research.
3.5. Left median frontal lobe and evaluation
A large number of studies using language stimuli have
shown activation in the left median frontal lobe. The
tasks used were relatively diverse and the location quite
variable, so that it is not clear whether the same function
should be attributed to all of these activations, Nevertheless some clear generalizations can be made, Generally four sorts of language studies (see Table 3) have
given rise to left median frontal activations: (1) those
comparing probabilistic reasoning to deductive reasoning, (2) those comparing anomaly detection with another
ARTICLE IN PRESS
8
L.A. Stowe et al. / Brain and Language xxx (2003) xxx–xxx
Table 3
Sentence and text level studies reporting activation in the median frontal lobe
Study
X
Y
Z
Comparison
Bottini et al. (1994)
Broere et al. (1997)
Caplan, Alpert, and Waters (1999)
Ferstl and von Cramon (2001)
Fletcher et al. (1995)
Gallagher et al. (2000)
Goel, Grafman, Sadato, and Hallett (1995)
Goel et al. (1995)
Goel, Gold, Kapur, and Houle (1997)
Gusnard, Akbudak, Shulman, and Raichle (2001)
Gusnard et al. (2001)
Maguire, Frith, and Morrism (1999)
Osherson et al. (1998)
Osherson et al. (1998)
Stromswold, Caplan, Alpert, and Rauch (1996)
)12
)6
)2
)4
)12
)8
)12
)4
)12
)9
)3
2
)12
)14
)2
54
44
18
58
36
50
38
52
54
39
23
42
24
46
30
20
28
48
13
36
10
32
24
24
42
54
)22
44
32
48
)6
55
13
Sentence–word list (Plausibility–lexical decision)
Plausibility judgment–grammaticality judgment
Auditory object clefts–subject clefts (anomaly detection)
Coherent paragraphs–incoherent
Theory of mind stories–physical stories
Theory of mind story–non-theory of mind
Theory of mind vs inference of own
theory of mind–semantic retrieval
Inductive–deductive
Judgment emotion vs episodic content
Judgment emotion vs episodic content
Coherent story–incoherent story
Probability reasoning–logic
Anomaly detection–logic and probability judgments
Center-embedded–pseudoword sentences
(plausibility–lexical decision)
Evaluative > episodic > rehearsal
Zysset, Huber, Ferstl, and von Cramon (2002)
Location of the maximal voxel for each activation is given for each study in Talairach and Tournoux coordinates (cf. Table 1) and the nature of
the comparison which elicited the activation is shown in the right hand column.
task such as ungrammaticality detection or lexical decision, (3) theory of mind stories in which subjects need to
infer motivations as well as understanding physical
events, and (4) several studies requiring judgment of
emotional response. The activated conditions all require
either some evaluation of plausibility or use of probabilistic inferences where the judgment is not very clear
(Ferstl & von Cramon, 2001). Gusnard et al. (2001)
suggest that the function of this area is to support recruitment of personal experience. In any case, the function of this area seems to support higher-level semantic
process involved in evaluation of plausibility, whether or
not it carries out such processes itself.
The current study does not involve any explicit
component of semantic evaluation. However, covertly,
it does require the evaluation of the relative plausibility
of the two potential structures to choose between them.
Given the clear evidence that this area supports evaluation, it seems likely that this is the contribution it
makes in the current task.
One issue is the extent to which the activations reported here are epiphenomena. This activation may reflect increased eye movement during more difficult
processing conditions (Stromswold et al., 1996). Most of
the activations reported here appear too anterior to involve median wall frontal eye fields (in a selection of
studies involving eye movements similar to that reported
for articulatory rehearsal studies, the center of supplementary motor activation due to eye movements ranged
from )14 to +17 mm y and from +42 to +58 mm z).
Additionally, eye movements would be expected to be
bilateral, while most of these activations are left lateralized. Indeed, many of them do not show much if any
extension to the right hemisphere (cf. the current study
as shown in Fig. 1). Lastly, in the current study and the
other from our lab (Broere et al., 1997), sentences were
presented one word at a time in the middle of the screen,
which should eliminate rescanning in difficult sentences.
4. Conclusion
We have considered several hypotheses about the involvement of areas that are not classically considered to
be language areas in sentence comprehension. The current study confirmed that the BG and CB can be actively
involved in sentence processing under some specific circumstances, as suggested by patient data. The cognitive
operations which are necessary to understand syntactically ambiguous sentences (selection of a structure on the
basis of semantic plausibility, detecting input which does
not match this choice and reanalysis of the structure) are
consistent with recent suggestions that the BG and CB
are involved in action selection and error detection within
cognitive processes as well as in motor processing and
that the left median frontal lobe is supports evaluation.
More globally, this conclusion is consistent with recent
suggestions that the involvement of a specific area in a
process depends more on the nature of the computation
which must be carried out than on the specific cognitive
domain of the process which must be carried out.
Acknowledgments
This stiudy was supported by a CBR grant from the
University of Groningen and the PIONIER project The
Neurological Basis of Language, granted by the Neth-
ARTICLE IN PRESS
L.A. Stowe et al. / Brain and Language xxx (2003) xxx–xxx
erlands Organization for Scientific Research (NWO) to
the first author Thanks to Bauke de Jong, Evelyn Ferstl
and two anonymous reviewers for careful comments on
the manuscript. Thanks to Paulien Rijkhoek and Jonneke Brouw for help with materials and running subjects, to Remi Schmeits for scanning, and Antoon
Willemsen for help with analysis, and the PET Center
for technical support.
References
Alexander, M. P., Naeser, M. A., & Palumbo, C. L. (1987).
Correlations of subcortical CT lesion sites and aphasia profiles.
Brain, 110, 961–991.
Awh, E., Jonides, J., Smith, E. E., Schumacher, E. H., Koeppe, R. A.,
& Katz, S. (1996). Dissociation of storage and rehearsal in verbal
working memory: Evidence from positron emission tomography.
Psychological Science, 7, 25–31.
Bergman, H., Feingold, A., Nini, A., Raz, A., Slovin, H., Abeles, M.,
& Vaadia, E. (1998). Physiological aspects of information processing in the basal ganglia of normal and parkinsonian primates.
Trends in Neuroscience, 21, 32–38.
Binkofsky, F., Amunts, K., Stephan, K. M., Posse, S., Schormann, T.,
Freund, H.-J., Zilles, K., & Seitz, R. J. (2000). Broca!s region
subserves imagery of motion: A combined cytoarchitectonic and
fMRI study. Human Brain Mapping, 11, 273–285.
Bottini, G., Corcoran, R., Sterzi, R., Paulesu, E., Schenone, P.,
Scarpa, P., Frackowiak, R. S. J., & Frith, C. D. (1994). The role of
the right hemisphere in the interpretation of figurative aspects of
language: A positron emission tomography activation study. Brain,
117, 1241–1253.
Broere, C. A. J., Stowe, L. A., Wijers, A. A., Mulder, G., Koster, J., &
Vaalburg, W. (1997). A complex syntax processing area in the
temporal cortex. Journal of Cerebral Blood Flow and Metabolism,
17(Suppl. 1), S265.
Caplan, D., Alpert, N., & Waters, G. (1998). Effects of syntactic
structure and propositional number on patterns of regional
cerebral blood flow. Journal of cognitive Neuroscience, 10, 541–552.
Caplan, D., Alpert, N., & Waters, G. (1999). PET studies of syntactic
processing with auditory sentence presentation. NeuroImage, 9,
343–351.
Caplan, D., & Waters, G. (2000). Verbal working memory and
sentence comprehension. Brain and Behavioral Science, 22, 77–126.
Celsis, P., Agniel, A., D"emonet, J.-F., & Marc-Vergnes, J. P. (1991).
Cerebral blood flow correlates of word list learning. Journal of
Neurolinguistics, 6, 253–272.
Cooke, A., Zurif, E. B., De Vita, C., Alsop, D., Koenig, P., Detre, J.,
Gee, J., Pi~
nango, M. M., Balogh, J., & Grossman, M. (2001).
Neural basis for sentence comprehension: Grammatical and
shorterm-memory components. Human Brain Mapping, 15, 80–94.
Copeland, D. A., Chenery, H. J., & Murdoch, B. E. (2001). Discourse
priming of homophones in individuals with dominant nonthalamic
subcortical lesions, cortical lesions and Parkinson!s disease. Journal
of Clinical and Experimental Neuropsychology, 23, 538–556.
Desmond, J. E., Gabrieli, J. D. E., & Glover, G. H. (1998).
Dissociation of frontal and cerebellar activity in a cognitive task:
Evidence for a distinction between selection and search. Neuroimage, 7, 368–376.
Doya, K. (2000). Complementary roles of basal ganglia and cerebellum in learning and motor control. Current Opinion in Neurobiology, 10, 732–739.
Fabbro, F., Clarici, A., & Bava, A. (1996). Effects of left basal ganglia
lesions on language production. Perceptual and Motor Skills, 82,
1291–1298.
9
Fabro, F., Moretti, R., & Bava, A. (2000). Language impairment in
patients with cerebellar lesions. Journal of Neurolinguistics, 13,
173–188.
Ferstl, E. C., & von Cramon, D. Y. (2001). The role of coherence and
cohesion in text comprehension: and event-related fMRI study.
Cognitive Brain Research, 11, 325–340.
Fiez, J. A., Petersen, S. E., Cheney, M. K., & Raichle, M. E. (1992).
Impaired non-motor learning and error detection associated with
cerevellar damage. Brain, 115, 155–178.
Fiez, J. A., Raife, E. A., Balota, D. A., Schwarz, J. P., & Raichle, M. E.
(1996). A positron emission tomography study of the short-term
maintenance of verbal information. Journal of Neuroscience, 16, 808–
822.
Fletcher, P. C., Happ"e, F., Frith, U., Baker, S. C., Dolan, R. J.,
Frackowiak, R. S. J., & Frith, C. D. (1995). Other minds in the
brain: a functional imaging study of ‘‘theory of mind’’in story
comprehension. Cognition, 57, 109–128.
Frazier, L. (1978). On comprehending sentences: Syntactic parsing
strategies. Bloomington, IN: Indiana University Linguistics Club.
Friederici, A. D., R€
uschemeyer, S.-A., Hahne, A., & Fiebach, C. J.
(2003). The role of left inferior frontal and superior temporal cortex
in sentence comprehension: Localizing syntax and semantic processes. Cortex, 13, 170–177.
Frisch, S., Kotz, S. A., von Cramon, D. Y., & Friederici, A. D. (2003).
Why the P600 is not just a P300: The role of the basal ganglia.
Clinical Neurophysiology, 114, 336–340.
Friston, K. J., Worsley, K. J., Frackowiak, R. S. J., Mazziotta, J.
C., & Evans, A. C. (1994). Assessing the significance of focal
activations using their spatial extent. Human Brain Mapping, 1,
214–220.
Friston, K. J., Ashburner, J., Frith, C. D., Poline, J. B., Heather, J. B.,
& Frackowiak, R. S. J. (1995a). Spatial registration and normalisation of images. Human Brain Mapping, 2, 165–189.
Friston, K. J., Holmes, A. P., Worsley, K. J., Poline, J. B., Frith, C.
D., & Frackowiak, R. S. J. (1995b). Statistical parametric mapping
and functional imaging: A general linear approach. Human Brain
Mapping, 2, 189–210.
Friston, K. J., Frith, C. D., Passingham, R. E., Liddle, P. F., &
Frackowiak, R. S. (1992). Motor practice and neurophysiological
adaptation in the cerebellum: A positron tomography study.
Proceedings of the Royal Society of London—Series B: Biological
Sciences, 48, 223–228.
^ , F., Brunswick, N., Fletcher, P. C., Frith,
Gallagher, H. L., HappeA
U., & Frith, C. D. (2000). Reading the mind in cartoons and
stories: an fMRI study of "theory of mind! in verbal and nonverbal
tasks. Neuropsychologia, 38, 11–21.
Gasparini, M., Di Piero, V., Ciccarelli, O., Cacioppo, M. M.,
Pantano, P., & Lenzi, L. (1999). Linguistic impairment after
right-cerevellar stroke: A case report. European Journal of
Neurology, 6, 353–356.
Godbout, L., & Doyon, J. (2000). Defective representation of
knowledge in Parkinson!s Disease: Evidence from a script-production task. Brain and Cognition, 44, 490–510.
Goel, V., Gold, B., Kapur, S., & Houle, S. (1997). he seats of reason?
An imaging study of deductive and inductive reasoning. NeuroReport, 8, 1305–1310.
Goel, V., Grafman, J., Sadato, N., & Hallett, M. (1995). Modeling
other minds. NeuroReport, 6, 1741–1746.
Grossman, M. (1999). Sentence processing Parkinson!s disease. Brain
and Cognition, 40, 387–413.
Gusnard, D. A., Akbudak, E., Shulman, G. L., & Raichle, M. E.
(2001). Median prefrontal cortex and self-referential mental
activity: Relation to a default mode of brain function. Proceedings of the National Academy of Sciences (USA), 98, 4259–
4274.
Hochsteinbach, J., Spaendonck, K. P. M., van Cools, A. R.,
Horstinck, M. W., & Mulder, T. (1998). Cognitive deficits
ARTICLE IN PRESS
10
L.A. Stowe et al. / Brain and Language xxx (2003) xxx–xxx
following stroke in the basal ganglia. Clinical Rehabilitation, 12,
514–520.
Jueptner, M., & Weiller, C. (1998). A review of differences between
basal ganglia and cerebellar control of movements as revealed by
functional imaging studies. Brain, 121, 1437–1449.
Koelsch, S., Giunter, T. C., von Cramon, D. Y., Zysset, S.,
Lohmann, G., & Friederici, A. D. (2002). Bach speakes: A
cortical ‘‘language-network’’serves the processing of music. Neuroimage, 17, 956–966.
Lieberman, P., Kako, E., Friedman, J., Tajchman, G., Feldman, L. S.,
& Jiminez, E. B. (1992). Speech production, syntax comprehension,
and cognitive deficits in Parkinson!s disease. Brain and Language,
43, 169–189.
Maguire, E. A., Frith, C. D., & Morrism, R. G. M. (1999). The
functional neuroanatomy of comprehension and memory: The
importance of prior knowledge. Brain, 122, 1839–1850.
Mari€en, P., Engelborghs, S., Fabbro, F., & De Deyn, P. P. (2001). The
lateralized linguistic cerebellum: a review and a new hypothesis.
Brain Language, 79, 580–600.
Molinari, M., Filippini, V., & Leggio, M. G. (2002). Neuronal
plasticity of interrelated cerebellar and cortical networks. Neuroscience, 111, 863–870.
Moretti, R., Torre, P., Antonello, R. M., Capus, L., Gioulis, M.,
Marsala, S. Z., Cazzato, G., & Bava, A. (2001). Effects on cognitive
abilities following subthalamic nucleus stimulation in Parkinson!s
disease. European Journal of Neurology, 8, 726–727.
Moro, A., Tettamanti, M., Perani, D., Donati, C., Cappa, S. F., &
Fazio, F. (2001). Syntax and the brain: Disentangling grammar by
selective anomalies. Neuroimage, 13, 110–118.
Nadeau, S. E., & Crosson, B. (1997). Subcortical aphasia. Brain and
Language, 58, 355–402.
Onlo-or, S., & Winstein, C. J. (2001). Function of the "direct! and
"indirect! pathways of the basal ganglia loop: Evidence from
reciprocal aiming movements in Parkinson!s disease. Cognitive
Brain Research, 10, 329–332.
Osherson, D., Perani, D., Cappa, S., Schnur, T., Grassi, F., & Fazio,
F. (1998). Distinct brain loci in deductive versus probabilistic
reasoning. Neuropsychologia, 36, 369–376.
Paulesu, E., Frith, C. D., & Frackowiak, R. S. J. (1993). The neural
components of the verbal component of working memory. Nature,
362, 342–344.
Pickett, E. R., Kuniholm, E., Protopapas, A., Friedman, J., &
Lieberman, P. (1998). Selective speech motor, syntax and cognitive
deficits associated with bilateral damage to the putamen and the
head of the caudate nucleus: A case study. Neuropsychologia, 36,
173–188.
Riva, D., & Giorgi, C. (2000). The cerebellum contributes to higher
functions during development: Evidence from a series of children
surgically treated for posterior fossa tumours. Brain, 123, 1051–1061.
Silveri, M. C., di Betta, A. M., Filippini, V., Leggio, M. G., &
Molinari, M. (1998). Verbal short-term store rehearsal system and
the cerebellum. Brain, 112, 2174–2187.
Smith, E. E., Jonides, J., & Koeppe, R. A. (1996). Dissociating verbal
and spatial working memory using PET. Cerebral Cortex, 6, 11–20.
Spivey, M. J., & Tanenhaus, M. K. (1998). Syntactic ambiguity
resolution in discourse: Modeling the effects of referential context
and lexical frequency. Journal of Experimental Psychology: Learning, Memory, Cognition, 24, 1521–1543.
Stowe, L. A. (1991). Ambiguity resolution: Behavioral evidence for a
delay. In Proceedings of the 13th annual meeting of the cognitive
science association..
Stowe, L. A., Broere, C. A. J., Paans, A. M. J., Wijers, A. A., Mulder,
G., Vaalburg, W., & Zwarts, F. (1998). Localizing cognitive
components of a complex task: Sentence processing and working
memory. NeuroReport, 9, 2995–2999.
Stromswold, K., Caplan, D., Alpert, N., & Rauch, S. (1996).
Localization of syntactic comprehension by positron emission
tomography. Brain and Language, 52, 452–473.
Talairach, J., & Tournoux, P. (1988). Co-Planar Stereotaxic Atlas of
the Human Brain. Thieme Medical Publishers..
Tanenhaus, M. K., Spivey-Knowlton, M. J., & Hanna, J. E. (2000).
Modeling thematic and discourse context effects with a multiple
constraints approach: Implications for the architecture of the
language comprehension system. In M. W. Crocker & M. Pickering
(Eds.), Architectures and mechanisms for language processing (pp.
90–118). New York, NY, US: Cambridge University Press.
Ullman, M. T., Corkin, S., Coppola, M., Hickok, G., Growden, J.
H., Koroshetz, W. J., & Pinker, S. (1997). A neural dissociation
within language: Evidence that the mental dictionary is part of
declarative memory, and that grammatical rules are processed by
the procedural system. Journal of Cognitive Neuroscience, 9, 266–
276.
Vakil, E., Kahan, S., Huberman, M., & Osimani, A. (2000). Motor
and non-motor sequence learning in patients with basal ganglia
lesions: The case of serial reaction time (SRT). Neuropsychologia,
38, 1–20.
Withaar, R. G. (2002). The role of the phonological loop in sentence
comprehension. Groningen: GRODIL (Groningen Dissertations in
Linguistics).
Zysset, S., Huber, O., Ferstl, E., & von Cramon, D. Y. (2002). The
anterior frontomedian cortex and evaluative judgment: An fMRI
study. NeuroImage, 15, 983–991.