www.elsevier.

com/locate/ynimg
NeuroImage 40 (2008) 1362 1368

Specialisation in Broca's region for semantic, phonological, and

syntactic fluency?
Stefan Heim,a,b, Simon B. Eickhoff,a,c and Katrin Amunts a,b,d
a

Research Centre Jlich, Institute of Neurosciences and Biophysics (INB-3), Jlich, Germany
Brain Imaging Center West (BICW), Jlich, Germany
c
C. & O. Vogt Institut fr Hirnforschung, Heinrich-Heine Universitt, Dsseldorf, Germany
d
Department of Psychiatry and Psychotherapy, RWTH Aachen University, Aachen, Germany
b

Received 21 August 2007; revised 28 December 2007; accepted 8 January 2008

Available online 19 January 2008
The literature suggests that that semantic and phonological fluency tasks
selectively activate left Brodmann's area (BA) 45 and 44, respectively, in
Broca's speech region. We used functional MRI to test this hypothesis.
Subjects performed a semantic and a phonological fluency task. In
addition, a syntactic fluency task (e.g. Generate nouns with masculine
gender) was included. Resting blocks were included as a low-level
control condition. The exact localisation of the effects was tested with
cytoarchitectonic probability maps of BA 44 and BA 45. Participants
generated fewer words in the syntactic than in both the semantic and the
phonological condition, which did not differ from each other. Compared
to rest, all language tasks activated the well-known language network in
the left hemisphere including both left BA 44 and BA 45. In the direct
contrasts between the different verbal fluency tasks, phonological
fluency activated BA 44 more strongly than semantic or syntactic
fluency. However, semantic fluency did not elicit higher activation than
the phonological fluency tasks in any part of Broca's region. No
differences were observed between syntactic and semantic fluency. Thus,
the activation in BA 45 observed during verbal fluency tasks seems to be
not restricted to semantic processing as suggested by the literature. In
contrast, phonological verbal fluency additionally involved the left BA
44. In conclusion, different parts of Broca's region support task-specific
and more general processes in verbal fluency.
2008 Elsevier Inc. All rights reserved.

Introduction
In verbal fluency tasks participants are required to produce
words according to a given criterion. Typical examples for verbal
fluency tasks are the generation of words of a particular semantic
category (e.g. animals; semantic fluency) or with a given letter or
sound (e.g. words starting with F; phonological fluency). The

Corresponding author. Research Centre Jlich, Institute of Neurosciences and Biophysics, 52425 Jlich, Germany. Fax: +49 2461 61 2820.
E-mail address: s.heim@fz-juelich.de (S. Heim).
Available online on ScienceDirect (www.sciencedirect.com).
1053-8119/$ - see front matter 2008 Elsevier Inc. All rights reserved.
doi:10.1016/j.neuroimage.2008.01.009

word generation may either be overt or covert. It can follow the

subject's own speed (i.e. self-paced) or be externally paced by the
experimenter (cf. Basho et al., 2007).
Verbal fluency tasks have since long (e.g. Burt, 1917; Lotsof,
1953; Rogers, 1953) been standard paradigms in psychology,
psychiatry, and neurology (e.g. Frith et al., 1995; Gaillard et al.,
2000; Monsch et al., 1992) for testing the cognitive abilities of
patients or for identifying language-related brain regions before
neurosurgery. The advent of neuroimaging techniques such as
functional magnetic resonance imaging (fMRI) revealed that
verbal fluency recruits the left inferior frontal gyrus (IFG)
including Broca's speech region. A recent meta-analysis (Costafreda et al., 2006) demonstrated that different types of verbal
fluency recruit different aspects of the left IFG: Whereas semantic
fluency tasks tended to activate a more ventral-anterior portion of
the IFG (roughly corresponding to Brodmann's area [BA] 45),
phonological fluency appeared to involve a more dorsal-posterior
aspect (approximately BA 44). This finding was seen in
accordance with recent neurocognitive models of language
processing describing a functional parcellation for semantic,
syntactic, and phonological processing in the IFG (e.g. Bookheimer, 2002; Friederici, 2002; Hagoort, 2005) and with patient
data (Szatkowska et al., 2000).
Although the results by Costafreda et al. (2006) are appealing,
some issues remain to be addressed. First, as the authors
acknowledged, none of the included studies directly compared
semantic and phonological fluency in the same subjects. This is
critical because only such direct comparison of two tasks allows a
precise assessment of activation differences. Second, the metaanalysis was performed on the locations of the peak activations,
which do not reflect the extent of the activation clusters. This
implies that the actual activations in the single studies may have
been more blurred than suggested by the location of the local
maxima. Third, the assignment of the peak coordinates to BA 44 or
BA 45 was only based on macroanatomical criteria, i.e. the sulcal
pattern in the IFG. However, as Amunts et al. (1999) demonstrated,
borders of cytoarchitectonically defined brain areas such as BA 44
or BA 45 do not necessarily coincide with the sulcal landmarks.

S. Heim et al. / NeuroImage 40 (2008) 13621368

1363

Thus, the assignment of the coordinates to BA 44 and BA 45

in the study by Costafreda et al. (2006) remains imprecise with
respect to the underlying cortical areas.
Therefore, we tested the hypothesis that semantic and
phonological fluency recruit left BA 45 or BA 44, respectively,
in a new functional imaging study that compared different types of
verbal fluency in the same subjects. The localisation of the effects
was assessed with cytoarchitectonic probability maps of Broca's
region (Amunts et al., 1999, 2004).
Materials and methods
Participants
28 healthy right-handed participants (mean age 29.4 years; 14
females) participated in the experiment. They were native German
speakers and had normal or corrected-to-normal vision. Participants
had no history of neurological or psychiatric disorders. Informed
consent was obtained from all participants. The experimental
standards were approved by the local ethics committee of the
University of Aachen.
Tasks
A total of four overt word generation tasks were applied in
German: semantic, syntactic, phonological, and free (no explicit
criterion). Each generation task was performed in six blocks (see
below). In the semantic fluency task subjects had to overtly
generate examples for the following six categories: birds,
mammals, food, weapons, tools, and toys. In the phonological
fluency task the participants were required to generate nouns
starting with the following phonemes: /b/, /f/, /k/, /m/, /sh/, and /t/.
In the syntactic fluency task, nouns with masculine, feminine, or
neuter gender had to be produced. The syntactic fluency task was
included since, according to the model of language production by
Levelt et al. (1999; see Fig. 1), it taps a level in the mental lexicon
(the lemma level) which is not targeted by the other two fluency
tasks. According to recent neurocognitive models of language
(Friederici, 2002; Hagoort, 2005) syntactic fluency should also
activate BA 44 and/or BA 45. A free generation task in which the
participants produced any nouns without a predefined criterion was
administered as a high-level baseline task controlling for the
retrieval and articulation of words. In addition, resting blocks were
included as a low-level baseline.
In the present paper the term verbal fluency task(s) will be
used when referring to the semantic, syntactic, or phonological
fluency task as opposed to the free word generation baseline. In
contrast, word generation (tasks) includes the verbal fluency
tasks and the free word generation baseline, and is used to contrast
these tasks requiring overt speech to the resting baseline.
Stimulus Presentation
Visual stimuli were presented as written strings in Helvetica
font at 48 pts via goggles (VisuaStim, Resonance Technology,
CA, USA). Stimulus presentation was controlled by a computer
placed in the control room using Presentation software (Neurobehavioral Systems, Albany, CA, USA).
The study employed a block design. There were six blocks for
each condition and 24 resting blocks separating the task blocks.
Before each task block a written instruction was presented for

Fig. 1. The serial model of language production by Levelt et al. (1999).

Semantics (concept), syntax (lemma), and phonology of a word are retrieved
during different stages of the production process. The fluency tasks tapping
the different stages are indicated.

6 seconds. The blocks started immediately after that instruction and

lasted for 20 seconds. The condition blocks were presented in a
pseudo-randomised order, with different randomisations for each
participant. The total duration of the experiment was 19 minutes.
Each block included ten repetitions of experimental trials
(generation of a single word). Each trial lasted 2 seconds. The
fMRI data were acquired in the first 1.04 seconds of each trial using a
bunched-early sequence (see below). After this time, a fixation cross
appeared for the remaining 0.96 seconds of silence, indicating the
subject that he or she could now utter the next word during a silent
period (see Fig. 2 and the section Data Acquisition and Analysis;
cf. Heim et al., 2002; de Zubicaray et al., 2001, 2002).
This procedure combines a number of advantages. First, it
prevents motion-induced susceptibility artefacts, since subjects
only speak when no fMRI data are recorded. Second, the scanner
noise is not superimposed on the verbal response. Consequently,
the subject's utterances can be better evaluated during or after the
experiment to assess the number of produced items. Finally, cueing
the overt speaking implies paced word generation, which has been
shown to reduce head motion during speaking as compared to
unpaced generation (Basho et al., 2007).
Speech recordings
The participants' speech production was recorded using the
microphone of the goggle system. The cable from the microphone
to the patient intercom in the MR control room was plugged there
into a splitter, from which one cable led to the intercom and one to
the line-in port of an external sound card attached to a Toshiba

1364

S. Heim et al. / NeuroImage 40 (2008) 13621368

calculated: Semantic N Rest, Syntactic N Rest, and Phonological N Rest.

Moreover, a conjunction analysis (Price and Friston 1997) was performed to assess brain regions significantly activated in all three
fluency conditions. Finally, the relative differences between the three
fluency tasks were tested in pair-wise comparisons.

Fig. 2. Speaking during scanning: A bunched-early EPI sequence was used for
the acquisition of the fMRI data. All slices were recorded in the first 1.04
seconds of the TR, resulting in a silent period of 0.96 seconds in the second half
of the TR. During this silent period, the participants generated the words. The
speech signal is clearly discernable, since it is not obscured by the scanner noise.

notebook used for digital recording. These recordings were used

for qualitative and quantitative analyses of the participants' overt
responses in each condition, yielding information about what they
said and how many items were generated.

Anatomical localisation of the fMRI data

For the anatomical localisation we used cytoarchitectonic
probability maps of BA 44 and BA 45 in Broca's region (Amunts
et al., 1999, 2004). These maps are based on an observer-independent
analysis of the cytoarchitecture in a sample of ten post-mortem brains
(Schleicher et al., 1999; Zilles et al., 2002). They provide information
about the location and variability of cortical regions in standard MNI
space. For the assignment of coordinates to cytoarchitectonically
defined regions, we used the SPM Anatomy Toolbox (Eickhoff et al.,
2005; available with all published cytoarchitectonic maps from www.
fz-juelich.de/ime/spm_anatomy_toolbox).
Since the present study was designed to test a hypothesis about
the differential involvement of the left BA 44 and BA 45 in different
verbal fluency tasks, the analysis was confined to these two areas in a
region of interest (ROI) analysis. The combined maximum
probability maps of BA 44 and BA 45 were used as the search
volume for small volume corrections (SVC) when differential task
effects in Broca's region were examined (Eickhoff et al., 2006).
Results

Data acquisition and analysis

Behavioural data
Functional and anatomical data
The fMRI experiment was carried out on a 3T Siemens Trio
scanner. A standard birdcage head coil was used with foam paddings
reducing head motion. The functional data were recorded from 17
sagittal slices in the left hemisphere using a gradient-echo EPI
sequence with echo time (TE) = 30 ms, flip angle = 90 degrees, and
repetition time (TR) = 2 s. The sagittal orientation of the slices was
chosen in order to correct head motion in-plane, which is highest in the
yz plane. Acquisition of the slices within the TR was arranged so that
all slices were acquired in the first 1040 ms, followed by a 960-ms
period of no acquisition to complete the TR during which the subjects
spoke. The field of view (FOV) was 200 mm, with an in-plane
resolution of 3.1 mm 3.1 mm. The slice thickness was 3 mm with an
inter-slice gap of 1 mm. In addition, anatomical T1-weighted MPRAGE images (resolution 1 mm 1 mm 1 mm, FOV= 256 mm,
TR = 2250 ms, TE = 3.03 ms, flip angle = 9 degrees) were also obtained.
The data processing was performed using MATLAB 6.5 (The
Mathworks Inc., Natick, USA), and SPM5 (Wellcome Department
of Cognitive Neurology, UK). Two dummy scans before the
beginning of the experiment were discarded to allow for magnetic
saturation. Data pre-processing included the standard procedures of
realignment, normalisation to the MNI single subject template, and
spatial smoothing (FWHM = 8 mm).
Head motion parameters were as follows: The mean translations
were 0.08 mm (x), 0.2 mm (y), and 0.5 mm (z), with corresponding
rotations of 0.04 degrees, 0.03 degrees, and 0.03 degrees, respectively.
For the statistical analysis at the single participant level, the block
functions for each word generation condition were convolved with a
canonical haemodynamic response function (HRF). For each
participant, the contrasts of each task vs. rest were calculated. For
the group analysis, the individual contrast images were entered into a
repeated-measures ANOVA as a second level random effects analysis.
From this ANOVA the following low-level baseline contrasts were

The subjects produced on average (standard error of mean)

46.8 (1.2) words in the semantic fluency task, 40.9 (1.7) words

Fig. 3. Top: Activation in the left cytoarchitectonic BA 44 and BA 45 in semantic

fluency (red), syntactic fluency (yellow), and phonological fluency (cyan). Bottom:
Conjunction analysis of the three fluency conditions and activation strength (beta
values) at the local maxima in BA 44 and BA 45. Abbreviations: L, left; R, right;
SEM, semantic fluency; SYN, syntactic fluency; PHO, phonological fluency.

S. Heim et al. / NeuroImage 40 (2008) 13621368

in the syntactic fluency task, 45.3 ( 1.3) words in the phonological

fluency task, and 56.5 (0.7) words in the free generation task.
The one-factorial repeated-measures ANOVA of the number of
generated items yielded a main effect for task (F1,29 = 97.3; p b .001).
The post-hoc contrasts revealed that the subjects produced significantly (all p b .001) more words in the free generation task than in any
of the fluency tasks (semantic vs. free: t29 = 10.8; syntactic vs. free:
t29 = 12.2; phonological vs. free: t29 = 9.2). Among the fluency
tasks, less words were produced in the syntactic condition than in the
semantic (t29 = 5.5; p b .001) or the phonological (t29 = 3.9; p = .001)
condition. No difference was present between the semantic and the
phonological condition (t29 = 1.4; p = .174).
In the phonological task subjects produced on average 8.1
words with /b/, 8.0 words with /t/, 7.6 words with /k/, 7.2 words
with /m/ and with /sh/, and 7.1 words with /f/. This ranking seems
to differ from the ranking according to the frequency of occurrence
of the corresponding letters in German, which is 1.89% for B,
6.15% for T, 1.21% for K, 2.53% for M, 7.27% for S/SH, and
1.66% for F (URL: http://de.wikipedia.org/wiki/Buchstabenh%
C3%A4ufigkeit#_note-1).
The qualitative analysis of the participants' verbal responses
revealed that only few errors were made, with highest error rates in
syntactic fluency (semantic: 1.1%; syntactic: 3.2%; phonological:
0.2%). However, it appeared that in the free generation condition the
participants engaged various strategies that were either phonological
(a series of words with the same initial phoneme), semantic (items
from related categories), or scenic/imagery (participants imagined e.g.
a room and named the objects in this room). Such strategies resulted in
a supposedly inconsistent free generation baseline which could
obscure task-related activation in the other three generation conditions
if these were compared to it. We therefore refrained from using the
free generation condition as a high-level baseline condition. Instead,
as planned, we compared the task-related activations against the
resting baseline and pair-wise with each other.

1365

Table 2
Localisation of the peak activations in the differential contrasts of the verbal
fluency tasks (at pcorr b.05 SVC for Broca's region, k N 10 voxels)
Coordinates
Contrast

BAcyto

Tmax

Semantic N Syntactic
Syntactic N Semantic
Semantic N Phonological
Phonological N Semantic
Syntactic N Phonological
Phonological N Syntactic

44

44

50

50

10

21

23

n.s.
n.s.
n.s.
6.93
n.s.
5.16

For details see the legend of Table 1. Further abbreviation: n.s., not significant.

cytoarchitectonic probability: 40%). Syntactic fluency also activated

both BA 44 (maximum cytoarchitectonic probability: 40%) and BA
45 (maximum cytoarchitectonic probability: 40%). Finally, phonological fluency also activated BA 44 (maximum cytoarchitectonic
probability: 50%) and BA 45 (maximum cytoarchitectonic probability: 40%). A conjunction analysis of all three fluencies (Fig. 3,
bottom row) revealed common significant activation in the left BA 44
(maximum cytoarchitectonic probability: 60%) and BA 45 (maximum
cytoarchitectonic probability: 40%).

fMRI data
The brain activation effects in Broca's region for contrasting each
single verbal fluency task against the resting baseline are displayed in
Fig. 3 (top row) and listed in Table 1 (all pcorr b.05 are FWE-corrected
for the search volume of BA 44 and BA 45, k N 10 voxels). Semantic
fluency elicited activation in the left cytoarchitectonic BA 44
(maximum cytoarchitectonic probability: 60%) and BA 45 (maximum

Table 1
MNI coordinates and T value (Tmax) at the local activation maxima in the
contrasts of each fluency task against the resting baseline (pcorr b.05 are
family-wise error corrected for the search volume of BA 44 and BA 45)
Coordinates
Contrast

BAcyto

Tmax

Semantic N Rest

44
45
44
45
44
45
44
45

52
44
52
44
52
44
52
44

2
26
2
26
6
28
2
28

23
25
23
25
23
23
23
23

8.30
5.55
8.00
6.39
8.76
5.61
8.00
5.49

Syntactic N Rest
Phonological N Rest
Conjunction analysis

The cytoarchitectonic Brodmann's area (BAcyto) at the local maximum (see

Amunts et al., 2004) is given.

Fig. 4. Top: Surface renderings of brain activation (at P b.05 whole-brain

FWE-corrected) of semantic (SEM) and phonological (PHO) fluency
compared to the resting baseline (REST) reveal comparable patterns.
Bottom: Differential fMRI effects (at P b.05 whole-brain FWE-corrected) of
semantic N phonological fluency and phonological N semantic fluency.

1366

S. Heim et al. / NeuroImage 40 (2008) 13621368

The differential contrasts between the three verbal fluency tasks

are reported in Table 2 (all pcorr b.05 are FWE-corrected for the
search volume of BA 44 and BA 45, k N 10 voxels). The activation
in the phonological vs. syntactic fluency and in the semantic vs.
syntactic fluency did not differ significantly. Phonological fluency
yielded higher activation than semantic fluency in BA 44 (maximum
cytoarchitectonic probability: 50%). The reverse contrast was not
significant in Broca's region. Phonological fluency also yielded
higher activation than syntactic fluency in BA 44 (maximum
cytoarchitectonic probability: 50%). Again, the reverse contrast did
not reach significance within Broca's region.
Outside Broca's region semantic fluency showed higher activation
than phonological fluency in the left middle frontal gyrus (60, 10,
23; t = 7.29) and the left fusiform gyrus (28, 38, 19; t = 5.71).
The reversed contrast yielded higher activation for phonological than
for semantic fluency in the left inferior parietal lobule (40, 40, 47;
t = 6.55) which was due to deactivation in the semantic rather than
activation in the phonological condition. These effects (all Pcorr b.05
whole-brain are FWE-corrected) are displayed in Fig. 4 (bottom).
Discussion
In the present study we tested the hypothesis that the left BA 44
supports phonological fluency whereas the left BA 45 is engaged
in semantic fluency. To this end we compared semantic, phonological, and syntactic verbal fluency in an fMRI study and superimposed the results on cytoarchitectonic maps of BA 44 and BA 45.
The data are only partly in line with this hypothesis. Compared to a
resting baseline, semantic fluency indeed activated BA 45. Moreover, phonological fluency yielded activation in BA 44 which was
higher than activation in BA 44 in any other task. So far, these data
corroborate the meta-analysis by Costafreda et al. (2006) and our
previous results that semantic fluency involves cytoarchitectonic
BA 45 (Amunts et al., 2004).
The observed pattern of activations, however, was more complex
than previously assumed, since activation in BA 45 was also present
during phonological and syntactic fluency. Likewise, the activation
in BA 44 was also observed for semantic and syntactic fluency.
These findings are not in line with the hypothesis that BA 44 and BA
45 are specialised for phonological vs. semantic fluency, respectively.
Therefore, alternative interpretations need to be considered.
The comparable effects in BA 45 in all three conditions may
reflect an unspecific aspect of verbal fluency shared by all three
fluency tasks rather than condition-specific demands. This idea is in
accordance with the findings by Rogers (1953) who observed in a
factor analysis of behavioural data from different fluency tasks one
factor for general verbal ability and another for oral verbal
ability. Interestingly, and supporting the observation of comparable
engagement of inferior frontal areas in the present study, no evidence
for separate factors for semantic, syntactic, or phonological fluency
was obtained in the Rogers (1953) study. Therefore, it would be
plausible to attribute the unspecific activation in BA 45 in all
fluency tasks to a general selection mechanism possibly related to
Rogers' (1953) oral verbal ability. Further support for our position
comes from a study by Gold and Buckner (2002) who also observed
comparable inferior frontal activation during the controlled processing of semantic and phonological information. Moreover, in
addition to the shared inferior frontal activation in both tasks, the
authors found a functional dissociation of semantic and phonological processing in inferior parietal and temporal regions. In
particular, the left inferior parietal lobule was activated more

strongly for phonological than for semantic processing. This finding

could be replicated when directly contrasting phonological vs.
semantic fluency in the present study.
Given the data from the Gold and Buckner (2002) study the
activation in the present study in BA 44 and BA 45 in all fluency
tasks may be related to the controlled retrieval of lexical information.
This argument is also in line with other studies showing that the left
IFG supports lexical selection processes (e.g. Badre et al., 2005; Kan
and Thompson-Schill, 2004; Snyder et al., 2007). When additionally
considering the behavioural results in the present study one may
even be more specific about the nature of the process reflected by the
activation in BA 44 and BA 45. If this activation was related to
lexical selection it should vary as a function of task difficulty, i.e., the
number of words produced in each condition. This is exactly what
we observed. There was no difference in the activation in semantic
and phonological fluency, which proved equally difficult when
considering the number of produced words in both conditions. In
contrast, the activation was significantly higher in the syntactic
fluency task which, according to the number of produced items and
also to the error rates, was the most difficult one. To conclude, the
activation data are compatible with the view that Broca's region
supports a rather unspecific selection mechanism that is shared by
semantic, phonological, and syntactic verbal fluency and that is
sensitive only to the difficulty of the selection.
Alternatively, one might suppose that the activation effects in
Broca's region do not reflect some unspecific selection mechanism,
but distinct processes which are specific for semantic fluency,
syntactic fluency, or phonological fluency, respectively.
The present study cannot provide ultimate evidence against this
view. However, the scenario is not very convincing given that the local
maxima in particular in BA 45 were at almost identical coordinates.
But even if one assumed some condition-specific function in BA 45 it
is nonetheless striking that semantic and phonological fluency showed
no differential effects, in particular given the large sample size in the
present study. To conclude, the observed effects are best understood as
reflecting lexical selection for verbal fluency in BA 45 which is not
specific for the actual type of fluency.
Another cognitive process that could be reflected by the activation
of BA 45 common to all three fluency conditions is working memory.
Verbal working memory has been reported to activate the dorsal
aspect of the posterior IFG (dorsal aspect of pars opercularis:
Zurowski et al., 2002; dorsal aspect of pars triangularis: see the metaanalysis by Vigneau et al., 2006). Thus, one could assume that a verbal
working memory component which is shared by the three fluency
tasks is reflected in the activation in BA 45. In particular, it has to be
considered that this study used a paced fluency paradigm in which the
participants had to utter one word every 2 seconds. The participants
might have internally generated words at a faster rate (e.g. a rate of 1
word every 1.5 seconds was used by Ischebeck et al., in press) and
rehearsed this item until the cue to speak aloud. However, the
combination of the imaging data with the behavioural data renders this
argumentation rather unlikely, for the following reason. If the
activation in BA 45 reflected verbal working memory the activation
should increase with the increasing number of generated words, since
a higher number of generated words implies a higher number of trials
during which these words must be kept in working memory, resulting
in a higher working memory load. However, no such pattern was
observed. Activation in BA 45 was equally high in all conditions
despite the fact that significantly different numbers of words were
produced. Therefore, although we cannot entirely rule out a
contribution of working memory processes, the idea that the

1367

S. Heim et al. / NeuroImage 40 (2008) 13621368

activation in BA 45 in all three fluency tasks was primarily related to

verbal working memory is unlikely given the behavioural results.
The activation effects in BA 44 were distinct from those in BA 45.
Although all three fluency tasks significantly activated BA 44, they did
not do so to the same extent. Rather, BA 44 yielded higher activation
for phonological than for semantic (or syntactic) fluency. On the one
hand, such pattern of effects provides an argument against the initial
hypothesis of a selective involvement of BA 44 in phonological
fluency, since semantic and syntactic fluency also recruited BA 44.
However, whereas the balance of all three types of activation was
similar to each other in BA 45, phonological fluency was emphasized
in BA 44, thus indicating its relevance in this particular area. An
explanation might incorporate the view that BA 44 supports two
different types of processes: (i) specific for phonological fluency and
(ii) relevant for all tested types of fluency. Two candidate processes are
controlled vs. automatic phonological processing. The relevance of
BA 44 for controlled phonological processing (e.g. during phonological decisions) has been demonstrated earlier (e.g. Burton et al., 2000;
Dmonet et al. 1992; Heim et al. 2003; Zatorre et al. 1992). Such
controlled processing is also required during phonological fluency
where phonological cues need to be used as criteria for word retrieval.
In addition to its role in controlled processing, it has also been shown
that BA 44 is involved in object naming during the automatic retrieval
of phonological information prior to articulation (for a review of object
naming studies cf. e.g. Price et al. 2005). Since controlled and
automatic phonological processing may occur independently from and
on top of each other in BA 44 (Noesselt et al. 2003), the effects in BA
44 observed in the present study could be explained as the summation
of activation from controlled and automatic phonological processing.
Such summation only occurs during phonologically cued word
generation but not during syntactic and semantic word generation,
where other than phonological cues are relevant. For the purpose of the
present study one may conclude that BA 44 is relevant for all tested
types of verbal fluency and, in particular, for phonological fluency.
The present study used cytoarchitectonically defined volumes of
interest in order to test the hypothesis of functional specialisation
within Broca's region for semantic and phonological verbal fluency.
Beside the obvious advantages of this method with respect to
(1) observer-independence and availability of quantitative data (e.g.
Amunts et al., 2004) and (2) the use of anatomically-informed
regions of interest rather than spheres or boxes (Eickhoff et al.,
2006), this approach might also have some drawbacks for the present
study. So far only maps of BA 44 and BA 45 are available. In
contrast, BA 47 which is supposed to be located in the pars orbitalis
of the IFG, i.e. its most rostro-ventral part, has not yet been mapped.
Thus, when using the maps of BA 44 and BA 45 for the present
analysis we might systematically have ignored more anterior
activation in the IFG. This would be of importance in the case of
the less strict anatomical hypothesis (e.g. Poldrack et al., 1999;
Vigneau et al., 2006) that phonological fluency involves the more
posterior part of the IFG (i.e. BA 44/45) whereas semantic fluency
recruits the more anterior portion (i.e. BA 45/47). However, the
whole-brain analyses of the semantic and the phonological fluency
tasks reveal that the anterior aspect of the left IFG was neither
activated in the tasks per se nor in the direct comparison of the two
tasks (Fig. 4 and Tables 2 and 3). Consequently, one may exclude the
possibility that the cytoarchitecture-based approach used in the
present study biased or obscured relevant data.
Finally, it should be noted that, even though the subjects produced
overt speech, the average head motion parameters were very small
and even below the values reported in earlier overt production

Table 3
Activations in the semantic and phonological fluency tasks in the left
hemisphere (Pcorr b.05 whole-brain are FWE-corrected)
Region

Tmax

Semantic N Rest
Postcentral gyrus
Cerebellum
Precentral gyrus
Caudate nuclues
Insula
Temporal pole
Inferior frontal gyrus
Fusiform gyrus
Middle temporal gyrus
Inferior frontal gyrus
Superior temporal gyrus

46
14
52
16
34
48
46
30
58
42
60

10
58
4
10
16
6
8
2
30
26
24

41
17
25
21
5
13
7
43
5
25
5

13.12
11.43
10.14
8.62
7.89
6.53
6.40
6.26
5.61
5.57
5.40

Phonological N Rest
Postcentral gyrus
Cerebellum
Precentral gyrus
Caudate nuclues
Insula lobe
Inferior frontal gyrus
Temporal pole
Inferior frontal gyrus
Putamen
Pallidum
Fusiform gyrus

48
16
50
14
34
48
48
40
18
50
30

10
58
0
0
16
8
6
28
6
6
2

41
19
45
17
5
7
13
23
9
5
43

12.20
10.27
9.46
8.53
7.93
7.34
5.89
5.84
5.43
5.34
5.27

experiments (e.g. Gracco et al., 2005; Heim et al., 2006a). This

finding again demonstrates that overt language production is a valid
and useful paradigm for fMRI studies.
Conclusion
The present study tested whether the left cytoarchitectonic BA
44 and BA 45 specifically support phonological and semantic
fluency, respectively. However, both regions were involved in all
tested types of verbal fluency. Thus, the present study challenges
the contemporary view of a clear-cut functional parcellation of
Broca's region during verbal fluency. It seems that basically the
general demands on the selection of entries from the mental
lexicon rather than the particular linguistic domain (semantic or
phonological) are reflected by the amplitude of the fMRI signal in
verbal fluency tasks. Future studies may build upon the findings
from the present and earlier studies (e.g. Thompson-Schill et al.,
1997) and parametrically vary the selection demands in phonological and semantic fluency.

Acknowledgments
This Human Brain Project/Neuroinformatics research is funded
by the National Institute of Biomedical Imaging and Bioengineering, the National Institute of Neurological Disorders and Stroke,
and the National Institute of Mental Health (KA). Further support
by Helmholtz-Gemeinschaft (VH-N6-012 to KA) and the Brain
Imaging Center West (BMBF 01GO0204) is gratefully acknowledged. We thank N. Jon Shah for the support of the NMR group at
the INB-3 during fMRI data acquisition, in particular Barbara
Elghahwagi for her assistance with fMRI data recording. More-

1368

S. Heim et al. / NeuroImage 40 (2008) 13621368

over, we appreciate the support of the Cognitive Neurology group

at the INB-3 relating to the peripheral stimulation devices. Finally,
we wish to thank Helen Schreiber for her assistance with the
analysis of the behavioural data.
References
Amunts, K., Schleicher, A., Burgel, U., Mohlberg, H., Uylings, H.B.M.,
Zilles, K., 1999. Broca's region revisited: Cytoarchitecture and
intersubject variability. J. Comp. Neurol. 412, 319341.
Amunts, K., Weiss, P.H., Mohlberg, H., Pieperhoff, P., Eickhoff, S., Gurd,
J.M., Marshall, J.C., Shah, N.J., Fink, G.R., Zilles, K., 2004. Analysis
of neural mechanisms underlying verbal fluency in cytoarchitectonically defined stereotaxic space The roles of Brodmann areas 44 and
45. Neuroimage 22, 4256.
Badre, D., Poldrack, R.A., Pare-Blagoev, E.J., Insler, R.Z., Wagner, A.D.,
2005. Dissociable controlled retrieval and generalized selection
mechanisms in ventrolateral prefrontal cortex. Neuron 47, 907918.
Basho, S., Palmer, E.D., Rubio, M.A., Wulfeck, B., Muller, R.A., 2007.
Effects of generation mode in fMRI adaptations of semantic fluency:
paced production and overt speech. Neuropsychologia 45, 16971706.
Bookheimer, S., 2002. Functional MRI of language: new approaches to
understanding the cortical organization of semantic processing. Annu.
Rev. Neurosci. 25, 151188.
Burt, C., 1917. The distribution and relations of educational abilities. King,
London.
Burton, M.W., Small, S.L., Blumstein, S.E., 2000. The role of segmentation
in phonological processing: an fMRI investigation. J. Cogn. Neurosci.
12, 679690.
Costafreda, S.G., Fu, C.H., Lee, L., Everitt, B., Brammer, M.J., David, A.S.,
2006. A systematic review and quantitative appraisal of fMRI studies of
verbal fluency: role of the left inferior frontal gyrus. Hum. Brain Mapp.
27, 799810.
Dmonet, J.F., Chollet, F., Ramsay, S., Cardebat, D., Nespoulous, J.L., Wise,
R., Rascol, A., Frackowiak, R., 1992. The anatomy of phonological and
semantic processing in normal subjects. Brain 115, 17531768.
de Zubicaray, G.I., Wilson, S.J., McMahon, K.L., Muthiah, S., 2001. The
semantic interference effect in the picture-word paradigm: an eventrelated fMRI study employing overt responses. Hum. Brain Mapp. 14,
218227.
de Zubicaray, G.I., McMahon, K.L., Eastburn, M.M., Wilson, S.J., 2002.
Orthographic/phonological facilitation of naming responses in the
picture-word task: an event-related fMRI study using overt vocal
responding. Neuroimage 16, 10841093.
Eickhoff, S.B., Stephan, K.E., Mohlberg, H., Grefkes, C., Fink, G.R.,
Amunts, K., Zilles, K., 2005. A new SPM toolbox for combining
probabilistic cytoarchitectonic maps and functional imaging data.
Neuroimage 25, 13251335.
Eickhoff, S.B., Heim, S., Zilles, K., Amunts, K., 2006. Testing anatomically
specified hypotheses in functional imaging using cytoarchitectonic
maps. Neuroimage 32, 570582.
Friederici, A.D., 2002. Towards a neural basis of auditory sentence
processing. Trends in Cogn. Sci. 6, 7884.
Frith, C.D., Friston, K.J., Herold, S., Silbersweig, D., Fletcher, P., Cahill, C.,
Dolan, R.J., Frackowiak, R.S., Liddle, P.F., 1995. Regional brain activity
in chronic schizophrenic patients during the performance of a verbal
fluency task. Br. J. Psychiatry 167, 343349.
Gaillard, W.D., Hertz-Pannier, L., Mott, S.H., Barnett, A.S., LeBihan, D.,
Theodore, W.H., 2000. Functional anatomy of cognitive development:
fMRI of verbal fluency in children and adults. Neurology 54, 180185.
Gold, B.T., Buckner, R.L., 2002. Common prefrontal regions coactivate
with dissociable posterior regions during controlled semantic and
phonological tasks. Neuron 35, 803812.
Gracco, V.L., Tremblay, P., Pike, B., 2005. Imaging speech production using
fMRI. Neuroimage 26, 294301.

Hagoort, P., 2005. On Broca, brain, and binding: a new framework. Trends
Cogn. Sci. 9, 416423.
Heim, S., Opitz, B., Friederici, A.D., 2002. Broca's area in the human brain
is involved in the selection of grammatical gender for language
production: evidence from event-related functional magnetic resonance
imaging. Neurosci. Lett. 328, 101104.
Heim, S., Opitz, B., Mller, K., Friederici, A.D., 2003. Phonological processing
during language production: fMRI evidence for a shared productioncomprehension network. Brain Res. Cogn. Brain Res. 16, 285296.
Heim, S., Amunts, K., Mohlberg, H., Wilms, M., Friederici, A.D., 2006a.
Head motion during overt language production in functional magnetic
resonance imaging. Neuroreport 17, 579582.
Ischebeck, A., Heim, S., Siedentopf, C., Zamarian, L., Schocke, M.,
Kremser, C., Egger, K., Strenge, H., Scheperjans, F., Delazer, M.,
in press. Are numbers special? Comparing the generation of verbal
materials from ordered categories (months) to numbers and other
categories (animals) in an fMRI study. Hum. Brain Mapp. (Electronic
publication ahead of print). doi:10.1002/hbm.20433.
Kan, I.P., Thompson-Schill, S.L., 2004. Effect of name agreement on
prefrontal activity during overt and covert picture naming. Cogn. Affect.
Behav. Neurosci. 4, 4357.
Levelt, W.J., Roelofs, A., Meyer, A.S., 1999. A theory of lexical access in
speech production. Behav. Brain Sci. 22, 138 discussion 3875.
Lotsof, E.J., 1953. Intelligence, Verbal Fluency, and the Rorschach Test.
J. Consult. Psychol. 17, 2124.
Monsch, A.U., Bondi, M.W., Butters, N., Salmon, D.P., Katzman, R., Thal,
L.J., 1992. Comparisons of verbal fluency tasks in the detection of
dementia of the Alzheimer type. Arch. Neurol. 49, 12531258.
Noesselt, T., Shah, N.J., Jncke, L., 2003. Top-down and bottom-up
modulation of language related areasan fMRI study. BMC Neurosci.
4, 13.
Poldrack, R.A., Wagner, A.D., Prull, M.W., Desmond, J.E., Glover, G.H.,
Gabrieli, J.D., 1999. Functional specialization for semantic and
phonological processing in the left inferior prefrontal cortex. Neuroimage 10, 1535.
Price, C.J., Friston, K.J., 1997. Cognitive Conjunction: a new approach to
brain activation experiments. Neuroimage 5, 261270.
Rogers, C.A., 1953. The structure of verbal fluency. Br. J. Psychol. 44, 368380.
Schleicher, A., Amunts, K., Geyer, S., Morosan, P., Zilles, K., 1999. Observerindependent method for microstructural parcellation of cerebral cortex: A
quantitative approach to cytoarchitectonics. Neuroimage 9, 165177.
Snyder, H.R., Feigenson, K., Thompson-Schill, S.L., 2007. Prefrontal
Cortical Response to Conflict during Semantic and Phonological Tasks.
J. Cogn. Neurosci. 19, 761775.
Szatkowska, I., Grabowska, A., Szymanska, O., 2000. Phonological and
semantic fluencies are mediated by different regions of the prefrontal
cortex. Acta Neurobiol. Exp. (Wars) 60, 503508.
Thompson-Schill, S.L., D'Esposito, M., Aguirre, G.K., Farah, M.J., 1997.
Role of left inferior prefrontal cortex in retrieval of semantic knowledge:
a reevaluation. Proc. Natl. Acad. Sci. U. S. A. 94, 1479214797.
Vigneau, M., Beaucousin, V., Herve, P.Y., Duffau, H., Crivello, F., Houde,
O., Mazoyer, B., Tzourio-Mazoyer, N., 2006. Meta-analyzing left
hemisphere language areas: phonology, semantics, and sentence
processing. Neuroimage 30, 14141432.
Zatorre, R.J., Evans, A.C., Meyer, E., Gjedde, A., 1992. Lateralization of
phonetic and pitch discrimination in speech processing. Science 256,
846849.
Zilles, K., Schleicher, A., Palomero-Gallagher, N., Amunts, K., 2002.
Quantitative analysis of cyto- and receptor architecture of the human
brain. In: Mazziotta Toga, J., Toga, A. (Eds.), Brain Mapping, the
methods. Academic Press, San Diego, pp. 573602.
Zurowski, B., Gostomzyk, J., Gron, G., Weller, R., Schirrmeister, H.,
Neumeier, B., Spitzer, M., Reske, S.N., Walter, H., 2002. Dissociating
a common working memory network from different neural substrates
of phonological and spatial stimulus processing. Neuroimage 15,
4557.