Cerebral Cortex February 2011;21:459--466

doi:10.1093/cercor/bhq108
Advance Access publication June 21, 2010

Neuroanatomical Prerequisites for Language Functions in the Maturing Brain

Jens Brauer, Alfred Anwander and Angela D. Friederici
Max Planck Institute for Human Cognitive and Brain Sciences, Department of Neuropsychology, 04103 Leipzig, Germany
Address correspondence to Dr Jens Brauer, Max Planck Institute for Human Cognitive and Brain Sciences, Department of Neuropsychology,
Stephanstrasse 1a, 04103 Leipzig, Germany. Email: brauer@cbs.mpg.de.

The 2 major language-relevant cortical regions in the human brain,

Brocas area and Wernickes area, are connected via the fibers of
the arcuate fasciculus/superior longitudinal fasciculus (AF/SLF).
Here, we compared this pathway in adults and children and its
relation to language processing during development. Comparison of
fiber properties demonstrated lower anisotropy in childrens AF/
SLF, arguing for an immature status of this particular pathway with
conceivably a lower degree of myelination. Combined diffusion
tensor imaging (DTI) data and functional magnetic resonance
imaging (fMRI) data indicated that in adults the termination of the
AF/SLF fiber projection is compatible with functional activation in
Brocas area, that is pars opercularis. In children, activation in
Brocas area extended from the pars opercularis into the pars
triangularis revealing an alternative connection to the temporal lobe
(Wernickes area) via the ventrally projecting extreme capsule fiber
system. fMRI and DTI data converge to indicate that adults make
use of a more confined language network than children based on
ongoing maturation of the structural network. Our data suggest
relations between language development and brain maturation and,
moreover, indicate the brains plasticity to adjust its function to
available structural prerequisites.
Keywords: development, DTI, fiber tractography, fMRI, language

Introduction
Language is a unique human capacity but its development in
the brain remains largely unexplained. Functional imaging
studies in adults show that sentence comprehension is
supported by a frontotemporal network (Friederici 2002;
Hashimoto and Sakai 2002; Hickok and Poeppel 2007; Tyler
and Marslen-Wilson 2008). The processing of grammatical
structures involves Brocas area, in particular Brodmann Area
(BA) 44 (Friederici, Bahlmann, et al. 2006) and the posterior
portion of Wernickes area (Bornkessel et al. 2005) and
a dynamic interplay between these 2 areas (Snijders et al.
2009). These 2 language relevant brain regions are connected
by ber bundles that guarantee information transmission. There
seem to be substantial phylogenetic differences in the white
matter ber pathways that connect the cortical circuits
underlying language functions in humans. Structural brain
imaging data suggest distinctions between human and nonhuman primates in ber tracts connecting the frontal and
temporal brain areas. By combining microelectrode recording
with anatomical tract tracing, 2 functionally different pathways
have been dened in the macaque, these being a dorsal and
a ventral route that connect the auditory cortex to the
prefrontal cortex (Romanski et al. 1999). Data revealed a ventral
pathway running from the inferior frontal gyrus (IFG) and
insular cortex to the superior temporal gyrus and sulcus (STG/
STS) as the dominant pathway in the macaque, whereas a dorsal
The Author 2010. Published by Oxford University Press. All rights reserved.
For permissions, please e-mail: journals.permissions@oup.com

pathway from BA 44, 45, and 47 in the IFG to the posterior

STG/STS and the middle temporal gyrus is dominant in humans
(Rilling et al. 2008). Nevertheless, both pathway connections
are existent in the human brain (Frey et al. 2008; Saur et al.
2008; Hua et al. 2009). The dorsal pathway (Friederici,
Bahlmann, et al. 2006; Rilling et al. 2008) is argued as a critical
language-relevant connection between the 2 brain regions that
have been shown to support the processing of syntactically
complex sentences (Constable et al. 2004; Bornkessel et al.
2005). Others take the ventral pathway to be the crucial one
for language comprehension (Saur et al. 2008). The distinctive
contributions of this dorsal and the ventral pathways to
language functions thus are still a matter of debate (Friederici
2009a, 2009b; Weiller et al. 2009).
Phylogenetic arguments derived from cross-species comparisons are important but can only contribute indirectly to our
understanding of the human language faculty (Ghazanfar 2008;
Petkov et al. 2009). An alternative route toward achieving more
profound comprehension of the brain basis of human language
can be taken by simultaneously investigating ontogenetic
development of brain structures and brain functions of
language in humans. The exact location and extent of human
white matter ber pathways can be identied noninvasively in
vivo by means of diffusion tensor imaging (DTI). Based on
structural properties of the tissue, DTI measures water
diffusion in the brain as well as its directional orientation
(fractional anisotropy [FA]) and thus provides useful information about the anatomy of the brains ber pathways.
Furthermore, DTI allows the explicit assessment of changes
in white matter maturation during ontogeny.
Structural maturation of ber pathways in the brain is
particularly characterized by increasing myelination of bers,
which is reected in DTI by an increase in FA during infancy
and childhood (Mukherjee et al. 2001; Barnea-Goraly et al.
2005; Dubois et al. 2008, Lebel et al. 2008). The process of
initial myelination during human development is most pronounced before the second year of life, but further condensation and agglomeration continue (Lenroot and Giedd 2006).
These structural changes are accompanied by concurrent
major developmental changes in motor, sensory, executive,
and cognitive functions.
The development of language functions is still under
investigation. Behavioral data, for example, indicate that the
processing of syntactically complex sentences in particular,
that is, sentences with a noncanonical word order, occurs late.
It has been demonstrated that object-rst sentences (Dittmar
et al. 2008) or passive sentences (Hahne et al. 2004) are not
fully understood before about age 7 years. From functional
magnetic resonance imaging (fMRI) data in adults, we know
that the processing of noncanonical sentences recruit Brocas
area (Stromswold 1996; Rogalsky et al. 2008), in particular BA
44, and the posterior STG/STS (Bornkessel et al. 2005;

Friederici, Fiebach, et al. 2006): 2 brain regions that are

connected via the dorsal pathway (Catani et al. 2005; Rilling
et al. 2008). Our knowledge concerning the functional
neuroanatomy of language during development is still sparse.
For auditory language comprehension, pediatric fMRI data
revealed functional and causal connectivity between frontal
and temporal regions (Wilke et al. 2009). Stronger activation in
the language relevant areas and beyond these for children
compared with adults has been reported for sentence
processing (e.g., Gaillard et al. 2000; Brauer and Friederici
2007). However, the pattern of development of functional
brain activation seems complex including increases and
decreases of functional activation across brain regions during
development (Rivera et al. 2005; Schlaggar and Church 2009)
and a ne tuning and/or focalization of activation in task
relevant areas (Casey et al. 2005; Cohen Kadosh and Johnson
2007).
Here, we hypothesized that the human ability to successfully
process sentences including their inherent grammatical and
semantic information might require a strong and fully developed dorsal connection via the arcuate fasciculus/superior
longitudinal fasciculus (AF/SLF) associating Brocas area to the
posterior portion of the temporal cortex. DTI data were
acquired from 7-year-old children and adults to identify brain
regions of ongoing maturation in language-relevant areas by
directly comparing ber tract anatomy. Additionally, fMRI data
on a well-established language comprehension paradigm were
acquired to identify the brain areas that support auditory
sentence processing in children and adults (Brauer and
Friederici 2007; Brauer et al. 2008). A combined analysis of
fMRI and DTI data converged functional and structural data.
Materials and Methods
Participants
Seven-year-old children, recruited from local kindergartens by open
letter, participated in the study. Interested parents were invited for an
informative meeting about the experiment and procedures. They gave
written informed consent, and children gave verbal assent prior to
assessment and scanning. Data from 12 children were obtained, 2 of
which had to be excluded from further analysis due to movement in
the scanner during data acquisition. Thus, data of 10 children (5 girls)
were available (mean age 7.0 years, standard deviation [SD] 1.1, range
5.6--8.7). Additionally, 10 adults (5 females, mean age 27.8 years, SD 2.7,
range 24.4--32.4) participated in the study after informed consent. No
participant had any history of linguistic, neurological, or psychiatric
disorders. Adult participants were students. Children were checked for
typical development in intellectual and language abilities. They scored
within the normal range on a nonverbal IQ test (Kaufman-Assessment
Battery for Children, Kaufman and Kaufman 1994) at IQ = 104.10
(SD 11.41), a syntactic/semantic subscale of a language ability test
(Heidelberger Sprachentwicklungstest, Grimm and Schoeler 1991) at
T = 55.20 (5.14), and the Child Behavior Checklist (Achenbach 1998)
at T = 50.11 (5.37) (IQ and T scores represent normalized test values
as a projection onto an (age matched) reference population described
by a mean IQ of 100 [SD 15] and a mean T of 50 [SD 10]). All
participants were right handed and German native speakers. Protocols
and procedures were approved by the Research Ethics Committee of
the University of Leipzig.
Materials
Stimulus material consisted of short German sentences in the active
voice from an established paradigm (Brauer and Friederici 2007; Brauer
et al. 2008). The material included correct and incorrect basic
sentences (e.g., The frog croaks, The lion in the zoo roars, The

460 Neuroanatomical Prerequisites for Language Functions

Brauer et al.

yoghurt in tastes good, or The stone bleeds). Participants evaluated

each sentence by an acceptability judgment via button press.
For the purpose of auditory presentation, items were spoken by
a trained female native speaker in a well-pronounced, child-directed
manner. All sentences were recorded and digitized at 44.1 kHz, 16-bit
mono. They had an average length of approximately 2 s. For adults, the
session contained 200 trials plus 25 null events, in which the blood
oxygen level--dependent response was allowed to return to baseline
state (Burock et al. 1998). For children, the session contained 120 trials
plus 15 null events. In all other ways, the procedure was the same as
that used for adults. Trials were presented every 8 s in a single session.
In order to rule out any effect of experiment length, a shorter
experiment in adults was simulated post hoc by truncating adults data
sets after 135 volumes. There were no relevant changes in activation
results, particularly not for activation in Brocas area. Thus, length of
experiment could not account for group differences in functional
activation.
While listening to stimuli and during the entire measurement,
participants could see a aquarium screensaver with shes swimming
calmly across the scene. In order to obtain increased signal-to-noise
ratio, no distinction between syntactic and semantic information
processing was made. Rather, language stimulation in general was
contrasted against resting baseline (null events) since our main interest
was general sentence processing (Brauer et al. 2008). Only correctly
answered trials entered subsequent analyses.
Magnetic Resonance Imaging
fMRI and DTI data were acquired in separate sessions on a whole-body
3-T Magnetom Trio scanner (Siemens). Diffusion-weighted data and
high-resolution T1-weighted images were obtained with an 8-channel
head array coil. Diffusion-weighted images were obtained with a twicerefocused spin echo-planar--imaging (EPI) sequence (Reese et al. 2003),
time echo (TE) 100 ms, time repetition (TR) 12 s, 128 3 128 image
matrix, and eld of view (FOV) 220 3 220 mm2, providing 60 diffusionencoding gradient directions with a b value of 1000 s/mm2. Seven
images without any diffusion weighting were obtained at the beginning
of the scanning sequence and after each block of 10 diffusion-weighted
images as anatomical reference for off-line motion correction. The
interleaved measurement of 72 axial slices with 1.7-mm thickness (no
gap) covered the entire brain. Random noise in the data was reduced by
averaging three acquisitions. Additionally, fat saturation was employed
together with 6/8 partial Fourier imaging, Hanning window ltering,
and parallel acquisition (generalized autocalibrating partially parallel
acquisitions with reduction factor 2).
For functional measurements, a gradient-echo EPI sequence was
applied (TE 30ms, ip angle 90, TR 2 s, bandwidth 100 kHz, matrix
64 3 64 voxels, FOV 192 mm, in-plane resolution 3 3 3 mm) for 20
slices (slice thickness 4 mm, 1 mm gap). T1-weighted modied driven
equilibrium Fourier transform images (Ugurbil et al. 1993), matrix
256 3 256, TR 1.3 s, TE 7.4 ms, with a nonslice-selective inversion pulse
followed by a single excitation of each slice (Norris 2000) were used
for registration. For anatomical data, a T1-weighted 3D magnetizationprepared rapid gradient echo sequence was obtained with magnetization preparation consisting of a nonselective inversion pulse (TI 650
ms, TR 1.3 s, snapshot FLASH 10 ms, TE 3.97 ms, angle 10, bandwidth
67 kHz, matrix 256 3 240, slab thickness 192 mm, sagittal orientation,
spatial resolution 1 3 1 3 1.5 mm). To avoid aliasing, oversampling was
performed in read direction (head--foot).
Data Analysis
Data were processed using the software packages LIPSIA (Lohmann
et al. 2001) and FSL (Smith et al. 2004). T1-weighted structural scans
were used for skull stripping, and the brain images were then
coregistered into Talairach space (Talairach and Tournoux 1988).
Twenty-one images without diffusion weighting distributed over the
entire sequence were used to estimate motion correction parameters
using rigid-body transformations (Jenkinson et al. 2002). Motion
correction for the 180 diffusion-weighted images was combined with
a global registration to the T1 anatomy computed using the same
method. The gradient direction for each volume was corrected using

the rotation parameters. The registered images were interpolated to

the new reference frame with an isotropic voxel resolution of 1 mm,
and the three corresponding acquisitions and gradient directions were
averaged. Finally, for each voxel, a diffusion tensor was tted to the
data.
Classical methods such as voxel-wise analysis of DTI data can be
affected by limitations, particularly in the context of investigations into
brain maturation. The registration method may be restricted in its
accuracy by low-dimensional transformations and the heterogeneity of
signal intensities in FA maps. Image smoothing and statistical thresholding are further sources of artifacts (Jones et al. 2005). Therefore, in
the present study tract-based spatial statistics (TBSS) were applied that
allows for the registering of data sets onto the core white matter of one
common target image (Smith et al. 2006).
A skeletonization algorithm that dened a group template for white
matter tracts was applied, and normalized individual maps were
projected onto this template. Skeletonization was employed for
cerebral white matter without the cerebellum and thalamus. The
cerebellum was omitted since it was not entirely included in the FOV
region during scanning. The thalamus was likewise omitted, as this
structure is an intermingled mixture of gray matter nuclei and smaller
white matter structures, and the focus of the study was the main ber
tract pathways in cerebral white matter, particularly frontotemporal
connections.
Voxelwise statistic on the white matter skeleton was used, and
correction for multiple comparisons across voxels was realized by
permutation testing generating corrected cluster-based P values.
Clusters of P < 0.05 were regarded as differing signicantly between
groups.
Anatomical connectivity and ber orientation in human brain white
matter was investigated by ber tracking to compute the connectivity
between cortical brain areas from the diffusion tensor maps (Anwander
et al. 2007). Mean DTI data averaged for each group were examined by
a whole-brain deterministic ber tracking. The measured diffusion
images for each group were aligned by nonlinear registration and
combined to one data set. A diffusion tensor was tted to the combined
data resulting in one averaged diffusion tensor of each voxel in each
group. In that way, the averaging is integrated implicitly in the tensortting procedure to avoid averaging of diffusion tensors. The reconstruction algorithm used the entire diffusion tensor to deect the
estimated ber trajectory (Lazar et al. 2003) as implemented in
MedINRIA according to Fillard et al. (2007). Fiber trajectories were
started in all voxels (voxel size: 1 mm3) with a FA > 0.13. All bers
crossing a seed volume were selected as white matter connections for
a single seed point.
T1 images were normalized on a common source space, and motion
artifacts were ruled out to contribute to any group differences in
functional or diffusion results (see Supplementary Material). Statistical
evaluation of functional activation was based on a general linear
regression with prewhitening (Worsley et al. 2002). Autocorrelation
parameters were estimated from the least squares residuals using Yule-Walker equations. These parameters were subsequently used to whiten
both data and design matrix. Finally, the linear model was re-estimated
using least squares on the whitened data to produce estimates of
effects and their standard errors. The design matrix was generated with
a synthetic hemodynamic response function (Josephs et al. 1997;
Friston et al. 1998) and its rst and second derivatives. Motion
correction parameters and stimulus duration were included as
regressors. For each participant, one contrast image was generated to
represent the main effect of sentence presentation versus baseline.
Individual functional data sets were aligned with the stereotactic
reference space.
Single-participant contrast images were entered into a second-level
random-effects analysis consisting of a 1-sample t-test across the
contrast images of all subjects to indicate whether observed differences
between conditions differed signicantly from zero. Subsequently,
t values were transformed into z scores. Group maps were thresholded
at z > 3.09 (P < 0.001, uncorrected). In order to control for cumulating
alpha errors, only clusters with a volume greater than 270 mm3
(equivalent to 10 voxels) were considered. Blobs of activation were
analyzed in more detail by extracting clusters around local maxima of

activity. Local maxima were identied by peak activation within

a search diameter of 6 voxels (18 mm) in any direction. For a separate
analysis, Brocas area was dened as a primary volume of interest (VOI)
for further investigation in order to dene seed voxels for the
reconstruction of ber bundles. Within this VOI, local maxima were
identied and beta values were compared across the 2 groups. A 2sample t-test was conducted for identifying signicant between-group
differences.

Results
DTI: TBSS
We rst explored differences in white matter FA between
adults and children on the basis of TBSS (Smith et al. 2006).
Direct comparison of FA between adults and children revealed
signicant differences in a number of regions including
perisylvian white matter pathways. Particularly along the dorsal
connection between frontal and temporal cortical regions (AF/
SLF), a number of locations indicated differences in FA between
groups (see Supplementary Figs 1 and 2, Supplementary
Table 1). Underlying language-relevant cortical areas, group
differences were observed in the white matter of the IFG and
the middle and posterior part of the STG/STS (see Fig. 1)
extending down to the border to MTG. In sum, these data argue
for weaker axonal ber connections and/or less myelination of
language-relevant bers in children compared with adults.
In order to identify affected ber pathways more specically,
the language-relevant regions of signicant FA differences
between groups (IFG, STG/STS) were selected as seed regions
of interest for a ber tracking analysis. Results revealed that the
2 regions are directly interconnected via 2 independent
connections, a dorsal pathway via the AF/SLF and a ventral
pathway via the extreme capsule ber system (ECFS) (Fig. 2).
An additional analysis also captured both connections, the AF/
SLF pathway and the ECFS pathway, when investigating the
2 groups separately.
fMRI Data
In a second step, these results of structural differences in white
matter between adults and children were compared with
functional activation patterns during language processing
obtained from the same participants in a separate session.
Behavioral task performance on a sentence comprehension
task revealed no signicant group differences but a trend in
response correctness. Both groups showed high performance
rates: adults 91.2% (SD 1.5), children 86.7% (SD 7.8), t1,9 = 2.18
(P = 0.06).
Functional data revealed activation in IFG and in STG/STS
and also in further brain regions in both groups (see Fig. 3 and
Supplementary Table 2A and 2B). IFG activation in adults in
Brocas area was located in the dorsal portion of BA 44 with
a maximum at 53, 13, 15 (Talairach coordinates, Talairach and
Tournoux 1988, see Fig. 3C, upper row). In children, IFG
activation included BA 44 and also the more anterior portion of
Brocas area BA 45 with a maximum located at 53, 22, 12 in BA
45 (see Fig. 3A, upper row). Direct comparison between
groups in BA 44 and in BA 45 demonstrated stronger activation
in children than in adults in BA 45 at 53, 22, 12 (t1,9 = 3.92, P <
0.001), while in BA 44 no difference was found (t1,9 = 1.01, P =
0.17), that is, adults did not show stronger involvement than
children in neither subregions of Brocas area (Fig. 3B).
Cerebral Cortex February 2011, V 21 N 2 461

Figure 1. TBSS results for differences between adults and children. The figure represents a mean FA image with the white matter skeleton superimposed (in green). Regions of
significant differences between groups are highlighted in red and filled into the local tract region of the mean FA image. They indicate lower FA in children compared with adults (P
\ 0.05, corrected) in the underlying white matter skeleton. Differences between groups are particularly found in Brocas (IFG) and Wernickes language regions (STG/STS down
to the MTG). The figure is displayed in 2 sagittal views and 1 axial view. Lines indicate location of the corresponding sections.

tary Figure 3). Furthermore, tractography using a crossing ber

model (Behrens et al. 2007) revealed again the same dominant
connections from BA 44 to AF/SLF and from BA 45 to the ECFS
(see Supplementary Fig. 4). Hence, the ber tracking reects
valid differences between adults and 7-year olds in the
underlying connectivity of functionally recruited cortical areas
in Brocas area (BA 44 and BA 45) that connect to STG/STS by
separate pathways.

Discussion

Figure 2. Fiber tracking of the DTI data averaged over both groups for the language
relevant regions in IFG (frontal cluster) and STG/STS down to the MTG (temporal
cluster) that showed significant group differences in FA between adults and children
in the white matter skeleton in TBSS. Fibers crossing both regions were regarded
interconnecting bundles. The tracking reveals that these regions are connected via 2
separate white matter pathways, a dorsal connection via AF/SLF (in yellow) and
a ventral connection via the ECFS (in blue). The pathways connecting the 2 significant
regions (red) are shown onto the transparent smoothed white matter skeleton. The
temporal cluster includes indications for the borders between STG, STS, and MTG.

Comparing DTI and fMRI Data

The involvement of Brocas area was further explored by DTIbased tractography seeded on the gray-to-white matter border
that was located closest to the functional activation maxima in
BA 44 and in BA 45 in the IFG. For the BA 44 activation
maximum retrieved from adults at 53, 13, 15, the tractography
seed point originated at 48, 10, 15. For the BA 45 activation
maximum retrieved from children and the direct group
comparison at 53, 22, 12, the seed point originated at 42,
22, 10. Tractography indicated a dorsal pathway from BA 44 via
the AF/SLF connecting the activated area in the IFG with the
STG/STS. For the BA 45 seeding, tractography revealed a ventral
pathway running from the IFG through the ECFS to the STG/
STS (see Fig. 3, lower row).
Fiber tracking was cross-validated for both seeds. In children
as well as in adults, tracking was seeded in both locations (BA
44 and BA 45) within Brocas area as described above. Results
captured for both groups the dorsal pathway via the AF/SLF
when tracking from the BA 44 seed and the ventral pathway via
the ECFS when tracking from the BA 45 seed (see Supplemen462 Neuroanatomical Prerequisites for Language Functions

Brauer et al.

The present ndings indicate that the brains language network

in children has not yet reached the same conned status as in
adults. Rather, when processing sentence, 7-year-old children
rely on alternative pathways connecting the cortical areas
involved in language processing. Functional activation during
auditory language comprehension demonstrates that adults
employ the posterior portion of Brocas area (BA 44), that is,
pars opercularis, while the main center of activation for
children is found in the anterior portion of Brocas area (BA
45), that is, pars triangularis. Direct comparison of functional
recruitment between adults and children reveals that BA 44 is
activated in both adults and children, but that BA 45 of Brocas
area is additionally activated in children. These results suggest
that in children, pars opercularis alone does not sufce to
accomplish the task of comprehending language. Rather,
children appear to need supplementary cortical involvement
within Brocas area. This is achieved by recruiting pars
triangularis. Thus, it appears that children make use of their
dorsal pathway to the temporal cortex, but since this is not
sufcient due to its immaturity, they extent their frontotemporal connection network by making additional use of a ventral
connection via the ECFS which already is more mature.
Previous DTI studies in adults reported convincing results by
tracking white matter connections for coordinates as reported
in existing functional studies on activation during language
tasks (e.g., Glasser and Rilling 2008). Our methodology is
supposedly even more accurate since we used functional and
diffusion data from the identical group of participants.
Tractography data of these centers of language processing in
both groups reveal that the observed activation differences
reect the use of 2 separate pathways connecting the frontal
and the temporal language areas. Children make additional use
of BA 45, which connects via the ventral pathway along ECFS to

Figure 3. Functional activation during auditory language processing and corresponding tractography of fiber tract connections underlying activated regions in Brocas area. Upper
row: activation z maps for children and adults during sentence comprehension versus resting baseline on a T1 reference brain. Both groups activated areas in the perisylvian
cortex in the IFG (Brocas area) and in the STG/STS (Wernickes area). Second-level analysis for both groups resulted with parametric maps of z scores showing local maxima
within the inferior frontal activation in BA 45 ( 53, 22, 12, blue quadrant) for children (A) and in BA 44 ( 53, 13, 15, yellow quadrant) for adults (C) (see also Supplementary
Table 2A and 2B). Direct comparison (2-sample t-test) between groups in Brocas area (B) revealed that children involved BA 45 ( 53, 22, 12) in addition to BA 44 activation,
while adults did not show any stronger activation than children. These results indicate that children make use of both BA 44 and BA 45 in Brocas area, while adults recruit BA 44
only. Lower row: tractography in group averaged DTI data based on functional activation in Brocas area with seed points (turquoise dots) in BA 45 (activation in children and
direct group contrast, A,B) and BA 44 (activation in adults, C). Tracking result of activation in children captured the ECFS (in blue), a ventral connection between inferior frontal
and superior temporal areas. Tracking result of activation in adults captured the AF/SLF (in yellow), a dorsal connection between these areas. In children (A), also the dorsal
connection is displayed as a result of a tractography seeded in the adults activation maximum. Functional data implicate that children make use of the ventral connection in
addition to the dorsal connection that was revealed for adults. The size of the seed voxels was 3 3 3 3 3 mm.

the temporal cortex. Adults who, in contrast to children,

primarily activate BA 44 rely on the dorsal pathway connecting
inferior frontal and superior temporal language areas via the
AF/SLF. In accordance with a previous study, this entire dorsal
pathway is labeled AF/SLF (Saur et al. 2008) since it is still not
yet possible to reliably distinguish between single bers that
are contained in the ber bundles of AF and of SLF that both
originate in Brocas area, but which appear to terminate in
different cortical regions (Catani et al. 2005; Rilling et al. 2008).
It is known from monkey and human data that in this dorsal
connection the SLF, probably with its subdivisions SLF II and
SLF III, coexists with the AF (Makris et al. 2005; Petrides and
Pandya 2006; Schmahmann et al. 2007; Frey et al. 2008). This
dorsal pathway, which is dominant in human adults, appears to
be functionally insufcient in children. The main reason for the
supplementary use of the ventral pathway in the language
network in young children might be the immature nature of
the main connection via the AF/SLF. Direct comparison of ber
tract anisotropy of white matter ber pathways supports this
view. Children show lower values of FA in several white matter
regions. For the perisylvian language-relevant region, these agerelated differences show up particularly along the dorsal
frontotemporal connection via the AF/SLF. Tractography
results suggest that the dorsal connection to the temporal
lobe includes direct connections to the STG/STS in both
groups. However, Figure 3 (lower panel) seems to suggest

between-group differences with much stronger terminations of

the dorsal connection to the STG/STS in adults. Perhaps, it
could be specically a missing branch of the ber tract that is
least developed in children (cf. Supplementary Figure 3). A
closer investigation of this nding, however, by tracking the 2
groups individually seeded from the whole FA difference
clusters did not reveal a strong support for such connectivity
differences between adults and children.
The ventral and the dorsal perisylvian pathways are
expressed differently in humans and other primates. A ventral
pathway along the ECFS has been described in the monkey
brain (Romanski et al. 1999) and in humans (Frey et al. 2008;
Hua et al. 2009; Makris and Pandya 2009). The dorsal pathway
differs between human and nonhuman primates. In the
monkey, the dorsal connection includes bers from different
subpathways of SLF and AF (Petrides and Pandya 2006;
Schmahmann et al. 2007). For the initial parts of the dorsal
connection, this appears to also be valid for the human brain
(Makris et al. 2005; Frey et al. 2008). However, humans differ
from nonhuman primates in that the former display stronger
terminations in the posterior temporal cortex (Rilling et al.
2008).
It is interesting to note that the dorsal pathway in humans
shows a very early asymmetry toward the language-dominant
hemisphere already in infants (Dubois et al. 2009), while on the
other hand it is among those white matter ber connections
Cerebral Cortex February 2011, V 21 N 2 463

that fully mature only very late in human ontogeny (Giorgio

et al. 2008). The present data argue for an immature AF/SLF in
7-year-old children compared with adults as measured by FA. It
is particularly this dorsal ber connection that shows agerelated FA differences in several sections along its pathway. The
structural distinction of the adult brain and the developing
brain converges with functional activation differences in
corresponding cortical language areas during sentence comprehension.
It has been shown that FA in language-related brain regions
can predict performance in language-related tasks (Floel et al.
2009). Reduced FA, as observed in children, is probably a result
of an immature axonal and myelination status (Paus 2010),
which is likely to hamper the speed and accuracy of information
transmission in these fasciculi (Jones 2004). Since impulse
synchrony is important for neuronal association (Hebb 1949)
and information processing (Gollisch and Meister 2008), timing
and precision parameters become crucial for optimal behavioral
and mental performance and also learning. Our results appear to
correspond to results reporting correlations between FA and
cognitive development (Schmithorst et al. 2005) and also to data
on cognitive decline in aging that show additional recruitment
of cortical areas and concurrent decline in white matter FA that
seem to associate to loss of cognitive capacities (Persson et al.
2006). From a distributed network perspective on brain
organization, more difcult information transfer between
cortical regions based on immature white matter would lead
to either smaller and more local networks in children (Fair et al.
2009) or, where this is not possible, to usage of additional
network connections as we observed in our study. Activation
differences between adults and children might reect effects of
task experience, attention, or strategy, even when behavioral
outcomes are comparable. As proposed in a model on development of functional brain activation, experience might be
a factor that interacts independently with both brain morphology and functional activation in relevant brain regions (Lu et al.
2009). Likewise, it is not possible to exclude differences in
maturational or environmental individual histories between our
samples of adults and children that could potentially contribute
to the group differences as described above.
Language processing involves whole networks of brain areas
(for reviews, see Vigneau et al. 2006; Hickok and Poeppel 2007;
Doehrmann and Naumer 2008). The meta-analysis of Vigneau
et al. (2006) indicates that the left IFG and the temporal cortex
are the regions that support sentence processing. By focusing
on activation peaks in Brocas area in our study, we leave other
areas also important for language comprehension undiscussed
(for comprehensive data, see Supplementary Table 2A and 2B).
Insofar, the activation clusters in BA 44 and BA 45 of Brocas
area are not the only areas involved in sentence processing but
certainly necessary parts of the network supporting the
processing of human language.
Our ndings raise the question of causal relations between
structure and function. On the one hand, functional processing
of language might be regarded as being dependent on existing
structural prerequisites, hence arguing for a structure-tofunction causal relationship. Causal inuences of white matter
structure on function can be concluded from studies on white
matter effects on neurotransmitter functions (Roy et al. 2007)
and from myelin-derived effects on axonal and synaptic growth
(McKerracher and Winton 2002). Such interpretations of
structure-to-function relations are, for example, entertained
464 Neuroanatomical Prerequisites for Language Functions

Brauer et al.

for the effects of structural corpus callosum connections on

the perception of speech (Westerhausen et al. 2009). Viewed
from this perspective, the present ndings would argue for
a deviating functional connection between Brocas and
Wernickes language regions in the developing brain. As long
as the main ber connection between these regions via the AF/
SLF is not yet fully mature, supplementary processing centers
(in Brocas area BA 45) and communication pathways (via the
ECFS) between the cortical areas involved in language
comprehension are needed.
On the other hand, changes and adjustments in brain
structure are subject to inuence by functional use. This might
also hold for processes of maturation and development, thus
arguing for a function-to-structure causal relationship. From
this perspective, the present ndings would indicate that
during development, increasing use of language-related brain
circuits and ongoing experience shape the language network
with the AF/SLF nally being the most important connection
between Brocas and Wernickes regions in the mature
language processing system. Such effects of practice and
experience on white matter brain structures during development were observed for nonhuman mammals (Markham and
Greenough 2004) and also for human infants (Als et al. 2004).
Myelinational plasticity of white matter structures could
therefore be regarded as providing a gliagenic mechanism of
behavioral adaptation and learning that corresponds to the
neuronal mechanism of synaptic plasticity. This viewpoint
rests upon empirical ndings supporting the assumption of
environmental effects on white matter morphology (Fields
2008).
Final conclusions on structure--function causalities, however,
are premature and need to be drawn with caution (Aslin and
Schlaggar 2006). Rather, a synthesis of both points of view,
structure-to-function versus function-to-structure causal relations, might be the basis for learning and adaptation throughout
the entire life cycle. During development and maturation, the
brain concentrates processing capacities onto functional
subsystems that are embodied in cortical regions and their
corresponding white matter connections that are hence
strengthened. These specialized regions and their connections
form networks that in turn allow faster and more efcient
information exchange. Beyond these considerations concerning the relationship of development and brain maturation, the
present results also emphasize the brains universal functional
exibility by dening alternative functional networks according
to the structural options at its disposal.

Supplementary Material
Supplementary material
.oxfordjournals.org/.

can

be

found

at:

http://www.cercor

Notes
The authors thank Marc Tittgemeyer for helpful comments on study
design and Yves von Cramon, Roberto Cozatl, and Karsten Muller for
helpful comments on a previous version of the manuscript. Conict of
Interest : None declared.

References
Achenbach TM. 1998. CBCL/4-18 Elternfragebogen uber das Verhalten von Kindern und Jugendlichen [Child Behavior Checklist

(CBCL/4-18)]. Cologne (Germany): Arbeitsgruppe Kinder- Jugendund Familiendiagnostik (KJFD). (Original work published 1991).
Als H, Duffy FH, McAnulty GB, Rivkin MJ, Vajapeyam S, Mulkern RV,
Wareld SK, Huppi PS, Butler SC, Conneman N, et al. 2004. Early
experience alters brain function and structure. Pediatrics.
113:846--857.
Anwander A, Tittgemeyer M, von Cramon DY, Friederici AD,
Knosche TR. 2007. Connectivity-based parcellation of Brocas area.
Cereb Cortex. 17:816--825.
Aslin RN, Schlaggar BL. 2006. Is myelination the precipitating neural
event for language development in infants and toddlers? Neurology.
66:304--305.
Barnea-Goraly N, Menon V, Eckert M, Tamm L, Bammer R,
Karchemskiy A, Dant CC, Reiss AL. 2005. White matter development
during childhood and adolescence: a cross-sectional diffusion tensor
imaging study. Cereb Cortex. 15:1848--1854.
Behrens TEJ, Berg HJ, Jbabdi S, Rushworth MFS, Woolrich MW. 2007.
Probabilistic diffusion tractography with multiple bre orientations:
what can we gain? Neuroimage. 34:144--155.
Bornkessel I, Zysset S, Friederici AD, von Cramon DY, Schlesewsky M.
2005. Who did what to whom? The neural basis of argument
hierarchies during language comprehension. Neuroimage.
26:221--233.
Brauer J, Friederici AD. 2007. Functional neural networks of semantic
and syntactic processes in the developing brain. J Cogn Neurosci.
19:1609--1623.
Brauer J, Neumann J, Friederici AD. 2008. Temporal dynamics of
perisylvian activation during language processing in children and
adults. Neuroimage. 41:1485--1493.
Burock MA, Buckner RL, Woldorff MG, Rosen BR, Dale AM. 1998.
Randomized event-related experimental designs allow for extremely rapid presentation rates using functional MRI. Neuroreport.
9:3735--3739.
Casey BJ, Tottenham N, Liston C, Durston S. 2005. Imaging the
developing brain: what have we learned about cognitive development? Trends Cogn Sci. 9:104--110.
Catani M, Jones DK, Ffytche DH. 2005. Perisylvian language networks of
the human brain. Ann Neurol. 57:8--16.
Cohen Kadosh K, Johnson MH. 2007. Developing a cortex specialized
for face perception. Trends Cogn Sci. 11:367--369.
Constable RT, Pugh KR, Berroya E, Mencl WE, Westerveld M, Ni WJ,
Shankweiler D. 2004. Sentence complexity and input modality
effects in sentence comprehension: an fMRI study. Neuroimage.
22:11--21.
Dittmar M, Abbot-Smith K, Lieven E, Tomasello M. 2008. German
childrens comprehension of word order and case marking in
causative sentences. Child Dev. 79:1152--1167.
Doehrmann O, Naumer MJ. 2008. Semantics and the multisensory brain:
how meaning modulates processes of audio-visual integration. Brain
Res. 1242:136--150.
Dubois J, Dehaene-Lambertz G, Perrin M, Mangin JF, Cointepas Y,
Duchesnay E, LeBihan D, Hertz-Pannier L. 2008. Asynchrony of the
early maturation of white matter bundles in healthy infants:
quantitative landmarks revealed noninvasively by diffusion tensor
imaging. Hum Brain Mapp. 29:14--27.
Dubois J, Hertz-Pannier L, Cachia A, Mangin JF, Le Bihan D, DehaeneLambertz G. 2009. Structural asymmetries in the infant language and
sensori-motor networks. Cereb Cortex. 19:414--423.
Fair DA, Cohen AL, Power JD, Dosenbach NUF, Church JA, Miezin FM,
Schlaggar BL, Petersen SE. 2009. Functional brain networks
develop from a local to distributed organization. PLoS Comput
Biol. 5:e1000381.
Fields RD. 2008. White matter in learning, cognition and psychiatric
disorders. Trends Neurosci. 31:361--370.
Fillard P, Pennec X, Arsigny V, Ayache N. 2007. Clinical DT-MRI
estimation, smoothing, and ber tracking with log-euclidean
metrics. IEEE Trans Med Imaging. 26:1472--1482.
Floel A, de Vries MH, Scholz J, Breitenstein C, Johansen-Berg H. 2009.
White matter integrity in the vicinity of Brocas area predicts
grammar learning success. Neuroimage. 47:1974--1981.

Frey S, Campbell JSW, Pike GB, Petrides M. 2008. Dissociating the

human language pathways with high angular resolution diffusion
ber tractography. J Neurosci. 28:11435--11444.
Friederici AD. 2002. Towards a neural basis of auditory sentence
processing. Trends Cogn Sci. 6:78--84.
Friederici AD. 2009a. Allocating functions to ber tracts: facing its
indirectness. Trends Cogn Sci. 13:370--371.
Friederici AD. 2009b. Pathways to language: ber tracts in the human
brain. Trends Cogn Sci. 13:175--181.
Friederici AD, Bahlmann J, Heim S, Schubotz RI, Anwander A. 2006. The
brain differentiates human and non-human grammars: functional
localization and structural connectivity. Proc Natl Acad Sci U S A.
103:2458--2463.
Friederici AD, Fiebach CJ, Schlesewsky M, Bornkessel ID, von
Cramon DY. 2006. Processing linguistic complexity and grammaticality in the left frontal cortex. Cereb Cortex. 1709--1717.
Friston KJ, Fletcher P, Josephs O, Holmes A, Rugg MD, Turner R. 1998.
Event-related fMRI: characterizing differential responses. Neuroimage. 7:30--40.
Gaillard WD, Hertz-Pannier L, Mott SH, Barnett AS, LeBihan D,
Theodore WH. 2000. Functional anatomy of cognitive development
fMRI of verbal uency in children and adults. Neurology. 54:180--185.
Ghazanfar AA. 2008. Language evolution: neural differences that make
a difference. Nat Neurosci. 11:382--384.
Giorgio A, Watkins KE, Douaud G, James AC, James S, De Stefano N,
Matthews PM, Smith SM, Johansen-Berg H. 2008. Changes in white
matter microstructure during adolescence. Neuroimage. 39:
52--61.
Glasser MF, Rilling JK. 2008. DTI tractography of the human brains
language pathways. Cereb Cortex. 18:2471--2482.
Gollisch T, Meister M. 2008. Rapid neural coding in the retina with
relative spike latencies. Science. 319:1108--1111.
Grimm H, Schoeler H. 1991. Heidelberger Sprachentwicklungstest
(HSET). 2nd ed. Gottingen (Germany): Hogrefe.
Hahne A, Eckstein K, Friederici AD. 2004. Brain signatures of syntactic
and semantic processes during childrens language development.
J Cogn Neurosci. 16:1302--1318.
Hashimoto R, Sakai KL. 2002. Specialization in the left prefrontal cortex
for sentence comprehension. Neuron. 35:589--597.
Hebb DO. 1949. The organization of behavior. New York: John Wiley
xix + 335 pp.
Hickok G, Poeppel D. 2007. Opinionthe cortical organization of
speech processing. Nat Rev Neurosci. 8:393--402.
Hua K, Oishi K, Zhang J, Wakana S, Yoshioka T, Zhang W, Akhter KD,
Li X, Huang H, Jiang H, et al. 2009. Mapping of functional areas in
the human cortex based on connectivity through association bers.
Cereb Cortex. 19:1889--1895.
Jenkinson M, Bannister P, Brady M, Smith S. 2002. Improved
optimization for the robust and accurate linear registration and
motion correction of brain images. Neuroimage. 17:825--841.
Jones DK, Symms MR, Cercignani M, Howard RJ. 2005. The effect of
lter size on VBM analyses of DT-MRI data. Neuroimage.
26:546--554.
Jones R. 2004. Myelin. made to measure. Nat Rev Neurosci. 5:343.
Josephs O, Turner R, Friston K. 1997. Event-related fMRI. Hum Brain
Mapp. 5:243--248.
Kaufman AS, Kaufman NL. 1994. Kaufman assessment battery for
children (K-ABC). [German version: Melchers P, Preuss U.] 2nd ed.
Lisse, Netherlands: Swets & Zeitlinger BV.
Lazar M, Weinstein DM, Tsuruda JS, Hasan KM, Arfanakis K,
Meyerand ME, Badie B, Rowley HA, Haughton V, Field A, et al.
2003. White matter tractography using diffusion tensor deection.
Hum Brain Mapp. 18:306--321.
Lebel C, Walker L, Leemans A, Phillips L, Beaulieu C. 2008.
Microstructural maturation of the human brain from childhood to
adulthood. Neuroimage. 40:1044--1055.
Lenroot RK, Giedd JN. 2006. Brain development in children and
adolescents: insights from anatomical magnetic resonance imaging.
Neurosci Biobehav Rev. 30:718--729.
Lohmann G, Muller K, Bosch V, Mentzel H, Hessler S, Chen L, Zysset S,
von Cramon DY. 2001. LIPSIAa new software system for the

Cerebral Cortex February 2011, V 21 N 2 465

evaluation of functional magnetic resonance images of the human

brain. Comput Med Imaging Graph. 25:449--457.
Lu LH, Dapretto M, OHare ED, Kan E, McCourt ST, Thompson PM,
Toga AW, Bookheimer SY, Sowell ER. 2009. Relationships between
brain activation and brain a structure in normally developing
children. Cereb Cortex. 19:2595--2604.
Makris N, Kennedy DN, McInerney S, Sorensen AG, Wang R,
Caviness VS, Pandya DN. 2005. Segmentation of subcomponents
within the superior longitudinal fascicle in humans: a quantitative,
in vivo, DT-MRI study. Cereb Cortex. 15:854--869.
Makris N, Pandya D. 2009. The extreme capsule in humans and
rethinking of the language circuitry. Brain Struct Funct.
213:343--358.
Markham JA, Greenough WT. 2004. Experience-driven brain plasticity:
beyond the synapse. Neuron Glia Biol. 1:351--363.
McKerracher L, Winton MJ. 2002. Nogo on the go. Neuron. 36:
345--348.
Mukherjee P, Miller JH, Shimony JS, Conturo TE, Lee BCP, Almli CR,
McKinstry RC. 2001. Normal brain maturation during childhood:
developmental trends characterized with diffusion-tensor MR
imaging. Radiology. 221:349--358.
Norris DG. 2000. Reduced power multislice MDEFT imaging. J Magn
Reson Imaging. 11:445--451.
Paus T. 2010. Growth of white matter in the adolescent brain: myelin or
axon? Brain Cogn. 72:26--35.
Persson J, Nyberg L, Lind J, Larsson A, Nilsson LG, Ingvar M, Buckner RL.
2006. Structure-function correlates of cognitive decline in aging.
Cereb Cortex. 16:907--915.
Petkov CI, Logothetis NK, Obleser J. 2009. Where are the human
speech and voice regions and do other animals have anything like
them? Neuroscientist. 15:419--429.
Petrides M, Pandya DN. 2006. Efferent association pathways originating
in the caudal prefrontal cortex in the macaque monkey. J Comp
Neurol. 498:227--251.
Reese TG, Heid O, Weisskoff RM, Wedeen VJ. 2003. Reduction of eddycurrent-induced distortion in diffusion MRI using a twice-refocused
spin echo. Magn Reson Med. 49:177--182.
Rilling JK, Glasser MF, Preuss TM, Ma XY, Zhao TJ, Hu XP, Behrens TEJ.
2008. The evolution of the arcuate fasciculus revealed with
comparative DTI. Nat Neurosci. 11:426--428.
Rivera SM, Reiss AL, Eckert MA, Menon V. 2005. Developmental changes
in mental arithmetic: evidence for increased functional specialization in the left inferior parietal cortex. Cereb Cortex. 15:
1779--1790.
Rogalsky C, Matchin W. 2008. Hickok G. 2008. Brocas area, sentence
comprehension, and working memory: an fMRI study. Front Hum
Neurosci. 2:14.
Romanski LM, Tian B, Fritz J, Mishkin M, Goldman-Rakic PS,
Rauschecker JP. 1999. Dual streams of auditory afferents target
multiple domains in the primate prefrontal cortex. Nat Neurosci.
2:1131--1136.
Roy K, Murtie JC, El Khodort BF, Edgar N, Sardi SP, Hooks BM, BenoitMarand M, Chen CF, Moore H, ODonnell P, et al. 2007. Loss of erbB
signaling in oligodendrocytes alters myelin and dopaminergic

466 Neuroanatomical Prerequisites for Language Functions

Brauer et al.

function, a potential mechanism for neuropsychiatric disorders.

Proc Natl Acad Sci U S A. 104:8131--8136.
Saur D, Kreher BW, Schnell S, Kammerer D, Kellmeyer P, Vry M-S,
Umarova R, Musso M, Glauche V, Abel S, et al. 2008. Ventral and dorsal
pathways for language. Proc Natl Acad Sci U S A. 105:18035--18040.
Schlaggar BL, Church JA. 2009. Functional neuroimaging insights into
the development of skilled reading. Curr Dir Psychol Sci. 18:21--26.
Schmahmann JD, Pandya DN, Wang R, Dai G, DArceuil HE, de
Crespigny AJ, Wedeen VJ. 2007. Association bre pathways of the
brain: parallel observations from diffusion spectrum imaging and
autoradiography. Brain. 130:630--653.
Schmithorst VJ, Wilke M, Dardzinski BJ, Holland SK. 2005. Cognitive
functions correlate with white matter architecture in a normal
pediatric population: a diffusion tensor MRI study. Hum Brain Mapp.
26:139--147.
Smith SM, Jenkinson M, Johansen-Berg H, Rueckert D, Nichols TE,
Mackay CE, Watkins KE, Ciccarelli O, Cader MZ, Matthews PM, et al.
2006. Tract-based spatial statistics: voxelwise analysis of multisubject diffusion data. Neuroimage. 31:1487--1505.
Smith SM, Jenkinson M, Woolrich MW, Beckmann CF, Behrens TEJ,
Johansen-Berg H, Bannister PR, De Luca M, Drobnjak I, Flitney DE,
et al. 2004. Advances in functional and structural MR image analysis
and implementation as FSL. Neuroimage. 23:S208--S219.
Snijders TM, Vosse T, Kempen G, Van Berkum JJA, Petersson KM,
Hagoort P. 2009. Retrieval and unication of syntactic structure in
sentence comprehension: an fMRI study using word-category
ambiguity. Cereb Cortex. 19:1493--1503.
Stromswold K. 1996. Genes, specicity, and the lexical/functional
distinction in language acquisition. Behav Brain Sci. 19:648ff.
Talairach J, Tournoux P. 1988. Co-planar stereotactic atlas of the human
brain. New York: Thieme.
Tyler LK, Marslen-Wilson W. 2008. Fronto-temporal brain systems
supporting spoken language comprehension. Philos Trans R Soc
Lond B Biol Sci. 363:1037--1054.
Ugurbil K, Garwood M, Ellermann J, Hendrich K, Hinke R, Hu XP,
Kim SG, Menon R, Merkle H, Ogawa S, et al. 1993. Imaging at high
magnetic eldsinitial experiences at 4 T. Magn Reson Q.
9:259--277.
Vigneau M, Beaucousin V, Herve PY, Duffau H, Crivello F, Houde O,
Mazoyer B, Tzourio-Mazoyer N. 2006. Meta-analyzing left hemisphere language areas: phonology, semantics, and sentence processing. Neuroimage. 30:1414--1432.
Weiller C, Musso M, Rijntjes M, Saur D. 2009. Please dont underestimate the ventral pathway in language. Trends Cogn Sci.
13:369--370.
Westerhausen R, Gruner R, Specht K, Hugdahl K. 2009. Functional
relevance of interindividual differences in temporal lobe callosal
pathways: a DTI tractography study. Cereb Cortex. 19:1322--1329.
Wilke M, Lidzba K, Krageloh-Mann I. 2009. Combined functional and
causal connectivity analyses of language networks in children:
a feasibility study. Brain Lang. 108:22--29.
Worsley KJ, Liao CH, Aston J, Petre V, Duncan GH, Morales F, Evans AC.
2002. A general statistical analysis for fMRI data. Neuroimage.
15:1--15.