RESEARCH REPORT

A TEMPORAL LOBE FACTOR IN VERB FLUENCY

Per Östberg*, Raffaella M. Crinelli*, Rimma Danielsson, Lars-Olof Wahlund, Nenad Bogdanovic
and Sven-Erik Fernaeus
(Section for Clinical Geriatrics, Neurotec Department, Karolinska Institutet, Karolinska University Hospital, Huddinge,
Stockholm, Sweden; *Equal contribution)

ABSTRACT

Verb fluency requires self-sustained verb retrieval. The brain correlates of this task are virtually unknown. We
investigated the relations between verb and noun (semantic) fluency and regional brain perfusion in subjects with varying
degrees of cognitive decline, ranging from very mild subjective impairment to Alzheimer’s disease (AD). Data consisted of
single-photon emission computed tomography (SPECT) data and temporally resolved verb and noun fluency scores from
93 participants. Impaired verb fluency was predicted by a temporal lobe hypoperfusion factor and low education, whereas
high age and low perfusion in the parietotemporal-occipital region predicted impaired noun fluency. Analysis of perfusion
within the temporal region indicated primary involvement of the temporal pole and medial temporal lobe in AD. This
might reflect pathology of the anterior parahippocampal region, which appears early in neurodegenerative disease. Although
temporal lobe structures have not usually been implicated in verb processing, early temporal pathology thus appears to
contribute to impaired verb fluency in cognitive decline.

Key words: dementia, language, mild cognitive impairment, nouns, verbs

INTRODUCTION lexical category: things, or more specifically,

Verb and noun deficits figure prominently among
neurocognitive dissociations. In fact, the double
dissociation of impaired verb/noun production was
already documented in the mid-18th century (Denes
and Dalla Barba, 1998; Östberg, 2003). The
linguistic basis for such observations goes back to
antiquity, however, with the division of words into
eight parts-of-speech by Dionysius Thrax (Robins,
1989). In Dionysius’ classification, a verb or rhēma
lacked case inflection but was inflected for tense,
person, and number. Semantically, verbs were
defined as signifying an activity or a process
performed or undergone. A noun or ónoma, in
contrast, was characterized by case inflection and by
signifying a concrete or abstract entity.
Today, verbs and nouns are considered
universally distinct lexical categories, although a
few languages appear to lack the distinction
between nouns and verbs (see Baker, 2003, for
discussion). The brain correlates of verb and noun
processing remain unclear. In particular, it remains
obscure whether verbs and nouns as lexical
categories depend on differential brain systems or
regions. Verb deficits are generally held to be
associated with frontal lobe lesions, whereas noun
deficits are associated with posterior lesions, in
particular of the left temporal lobe. This general
conclusion, however, is weakened by the tacit
assumption of a one-to-one relationship between
lexical category and conceptual category.
Experimental stimuli have typically been drawn
from conceptual categories associated with each
physical objects for nouns; and (human) activities
for verbs. Nouns certainly express all physical
object concepts but can also express virtually any
other conceptual category, including state or
activity concepts normally associated with verbs.
PET studies with conceptually matched noun and
verb stimuli for lexical decision have yielded
largely overlapping or identical brain activation
patterns (Perani et al., 1999; Tyler et al., 2001). It
would appear that the complex interrelationships
between lexical category, conceptual structure, and
phonological structure in lexical items preclude a
straightforward neural localization of noun and
verb processing (Black and Chiat, 2003). It has
been proposed, however, that verb processing
requires more executive resources than noun
processing, implying that disproportionate problems
with verbs are to be expected in various brain
pathologies. Action naming was indeed more
impaired than object naming in frontotemporal
dementia, and correlated with executive
dysfunction rather than linguistic dysfunction as in
Alzheimer’s disease (AD) (Silveri et al., 2003; see
also Cappa et al., 1998).
Recent studies have employed the verb fluency
task as a measure of verb processing (Piatt et al.,
1999a, 1999b, 2004; Östberg et al., 2005). Verb
fluency requires self-sustained enumeration of
activity verbs during one minute. It may be
construed as a “pure” measure of verb retrieval in
contrast to action picture naming, which involves
visual perception and access to semantics from the
visual system, and in contrast to verb generation in

Cortex, (2007) 43, 607-615

608 Per Östberg and Others
which verbs are produced in response to stimulus
nouns. In contrast to lexical decision, verb fluency
also engages the speech production system. It is
disproportionately impaired, as compared to noun
and letter-based word fluency, in Parkinson’s
disease (PD) with dementia (Piatt et al., 1999a).
Piatt et al. (1999a) suggested that the task is
sensitive to the striatal loop pathology associated
with PD. A comparable verb fluency deficit was
observed in persons with HIV-1 disease, which is
also associated with striatal loop pathology (Woods
et al., 2005). Verb fluency has been studied in
healthy elderly subjects (Piatt et al., 1999b, 2004).
It was found to be in modest to moderate relation
to tests of executive function, but unrelated to
episodic memory and picture (i.e. noun) naming.
This suggests that verb fluency taps aspects of
executive function not measured by standard tests.
Further statistical evidence for distinctness of the
verb fluency test comes from factor analyses of
verb, letter-based, and noun (category) fluency, in
which verb fluency loads on a separate factor
(Östberg et al., 2005).
Verb fluency contrasts with letter-based (or
phonemic) fluency, that requires the enumeration of
words that begin with a given letter, and with noun
fluency (or category fluency or semantic fluency),
that requires subjects to enumerate nouns from a
taxonomy such as animals or a semantic field such
as supermarket items. Noun fluency is particularly
impaired in AD (Monsch et al., 1997), semantic
dementia (Hodges et al., 1992), schizophrenia
(Kremen et al., 2003), depression (Fossati et al,
2003; c.f., Ravdin et al., 2003), and – as compared
to letter-based fluency – PD (Henry and Crawford,
2004).
In contrast to verb fluency, noun fluency and
letter-based fluency are widely used in the
assessment of dementia and other cognitive
disorders. Although both verb and noun fluency are
semantically oriented fluency tasks, they differ in
important regards. First, nouns are typically used
for referring, whereas verbs are typically used to
predicate. The task of enumerating single verbs
may therefore be considered less “natural” for the
speaker than enumerating single nouns from a
well-defined category. Secondly, a nominal
category such as animals represents a taxonomy.
This means that a particular hierarchical sense
relation, taxonymy (Cruse, 1986), holds between
the target words. The targeted word set is therefore
characterized by a relatively “tight” configuration.
The word set targeted by verb fluency, in contrast,
is less tightly defined (‘human activities’), with
limited taxonomical structure. Some taxonomical
structure or troponymy does exist among verbs but
is more shallow and “bushy” than taxonymy
among nouns (Fellbaum, 1998). Verb taxonomies
thus tend to have fewer hierarchical levels than
noun taxonomies, with a relatively large number of
subordinate verbs (troponyms) for each
superordinate. Thirdly, the “lead-in” concepts for
verb and noun fluency differ in clarity: compare
“things that people do” with “animals”.
PD and AD are neurodegenerative diseases that
may have relatively long preclinical and
oligosymptomatic phases. Patients do not meet
clinical diagnostic criteria until the
neuropathological process has spread into several
brain regions and caused marked devastation in
structures that are affected early (Braak and Braak,
1991; Braak et al., 2003). Up to now, however, the
brain correlates of verb and noun fluency deficits
have not been studied in subjects with milder
grades of cognitive decline, i.e., with less severe
brain involvement. The neuropathological lesions
of Alzheimer’s and Parkinson’s diseases appear
very early in the anterior parahippocampal region
including the perirhinal and entorhinal cortex.
Severe neuronal loss in this region can precede
dementia with several years, and in PD
parahippocampal atrophy also appears in the same
stage of the disease process as the well-known
nigrostriatal pathology (Braak et al., 2003). Since
lesions in the parahippocampal region not only
produce deficits in recognition memory but also in
executive function (Davachi and Goldman-Rakic,
2001; Chavoix et al., 2002), it is possible that verb
fluency is affected not only by striatal loop
pathology but also by early temporal lobe
neurodegenerative changes that isolate the higher
centers of the limbic loop (Östberg et al., 2005).
The present study was aimed at investigating a
possible relation between impaired performance in
verb and noun fluency tasks and perfusion deficits
in specific brain areas in subjects with different
levels of cognitive impairment. In particular, we
hypothesized that signs of temporal lobe pathology
are associated with decreased verb fluency. Toward
that end, we used regional hypoperfusion factors
derived from single-photon emission computed
tomography (SPECT) and temporally resolved verb
and noun fluency scores from participants assessed
at a memory disorders clinic, ranging from very
mild subjective cognitive decline to mild (objective)
cognitive impairment and AD. Mild cognitive
impairment (MCI) is associated with a strongly
increased risk for dementia of the Alzheimer’s type;
indeed up to 84% of MCI patients show Alzheimer-
type changes on neuropathological examination
(Morris et al., 2001).
MATERIAL AND METHOD
Participants
The participants (n = 93) were examined at the
memory clinic, Department of Geriatrics,
Karolinska University Hospital, Huddinge, because
of cognitive complaints. This is a secondary/tertiary
referral center with referrals from e.g. occupational
 Temporal factor in verb fluency
TABLE I
Means and standard deviations in age, years of education and
MMSE scores for the three diagnostic groups in this study
Age Education MMSE
AD 64.8; 9.4 11.6; 4.0 23.4; 3.4
MCI 60.5; 8.7 12.5; 4.0 27.6; 2.1
SCI 56.9; 7.4 12.9; 2.7 29.3; 0.8
physicians, general practitioners, and neurologists.
Patients are relatively young and in general do not
have cerebrovascular disease (see below). The
examination included physical and psychiatric
evaluations, computed tomography or magnetic
resonance imaging scans, electroencephalography,
cerebral single-photon emission computed
tomography, lumbar puncture, blood analyses, and
neuropsychological assessment. All participants
spoke Swedish. Participants were drawn from three
diagnostic categories reflecting three levels of
cognitive functioning: Subjective Cognitive
Impairment (n = 30), Mild Cognitive Impairment (n
= 30), and Alzheimer’s disease (n = 33).
Descriptive statistics for the participants’ age,
education, and Mini-Mental State Examination
(MMSE) scores are given in Table I.
Subjective Cognitive Impairment (SCI)
SCI was diagnosed in cases of memory
complaint without objective signs of cognitive
decline. These patients had been referred for
cognitive complaints and had been thoroughly
assessed without any signs of cognitive impairment
being found. SCI corresponds to level 2 of the
Global Deterioration Scale, i.e. very mild cognitive
decline or Age-Associated Memory Impairment,
although objective impairment was ruled out by
neuropsychological assessment in our patients. SCI
is likely to contain many normal individuals but
might also possibly reflect very early neuronal loss
or Alzheimer-type changes in the anterior medial
temporal lobe (transentorhinal stages I or II, Braak
and Braak, 1991; low limbic stage, Mesulam, 2000).
Mild Cognitive Impairment (MCI)
MCI was diagnosed according to the following
criteria (Wahlund et al., 2003): (a) subjective
memory complaint; (b) objective signs (1.5
standard deviation units below age-matched
controls) of decline in any cognitive domain; (c)
intact activities of daily living; (d) not fulfilling the
DSM-IV or ICD-10 criteria for dementia. This
definition of MCI might also cover preclinical
stages of non-Alzheimer neurodegenerations;
patients that fulfilled criteria for frontotemporal
lobar degeneration or PD were not diagnosed as
MCI, however. Pathoanatomically, MCI may reflect
more severe Alzheimer-type changes in the
temporal lobe than in SCI (limbic stages III and IV,
609
Braak and Braak, 1991; high limbic stage,
Mesulam, 2000). Magnetic resonance imaging
disclosed two small cerebral infarctions (right
parietal and left temporal) in one MCI case; this
was regarded a case of vascular cognitive
impairment. The other participants with MCI did
not show evidence of cerebrovascular pathology.
Alzheimer’s Disease (AD)
AD was diagnosed according to the International
Classification of Diseases, tenth edition (ICD-10).
AD is usually diagnosed when the Alzheimer-type
changes have caused widespread destruction of
neocortical areas, i.e. in advanced stages of
pathological changes (neocortical stages V and VI,
Braak and Braak, 1991; low and high neocortical
stages of Mesulam, 2000). The overall cognitive
impairment in this group with a clinical diagnosis of
AD was mild (mean MMSE score 23.4).
It should be noted that there were significant
differences between the three diagnoses in age, F (2,
90) = 6.58, p < .01, and in MMSE scores, F (2, 89) =
50.72, p < .001, but not in years of education. Fisher
least significant difference (LSD) post-hoc analysis
showed significant differences in age between AD
and SCI. Moreover, there were significant
differences in MMSE scores between AD and MCI,
AD and SCI, and between MCI and SCI.
Cerebral Single-Photon Emission Computed
Tomography (SPECT)
Acquisition
All participants underwent a SPECT scan at the
Department of Radiology, Karolinska University
Hospital, Huddinge. The scan was performed as
part of a routine work-up for dementia. Data were
acquired as follows. Each participant was injected
with 1000 Mbq Tc-99m hexamethylpropylenamine
oxime (HMPAO; Ceretec, Amersham Ltd). Data
acquisition started 30 minutes after the injection. A
single-headed rotating gamma camera (Siemens
Diacam) was used for collecting data in 64
projections, evenly spread through 360°. Total
acquisition time was 32 minutes. Tomographic
slices were reconstructed using an iterative
algorithm (Hosem, Nuclear Diagnostics AB,
Sweden) with Chang attenuation correction. The
attenuation coefficient was 0.12 cm– 1. The data
were formatted as a 3D data set with 64 × 64 × 64
cubic voxels with 3.5 mm sides. The resolution in
a tomographic slice was 10.2 mm full width at half
maximum (FWHM). The data sets were post-
filtered with a Butterworth filter, cutoff 1.0 cm– 1.
BRASS Analysis
The BRASS software was used for image
registration and quantification (Radau et al., 2001).
610 Per Östberg and Others

The data sets were iteratively registered using 9 TABLE II

parameter linear registration to a normal template, SPECT variables and abbreviations

using normalized mutual information as similarity Variable Abbreviation
function. The relative cerebral blood flow in the
Left lentiform nucleus* LNL
selected regions was calculated as cerebellar ratios Right lentiform nucleus* RNL
(mean count per voxel of region/mean count per Left caudate nucleus LCN
voxel of bilateral cerebellar cortex). The Right caudate nucleus RCN
Left thalamus LT
information obtained from the SPECT scans Right thalamus RT
consisted of measurements of relative blood flow Left sensorimotor cortex LSCTX
(cerebellar ratios) in 42 brain areas normalized to Right sensorimotor cortex RSCTX
four regions of the cerebellum (see Table II with Left occipital cortex LOCTX
Right occipital cortex ROCTX
abbreviations). The use of the cerebellum as a Left superior parietal lobule LSPL
reference region is a better choice when Right superior parietal lobule RSPL
investigating AD subjects than global cerebral blood Left anterior dorsal frontal cortex LADFCTX
flow. This is because large parts of the brain can be Right anterior dorsal frontal cortex RADFCTX
Left posterior dorsal frontal cortex LPDFCTX
affected in AD, thus lowering the global ratio (Syed Right posterior dorsal frontal cortex RPDFCTX
et al., 1992). Four of the 42 regions did not show Left anterior orbital frontal cortex LAOFCTX
hypoperfusion in any of the subjects: left and right Right anterior orbital frontal cortex RAOFCTX
lentiform nuclei and left and right anterior cingulate Left posterior orbital frontal cortex LPOFCTX
Right anterior orbital frontal cortex RPOFCTX
gyri. These were excluded from further analysis.
Left parietotemporal cortex LPTCTX
Right parietotemporal cortex RPTCTX
Word Fluency Tasks Left medial temporal lobe LMTL
Right medial temporal lobe RMTL
Participants performed two fluency tasks: noun Left lateral temporal lobe LLTL
Right lateral temporal lobe RLTL
fluency and verb fluency. Scores were divided into Left posterior temporal lobe LPTL
six 10-seconds intervals as described in Fernaeus and Right posterior temporal lobe RPTL
Almkvist (1998), according to which the number of Left temporal pole LTP
appropriate words was recorded at each interval. Right temporal pole RTP
Left insular cortex LICTX
Noun fluency was measured using the animal Right insular cortex RICTX
naming task. Here, subjects were requested to name Left anterior cingulate gyrus* LACG
as many sorts of animals as possible during one Right anterior cingulate gyrus* RACG
minute. Verb fluency was measured by the verb or Left posterior cingulate gyrus LPCG
Right posterior cingulate gyrus RPCG
action fluency task (Piatt et al., 1999a, 1999b).
Subjects were instructed to tell by single words as Pons and midbrain P&M
Left anterior subcortical LAS
many things as possible that a person can do, Right anterior subcortical RAS
exemplified by “swim” and “smell”. Activity verbs Left posterior subcortical LPS
Right posterior subcortical RPS
such as “cut” but also state verbs such as “love” were Other subcortical OS
scored as appropriate responses. Nouns derived from
*Not analyzed further.
verbs and phrases using the same verb were not
accepted (for example, ‘drive a car’, ‘drive a train’).
analyses were applied, with age, education and
Statistical Analysis mean hypoperfusion factor scores as predictor
variables and verb fluency and noun fluency mean
Statistical analyses were conducted using the scores per interval as dependent variables. This was
Statistica package (4.1) for PC. Principal factor done in order to find out whether specific
analysis with Varimax rotation was undertaken (i) hypoperfusion factors as well as age and education
on noun and verb fluency interval scores (12 would predict noun or verb fluency performance.
variables in sum) to determine the number of
factors involved in verb and noun fluency Ethical Approval
performance in the sample and (ii) on SPECT
regional perfusion quotes to uncover patterns of The Ethical Committee at Karolinska Institutet,
decreased perfusion spanning larger anatomical Karolinska University Hospital, Huddinge,
regions. Multivariate analysis of variance approved this study (registration number 342/03).
(MANOVA) was applied in order to assess (i)
differences in hypoperfusion factor scores between
the three diagnostic groups; (ii) effects of RESULTS
diagnostic group and type of fluency test (noun vs.
verb); (iii) effects of diagnostic group and fluency Factor Analysis of Perfusion Patterns
score interval (noun 1-6, verb 1-6). Fisher LSD
post-hoc analysis was used to assess whether Principal factor analysis of the SPECT data
differences were significant. Multiple regression revealed seven factors with eigenvalues above one.
 Temporal factor in verb fluency 611

TABLE III

Varimax rotated factor loadings based on SPECT data

Hypoperfusion factors
SBC DF-C OF TL PTO BS SP-C
Variable
LCN – .74 – .13 .22 – .22 – .02 – .17 – .21
RCN – .81 – .09 .18 – .14 – .07 – .22 – .16
LT – .78 – .22 .10 – .30 – .05 – .07 – .18
RT – .76 – .12 .05 – .29 – .07 – .15 – .20
LSCTX .04 – .73 – .12 – .11 – .17 .05 .55
RSCTX – .02 – .76 – .06 – .09 – .16 .02 .56
LOCTX – .04 – .18 – .21 – .24 – .75 .07 .30
ROCTX – .09 – .24 – .19 – .14 – .77 .21 .25
LSPL .24 – .44 – .14 – .03 – .23 – .03 .75
RSPL .18 – .45 – .07 – .01 – .28 .02 .75
LADFCTX – .10 – .88 – .23 .06 – .08 – .00 – .04
RADFCTX – .07 – .87 – .19 .01 – .17 – .05 – .06
LPDFCTX – .17 – .90 – .10 – .03 – .13 – .01 .19
RPDFCTX – .18 – .90 – .07 – .11 – .09 – .03 .19
LAOFCTX .11 – .20 – .89 .01 – .15 .09 – .02
RAOFCTX .11 – .18 – .89 .03 – .25 .02 .04
LPOCTX – .25 – .26 – .72 – .37 .07 .10 .10
RPOCTX – .27 – .23 – .75 – .36 .08 .02 .20
LPTCTX – .44 – .37 – .01 – .35 – .59 – .07 .11
RPTCTX – .35 – .46 .05 – .32 – .62 .01 – .00
LMTL – .25 .22 – .44 – .66 – .12 .04 .10
RMTL – .26 .12 – .42 – .61 – .06 .10 .09
LLTL – .34 – .21 – .17 – .75 – .29 .10 – .01
RLTL – .31 – .13 – .06 – .78 – .32 .04 – .02
LPTL – .52 – .16 – .00 – .58 – .41 .14 – .07
RPTL – .18 – .14 .07 – .58 – .36 .08 – .22
LTP – .22 – .00 – .02 – .77 .00 .14 .02
RTP – .26 .06 – .01 – .79 – .05 – .02 .14
LICTX – .60 – .31 – .19 – .50 .02 .19 .05
RICTX – .62 – .26 – .20 – .47 .12 .06 .01
LPCG – .58 – .08 – .16 – .06 – .35 .11 .32
RPCG – .75 .07 – .16 – .11 – .25 .24 .16
P&M – .21 .02 – .20 – .15 – .08 .85 .08
LAS – .86 – .14 – .17 – .23 .11 .12 .06
RAS – .87 – .06 – .18 – .21 .11 .05 .05
LPS – .87 .0 – .06 – .23 – .27 .05 – .05
RPS – .85 .14 – .07 – .21 – .28 .07 – .07
OS .18 .03 .02 – .16 – .07 .90 – .06
Variance .22 .15 .10 .14 .08 .05 .05
Note. Loadings ≥ .55 are boldface. For abbreviations, see Table II.

These factors were Varimax rotated and labeled by
their highest loadings from the anatomical regions;
they are listed below (see Table III for factor
loadings).
– The Subcortical (SBC) Factor included: Left
and right caudate nucleus, left and right thalamus,
left and right insular cortex, left and right posterior
cingulate gyrus, left and right anterior subcortical,
and left and right posterior subcortical.
– The Dorsofrontal-Central (DF-C) Factor
included: Left and right sensorimotor cortex, left
and right anterior dorsal frontal cortex, and left and
right posterior dorsal frontal cortex.
– The Orbitofrontal (OF) Factor included: Left
and right anterior orbital frontal cortex and left and
right posterior orbital frontal cortex.
– The Temporal Lobe (TL) Factor included:
Left and right medial temporal lobe, left and right
lateral temporal lobe, left and right posterior
temporal lobe, and left and right temporal pole.
– The Parietotemporal-Occipital (PTO) Factor
included: Left and right occipital cortex and left
and right parietotemporal cortex.
– The Brainstem (BS) Factor included: Pons
and medulla and “other subcortical”.
– Finally, the Superoparietal-Central (SP-C)
Factor included: Left and right sensorimotor cortex
and left and right superior parietal lobule.
As can be seen in Table III there was decreased
perfusion in both hemispheres for all seven factors.
Diagnostic Differences in Large-Scale
Hypoperfusion Patterns
Hypoperfusion factors scores for each subject
were derived on the basis of the seven-factor
solution. Further analysis of the hypoperfusion
factor scores using MANOVA revealed significant
differences between AD on one hand and SCI and
MCI on the other in SPECT factor scores overall,
F (2, 89) = 11.92, p < .001. Post-hoc analysis
showed significant differences between SCI on one
hand and MCI and AD on the other in the SBC
factor (see Figures 1 and 2; AD vs. SCI: LSD, p <
.001; MCI vs. SCI: LSD, p < .02) and significant
differences between AD and SCI in the TL factor
612 Per Östberg and Others

TABLE IV
Factor loadings derived from Principal Factors analysis of noun
fluency (ANIM1-6) and verb fluency (VERB1-6) interval scores

Variable NF factor VF factor

ANIM1 .58 .40
ANIM2 .72 .35
ANIM3 .81 .19
ANIM4 .76 .21
ANIM5 .67 .18
ANIM6 .71 .29
VERB1 .39 .64
VERB2 .35 .62
VERB3 .32 .68
VERB4 .24 .68
Fig. 1 – Mean factor scores in the diagnostic groups AD, VERB5 .28 .66
MCI and SCI across the seven hypoperfusion factors. VERB6 .04 .78
1: Subcortical; 2: Dorsofrontal-Central; 3: Orbitofrontal; 4:
Temporal Lobe; 5: Parietotemporal-Occipital; 6: Brainstem; 7: Variance 30 27
Superoparietal-Central. Note. Loadings > .55 are boldface.

lobe, and the posterior temporal lobe. A significant

interaction was found, Rao R (6, 176) = 2.47, p <
.05. This indicated significant differences in the
temporal pole and medial temporal lobe between
SCI and MCI on one hand and AD on the other,
with decreased perfusion in AD. There was also a
significant difference between MCI and SCI in the
temporal pole, with decreased perfusion in MCI
(LSD, p < .01).

Factor Analysis of Noun and Verb Fluency

Fig. 2 – Mean scores in hypoperfusion for the TL (Temporal
Lobe) factor in the three diagnostic groups. Fluency data were also factor-analyzed using
the six interval scores from noun fluency and verb
fluency, i.e. 12 variables in all as input to the
Principal Factors analysis. Two factors, the noun
fluency factor and the verb fluency factor, emerged
after Varimax rotation. Table IV shows that the two
factors correspond to the two types of test (noun
vs. verb); the factors explain 30 and 27 percent
each of the total variance.

Effects of Diagnostic Group, Fluency Type, and

Fluency Interval

There was a significant main effect of diagnostic

Fig. 3 – Mean scores in noun and verb fluency.
(LSD, p < .05). There were no significant
differences between MCI and the other diagnoses
in the TL factor, see Figures 1 and 3. In the PTO
factor there was a significant difference between
SCI and MCI (LSD, p < .05) and between AD and
SCI on the other, see Figure 1 (LSD, p < .001).
Effects of Diagnostic Group on Regional Temporal
Lobe Perfusion
The eight temporal SPECT variables loading on
the TL factor were included in a MANOVA with
diagnostic group (AD, MCI, SCI) as independent
variable. The dependent variables were the SPECT
scores (for each hemisphere) from the temporal
pole, the medial temporal lobe, the lateral temporal
group in fluency performance, F (2, 89) = 46.7, p <
.001. There was also a significant effect of type of
test, F (1, 90) = 63.56, p < .001, indicating higher
performance in noun fluency than in verb fluency.
Moreover, there was a significant interaction
between type of fluency and diagnostic group, F (2,
90) = 7.22, p < .01, indicating a larger difference
between SCI and MCI on one hand and AD on the
other in noun fluency, see Figure 3. There was also
a significant effect of interval, F (5, 450) = 119.85,
p < .001, and a significant interaction between type
of test and interval, F (5, 450) = 2.40, p < .05,
indicating higher scores in the initial intervals in
noun fluency. Fisher LSD post-hoc analysis showed
a significant difference in noun fluency tasks
between SCI and MCI on one hand and AD on the
other (p < .01) and a significant difference in verb
fluency between AD and SCI (p < .001).
 Temporal factor in verb fluency
TABLE V
Prediction of fluency factors based on multiple regression analysis
a. Prediction of noun fluency
Variable BETA St. Err. of BETA
Age – .44 .09
PTO – .21 .09
b. Prediction of verb fluency
Variable BETA St. Err. of BETA
Education .42 .09
TL – .24 .09
Prediction of Verb and Noun Fluency Using
Hypoperfusion Factors, Age, and Education
Multiple regression analyses were applied
including the noun fluency and verb fluency factors
as predicted variables and age, education and the
hypoperfusion factors as predictor variables. With a
strict criterion of including variables only with F-
values for beta weights > 4, two variables
predicted noun fluency, R = .47, p < .001, namely
age and the PTO factor (see Table Va).With the
application of the same criterion, two variables
predicted verb fluency, R = 49, p < . 001, namely
education and the TL factor (see Table Vb).
DISCUSSION
This study used factor analysis as a method of
distinguishing large-scale patterns of decreased
brain perfusion and two types of word fluency.
Temporally resolved verb and noun fluency scores
loaded on different factors, indicating that these
tasks are indeed distinct as proposed by Piatt et al.
(1999a, 1999b). Verb and noun fluency were also
associated with partly different anatomical regions.
Our data did not point toward a lateralized neuronal
substrates of verb and noun fluency. This is not to
be expected, however, as semantic neuronal sets are
thought to be bihemispherically represented
(Pulvermüller, 2003) and the statistical techniques
used here detect correlations rather than differences.
Decreased noun fluency was predicted by
higher age (see Kozora and Cullum, 1995) and
decreased perfusion in the parietotemporal-occipital
regions. This makes sense, since these regions
support object knowledge and thus concrete noun
processing (Pulvermüller, 2003; Fuster, 2003) and
degenerate heavily in PD (Braak and Braak, 1991).
Verb fluency, in contrast, was predicted by
education (see Piatt et al., 2004) and a temporal
hypoperfusion factor. Awareness of verbs as a
grammatical category is obviously not necessary
for producing verbs in response to the instruction
“things that people do”. Well-educated subjects
may however be better prepared for the task by
using a grammatical frame such as the sign of the
infinitive for self-prompting; this might be one
613
B St. Err. of B t(92) p-level
– .05 .01 – 4.75 .001
– .21 .09 – 2.38 .020
B St. Err. of B t(92) p-level
.12 .03 4.6 .001
– .24 .09 – 2.7 .010
explanation for the positive influence of
educational level on verb fluency. Within the
temporal region, the temporal poles and the medial
temporal lobes were primarily affected in MCI and
AD, with a significant decrease in perfusion in the
temporal poles already in MCI. This temporal
subregion contains the anterior parahippocampal
region, including the perirhinal and entorhinal
cortices that are affected early and severely by the
neuropathological changes seen in Alzheimer’s and
PD (Braak and Braak, 1991; Mesulam, 2000;
Braak et al., 2003). These areas are not discernible
on SPECT scans, but their involvement can be
inferred by hypoperfusion observed in adjacent
areas. The anterior parahippocampal region has not
been implicated in verb processing previously.
Rather, verbs are held to depend on frontal
networks, and verb deficits are typically observed
in subjects with frontal lesions (Damasio and
Tranel, 1993).
One explanation for the temporal lobe factor in
verb fluency may be inherent task differences
between verb and noun fluency. Such differences
exist between the word relations specific to each
task. Thus, noun fluency – as represented by the
animal naming task – targets a taxonomy, a word
set that is familiar, hierarchically structured, and
easily defined. The semantic field targeted by verb
fluency is less coherent, requiring actively
structured retrieval. The perirhinal cortex is
important in this regard through its connections to
the frontal motor cortex (Kyuhou and Gemba,
2002) and subcortical projections to the basal
ganglia, basal forebrain, and amygdala (for review
see Burwell, 2000). It is particularly noteworthy
that accumulation of the pathological form of α-
synuclein (a pathognomonic marker for PD) takes
place in the deep neurons of the perirhinal cortex
in PD (Braak et al., 2003). These neurons are the
main source of subcortical and cortical projections.
What looks like frontal-subcortical deficits in
cognitive functioning may thus derive to some
extent from lesions in specific areas of the
temporal lobe, resulting in disconnection; this
appears plausible also for the verb processing
deficits observed in PD. The fact that limbic loop
and striatal loop deficits may coexist in certain
disorders like PD (c.f., HIV-1 infection; Woods et
614 Per Östberg and Others
al., 2005) clearly complicates inferences about
lexical processing based on a temporolimbic-
frontostriatal dichotomy. It is possible, however,
that more detailed analysis of word production
might reveal more fine-grained clinico-pathological
correlations. For instance, the syntactic valence of
verbs (i.e., how many syntactic arguments a verb
takes) does not appear to influence the verb deficit
in PD (Kim and Thompson, 2004). This is in
contrast with agrammatic subjects, who typically
have frontal lobe lesions. As verb fluency does not
involve syntactic processing, at least not overtly, it
does not seem very likely that problems with the
lexical-syntactic structure of verbs would influence
performance on this task. One way to assess this
would be to investigate what types of verbs are
actually produced by different diagnostic
categories. For instance, assuming that syntactic
processing depends largely on frontal networks,
producing verbs with high syntactic valence might
be associated with an increased “cost” for patients
with frontal network deficits, causing them to
“avoid” them in verb fluency although the task
does not require grammatical encoding.
To conclude, factor analysis disclosed large-
scale patterns of decreased perfusion in subjects
with different levels of cognitive decline. Noun and
verb fluency impairments were distinguished by
decreased perfusion in partly different brain
regions, parietotemporal-occipital versus temporal,
respectively. Decreased verb fluency in cognitive
decline may thus in part be associated with
temporal lobe dysfunction.
Acknowledgments. This study was supported by a grant
from Alzheimerfonden to Professor Lars-Olof Wahlund.
REFERENCES
BAKER MC. Lexical Categories: Verbs, Nouns, and Adjectives.
Cambridge, UK: Cambridge University Press, 2003.
BLACK M and CHIAT S. Noun-verb dissociations: A multi-faceted
phenomenon. Journal of Neurolinguistics, 16: 231-250, 2003.
BRAAK H and BRAAK E. Neuropathological stageing of Alzheimer-
related changes. Acta Neuropathologica, 82: 239-259, 1991.
BRAAK H, DEL TREDICI K, RÜB U, DE VOS RA, JANSEN STEUR EN
and BRAAK E. Staging of brain pathology related to sporadic
Parkinson’s disease. Neurobiology of Aging, 24: 197-211,
2003.
BURWELL RD. The parahippocampal region: Corticocortical
connectivity. Annals of the New York Academy of Sciences,
911: 25-42, 2000.
CAPPA SF, BINETTI G, PEZZINI A, PADOVANI A, ROZZINI L and
TRABUCCHI M. Object and action naming in Alzheimer’s
disease and frontotemporal dementia. Neurology, 50: 351-355,
1998.
CHAVOIX C, BLAIZOT X, MEGURO K, LANDEAU B and BARON J.
Excitotoxic lesions of the rhinal cortex in the baboon
differentially affect visual recognition memory, habit memory
and spatial executive functions. European Journal of
Neuroscience, 15: 1225-1236, 2002.
CRUSE DA. Lexical Semantics. Cambridge, UK: Cambridge
University Press, 1986.
DAMASIO A and TRANEL D. Nouns and verbs are retrieved with
differently distributed neural systems. Proceedings of the
National Academy of Sciences of the USA, 90: 4957-4960,
1993.
DAVACHI L and GOLDMAN-RAKIC PS. Primate rhinal cortex
participates in both visual recognition and working memory
tasks: Functional mapping with 2-DG. Journal of
Neurophysiology, 85: 2590-2601, 2001.
DENES G and DALLA BARBA G. G.B. Vico, precursor of cognitive
neuropsychology? The first reported case of noun-verb
dissociation following brain damage. Brain and Language,
62: 29-33, 1998.
FELLBAUM C. A semantic network of English verbs. In Fellbaum C
(Ed), Wordnet: An Electronic Lexical Database. Cambridge,
MA: MIT Press, 1998.
FERNAEUS SE and ALMKVIST O. Word production: Dissociation of
two retrieval modes of semantic memory across time. Journal
of Clinical and Experimental Neuropsychology, 20: 137-143,
1998.
FOSSATI P, LE BASTARD G, ERGIS AM and ALLILAIRE JF.
Qualitative analysis of verbal fluency in depression.
Psychiatry Research, 117: 17-24, 2003.
FUSTER JM. Cortex and Mind: Unifying Cognition. Oxford:
Oxford University Press, 2003.
HENRY JD and CRAWFORD JR. Verbal fluency deficits in
Parkinson’s disease: A meta-analysis. Journal of the
International Neuropsychological Society, 10: 608-610, 2004.
HODGES JR, PATTERSON K, OXBURY S and FUNNELL E. Semantic
dementia: Progressive fluent aphasia with temporal lobe
atrophy. Brain, 115: 1783-1806, 1992.
KIM M and THOMPSON CK. Verb deficits in Alzheimer’s disease
and agrammatism: Implications for lexical organization. Brain
and Language, 88: 1-20, 2004.
KOZORA E and CULLUM CM. Generative naming in normal aging:
Total output and qualitative changes using phonemic and
semantic constraints. Clinical Neuropsychologist, 9: 313-325,
1995.
KREMEN WS, SEIDMAN LJ, FARAONE SV and TSUANG MT. Is there
disproportionate impairment in semantic or phonemic fluency
in schizophrenia? Journal of the International
Neuropsychological Society, 9: 79-88, 2003.
KYUHOU S and GEMBA H. Projection from the perirhinal cortex to
the frontal motor cortex in the rat. Brain Research, 929: 101-
104, 2002.
MESULAM MM. Aging, Alzheimer’s disease, and dementia. In
Mesulam MM (Ed), Principles of Behavioral and Cognitive
Neurology. Oxford: Oxford University Press, 2000.
MONSCH AU, SEIFRITZ E, TAYLOR KI, ERMINI-FÜNFSCHILLING D,
STÄHELIN R and SPIEGEL R. Category fluency is also
predominantly affected in Swiss Alzheimer’s disease patients.
Acta Neurologica Scandinavica, 95: 81-84, 1997.
MORRIS JC, STORANDT M, MILLER JP, MCKEEL DW, PRICE JL,
RUBIN EH and BERG L. Mild cognitive impairment represents
early-stage Alzheimer disease. Archives of Neurology, 58:
397-405, 2001.
ÖSTBERG P. 18th century cases of noun-verb dissociation: The
contribution of Carl Linnaeus. Brain and Language, 84: 448-
450, 2003.
ÖSTBERG P, FERNAEUS SE, HELLSTRÖM Å, BOGDANOVIC N and
WAHLUND L-O. Impaired verb fluency: A sign of mild cognitive
impairment. Brain and Language, 95: 273-279, 2005.
PERANI D, CAPPA SF, SCHNUR T, TETTAMANTI M, COLLINA S, ROSA
MM and FAZIO F. The neural correlates of verb and noun
processing: A PET study. Brain, 122: 2337-2344, 1999.
PIATT AL, FIELDS JA, PAOLO AM, KOLLER WC and TRÖSTER AI.
Lexical, semantic, and action verbal fluency in Parkinson’s
disease with and without dementia. Journal of Clinical and
Experimental Neuropsychology, 21: 435-443, 1999a.
PIATT AL, FIELDS JA, PAOLO AM and TRÖSTER AI. Action (verb
naming) fluency as an executive function measure:
Convergent and divergent evidence of validity.
Neuropsychologia, 37: 1499-1503, 1999b.
PIATT AL, FIELDS JA, PAOLO AM and TRÖSTER AI. Action verbal
fluency normative data for the elderly. Brain and Language,
89: 580-583, 2004.
PULVERMÜLLER F. The Neuroscience of Language. Cambridge,
UK: Cambridge University Press, 2003.
RADAU PE, SLOMKA PJ, JULIN P, SVENSSON L and WAHLUND L-O.
Evaluation of linear registration algorithms for brain SPECT
and the error due to hypoperfusion lesions. Journal of Medical
Physics, 28: 1660-1668, 2001.
RAVDIN LD, KATZEN HL, AGRAWAL P and RELKIN NR. Letter and
semantic fluency in older adults: effects of mild depressive
symptoms and age-stratified normative data. Clinical
Neuropsychologist, 17: 195-202, 2003.
ROBINS RH. A Short History of Linguistics (third ed.). London:
Longman, 1989.
SILVERI MC, SALVIGNI BL, CAPPA A, DELLA VEDOVA C and
PUOPOLO M. Impairment of verb processing in frontal variant-
 Temporal factor in verb fluency 615

frontotemporal dementia: A dysexecutive symptom. Dementia
and Geriatric Cognitive Disorders, 16: 296-300, 2003.
SYED GM, EAGGER S, TOONE BK, LEVY R and BARRET
JJ. Quantification of regional cerebral blood flow (rCBF)
using 99mTC-HMPAO and SPECT: Choice of the reference
region. Nuclear Medicine Communications, 13: 811-816,
1992.
TYLER LK, RUSSELL R, FADILI J and MOSS HE. The neural
representation of nouns and verbs: PET studies. Brain, 124:
1619-1634, 2001.
WAHLUND L-O, PIHLSTRAND E and ERIKSDOTTER JÖNHAGEN M.
Mild cognitive impairment: Experience from a memory clinic.
Acta Neurologica Scandinavica, 179: S21-S24, 2003.
WOODS SP, CAREY CL, TRÖSTER AI and GRANT I. Action (verb)
generation in HIV-1 infection. Neuropsychologia, 43: 1144-
1151.
Nenad Bogdanovic, Section for Clinical Geriatrics, Neurotec Department, Karolinska
Institutet, Novum, Floor 5, Karolinska University Hospital, Huddinge, SE-141 86
Stockholm, Sweden. e-mail: nenad.bogdanovic@ki.se

(Received 12 October 2004; reviewed 14 December 2004; revised 18 October 2005; accepted 15 February 2006; Action
Editor Stefano Cappa)