Neuropsychologia 37 (1999) 1499±1503

www.elsevier.com/locate/neuropsychologia

Note

Action (verb naming) ¯uency as an executive function measure:

convergent and divergent evidence of validity
Andrea L. Piatt 1, Julie A. Fields, Anthony M. Paolo, Alexander I. TroÈster*
Department of Neurology, University of Kansas Medical Center, 3901 Rainbow Blvd, Kansas City, KS 66160±7314, USA
Received 2 June 1998; received in revised form 26 February 1999; accepted 2 March 1999

Abstract

Numerous studies have demonstrated dissociable neuroanatomic underpinnings for the retrieval of grammatical classes of
words such as nouns and verbs. Whereas retrieval of common and proper nouns is primarily mediated by posterior and anterior
temporal regions, respectively, verb retrieval is primarily mediated by frontal regions. The majority of studies evaluating verb
production have utilized tasks requiring subjects to name a graphically depicted action (i.e. action naming), leaving tests of verb
generation in the absence of prompting stimuli (i.e. action ¯uency) largely unexamined. In a recent study, Piatt, Fields, Paolo,
Koller and TroÈster (in press) found that an action ¯uency task discriminated demented Parkinson's disease (PD) patients from
non-demented PD patients and healthy control subjects, whereas lexical and categorical ¯uency tasks did not. These authors
suggested that action ¯uency was sensitive to the fronto±striatal pathophysiology associated with PD dementia, and thus, that
action ¯uency might serve as an indicator of executive functioning. This study was undertaken to evaluate the construct validity
of action ¯uency as an executive function measure in a group of healthy elderly control subjects. Findings revealed modest to
moderate relationships between action ¯uency and several putative executive measures. Action ¯uency was unrelated to indices
of semantic and episodic memory. Results support the construct validity of action ¯uency as an executive function measure and
suggest that this task may provide some unique information not tapped by traditional executive function tasks. # 1999 Elsevier
Science Ltd. All rights reserved.

Keywords: Verbal ¯uency; Executive functioning; Neuropsychological assessment

Numerous studies have demonstrated dissociable
neuroanatomic underpinnings for the retrieval of
grammatical classes of words such as nouns and verbs.
An accumulation of evidence supports the contention
that common and proper nouns are represented in pos-
terior and anterior temporal regions, respectively,
whereas verb retrieval is strongly mediated by frontal
neural circuits [1,3,5±9,11,14,15,19]. Evidence for the
prominent role of the frontal lobes in verb retrieval
comes predominantly from studies comparing noun
and verb retrieval in patients with known or suspected
* Corresponding author. Tel.: +1-913-588-7187; fax: 1-913-588-
6965.
E-mail address: atroster@kumc.edu (A.I. TroÈster)
1
Present address Department of Psychiatry and Human Behavior,
Brown University School of Medicine
frontal and posterior lesions. For example, in a study
involving patients with aphasia syndromes suggestive
of neuroanatomical dysfunction in frontal (agrammatic
aphasics) and temporal (anomic aphasics) regions,
Miceli et al. [15] found that agrammatic aphasics
demonstrated notable de®cits in verb production,
despite preservation of noun retrieval. In contrast, the
anomic aphasic subjects demonstrated greater diculty
with noun retrieval in comparison to verb production.
Similar ®ndings were reported by Damasio and Tranel
[7] who found that patients with frontal lobe injury
demonstrated notable diculty with verb but not noun
retrieval. In contrast, patients with posterior lesions
demonstrated the opposite pattern of preserved verb
retrieval in the presence of poor noun generation.
Based on this double-dissociation, Damasio and
Tranel asserted that verb naming is mediated by the

0028-3932/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 2 8 - 3 9 3 2 ( 9 9 ) 0 0 0 6 6 - 4
1500 A.L. Piatt et al. / Neuropsychologia 37 (1999) 1499±1503
frontal regions, whereas noun retrieval is mediated by
the dominant, anterior temporal lobe. In a recent
study by Cappa et al. [2], similar dissociations were
observed among patients with Alzheimer's disease
(AD) and fronto±temporal dementia (FTD).
Speci®cally, Cappa et al. reported that both the AD
and the FTD subjects exhibited inferior object and
action naming relative to normal controls. However,
the patients with FTD, relative to patients with AD,
performed signi®cantly better on the object naming
test. Cappa et al. interpreted these ®ndings as
suggesting that the retrieval de®ciencies noted in the
AD subjects re¯ect more general loss of semantic in-
formation, whereas the dissociation between object
and action naming observed in patients with FTD
underscores the importance of the frontal lobes in the
retrieval of verbs. In a similar vein, Williamson et al.
[18] reported that patients with AD performed signi®-
cantly more poorly on object than action naming, even
once word-frequency e€ects were controlled for.
Although these and other studies [1,3,5,6,8,11,14,19]
support the hypothesized di€erential neuroanatomic
substrates of noun and verb retrieval, functional ima-
ging (PET) studies have suggested a more widely dis-
tributed representation of verb retrieval including
heavy reliance on the dominant temporal and parietal
regions [17].
To date, the assessment of verb retrieval abilities has
been generally limited to action naming tasks that
require subjects to identify the verb associated with a
graphically depicted image, similar to the action nam-
ing subtest of the Boston Diagnostic Aphasia
Examination (BDAE) [10], see also Refs.
[1,3,11,14,15,19]. In contrast, the ability to generate
verbs in the absence of prompting stimuli (i.e. action
verbal ¯uency) had been virtually unexamined until
recently. In a study assessing the utility of an action
¯uency task (rather than action naming) as an indi-
cator of frontal±subcortical pathophysiology, Piatt et
al. [16] found that action ¯uency, but not lexical or
semantic ¯uency scores, discriminated patients with
Parkinson's disease (PD) and dementia from non-
demented PD and healthy elderly control groups.
Based on their ®ndings, Piatt et al. concluded that
action ¯uency was particularly sensitive to the fronto±
striatal pathophysiology of PD with dementia,
suggesting that action ¯uency may have utility as an
indicator of frontal lobe functioning in clinical popu-
lations.
Findings across clinical populations and assessment
methods (i.e. task) suggest that verb retrieval abilities
re¯ect the underlying integrity of frontal lobe neural
circuitry, and can be considered to re¯ect the cognitive
construct of executive function. In a recent report,
Grossman [12] discussed the hypothesis that verb
retrieval de®cits re¯ect underlying executive system
dysfunction that undermines the ability to mentally
coordinate and manipulate the diverse range of infor-
mation that may be associated with a verb. Although
there is ample evidence to support the contention that
the retrieval of verbs re¯ects executive functioning, and
possibly a unique aspect of executive functioning, these
hypotheses have not been directly evaluated in non-
clinical populations. Moreover, investigations of verb
retrieval employing a ¯uency-based task are extremely
rare. The absence of both ¯uency-based methods of
assessing verb retrieval and the use of such measures
as indicators of frontal lobe and/or executive function-
ing is surprising given the notably wide use of lexical
¯uency as an executive function test.
This study was undertaken to evaluate the construct
validity of an action ¯uency task as a measure of
executive functioning in a sample of healthy elderly
subjects. Given the prominent role ascribed to the
frontal lobes in verb retrieval, it was hypothesized that
action ¯uency would be signi®cantly related to puta-
tive measures of executive functioning (convergent val-
idity), but not signi®cantly related to measures tapping
semantic and episodic memory (divergent validity).
However, based on evidence that verb retrieval
involves neural circuits distinct from those involved in
traditional executive function tasks, we anticipated
only modest relationships between action ¯uency and
the executive function measures.
1. Method
1.1. Participants
Participants consisted of 67 healthy elderly control
subjects recruited as caregivers of patients enrolled in
studies at the Neurodegenerative Disease Research
Center at the University of Kansas Medical Center.
Forty-two (63%) of the participants were female.
Sixty-four (96%) of the subjects were right-handed.
Additional sample characteristics are presented in
Table 1. All subjects were free of histories of neuro-
logic disease or injury, psychiatric illness including
substance abuse or dependence, or self-reported
speci®c developmental disorders. All subjects under-
went comprehensive interviews, neuropsychological
testing, and neurologic examinations.
1.2. Materials and procedures
As part of a larger test battery, subjects were admi-
nistered the Dementia Rating Scale (DRS), Wisconsin
Card Sorting Test (WCST), Trail Making Tests
(Trails), Boston Naming Test (BNT), California
Verbal Learning Test (CVLT), and the Wechsler
Memory ScaleÐRevised (WMS-R), all in standard
 A.L. Piatt et al. / Neuropsychologia 37 (1999) 1499±1503 1501

Table 1 total time to complete Trails B all served as measures

Demographic characteristics and Neuropsychological Test
of executive functioning.
Performance (Mean, SD, and Range)

Mean (SD) Range 1.4. Memory and language measures

Age 71.39 (6.45) 57±84 The following measures of episodic memory were
Education 14.96 (2.31) 12±20 included: The number of words recalled on CVLT
Dementia Rating Scale total (/144) 139.37 (4.05) 128±144
Action ¯uency (words/min) 16.75 (4.73) 8±30 trials one through ®ve and the WMS-R Logical
Dementia Rating Scale I/P (/37) 36.49 (1.30) 31±37 Memory (WMS-R LM) immediate recall raw score.
Dementia Rating Scale CNC (/39) 37.45 (2.36) 30±39 The BNT total correct score was used as a measure of
Trails B (total time) 83.00 (33.32) 37±218 semantic memory.
WCST-Categories (/6) 4.77 (2.13) 0±6
WCST-Conceptual Responses (%) 66.15 (24.34) 5±91
WCST-Perseverative Responses 15.98 (14.56) 3±65
WCST-Perseverative Errors 14.26 (11.94) 3±51
Boston Naming Test (/60) 57.09 (3.34) 36±60
CVLT total (/80) 50.91 (10.45) 24±70 2. Results
WMS-R Logical Memory Immediate (/50) 27.15 (5.53) 10±36

Table 1 presents the means and standard deviations

fashion. Detailed descriptions of these tests are pre-
sented in Lezak [13]. Instructions for the Action
Fluency task were as follows: I'd like you to tell me as
many di€erent things as you can think of that people do.
I don't want you to use the same word with di€erent
endings, like eat, eating, eaten. Also, just give me single
words such as eat, or smell, rather than a sentence. Can
you give me an example of something that people do? If
the response was acceptable participants were further
instructed: That's the idea. Now in one minute, tell me
as many di€erent things as you can think of that people
do. If the subject had diculty understanding the task,
the words `verb' or `action' were used for clari®cation.
The score for this task was the total number of unique
words produced in 60 s. Given the marked similarity
in method (monomethod) used in the assessment of
action, lexical and semantic ¯uency, the latter two
were not included in the present study. Although re-
lationships between the three ¯uency measures may
have been identi®ed, whether such relationships would
have re¯ected similarity of underlying traits or simply
comparability of methodology would have been quite
dicult to ascertain with any degree of certainty. As
such, a multi-trait/multi-method approach was utilized
in this study.
1.3. Executive function measures
In order to provide an adequate representation of
executive function abilities, several scores from various
measures were included. Speci®cally, the Initiation/
Perseveration and Conceptualization subtest scores
from the DRS (DRS±I/P and DRS±CNC), the Percent
Conceptual Level Responses (WCST±CR),
Perseverative Responses (WCST±PR), Perseverative
Errors (WCST±PE) and Categories Completed
(WCST±Categories) scores from the WCST, and the
for action ¯uency and the various neuropsychological
measures. Because not all variables were normally dis-
tributed, nonparametric Spearman's rho correlations
were utilized for relational analyses, with attendant
results depicted in Table 2. As indicated, there is a sig-
ni®cant positive relationship between education and
action ¯uency scores, whereas age is unrelated to
action ¯uency scores. Modest negative correlations
were noted between action ¯uency performance and
Trails B, WCST±PR, and WCST±PE scores, whereas
moderate positive correlations were noted between
action ¯uency and WCST±CR and WCST±Categories.
A modest positive correlation between action ¯uency
scores and DRS±CNC scores was observed, but no
signi®cant relationship was found between the DRS±I/
P scores and action ¯uency, despite a signi®cant re-
lationship between DRS±I/P scores and Trails B and
WCST±PE. In order to further clarify the relationships
between action ¯uency and the executive measures, a
step-wise multiple regression analysis was conducted
with six executive measures (the four WCST scores,
two DRS scores, and Trails B) and education serving
as predictor variables and action ¯uency serving as the
criterion variable. Results yielded a signi®cant re-
gression equation that accounted for slightly more
than 28% of the variance in action ¯uency scores
(adjusted R 2=0.28), F (3,61)=9.47, P < 0.001.
WCST±PR was the ®rst variable to enter the equation,
accounting for approximately 15% of the variance in
action ¯uency scores (R 2=0.145), followed by edu-
cation which accounted for an additional 11% of the
variance in action ¯uency scores (R 2 change=0.106),
while DRS±I/P was the last variable to enter the
equation, accounting for another 5% of variance (R 2
change=0.053). The assumptions of multiple re-
gression were met as indicated by observation of stan-
dardized residual plots and by results of a Durbin±
Watson test that indicated independence of errors.
1502 A.L. Piatt et al. / Neuropsychologia 37 (1999) 1499±1503

3. Discussion

Education DRS±I/P DRS±CNC Trails B WCST±PR WCST±PE WCST±CR WCST±Categories CVLT total WMS-R LM BNT total

1.0
As hypothesized, action ¯uency was signi®cantly,
but moderately, related to several putative executive
function measures. Although no signi®cant relation-
ship was observed between action ¯uency and DRS±I/

0.18
P scores, this may have re¯ected a marked restriction

1.0
of range in DRS±I/P scores (60 of the 67 subjects
obtained scores of 36 or 37/37). However, the signi®-
cant relationships between DRS±I/P scores and Trails

0.37b
0.25a
B and WCST±PE argue against this explanation.

1.0
Another possibility is that the low correlation between
DRS±I/P and action ¯uency stems from the fact that
the DRS±I/P score is heavily in¯uenced by category
(semantic) verbal ¯uency (up to 20 of 37 points), and
action and semantic ¯uency tasks tap somewhat di€er-
0.42b
0.30a

0.01
ent retrieval processes (see [16]). Of the executive
1.0

measures, WCST±PR was the most predictive of

action ¯uency scores, whereas DRS±I/P scores
Spearman's rho correlation coecients between action ¯uency scores, demographic variables, and neuropsychological test scores

accounted for a small portion of variance in action ¯u-

0.78b
0.42b
0.39b
0.26a

ency scores not tapped by WCST±PR. As the remain-

1.0

ing executive measures failed to enter the regression

equation, it appears that they tapped redundant
ÿ0.95b
ÿ0.76b
ÿ0.35b
ÿ0.40b

aspects of executive functioning. Although action ¯u-

ÿ0.22
1.0

ency shares common variance with other executive

tests, it appears to also measure a component of execu-
tive functioning not measured, or measured to a lesser
0.99b
ÿ0.94b
ÿ0.73b

ÿ0.40b
ÿ0.35a

ÿ0.22

extent, by more traditional tasks. Such an hypothesis

1.0

is supported by the rather modest relationships

observed between action ¯uency and the executive
ÿ0.32b
0.25a
ÿ0.28a

ÿ0.31a
0.23

ÿ0.12
ÿ0.21

tests utilized in the present study. This ®nding is con-

1.0

sistent with the report of Piatt et al. [16] who found

action ¯uency to be uniquely sensitive and speci®c to
the advanced frontostriatal pathophysiology of
0.36b
0.30a
ÿ0.03
ÿ0.16
ÿ0.20

0.18
0.11
0.09

Parkinson's disease with associated dementia, while

1.0

lexical and semantic ¯uency were not.

It is important to note that action ¯uency scores
ÿ0.27a

ÿ0.25a

were not signi®cantly related to BNT or Logical

ÿ0.03

ÿ0.24

0.22
0.20
0.13
0.03
0.15
1.0

Memory scores, arguing against the likelihood that

retrieval of verbs re¯ects only aspects of more general
semantic retrieval or components of memory. In the
0.25a

0.29a
ÿ0.01
0.03
0.04
ÿ0.15
ÿ0.16
0.14

ÿ0.02

0.03

presence of signi®cant relationships between action ¯u-

1.0

ency and several executive function measures, the lack

of association between action ¯uency and the memory
0.38b
0.35b
0.35b
ÿ0.37b
ÿ0.35b
ÿ0.39b
ÿ0.38b
ÿ0.05
ÿ0.23
ÿ0.19

ÿ0.22
1.0
Action ¯uency Age

measures o€ers further support to the purported sensi-

tivity of action ¯uency as an executive function
measure. These ®ndings also are consistent with a
number of studies that have identi®ed single or
0.41b

ÿ0.37b
ÿ0.38b
0.39b
0.40b
0.26a
ÿ0.28a

double-dissociations between noun and verb retrieval

ÿ0.10

0.24

0.04
0.18
0.19
1.0

abilities implicating the anterior brain region in verb

retrieval [2,7,15].
WCST±Categories

Overall, results provide evidence for the convergent

Action Fluency

P < 0.05,.
P < 0.01.

and divergent validity of action ¯uency as an executive

WMS-R LM
CVLT total
WCST±CR
DRS±CNC

WCST±PR
WCST±PE
Education

function measure in healthy elderly subjects. These

DRS±I/P

Trails B
Table 2

®ndings, in combination with the previously reported

BNT
Age

a
b

ability of action ¯uency to discriminate groups with

 A.L. Piatt et al. / Neuropsychologia 37 (1999) 1499±1503
known frontal±subcortical involvement from those
with only suspected or no frontal±subcortical involve-
ment [16], suggest that action ¯uency may have clinical
utility as a measure of executive functioning. It is also
of note that while action ¯uency was not related to in-
dices of semantic or episodic memory, Trails B and
WCST±CR were signi®cantly related to BNT scores,
and all four WCST scores were signi®cantly related to
both the CVLT Total Learning score and the WMS-R
Logical Memory I scores. These ®ndings could parallel
earlier accounts of hippocampal involvement in per-
formance on executive functioning tasks (e.g. card
sorting) [4], and, to the extent that this is the case, one
could argue that action ¯uency is a frontally mediated
task that is less susceptible to hippocampal in¯uence.
Moreover, it appears that action ¯uency may tap
aspects of executive functioning not measured by tra-
ditional executive instruments, thus perhaps adding to
what is considered by many to be a very limited pool
of sensitive and speci®c executive function tests [13].
Although these initial ®ndings are promising, the lit-
erature examining verb retrieval generally, and action
¯uency speci®cally, remains sparse. Additional research
examining the construct validity and clinical utility of
action ¯uency with a range of clinical populations will
be needed in order to establish action ¯uency as an
executive functioning measure in its own right.
Nonetheless, should research continue to support the
ability of action ¯uency to assess unique aspects of
executive functioning, action ¯uency may prove to be
a useful and easily administered instrument in the
daily practice of clinical neuropsychology.
References
[1] Baxter DM, Warrington EK. Category-speci®c phonologic dys-
graphia. Neuropsychologia 1985;23:653±66.
[2] Cappa SF, Binetti G, Pezzini A, Padovani A, Ruzzini L,
1503
Trabucci M. Object and action naming in Alzheimer's disease
and frontotemporal dementia. Neurology 1998;50:351±5.
[3] Caramazza A, Hillis AE. Lexical organization of nouns and
verbs in the brain. Nature 1991;349:788±90.
[4] Corcoran R, Upton D. A role for the hippocampus in card
sorting? Cortex 1993;29:293±304.
[5] Damasio AR. Aphasia. New England Journal of Medicine
1992;326:531±9.
[6] Damasio AR, Damasio H. Brain and language. Scienti®c
American 1992;267:63±71.
[7] Damasio AR, Tranel D. Nouns and verbs are retrieved with dif-
ferentially distributed neural systems. Proceedings of the
National Academy of Science, USA 1993;90:4957±60.
[8] Damasio AR, Tranel D, Damasio H. Verbs but not nouns:
Damage to left temporal cortices impairs access to nouns but
not verbs. Society for Neuroscience Abstracts 1992;18:387.
[9] Daniele A, Giustolisi L, Silveri MC, Colosimo C, Gainotti G.
Evidence for a possible neuroanatomical basis for lexical proces-
sing of nouns and verbs. Neuropsychologia 1994;32:1325±41.
[10] Goodglass H, Kaplan E. Boston diagnostic aphasia examin-
ation, 2nd ed. Philadelphia: Lea and Febinger, 1987.
[11] Goodglass H, Klein B, Carey P, Jones K. Speci®c semantic
word categories in aphasia. Cortex 1966;2:74±89.
[12] Grossman M. Not all words are created equal: category-speci®c
de®cits in central nervous system disease. Neurology
1998;50:324±5.
[13] Lezak M. Neuropsychological assessment, 3rd ed. Oxford:
Oxford University Press, 1995.
[14] McCarthy RA, Warrington EK. Category-speci®city in an
agrammatic patient: the relative impairment of verb retrieval
and comprehension. Neuropsychologia 1985;23:709±27.
[15] Miceli G, Silveri MC, Villa G, Caramazza A. On the basis of
the agrammatic's diculty in producing verbs. Cortex
1984;20:207±20.
[16] Piatt AL, Fields JA, Paolo A, Koller WC, TroÈster AI. (in
press). Lexical, semantic, and action ¯uency in Parkinson's dis-
ease with and without dementia. Journal of Clinical and
Experimental Neuropsychology.
[17] Warburton E, Wise RJS, Price CJ, Weiller C, Hadar U,
Ramsay S, Frackowiak RSJ. Noun and verb retrieval by normal
subjects: Studies with PET. Brain 1996;119:159±79.
[18] Williamson DJG, Adair JC, Raymer AM, Heilman KM. Object
and action naming in Alzheimer's disease. Cortex 1998;34:601±
10.
[19] Zingeser LB, Berndt RS. Retrieval of nouns and verbs in
agrammatism and anomia. Brain and Language 1990;39:14±32.