Brain and Cognition 55 (2004) 535–544

www.elsevier.com/locate/b&c

Verbal fluency over time as a measure of automatic and

controlled processing in children with ADHD
P.P.M. Hurks,a,* J.G.M. Hendriksen,a,b,c J.S.H. Vles,b A.C. Kalff,a F.J.M. Feron,d
M. Kroes,b T.M.C.B. van Zeben,e J. Steyaert,f and J. Jollesa
a
Department of Neuropsychology and Biopsychology, University of Maastricht, The Netherlands
b
Department of Neurology, Academic Hospital Maastricht, The Netherlands
c
Child Rehabilitation Center Franciscusoord, Valkenburg, The Netherlands
d
Youth Health Care of the Municipal Health Center, Maastricht, The Netherlands
e
Department of Pediatrics, Academic Hospital Maastricht, The Netherlands
f
Department of Psychiatry and Neuropsychology, Academic Hospital Maastricht, The Netherlands
Accepted 26 March 2004
Available online 5 May 2004

Abstract

The performance of ADHD children on semantic category fluency (SCF) versus initial letter fluency (ILF) tasks was examined. For
each participant, word production was recorded for each 15-s time slice on each task. Performance on both fluency tasks was
compared to test the hypothesis that children with ADHD are characterized by a performance deficit on the ILF task because per-
formance on this task is less automated than performance on the SCF. Children classified with ADHD (N ¼ 20) were compared to
children with other psychopathology (N ¼ 118) and healthy controls (N ¼ 130). Results indicated that the groups could not be
differentiated by the total number of words produced in 60 s in either fluency task. As hypothesized, a significant interaction of group
by productivity over time by type of fluency task was found: ADHD children had more problems finding words in the first 15 s of the
IFL than did children in the other two groups, and as compared with their performance on the SCF. Results were taken to indicate
that children with ADHD symptoms show a delay in the development of automating skills for processing abstract verbal information.
Ó 2004 Elsevier Inc. All rights reserved.

Keywords: ADHD; Automatic-controlled processing; Verbal fluency; Children

1. Introduction 1994), a syndrome that affects approximately 1–5% of

children (Barkley, 1998). Kroes et al. (2001) have esti-
Impulsive behavior, inattention, and overactivity in mated the prevalence of ADHD within a population of
young children can be a risk factor for abnormal psy- Dutch school-aged children at 3.8%.
chological development later (Sagvolden & Sergeant, Over the years, several causes of ADHD have been
1998; Taylor, Chadwick, Heptinstall, & Danckaerts, proposed. For instance, neurological evidence is point-
1996; Weiss & Trokenberg Hechtman, 1993). If these ing towards an involvement of the frontal-striatal cir-
behaviors cause significant impairments in social or ac- cuits in ADHD (Barkley, 1998; Oades, 1998; Sagvolden
ademic functioning and are inconsistent with the de- & Sergeant, 1998). Furthermore, from a neuropsycho-
velopmental level of the child, they may be indicative of logical perspective, ADHD is associated with deficits in
the clinical diagnosis ‘Attention-Deficit/Hyperactivity well-defined cognitive domains, including sustained at-
Disorder’ (ADHD; American Psychiatric Association, tention and executive functioning (Barkley, 1998; Pen-
nington & Ozonoff, 1996). According to Lezak (1995),
the term ‘executive functioning’ encompasses those ca-
*
Corresponding author. Fax: +31-43-388-4092. pacities that enable a person to engage successfully in
E-mail address: [email protected] (P.P.M. Hurks). independent, purposive, self-serving behavior. Examples

0278-2626/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.bandc.2004.03.003
536 P.P.M. Hurks et al. / Brain and Cognition 55 (2004) 535–544

of executive functioning are non-verbal working mem- Both automatic and controlled information process-
ory, internalization of speech, self-monitoring, and self- ing have been investigated in studies on the cognitive
regulation (Barkley, 1998; Perugini, Harvey, Lovejoy, performance of children with ADHD, but results are
Sandstrom, & Webb, 2000). Another cognitive function inconsistent. Several authors have shown that children
highly correlated with executive functioning and atten- with ADHD perform less well than controls in situa-
tion capacity, which is often used to evaluate cognitive tions demanding automatic and/or more controlled
functioning in ADHD, both clinically and experimen- processing strategies (Ackerman, Anhalt, Holcomb, &
tally, is the fluency of speech (Barkley, 1998; Monsch Dykman, 1986; Borcherding et al., 1988; Hazell et al.,
et al., 1994; Rosser & Hodges, 1994). This function is 1999), whereas other authors have not (Van der Meere
usually defined as the number of words produced, usu- & Sergeant, 1988). Comparison of children with ADHD
ally within a restricted category and over a limited pe- and controls on both types of fluency, evaluated by their
riod of time (usually 60 s) (Lezak, 1995). Two major performance over time, may help clarify these conflict-
categories of verbal fluency tasks can be distinguished, ing results. Given the deficits in executive functioning in
namely (A) semantic category fluency (SCF; that is, ADHD, we hypothesize that children with ADHD
recitation of examples of a given category) and (B) initial perform significantly worse on both types of informa-
letter fluency (ILF; that is, generating words beginning tion processing (automatic vs. controlled) than do
with a given initial letter). healthy control subjects.
Earlier studies of ADHD and verbal fluency provided In the present study, we evaluated verbal information
inconclusive results. A number of studies found signifi- processing in ADHD, using an extensive, two-fold
cant differences between children with ADHD and controlled design. Children with ADHD were compared
controls on either the SCF or the ILF, in that children to healthy children and to a group of children with at
with ADHD performed significantly worse (Grodzinsky least one DSM-based disorder other than ADHD. The
& Barkley, 1999; Grodzinsky & Diamond, 1992; Klor- second control group was included in view of recent
man et al., 1999; Koziol & Stout, 1992; Loge, Station, & evidence that children with ADHD are at risk of de-
Beatty, 1990; Pineda, Ardila, & Rosselli, 1999; Schuer- veloping other psychiatric disorders and learning dis-
holz, Singer, & Denckla, 1998). In contrast, also a large abilities (Barkley, 1998). Thus, we also compared the
number of studies found no differences between these ADHD group with psychiatric controls, to determine
groups (Barkley, Grodzinsky, & DuPaul, 1992; Fischer, the specificity of possible ADHD-related findings. Fi-
Barkley, Edelbrock, & Smallish, 1990; Grodzinsky & nally, because the behavioral expression of ADHD
Diamond, 1992; Kusche, Cook, & Greenberg, 1993; changes over time, and at least some symptoms are
Loge et al., 1990; Pineda et al., 1999; Reader, Harris, present before age 7 years (American Psychiatric Asso-
Schuerholz, & Denckla, 1994; Weyandt & Willis, 1994), ciation, 1994), and because the period between the ages
with the observation that the ILF tended to discriminate 5 and 7 years is believed to be critical for later cognitive
somewhat better between ADHD and controls than the (and more specifically language) development (Bern-
SCF (Sergeant, Geurts, & Oosterlaan, 2002). However, stein, 1989; Riva, Nichelli, & Devoti, 2000), language
these studies included only the total number of words development may be influenced most by the early ex-
produced over a set period of 60–90 s as the main out- pression of ADHD (Levy, Hay, McLaughlin, Wood, &
come variable, whereas recent experiments have shown Waldman, 1996). For this reason, we decided to relate
that, in healthy adults, the pattern of word production child psychiatric data gathered when the child was ap-
over time is relevant (Troyer, 2000). For example, ex- proximately 6–7 years to later cognitive functioning at
periments have shown in healthy adults and children an age when the child had been to school for 2–3 years.
that the effectiveness of word production changes from In sum, we investigated the performance of children
approximately the first 15–20 s to the last 40–45 s of a 1- with ADHD in a controlled design on two types of the
min task. In the first period, a ready pool of frequently fluency task and as a function of time.
used words appears to be available and is automatically
activated for production. As time passes, the pool be-
comes exhausted and the search for new words becomes 2. Methods
both more effortful and less productive (Crowe, 1998;
Hurks et al., in press). By measuring performance over 2.1. Subjects and procedure
time, the fluency task can be used to measure the effec-
tiveness of both automatic and controlled processing, in The present study is based on data collected within a
which automatic processing is believed to be generally research program entitled ‘Study of Attention Disorders
fast and relatively unconscious and controlled process- Maastricht (SAM)’ and is embedded in an extensive,
ing slow, effortful, and attention-demanding (Fodor, longitudinal design. The SAM study consists of three
1983; Schneider & Shiffrin, 1977; Shiffrin & Schneider, separate phases (for an extensive description of the
1977). study see Kalff et al., 2001; Kroes et al., 2001).
P.P.M. Hurks et al. / Brain and Cognition 55 (2004) 535–544 537

Stage 1 (months 1–9): selection of subjects. During and any other psychopathology were submitted to
this phase, all caregivers of children frequenting the group a.
second grade of normal kindergarten were asked to give Furthermore, in stage 3 (months 39–49), all selected
permission for participation in the SAM study children were again asked to participate in the SAM
(n ¼ 2300). In the Netherlands, the second grade pre- study. In total, 284 of the 403 children for whom we had
cedes the first class of elementary school in which chil- ADIKA results in phase 2 agreed to participate in the
dren learn to read and write. Response rate was 57.5% follow-up (70.5%). Because of the relative large dropout
(n ¼ 1317). The children were examined as part of the of participants in stage 3, responders and non-re-
routine health examination carried out by the school sponders of stage 3 were tested for group differences
doctors. By law, these physicians are allowed to use in terms of sex, parental occupation, and ADIKA
medical information anonymously for epidemiological results (stage 2). Cramer’s V testing revealed no signifi-
purposes. In this way, it was possible to compare two cant differences between responders and non-respond-
random samples of 200 non-responders with 200 re- ers on these variables (sex: value ¼ .002, p ¼ :965;
sponders with regard to child characteristics (sex, age), parental occupation: value ¼ .083, p ¼ :258; ADIKA re-
family variables (parental occupation, nationality, and sults: value ¼ .109, p ¼ :091). Next, all children were
family structure), and environmental variables (living tested neuropsychologically by means of an extensive
area), collected from the medical status of the Youth test protocol including assessment tools such as the Ver-
Health Care. This comparison was necessary because of bal Fluency Test, Vocabulary scores, and the Dutch
the relative large percentage of non-responders; how- Klepel Reading Test, which is a test for reading non-
ever, no significant differences between the groups were words. A neuropsychologist or one of three well-trained
found (for a full description, see Kroes et al., 2001). assistants administered the test battery. Testing took
Next, based on the Dutch version of the Child Behavior place in a room at the child’s school. The examiner was
Checklist (CBCL; Achenbach & Edelbrock, 1983; Ver- blind to group membership of the children.
hulst, Koot, & Van der Ende, 1996), three groups were Next, of the 284 cases included, fifteen children had
selected from the responders group for the second stage. to be excluded, because of: (a) known use of Methyl-
Group E (externalizing group; n ¼ 173, that is, ap- phenidate or Pipamperon (ADHD group n ¼ 5; psy-
proximately 7.5% of the total sample) consisted of chiatric control group n ¼ 4; and control group n ¼ 2)
children who scored high either on the CBCL external- or (b) missing data in relation to LOA-codes (n ¼ 4) or
izing broad-bent scale (>90th percentile) or on the Klepel reading scores (n ¼ 1). After this deletion of
CBCL Attention problems subscale (>95th percentile). cases, data from 268 children were still available for the
According to Chen, Faraone, Biederman, and Tsuang analyses: 20 children in the ADHD group, 118 children
(1994), this selected group contains children with a pu- in the psychiatric control group, and 130 normal con-
tative risk for the development of ADHD. Next, group I trols. Variables influential on research outcome were
(internalizing group; n ¼ 59, that is, approximately 2.6% summed in Table 1, including age (as measured at stage
of the total sample) contained children scoring within a 3), sex, the level of occupational achievement of the
clinical range on the CBCL Internalizing scale (>90th caregiver (LOA), and an estimate of verbal abilities and
percentile), but did not fulfill the criteria for group E reading skills.
membership. Finally, a matched control group (n ¼ 220, Additionally, in Table 2, the distribution of psycho-
that is, approximately 9.6% of the total sample) was pathology other than ADHD was provided for each
formed consisting of children with low CBCL total group (ADHD, psychiatric controls, and normal con-
problem scores (<90th percentile) and who were mat- trols). No children were excluded from the ADHD-
ched to groups E and I in terms of age (2 months), sex group and the psychiatric control group because of the
and school (urban vs. rural). (co-) occurrence of specified psychopathology.
Stage 2 (months 15–25): caregivers of 403 children of
the originally selected group agreed to a semi-structured, 2.2. Measurements
psychiatric interview, using the Amsterdam Diagnostic
Interview for Children and Adolescents (ADIKA). This The Amsterdam Diagnostic Interview for Children and
instrument assesses the presence of several DSM classi- Adolescents (ADIKA; Kortenbout van der Sluijs, Lev-
fications for childhood psychopathology, including ita, Manen, & Defares (1997): this is the Dutch trans-
ADHD, using a systematic, horizontal approach. Based lation of the Diagnostic Interview for Children and
on these ADIKA results, three independent groups were Adolescent (DICA; Ezpeleta et al., 1997; Granero Perez,
formed: (a) children, who met during phase 2 all DSM- Ezpeleta Ascaso, Domenech Massons, & De la Osa
IV criteria for ADHD (ADHD group), (b) children, who Chaparro, 1998; Kortenbout van der Sluijs et al., 1997).
met in stage 2 the DSM-criteria for psychopathology The ADIKA is a semi-structured psychiatric interview
(other than ADHD; psychiatric control group), and (c) that yields scores for several child psychiatric syndromes
controls (normal controls). Children with both ADHD according to DSM-III-R guidelines and was adapted by
538 P.P.M. Hurks et al. / Brain and Cognition 55 (2004) 535–544

Table 1
Group characteristics and diagnostic data of the sample (n ¼ 268)
ADHD (n ¼ 20) Psychiatric (n ¼ 118) Controls (n ¼ 130) Contrasts
Sex (boys:girls): 14:6 71:47 65:65 n.s.
Mean SD Mean SD Mean SD
Age 9.2 0.3 9.2 0.4 9.1 0.4 n.s.
Vocabulary-score: 9.0 2.6 8.9 2.5 8.8 2.9 n.s.
Klepel-score: 8.8 3.6 10.7 2.9 10.7 2.6 p ¼ :019
LOA-score: 3.0 1.7 3.7 2.0 4.5 1.9 p < :001
Associations between ordinal and/or interval variables (group membership vs. age, Block design scores, and LOA) were calculated by use of
ANOVA. In contrast, v2 -tests were used to measure the associations between nominal (sex) and ordinal variables (groups).

Table 2
Distribution of psychopathology, other than ADHD, found in the three groups (ADHD, psychiatric controls, and normal controls)
Type of comorbidity (number of cases) ADHD Psychiatric Controls
No psychopathology — — 130
No psychopathology other than ADHD 1 — —
Anxiety disorders 2 42 —
Mood disorders — 5 —
Mood disorders and anxiety disorders 1 7 —
Conduct disorders 4 17 —
Conduct disorders and anxiety disorders 3 12 —
Conduct disorders, mood disorders, and anxiety disorders — 1 —
Conduct disorders, mood disorders, and anxiety disorders 4 3 —
Disorders of elimination 2 19 —
Disorders of elimination and anxiety disorders — 3 —
Disorders of elimination and mood disorders — 2 —
Disorders of elimination, mood disorders, and anxiety disorders — 1 —
Disorders of elimination and conduct disorders 1 5 —
Disorders of elimination, conduct disorders, and anxiety disorders 1 — —
Disorders of elimination, conduct disorders, and mood disorders — 1 —
Disorders of elimination, conduct disorders, mood disorders, and anxiety disorders 1 — —

Number of cases 20 118 130

using the criteria of DSM-IV for diagnosing ADHD liability and moderate correlations with clinician based
(American Psychiatric Association, 1987; American diagnoses (Ezpeleta et al., 1997; Welner, Reich, Herj-
Psychiatric Association, 1994). In line with these crite- anic, Jung, & Amado, 1987).
ria, children were classified as ADHD if they showed a Verbal fluency test (Korkman, Kirk, & Kemp, 1998):
persistent pattern of inattention and/or hyperactivity– this test is a subtest of the NEPSY-battery and is be-
impulsivity that was more frequent and severe than that lieved to be a measure of retrieval from semantic
typically observed in individuals of comparable devel- memory. It includes two independent types of verbal
opment (American Psychiatric Association, 1994). fluency, namely, semantic category fluency (SCF) and
Children were included in the ‘psychiatric control group’ initial letter fluency (ILF). SCF involves recitation of
if they fulfilled the criteria for at least one child psy- examples of a given category, whereas ILF involves the
chiatric syndrome (with the exception of ADHD).1 Al- generation of words beginning with a given initial letter.
though the Dutch version of the DICA-III-R interview The verbal fluency test consisted of: (A) two trials of a
has not been separately validated, DICA and DICA-III- SCF type task (e.g., to name as: (1) many animals and
R have been demonstrated to have high test–retest re- (2) things you can eat or drink) and (B) two trials of an
ILF type task (e.g., including the letters ‘M’ and ‘S’
[subtasks 3 and 4]). According to the Van Loon list
1
Such as mood disorders (e.g., major depression, dysthemia, and presenting imageability ratings of Dutch words (Van
bipolar syndrome), anxiety disorders (e.g., separation anxiety, phobia, Loon, 1985), words that begin with the letters M or S
overanxious disorder, and avoidant disorder), conduct disorders (e.g.,
oppositional disorder), disorders of elimination (e.g., functional
are common in the Dutch language (respectively 4.3 and
encopresis and functional enuresis), obsessive compulsive disorder, 11.7% of the Dutch words begin with these letters).
post traumatic stress disorder, and pervasive developmental disorders. Consistent with standard instructions, participants were
P.P.M. Hurks et al. / Brain and Cognition 55 (2004) 535–544 539

asked each time to generate over 60 s as many words as ates for they accounted for at least some variance in the
possible of the above-mentioned categories, excluding verbal fluency scores. Preliminary experiments showed
the names of humans and cities. For each subtask, total that the age of the child did not correlate with test
number of correct responses, incorrect responses (e.g., performance over all groups, and thus age was not in-
words not beginning with the appropriate letter or not cluded as a covariate in the analyses. Corrections were
an exemplar of the category), and perservations (e.g., made for unequal sample sizes within a non-experi-
repetitions of correct words, morphological variants [for mental research design. The critical value for rejecting
instance when a child says car and cars]) were recorded the null hypotheses was chosen at p < :05.
over (A) 1–15 s, (B) 16–30 s, (C) 31–45 s, and (D) 46– Significant main effects were found for type of fluency
60 s. To reduce error variances, data were averaged over task (Type: F ð1; 261Þ ¼ 1138:80, p < :001; partial eta
the time samples A, B, C, and D for each type of fluency. squared ¼ .814), word production as a function of
Vocabulary test: this subtest of the Wechsler Intelli- time (Time: F ð3; 259Þ ¼ 982:67, p < :001; partial eta
gence Scales for Children—Dutch Version (WISC— squared ¼ .919), and the interaction between these fac-
Dutch Version by De Bruyn et al., 1986) was used to tors (Type Time: F ð3; 259Þ ¼ 159:20, p < :001; partial
provide an estimate of general ability (Lezak, 1995). The eta squared ¼ .648). Overall, children produced signifi-
examiner asked the child to explain the meaning of cantly more words in the SCF (unadjusted for covariates:
certain words. The complexity of words increased with meanSCF ¼ 14.4 [SD ¼ 3:6]) than on the ILF (mean-
each item. Range of standard scores is 1–19 (mean ILF ¼ 7.1 [SD ¼ 2:8]). Secondly, word productivity de-
score ¼ 10, SD ¼ 3). Reliability and validity are believed creased significantly with each succeeding quartile of
to be average to good (De Bruyn et al., 1986). 60 s, averaged over types of fluency (unadjusted for co-
Klepel reading test (Van den Bos, Lutje Spelberg, variates: mean1–15 s ¼ 4.6 [SD ¼ 1:0]; mean16–30 s ¼ 2.6
Scheepstra, & De Vries, 1994): this test provides a [0.9]; mean31–45 s ¼ 1.9 [0.8]; mean46–60 s ¼ 1.6 [0.8]).
standardized measure of non-word decoding. For 2 min, Thirdly, the significant interaction of Type by Time in-
the child had to read as many words as possible from a dicated that the decrease in word production was not
list of non-words. With each item the complexity of parallel in the SCF and ILF, although in both tests the
words increased. Standard scores range from 1 to 19 number of words produced decreased with time.
(mean score ¼ 10, SD ¼ 3). This task was included as a Fourthly, with regard to the covariates, significant in-
covariate because of the assumed relation between the teractions were found for Sex by Type of test
development of the ability to organize and retrieve (F ð1; 261Þ ¼ 4:23, p ¼ :041; partial eta squared ¼ .016),
words phonemically and reading skills (Riva et al., Sex by Time samples (F ð3; 259Þ ¼ 2:65, p ¼ :049; partial
2000). eta squared ¼ .030), and Klepel reading test scores by
Level Occupational Achievement of the caregiver Time samples (F ð3; 259Þ ¼ 4:38, p ¼ :005; partial eta
(LOA): this variable was based on the full description of squared ¼ .048). In addition, significant between-subjects
the parental occupation and was originally scored on a effects were found for LOA (F ð1; 261Þ ¼ 25:43, p < :001;
7-point-scale, ranging from unskilled to scientific skilled partial eta squared ¼ .089), Klepel reading scores
labor (Directoraat-Generaal voor de Arbeidsvoorzien- (F ð1; 261Þ ¼ 22:35, p < :001; partial eta squared ¼ .079),
ing, 1989). Housewives and househusbands were coded and Vocabulary scores (F ð1; 261Þ ¼ 13:12, p < :001;
as a separate category. When the LOA differed between partial eta squared ¼ .048).
mother and father, the highest score was chosen. Fi- Fifth, using Pillai’s trace criterion, the factor ‘group’
nally, if data was missing for one caregiver, the score (ADHD, psychiatric controls, and controls) did not
available was chosen for the analyses. differ in the total number of words produced over 60 s,
irrespective of the test used (Group Type: F ð2; 261Þ ¼
2:27, p ¼ :106; partial eta squared ¼ .017). Furthermore,
3. Results no interaction was found between group membership
and number of words produced in different time inter-
3.1. Word production over time and type of fluency vals (Group Time: F ð6; 520Þ ¼ 1:81, p ¼ :094; partial
eta squared ¼ .021). In contrast, there was an interaction
Group performance on the fluency task, in terms of of Group by Time by Type of test (F ð6; 520Þ ¼ 2:09,
word production over time for both SCF and ILF, was p ¼ :052; partial eta squared ¼ .024). This finding indi-
analyzed in a 2 4 (Type of fluency task Time inter- cated that performance of groups was not linear across
val) General Linear Model (GLM) repeated measures different time intervals and types of fluency.
design. The grouping variable consisted of three inde- Additional GLM multivariate testing revealed only a
pendent categories, namely: (1) children with ADHD, significant interaction between groups and performance
(2) psychiatric controls, and (3) normal controls. The (words produced) as a function of time on the ILF task
child’s sex, Klepel reading scores, vocabulary scores, (ILF: Group Time F ð6; 520Þ ¼ 3:53, p ¼ :002; partial
and the LOA of the caregiver were included as covari- eta squared ¼ .039). This interaction could primarily be
540 P.P.M. Hurks et al. / Brain and Cognition 55 (2004) 535–544

Fig. 1. Productivity-scores over different time samples, groups, and type fluency tests (semantic vs. initial letter fluency). Note. Evaluated at covariates
appeared in the model: LOA, Klepel, Vocabulary test, and sex.

explained by the large group differences in test perfor- letter or words containing the initial letter but not
mance over the first 15 s of the ILF. Children diagnosed starting with it). Error scores for each individual were
with ADHD in phase 2 produced significantly fewer transformed to display a bimodal distribution (i.e., in-
words during the first 15 s of the ILF than did children dividuals either made errors of a certain type or they did
in the control groups (F ð2; 261Þ ¼ 5:62, p ¼ :004, partial not). For each type of errors, the raw data for errors
eta squared ¼ .041; after Bonferoni correction, A vs. C: were collapsed over all the time intervals and across the
p ¼ :022; A vs. B: p ¼ :003; B vs. C: p ¼ :805). Over the same type of fluency. The data within each of these error
succeeding time intervals, all groups performed at a conditions were compared for group performance by use
comparable level (16–30 s: F ð2; 261Þ ¼ 1:03, p ¼ :357, of the Pearson v2 test. When there were less than five
partial eta squared ¼ .008; 31–45 s: F ð2; 261Þ ¼ 1:77, cases in a cell, the Fisher’s Exact Test was chosen.
p ¼ :172, partial eta squared ¼ .013; 46–60 s: F ð2; 261Þ ¼ Again, the grouping variable consisted of three inde-
1:43, p ¼ :241, partial eta squared ¼ .011). Fig. 1 pre- pendent categories, namely: (1) children with ADHD,
sents group performance for each type of fluency (ILF (2) psychiatric controls, and (3) normal controls.
and SCF). For all analyses, Levene’s Test of Equality of On the ILF, a trend was found for children with
Error Variances was not significant. ADHD to make more errors of the miscues-type com-
To examine additionally whether these last-men- pared to the control children (v2 ¼ 4:84, p ¼ :089; Table
tioned results are primarily attributable to concurrent 3). No group differences were found in terms of intru-
phonological impairments, the relation between fluency, sions (v2 ¼ 3:11, p ¼ :191), perseverations (v2 ¼ 0:20,
Klepel reading test scores, and group membership p ¼ :914), and non-words (v2 ¼ 0:27, p ¼ :913). On the
(ADHD, psychiatric controls, and healthy controls) was SCF, however, children with ADHD tended to make
further examined. For this purpose, zero-order correla- more intrusions compared to the controls (intrusions:
tions between the variables were calculated. It was v2 ¼ 5:63, p ¼ :043; perseverations: v2 ¼ 0:06, p ¼ :968;
found that the correlations between the variable ‘group’ non-words: v2 ¼ 2:26, p ¼ :317). Caution is necessary
and the dependent variables (type of fluency time when interpreting these last results, because of the
sample) before correcting for Klepel scores were highly overall low number of errors.
comparable with the correlations the variable ‘group’
and the dependent variables (type of fluency time
sample) after correction for Klepel scores. Based on 4. Discussion
these findings, it can be assumed that the results found
in this study can not be solely described to phonological In a clinical setting, verbal fluency tests are used to
or reading impairments. evaluate the cognitive performance of children with
ADHD. Successful performance on these tasks relies
3.2. Error analyses strongly on attention capacity and executive functioning
(Monsch et al., 1994; Rosser & Hodges, 1994), which are
Errors were coded as one of five types: (1) intrusions aspects of cognition frequently mentioned in relation to
(i.e., noncategorical or non-initial letter errors, and ADHD (Barkley, 1998). Unfortunately, evidence is still
names of people or cities), (2) perseverations, (3) non- inconclusive with regard to the influence of ADHD on
words, (4) miscues (i.e., words starting with a non-initial the performance on the verbal fluency tasks (e.g., Fel-
letter that is phonologically comparable to the initial- ton, Wood, Brown, Campbell, & Harter, 1987; Loge
P.P.M. Hurks et al. / Brain and Cognition 55 (2004) 535–544 541

Table 3
Error analyses
Type of error SCF ILF
ADHD Psychiatric Controls ADHD Psychiatric Controls
n % n % n % n % n % n %
Intrusions
0 intrusion 18 90 116 98 129 99 18 90 108 92 125 96
1 intrusion 2 10 2 2 1 1 2 10 10 8 5 4

Perseverations
0 perseveration 14 70 85 72 92 71 16 80 97 82 105 81
1 perseveration 6 30 33 28 38 29 4 20 21 18 25 19
Non-words
0 non-word 20 100 117 99 125 96 18 90 103 87 116 89
1 non-word 0 0 1 1 5 4 2 10 15 13 14 11
Miscues
0 miscue 20 100 118 100 130 100 12 60 96 81 104 80
1 miscue 0 0 0 0 0 0 8 40 22 19 26 20

et al., 1990). Many studies used the total number of one hand and psychiatric and healthy controls on the
correct words generated over a set period of 60–90 s as other. In contrast, there was a significant interaction of
the main outcome variable, whereas, in theory, the groups by word production over time and as a function
pattern of word production over time may provide ad- of type of fluency, after correction for sex, LOA, vo-
ditional information about the cognitive processes un- cabulary, and reading skills. Analyses indicated that
derlying fluency performance (Troyer, 2000). children classified as having ADHD earlier in life per-
In our study, the number of correct words produced formed significantly less well than their controls (psy-
on both types of verbal fluency (semantic category flu- chiatric or healthy) over the first 15 s of a complex, ILF
ency [SCF] and initial letter fluency [ILF]) decreased task. In contrast, no significant interaction was found
significantly with time (defined in quartiles of 15 s) in all between groups, word production, and time samples on
groups of children. This is consistent with findings from the SCF task.
previous studies of healthy adult participants (Crowe, Several explanations can be proposed based on these
1998; Monsch et al., 1994; Rosser & Hodges, 1994). This findings. For one, children who show behavior related
decrease can be explained in terms of the model of lex- to ADHD at age 7 years (that is, at a phase during
ical organization (Crowe, 1996; Smith & Claxton, 1972), which brain maturation occurs) may show a delay in
which states that there are two types of stores, namely: performance on tests measuring recently acquired au-
(1) a long-term store (‘topicon’) which is readily acces- tomatic processes at age 9, when compared with control
sible and contains common words, and (2) a more ex- children and children with one or more psychiatric
tensive lexicon which is searched after the ‘topicon’ is problems other than ADHD. The idea of impaired au-
exhausted. It can thus be hypothesized that, during the tomation of skills in ADHD is supported by research of
first time period (1–15 s) of either type of fluency task, a Dykman et al. (in Ackerman et al., 1986). Moreover, the
ready pool of frequently used words is available and disturbed pattern in performance over time seemed
automatically activated for production. As time passes specific for ADHD: children classified as having ADHD
the pool becomes exhausted and production becomes performed significantly less well on the initial letter flu-
both more effortful and less productive (Crowe, 1998). ency than children assigned to the psychiatric or the
Therefore to examine the link between test performance healthy control groups.
and ADHD, it may be relevant to include variables, Also, when comparing an initial letter to semantic
such as automatic and controlled processes, as a func- fluency tasks the difference may not only be in how
tion of time in addition to the overall performance on a much effort the task ‘requires’ but also in how ‘difficult’
fluency task. the task is. Priming studies suggest that language is
In this context, children with ADHD, psychiatric represented semantically (Collins & Loftus, 1975; Je-
controls and controls were found in the study presented scheniak & Levelt, 1994; Mercer, 1976), so when asked
in this article to produce over 60 s a similar total number to generate words according to a category, this matches
of words on both verbal fluency tasks. Thus, the normal with the way in which language is stored. An initial
way of scoring performance on the verbal fluency task letter fluency task is different in that language is not
could not discriminate between ADHD children on the ‘organized’ alphabetically, thus this type of task is
542 P.P.M. Hurks et al. / Brain and Cognition 55 (2004) 535–544

relatively novel to most participants and likely requires cessing, and noted that the development of response
more executive function than does a semantic fluency organization largely accounted for the improvement in
task. For deficits in executive functioning are often information processing efficiency between 7 and 9 years
mentioned in relation to ADHD, this may well add to of age. Cross-sectional studies indicate that word fluency
the explanation of the finding that children suffering improves with increasing age to at least 13 years, with
from ADHD perform less well than controls on the written word fluency increasing up to 18 years (Levin,
initial letter fluency tests and no differences were found Song, Ewing-Cobbs, Chapman, & Mendelsohn, 2001).
between groups on the semantic category fluency test. Furthermore, significant deviations were found between
Additionally, these results emphasize the importance the test performance of healthy children aged 9 and
of ‘testing the limits’ when clinically assessing a child adults (Hurks et al, in press). Over 60 s, healthy children
with ADHD characteristics. From the present results, it produced more correct responses on the SCF task than
can be concluded that children classified as having on the ILF task, while with adults this relation was in-
ADHD are able to achieve on a similar level of per- verted. Over 60 s, adults appeared to be more successful
formance as controls, if they are given sufficient time to in producing words in the ILF task (Butters, Granholm,
do a task. Thus, the total time on task is relevant, in a Salmon, Grant, & Wolfe, 1987). Therefore, at the mo-
way that children with ADHD seem to need extra time ment of neuropsychological testing (age 9), the cognitive
to ‘start the engine.’ However, because of relative small development of the children in this study may have
samples and effect sizes additional research is needed to passed or was about to pass an important milestone in
test the above-mentioned assumptions. terms of controlled processing. From this perspective,
Finally, while studying the errors made by the sub- these children should be tested again when they are
jects, a trend was found for children with ADHD to older to see whether there are differences in, for example,
make more errors of the miscues-type (i.e., words automatic and effortful processing.
starting with a non-initial letter that is phonologically
comparable to the initial-letter or words containing the
initial letter but not starting with it) on the ILF than the References
controls. These results may indicate that ADHD chil-
dren use search strategies (i.e., the child may either use Achenbach, T. M., & Edelbrock, C. (1983). Manual for the child
the sound of the letter to retrieve other words or use its behavior checklist and revised child behavior profile. Burlington:
University of Vermont Department of Psychiatry.
visual image) inefficiently. Using only the sound of the Ackerman, P. T., Anhalt, J. M., Holcomb, P. J., & Dykman, R. A.
letter to retrieve words beginning with for instance the (1986). Presumably innate and acquired automatic processes in
letter ‘s’ may give rise to miscues such as words starting children with attention and/or reading disorders. Journal of Child
with ‘z’ or ‘c’ or words containing but not starting with Psychology and Psychiatry, 27, 513–529.
an ‘s.’ More research in this area is still needed to study American Psychiatric Association (1987). Diagnostic and statistical
manual of mental disorders (3rd ed.). Washington, DC: American
the primary approach used by the ADHD group as Psychiatric Association.
opposed to the other groups. American Psychiatric Association (1994). Diagnostic and statistical
The present study includes some limitations. For in- manual of mental disorders (4th ed.). Washington, DC: American
stance, the DSM classifications were measured solely on Psychiatric Association.
basis of the outcome of a semi-structured interview with Barkley, R. A. (1998). Attention-deficit/hyperactivity disorder: A
handbook for diagnosis and treatment (2nd ed.). New York: The
the caregivers of the child. Cross-informant reports, like Guilford Press.
Teacher Reports, were not included here. Therefore, it is Barkley, R. A., Grodzinsky, G., & DuPaul, G. J. (1992). Frontal lobe
possible that the number of children with externalizing functions in attention deficit disorder with and without hyperac-
behavior was overestimated. However, Barkley (1998) tivity: A review and research report. Journal of Abnormal Child
reported evidence of a 90% overlap between parent re- Psychology, 20, 163–188.
Bernstein, D. (1989). Language development: The school age years. In
ports and teacher reports. An additional limitation is E. Tiegermen (Ed.), Language and communication disorders in
that the exclusion and drop out of children in the children. Columbus, OH: Merrill/MacMillon.
analyses (37% of the selected sample) could limit the Borcherding, B., Thompson, K., Kruesi, M., Bartko, J., Rapoport, J.
generalizability of the results to the population. L., & Weingartner, H. (1988). Automatic and effortful processing
Also, no measurement of broader based language in attention deficit/hyperactivity disorder. Journal of Abnormal
Child Psychology, 16(3), 333–345.
abilities was included in the test protocol (e.g., standard Butters, N., Granholm, E., Salmon, D., Grant, L., & Wolfe, J. (1987).
test of semantic abilities and of general word retrieval Episodic and semantic memory: A comparison of amnestic and
abilities under a non-speeded condition): a limitation that demented patients. Journal of Clinical and Experimental Neuropsy-
is of concern given the known impairments in verbal flu- chology, 9, 479–497.
ency associated with specific language impairments. Chen, W. J., Faraone, S. V., Biederman, J., & Tsuang, M. T. (1994).
Diagnostic accuracy of the child behavior checklist scales for
Lastly, the study used only a 2-year follow-up. attention-deficit hyperactivity disorder: A reciever-operating char-
McKay, Halperin, Schwartz, & Sharma (1994) reported acteristic analysis. Journal of Consulting and Clinical Psychology,
that response organization was pivotal to output pro- 62, 1017–1025.
P.P.M. Hurks et al. / Brain and Cognition 55 (2004) 535–544 543

Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory Korkman, M., Kirk, U., & Kemp, S. L. (1998). NEPSY. A
of semantic processing. Psychological Review, 82, 407–428. developmental neuropsychological assessment. San Antonio, TX:
Crowe, S. F. (1996). The performance of schizophrenic and depressed The Psychological Corporation.
participants on tests of fluency: Support for a compromise in Kortenbout van der Sluijs, M. J., Levita, D. J., Manen, R., & Defares,
dorso-lateral pre-frontal functioning. Australian Psychologist, 31, P. B. (1997). Test bespreking: Amsterdams diagnostisch interview
204–209. voor Kinderen (ADIKA). Nederlands Tijdschrift voor de Psychol-
Crowe, S. F. (1998). Decrease in performance on the verbal fluency test ogie, 52, 135–138.
as a function of time: Evaluation in a young healthy sample. Journal Koziol, L. F., & Stout, C. E. (1992). Use of a verbal fluency measure in
of Clinical and Experimental Neuropsychology, 20(3), 391–401. understanding and evaluation ADHD as an executive function
De Bruyn, E. E. J., Van der Steene, G., Van Haasen, P. P., Coetsier, P., disorder. Perceptual and Motor Skills, 75, 1187–1192.
Pijl, Y. J., Spoelders-Claes, R., Poortinga, Y. H., Stinissen, J., & Kroes, M., Kalff, A. C., Kessels, A. G. H., Steyaert, J., Feron, F., Van
Lutje Spelberg, H. C. (1986). Wechsler intelligence scale for Someren, A. J. W. G. M., Hurks, P. P. M., Hendriksen, J. G. M., Van
children—revised (WISC-R). Lisse: Swets & Zeitlinger. Zeben, T. M. C. B., Rozendaal, N., Crolla, I. F. A. M., Troost, J.,
Directoraat-Generaal voor de Arbeidsvoorziening (1989). Handleiding Jolles, J., & Vles, J. S. H. (2001). Child psychiatric diagnoses in a
voor de functieanalyse. Den Haag: SDU uitgeverij. population of Dutch schoolchildren aged 6 to 8 years. Journal of the
Ezpeleta, L., De la Osa, N., Domenech, J. M., Navarro, J. B., Losilla, American Academy of Child and Adolescent Psychiatry, 40(12), 1401–
J. M., & J udez, J. (1997). Diagnostic agreement between clinicians 1409.
and the diagnostic interview for children and adolescents-DICA-R- Kusche, C. A., Cook, E. T., & Greenberg, M. T. (1993). Neuropsy-
in an outpatient sample. Journal of Child Psychology and Psychi- chological and cognitive functioning in children with anxiety,
atry, 38(4), 431–440. externalizing, and comorbid psychopathology. Journal of Clinical
Felton, R. H., Wood, F. B., Brown, I. S., Campbell, S. K., & Harter, Child Psychology, 22, 172–195.
M. R. (1987). Separate verbal memory and naming deficits in Mercer, N. (1976). Frequency and availability in the encoding of
attention deficit disorder and reading disabilities. Brain and spontaneous speech. Language and Speech, 19, 129–143.
Language, 31, 171–184. Levin, H. S., Song, J., Ewing-Cobbs, L., Chapman, S. B., &
Fischer, M., Barkley, R. A., Edelbrock, C. S., & Smallish, L. (1990). Mendelsohn, D. (2001). Word fluency in relation to severity of
The adolescent outcome of hyperactive children diagnosed by closed head injury, associated frontal brain lesions, and age at
research criteria: II. Academic, attentional, and neuropsycholog- injury in children. Neuropsychologia, 39, 122–131.
ical status. Journal of Consulting and Clinical Psychology, 58, Levy, F., Hay, D., McLaughlin, M., Wood, C., & Waldman, I. (1996).
580–588. Twin sibling differences in parental reports of ADHD, speech,
Fodor, J. (1983). The modularity of mind. Cambrigde, MA: MIT Press. reading and behaviour problems. Journal of Child Psychology and
Granero Perez, R., Ezpeleta Ascaso, L., Domenech Massons, J. M., & Psychiatry, 37(5), 569–578.
De la Osa Chaparro, N. (1998). Characteristics of the subject and Lezak, M. D. (1995). Neuropsychological assessment (3rd ed.). Oxford:
interview influencing the test–retest reliability of the diagnostic Oxford University Press.
interview for children and adolescents—revised. Journal of Child Loge, D. V., Station, R. D., & Beatty, W. W. (1990). Performance of
Psychology and Psychiatry, 39(7), 963–972. children with ADHD on tests sensitive to frontal lobe dysfunction.
Grodzinsky, G. M., & Barkley, R. A. (1999). Predictive power of Journal of the American Academy of Child and Adolescent Psychi-
frontal lobe tests in the diagnosis of attention deficit hyperactivity atry, 29, 540–545.
disorder. Clinical Neuropsychologist, 13, 12–21. McKay, K. E., Halperin, J. M., Schwartz, S. T., & Sharma, V.
Grodzinsky, G. M., & Diamond, R. (1992). Frontal lobe functioning (1994). Developmental analysis of three aspects of information
in boys with attention deficit hyperactivity disorder. Journal of processing: Sustained attention, selective attention, and response
Clinical Psychology, 45, 427–446. organization. Developmental Neuropsychology, 10, 121–132.
Hazell, P. L., Carr, V. J., Lewin, T. J., Dewis, S. A. M., Heathcote, D. Monsch, A. U., Bondi, M. W., Butters, N., Paulsen, J. S., Salmon, D.
M., & Brucki, B. M. (1999). Effortful and automatic information P., Brugger, P., & Swenson, M. R. (1994). A comparison of
processing in boys with ADHD and specific learning disorders. category and letter fluency in Alzheimer’s disease and Huntington’s
Journal of Child Psychology and Psychiatry, 40, 275–286. disease. Neuropsychology, 8, 25–30.
Hurks, P. P. M., Vles, J. S. H., Hendriksen, J. G. M., Kalff, A. C., Oades, R. D. (1998). Frontal, temporal and lateralized brain function
Feron, F. J. M., Kroes, M., Van Zeben, T. M. C. B., Steyaert, J., & in children with attention-deficit hyperactivity disorder: A psycho-
Jolles, J. (in press). Semantic category fluency versus initial letter physiological and neuropsychological viewpoint on development.
fluency over a 60-second interval as a measure of automatic and Behavioural Brain Research, 94, 83–95.
controlled processing in healthy school-aged children. Journal of Pennington, B. E., & Ozonoff, S. (1996). Executive functions and
Clinical and Experimental Neuropsychology. developmental psychopathology. Journal of Child Psychology and
Jescheniak, J. D., & Levelt, W. J. M. (1994). Word frequency effects in Psychiatry and Allied Disciplines, 37, 51–87.
speech production: Retrieval of syntactic information and of Perugini, E. M., Harvey, E. A., Lovejoy, D. W., Sandstrom,
phonological form. Journal of Experimental Psychology: Learning, K., & Webb, A. H. (2000). The predictive power of
Memory and Cognition, 20, 824–843. combined neuropsychological measures for attention-deficit/
Kalff, A. C., Kroes, M., Vles, J. S. H., Bosma, H., Feron, F. J., hyperactivity disorder in children. Child Neuropsychology,
Hendriksen, J. G., Steyaert, J., Van Zeben, T. M., Crolla, I. F., & 6(2), 101–114.
Jolles, J. (2001). Factors affecting the relation between parental Pineda, D., Ardila, A., & Rosselli, M. (1999). Neuropsychological and
education as well as occupation and problem behaviour in Dutch 5- behavioural assessment of ADHD in 7–12-year-old children: A
to 6-year-old children. Social Psychiatry and Psychiatric Epidemi- discriminant analysis. Journal of Learning Disabilities, 32(2), 159–
ology, 36, 324–331. 173.
Klorman, R., Hazel-Fernandez, L. A., Shaywitz, S. E., Fletcher, J. M., Reader, M. J., Harris, E. L., Schuerholz, L. J., & Denckla, M. B.
Marchione, K. E., Holehan, J. M., Stuebing, K. K., & Shaywitz, B. (1994). Attention deficit hyperactivity disorder and executive
A. (1999). Executive functioning deficits in attention deficit/ dysfunction. Developmental Neuropsychology, 10, 493–512.
hyperactivity disorder are independent of oppositional defiant or Riva, D., Nichelli, F., & Devoti, M. (2000). Developmental aspects of
reading disorder. Journal of the American Academy Child and verbal fluency and confrontation naming in children. Brain and
Adolescent Psychiatry, 38, 1148–1155. Language, 71, 267–284.
544 P.P.M. Hurks et al. / Brain and Cognition 55 (2004) 535–544

Rosser, A., & Hodges, J. (1994). Initial letter and semantic category development. Journal of the American Academy of Child Psychiatry,
fluency in Alzheimer’s disease, Huntington’s disease, and progres- 35, 1213–1226.
sive supranuclear palsy. Journal of Neurology, Neurosurgery and Troyer, A. K. (2000). Normative data for clustering and switching on
Psychiatry, 57, 1389–1394. verbal fluency tasks. Journal of Clinical and Experimental Neuro-
Sagvolden, T., & Sergeant, J. A. (1998). Attention deficit/hyperactivity psychology, 22(3), 370–378.
disorder—from brain dysfunctions to behaviour. Behavioural Brain Van den Bos, K. P., Lutje Spelberg, H. C., Scheepstra, A. J. M., & De
Research, 94, 1–10. Vries, J. R. (1994). De Klepel. Lisse: Swets and Zeitlinger.
Schneider, W., & Shiffrin, R. M. (1977). Controlled and automatic Van der Meere, J., & Sergeant, J. (1988). Acquisition of attention skill
human information processing: I. Detection, search, and attention. in pervasively hyperactive children. Journal of Child Psychology and
Psychological Review, 84, 1–66. Psychiatry, 29(3), 301–310.
Schuerholz, L. J., Singer, H. S., & Denckla, M. B. (1998). Gender Van Loon, A. (1985). Voorstelbaarheidswaarden van Nederlandse
study of neuropsychological and neuromotor function in children woorden. Lisse, The Netherlands: Swets and Zeitlinger.
with Tourette syndrome with and without attention-deficit hyper- Verhulst, F. C., Koot, J. M., & Van der Ende, J. (1996). Handleiding
activity disorder. Journal of Child Neurology, 13, 277–282. voor de CBCL/4-18. Rotterdam, The Netherlands: Afdeling Kinder-
Sergeant, J. A., Geurts, H., & Oosterlaan, J. (2002). How specific is a en Jeugdpsychiatrie, Sophia Kinderziekenhuis/Academisch Zie-
deficit of executive function for attention-deficit/hyperactivity kenhuis Rotterdam/Erasmus Universiteit Rotterdam.
disorder? Behavioural Brain Research, 130, 3–28. Weiss, G., & Trokenberg Hechtman, L. (1993). Hyperactive children
Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic grow up: ADHD in children, adolescents and adults (2nd ed.). New
human information processing: II. Perceptual learning, automatic York: The Guilford Press.
attenting, and a general theory. Psychological Review, 84, 127– Welner, Z., Reich, W., Herjanic, B., Jung, K. G., & Amado, H. (1987).
190. Reliability, validity, and parent–child agreement studies of the
Smith, P. T., & Claxton, G. L. (1972). Lexical search and phonemic diagnostic interview for children and adolescents (DICA). Journal of
organisation in memory. Paper presented at the Paper presented to the American Academy Child and Adolescent Psychiatry, 26, 649–653.
the Experimental Psychology Society London. Weyandt, L. L., & Willis, W. G. (1994). Executive functions in school-
Taylor, E., Chadwick, O., Heptinstall, E., & Danckaerts, M. (1996). aged children: Potential efficacy of tasks in discriminating clinical
Hyperactive and conduct problems as risk factors for adolescent groups. Developmental Neuropsychology, 10(1), 27–38.