Child Neuropsychology, 13: 319–332, 2007

http://www.psypress.com/childneuropsych
ISSN: 0929-7049 print / 1744-4136 online
DOI: 10.1080/09297040600837370

SPEECH RATE AND FLUENCY IN CHILDREN

AND ADOLESCENTS

Isabel Pavão Martins,1 Rosário Vieira,1 Clara Loureiro,1 and

M. Emilia Santos2
1
Language Research Laboratory, I.M.M., Lisbon Faculty of Medicine, Hospital de
Sta Maria, Lisboa, Portugal, and 2Escola Superior de Saúde do Alcoitão, Alca-
bideche, Portugal

Reduced speech fluency is frequent in clinical paediatric populations, an unexplained

finding. To investigate age related effects on speech fluency variables, we analysed samples
of narrative speech (picture description) of 308 healthy children, aged 5 to 17 years, and
studied its relation with verbal fluency tasks. All studied measures showed significant devel-
opmental effects. Speech rate and verbal fluency scores increased, while pauses, repetitions
and locution time declined with age. Speech rate correlated with semantic fluency tasks
suggesting that it also depends upon the efficacy of lexical retrieval. These results indicate
that the interpretation of disorders of speech fluency in childhood must incorporate age
appropriate norms.

Keywords: speech fluency in children, speech rate, acquired aphasia in children

INTRODUCTION
Mutism and nonfluent types of speech are often observed in different pediatric clin-
ical populations, namely in acquired childhood aphasia, following cerebellar surgery, in
traumatic brain injury and in particular social contexts (Gordon, 2001; Dayer, Roulet,
Maeder, Deonna, 1998; Riva & Giorgi, 2000). This age specific prevalence of nonfluent
types of speech has not been fully explained.
Assessment of speech fluency is usually performed through the analysis of sponta-
neous conversational/narrative speech. However, the analysis of the different aspects of
speech fluency in children have not received much attention despite their clinical interest,
since they are among the major factors responsible for the appropriate classification of
aphasia and other acquired disorders of speech and language.
Indeed, the fluency dimension of speech depends upon multiple factors namely
speech rate, phrase length, articulatory agility, effort in speech production, number

The authors are indebted to Dr Klaus Wilms, of the University of Aachen, for his help in the statistical
analysis of the data. Part of the results presented in this study were presented as a Speech Therapy Dissertation
by Rosário Vieira (Abstract : Jornal da Associação Portuguesa de Terapeutas da Fala, 1999). Definite results
were presented at the Meeting of the Portuguese Society of Neurology in 2003.
Address correspondence to Isabel Pavão Martins, MD, PhD, Laboratório de Estudos de Linguagem,
Centro de Estudos Egas Moniz, IMM, Lisbon Faculty of Medicine, Hospital Sta Maria, 1648-028 Lisbon,
Portugal. Fax: 351-21-7934480. E-mail: isabel_martins@fm.ul.pt

© 2007 Psychology Press, an imprint of the Taylor & Francis Group, an Informa business
320 I. P. MARTINS ET AL.

and duration of pauses, complexity of syntax, and the integrity of the prosodic line
(Deloche, Jean-Louis, & Seron, 1979; Feyereisen, Pillon, & Partz, 1991; Goodglass &
Kaplan, 1983; Greenwald, Nadeau, & Rothi, 2000; Kerschensteiner, Poeck, & Brunner,
1972; Klein, Masur, Farber, Shinnar, & Rapin, 1992). Some authors recommend the
use of total speaking time and the number of words produced to tell a story as addi-
tional measures of fluency (Kreindler, Mihailescu, & Fradis, 1980). Because of its
multivariable composition, speech fluency is particularly difficult to quantify. Speech
rate and pauses are considered powerful discriminators between nonfluent and fluent
speech (Kerschensteiner et al., 1972). Speech rate, a measure of processing and pro-
duction speed, is commonly expressed by the number of words produced per minute
(WPM) or syllables per minute. In adults, its normal range varies between 110 to
175WPM or 50–150WPM, according to different authors (Greenwald et al., 2000;
Kerschensteiner et al., 1972). There is little knowledge about the normal development
of narrative speech fluency measures in children, and age appropriate norms are usu-
ally not taken into account in the classification of acquired childhood aphasias.
Although there is compelling evidence that fluent aphasias do occur in childhood, jargon
aphasia is a rarity, especially in patients younger than 7 years of age (Klein et al., 1992;
Paquier & Van Dongen, 2000), and transient mutism and nonfluent speech predomi-
nate in many series of children with aphasia (Cranberg, Filley, Hart, & Alexander,
1987; Loonen & Van Dongen, 1990; Martins & Ferro, 1993; Martins, 1997; Satz,
1991; Woods & Teuber, 1978).
Given that language is “a moving target” (Bates et al., 2001), in continuous acquisi-
tion, the fluency dimension of speech must be compared to the expected level of develop-
ment. Indeed, in a very detailed study of expressive language in children with perinatal
brain lesions, it was found that the profile of language impairment changed considerably
when the results were corrected for age (Bates et al., 2001).
Tests of verbal fluency (VFT) require the production of words, of a specific
semantic or phonemic category, during one minute. Although they provide one mea-
sure of speech fluency, they are primarily used to assess executive functions, for they
require, in addition to lexical knowledge, an orchestration of different capacities
mainly subserved by the frontal lobes: planning a search strategy, maintaining the
instructions and retrieved items on line, inhibiting repetitions, automatic responses and
violations of the rules, and alternating search strategies (switching) between possible
semantic or phonological entries. Given the late maturation of the frontal lobes and
executive functions during development (Anderson, 2002; Anderson, Anderson,
Northam, Jacobs, & Catroppa, 2001; Thatcher, 1991), these, and other executive func-
tions tests, can be particularly useful to evaluate children and adolescents during spe-
cific periods of maturation to understand normal development and its disturbances.
Functional brain imaging studies have demonstrate the participation of the prefrontal
cortex in those tasks, since they are associated to an activation of the dorsolateral pre-
frontal gyrus, both in adults and children (Frith, Friston, Liddle, & Frackowiak, 1991;
Gaillard et al., 2000, 2003).
Verbal fluency tests are included in comprehensive neurobehavioral batteries
designed for children, both in Europe and in North America, and there are norms for
different ages and cultural/linguistic backgrounds (Halperin, Healey, Zeitchik, Ludman, &
Weinstein, 1989; De Agostini et al., 1998; Sauzéon, Lestage, Raboutet, N’Kaoua, &
Claverie, 2004; Riva, Nichelli, & Devoti, 2000; Delis, Kaplan, & Kramer, 2001; Korkman,
Kirk, & Kamp, 1997).
 SPEECH RATE AND FLUENCY IN CHILDREN 321

To understand the development of speech fluency in childhood, we analyzed sev-

eral measures of narrative speech (speech rate, total speaking time, number of pauses and
repetitions) in healthy children and adolescents and studied its relation with age and with
lexical retrieval efficacy (verbal fluency tasks). Our hypothesis was that the measures
usually incorporated in the fluency classification of speech change with age, thus provid-
ing a phenomenological explanation for the high prevalence of nonfluency of childhood
aphasia.

METHOD
Subjects
Three hundred and eight children (152 boys and 156 girls) participated in the study.
Children were selected from five regular schools in urban districts (kindergarten, primary
and secondary schools), after parents gave their written consent to a letter describing the
aims and the procedures of the study. It was decided to test children at 2 year intervals
(uneven ages). In each school alternate grades were selected and all children attending that
grade, with the same age (in years), were tested. Age selected in each grade is the usual age
of children attending that grade at mid to late term. Children aged 5 years old (N = 59) were
selected among preschoolers, children aged 7 years old from the First Grade (N = 64), 9
years old from Third Grade (N = 55), 11 years old from Fifth Grade (N = 33), 13 years old
from Seventh Grade (N = 31), 15-year-old children among Ninth Grades (N = 34), and the
17-year-old participants were selected among eleventh graders (N = 32) (Table 1). All were
native speakers of European Portuguese. Data regarding parents occupation was collected
and participants were classified in a four-point socioeconomic status (SES) scale accord-
ingly (considering the parent with the highest SES): 3 - unskilled manual labor; 2 - skilled
manual work, 1 - clerk or technical work and small trade, 0 - professionals, large trade, or
land owners. Exclusion criteria were the presence of any of the following (reported by
teachers or families): deafness, developmental speech and language disorders (including
stuttering), Portuguese as the second language, or refusal to participate.

Table 1 Population (N = 308).

Age Groups

5-yr-old 7-yr-old 9-yr-old 11-yr-old 13-yr-old 15-yr-old 17-yr-old Total

N (%) 59 64 55 33 31 34 32 308

Gender
F (50.6) 30 33 28 18 16 16 15 156
M (49.4) 29 31 27 15 15 18 17 152
Social Economic Status (SES)
0 (30.5) 19 19 18 8 8 13 9 94
1 (22.7) 4 7 8 16 14 11 10 70
2 (21.1) 15 12 9 9 5 8 7 65
3 (25.6) 21 26 20 0 4 2 6 79

Age x Gender (χ2 = 0.052 n.s); Age x SES (χ2 = 10.65, p = .001).
Distribution of the population by age groups and gender. No significant differences were found, Age x Gender
(χ2 = 0.052, n.s.).
322 I. P. MARTINS ET AL.

Procedures
Speech samples. The assessment was performed individually, in a room set
apart from the class, during school hours, between mid and late term by one of the authors.
Testing procedure was explained to each child. Narrative speech was elicited through
the use of the “Cookie theft” picture from the Boston Diagnosis Aphasia Examination
(Goodglass et al., 1983). Children were shown the picture and asked to say as much as
possible about it, with no time limits. The examiner did not interrupt the child. When the
child finished the examiner asked “Do you see anything else?” and if there was any further
information, it was also included in the speech sample.
All speech samples were audiotape recorded and transcribed verbatim using
conventional spelling. All words produced were counted (content and function words,
including word repetitions). Four aspects of the narrative discourse were analyzed: 1)
Total Locution Time (TLT): audio-playbacks of speech samples of each child were timed
in minutes and seconds, with a stopwatch. Any intervention of the examiner was sub-
tracted, according to the methodology described by Van Dongen, Paquier, Raes, & Creten
(1994), to calculate the TLT in seconds; 2) Speech rate: this measure was obtained as
described by Shipley & McAfee (1992), by the quotient between all words produced and
TLT, and estimated for 60 seconds when the child spoke for less than one minute. The
result is expressed as the average number of words produced per minute (WPM); 3) Num-
ber of pauses: a pause was defined as any interruption of the flow of speech of 4 seconds
or more, because hesitation pauses in adults and adolescents are not longer than 3 seconds
(Goldman-Eisler, 1968), and, besides, it is difficult to quantify manual pauses inferior to
4 seconds; and 4) Number of Repetitions. These were used as a measure of disfluency
(McLaughlin & Cullinan, 1989) and included the total number of part-word, whole word,
or phrase repetition.
To validate the measures obtained, samples of speech (representing about 10% of
the total sample) were randomly selected and analyzed by a naïve observer, blinded for the
initial results. Interjudge reliability was estimated by correlation analysis.

Verbal fluency tasks. Elder children (age groups 11, 13, 15 and 17 years old)
were also evaluated by five verbal fluency tasks using the methodology described by
Baker et al. (2001). In the two semantic fluency tasks participants were requested to pro-
duce as many words as possible belonging to the animals or food category, during one
minute. In the three phonemic fluency tasks the participants were asked to produce words
beginning with the letters P, M, and R. The choice of these letters, instead of the usual
F, A, and S (Borkowsky, Benton, & Spreen, 1967), was based both on: a) their ortho-
graphic transparency and b) the number of words in the lexicon (entries in a Portuguese
Language Dictionary) beginning by them. According to the latter, P is the easiest and R
the most difficult of these tasks. The score obtained in fluency tasks corresponds to the
total number of words correctly produced. Repetitions, proper names, and wrong items
were scored as errors.

Statistical analysis. Statistical analysis was performed using a SPSS Base system
12.0 for Windows XP Professional Software. Our primary hypothesis was that narrative
speech and verbal fluency measures were subject to developmental effects and would
improve with age. Our secondary hypothesis was that lexical retrieval ability (assessed by
VFTs) would correlate with speech rate, pauses, and repetitions on narrative speech.
 SPEECH RATE AND FLUENCY IN CHILDREN 323

Chi-square test (χ2) was used to analyze differences in frequency distribution (by
gender and SES) among age groups. Differences between demographic factors (age
groups, gender, SES) for the dependent variables (speech rate, TLT, pauses, repetitions,
and scores of VFTs) were performed by Student t-test or Mann-Whitney tests (gender dif-
ferences) or ANOVA or Kruskall-Wallis test (K-W). The effect size was measured by
eta2. Regression analysis was performed to understand the effect of age and SES in narra-
tive speech measures and on VFTs. Furthermore a partial correlation matrix, controlled
for age, was computed to study the relationships between the different fluency measures.

RESULTS
There were 308 children. Distribution by age, gender, and SES is presented in Table 1.
No differences were observed in gender distribution by age group (χ2 = .05, n.s.), but
there were significant differences concerning the distribution of SES x Age (χ2 = 10.6; df
= 1; p = .001).

Speech measures
The distribution of speech rate in this population followed a normal curve (One
sample Kolgomorov-Smirnov test, Z = .93 (2-tailed), p > .10), but the other speech vari-
ables did not follow a normal distribution and were analyzed by nonparametric tests.
Considering all children tested, speech rate was on average 91.2 WPM (words per
minute) with a standard deviation of 39.4, ranging from 17.9 to 221.3. Children made few
pauses (median = 3) and few repetitions (median = 1) during narrative speech and their
median locution time was 39 seconds, ranging from 13 to 381 seconds.
There was a significant interrater correlation of all measures: total locution time
(Spearman rho = .99, p = < .000), number of repetitions (Spearman rho = .99, p = < .000),
speech rate (r = .99, p = < .000), and number of pauses (Spearman rho = .98, p = < .000).
There were no significant gender differences in speech rate (t = −.17 (306), p > .10,
n.s.), number of pauses (Z = −.18, p > .10, n.s.), repetitions (Z = −1.44, p > .10, n.s.) or
total locution time (TLT) (Z = −.34, p > .10, n.s.) considering all participants. Further-
more, there was no gender effect on each age group, nor age group x gender interaction, in
a two-factor ANOVA (gender and age group) for speech rate.
There was a significant age effect for all measures studied: speech rate (F[6,301] =
22.89; p < .0001) (one-way ANOVA) (eta2 = .31); total locution time (K-W = 28.84; 6;
p < .000); number of pauses (K-W = 43.61; 6; p < .000); and number of repetitions (K-W
= 20.79; 6; p < .002). Speech rate increased while TLT, pauses, and repetitions decreased
significantly with age. Younger children spoke for longer periods of time and made more
detailed, enumerative descriptions, with more repetitions, hesitations or pauses, than older
children. Mean or median values, standard deviations, minimum and maximum values,
and extreme percentile values (that may be used as cutoff points between normal and
extreme values) are presented for each speech measure in each age group (Table 2).
Newman-Keuls post hoc analysis was carried out to analyze age group differences
in speech rate (level of significance = .01). Five different groups were found: participants
at 5 and 7 years of age had a significantly lower speech rate than children over 7, those
aged 7 had lower rates than those 11 years or older, 9-year-old participants had rates
below those aged 13–17, and 17-year-old participants had higher rates than those aged 11
or younger.
 Table 2 Speech Variables in Different Age Groups.

Age Groups

5-yr-old 7-yr-old 9-yr-old 11-yr-old 13-yr-old 15-yr-old 17-yr-old

(N = 59) (N = 64) (N = 55) (N = 33) (N = 31) (N = 34) (N = 32) Test Df p=

Speech rate
Mean (SD) 64.1(28.3) 73.2 (38.7) 88.8 (34.6) 101.3 (33.2) 114.2 (30.5) 115.2 ( 23.1) 125.1 (26.4) F=22.89 .000
Min-Max 22.2–129.0 20.5–207.7 17.9–207.7 47.0–175.3 52.2–166.9 67.6–164.5 61.9–181.1 (6,301)
10th percentile 30.0 34.4 53.4 57.5 69.1 84.4 94.4
5th percentile 22.7 26.1 42.3 48.8 60.1 76.7 75.8
Number of Pauses K-W=36.28
Median 4.0 3.0 2.0 2.0 2.0 3.0 2.5 .000
Min-Max 0–13 0–15 0–9 0–6 0–8 1–10 0–6 (2)

324
90th percentile 9.0 9.0 5.0 4.0 5.6 5.5 5.0
95th percentile 10.0 10.5 6.2 5.3 6.8 9.3 5.4
Total Locution Time K-W=26.43
Median 51 49.5 38.0 38.0 32.0 35.0 35.0 .000
Min-Max 14–186 13–381 13–149 13–59 17–87 23–141 14–71 (2)
Repetitions K-W=12.16
Median 1.0 2.0 0.0 1.0 1.0 1.0 1.0 .005
Min-Max 0–11 0–9 0–13 0–9 0–6 0–3 0–7 (2)
90th percentile 6.0 6.0 5.0 5.6 4.0 2.0 2.7
95th percentile 7.0 8.8 7.0 8.3 6.0 3.0 5.1

Mean values, standard deviations and minimum and maximum values are presented for each speech measure in each age group. Speech rate increased while TLT, pauses and
repetitions decreased significantly with age.
 SPEECH RATE AND FLUENCY IN CHILDREN 325

Comparing the speech rate produced by this sample of healthy children, with scores
obtained by adults performing the same task (using the same methodology) (Afonso, 1994),
we found that 45.7% of children performed below minimum adult range and 78.9% scored
two or more standard deviations below mean adults scores (of whom, 18.2% in the 5- year-
old group, 18.5% of children aged 7 years, 16.2% of 9-year-olds, 7.8% of 11-year-olds,
5.5% of 13-year-olds, 7.1% of 15-year-olds, and 5.5% of 17-year-old participants). The
effect of SES on the different variables was analyzed. SES had a significant effect (F[3,304]
= 8.22; p < .000) on speech rate, although it was not linear, for children belonging to SES
class 1 had a higher speech rate (109.6 WPM) than those of class 0 (90.5 WPM), class 2
(84.9 WPM), and class 3 (81.6 WPM) on Newman-Keuls post hoc analysis (level of signifi-
cance = .01). SES was significantly related to TLT (K-W = 14.99; 3; p = .002) and to the
number of pauses produced (K-W = 10.33; 3; p = .02) but not with the repetitions.
Since there was a significant interaction between age groups and SES, and both had
impact on measures of narrative speech, a linear regression analysis (stepwise procedure)
was carried out for each of the speech variables (speech rate, TLT, and pauses) as depen-
dent variables. The independent variables were age and SES. The only factor retained in
the model was age. The following adjusted R square was obtained: Speech rate (β =
.55, t = 11.59, p < .000; r2 aj = .30, F[306,1] = 133.59, p < .000), TLT (β = −.24, t = −4.39,
p < .000; r2 aj = .06, F[307,1] = 19.29, p < .000), pauses (β = −.28, t = −5.15, p <
.000; r2 aj = .08, F[307,1] = 26.49, p < .000). We searched for a possible colinearity
between variables included in the regression model (age and SES). There was no indica-
tion of colinearity, since the highest condition index only had a value of 6.65.

Verbal fluency
Scores obtained in all VFT followed a normal distribution: animal fluency (Z =
1.34 [2-tailed], p = .06), food fluency (Z = 1.15 [2-tailed], p > .10), P fluency (Z = .87
[2-tailed], p > .10), M fluency (Z = .90 [2-tailed], p > .10), and R fluency (Z = 1.18 [2-
tailed], p > .10). Considering all children tested (from 11 to 17 years old), average scores
were higher for semantic than phonemic fluency tasks. Mean scores were better for animal
category (20.00 + 5.29) than for food category (17.25 + 5.75), followed by P fluency
(10.36 + 4.42), M fluency (9.73 + 4.22), and R fluency (9.14 + 3.67).
Scores in all verbal fluency tasks increased significantly with age: (one-way
ANOVA): animals (F[3,126] = 4.03; p = .009) (eta2 = .09); food (F[3,126] = 4.18; p =
.007) (eta2 = .10); P fluency (F[3,116] = 6.06; p = .001) (eta2 = .14); M fluency (F[3,116] =
3.34; p = .022) (eta2 = .08), and R fluency (F[3,116] = 4.26; p= .007) (eta2 = .10). Mean
scores, standard deviations and ranges, for each age group are presented in Table 3. There
were no significant gender effects, nor gender x age group interaction, on these tasks in any
age group (2x2 ANOVA). Post hoc tests (Student Newman-Keuls test, with a level of signif-
icance = .05) identified two different groups for animal fluency and P fluency. In both cases
group 11 years had significantly lower scores than groups 15 and 17 years of age. In the
semantic fluency test-animals group 11 years also differed from group 13 years.
There were no significant differences in any of the VFT scores by SES.
To explore the effect of age, in the difference between semantic and phonemic
performance, composite scores were computed for each type of fluency: the mean seman-
tic fluency score (MSF) and the mean phonemic fluency score (MPF), representing the
average scores obtained by each participant on the two semantic VFTs and the three
phonemic VFTs, respectively. A difference score, MSF-MPF, was also calculated. All
326 I. P. MARTINS ET AL.

Table 3 Verbal Fluency Tasks and Age.

Age groups

11-yr-old 13-yr-old 15-yr-old 17-yr-old F = df p=

Semantic fluency
Animals
Mean (SD) 17.36 (4.71) 21.3 (5.2) 20.82 (4.47) 20.56 (6.00) 4.03 .009
Min-Max (8–29) (10–34) (11–29) (12–37) (3,126)
Food
Mean (SD) 14.30 (4.55) 18.32 (6.45) 18.00 (5.14) 18.44 (5.92) 4.18 .007
Min-Max (7–24) (8 –33) (8–28) (10–31) (3,126)
Phonemic fluency
P
Mean (SD) 7.93 (3.74) 9.83 (4.01) 11.87 (3.49) 11.80 (5.20) 6.06 .001
Min-Max (2–16) (2–18) (5 –17) (2–24) (3,116)
M
Mean (SD) 7.70 (4.12) 10.17 (3.61) 10.70 (3.96) 10.33 (4.64) 3.34 .022
Min-Max (0 –19) (4–16) (3–17) (4–21) (3,116)
R
Mean (SD) 7.30 (3.57) 9.03 (3.26) 10.17 (3.34) 10.07 (3.90) 4.26 .007
Min-Max (1 –17) (1–17) (4–17) (4–22) (3,116)

Results obtained in each verbal fluency task (mean scores, SD, and range of values) in each age group. Scores
in all verbal fluency tasks increased significantly with age.

scores followed a normal distribution. There was an age effect on MSF (F[3,126] = 5.44;
p < .002) (one-way ANOVA) (eta2 = .12); and MPF (F[6,116] = 5.95; p < .001) (one-
way ANOVA) (eta2 = .13); whereby 11 year old children had significantly lower scores
than elder participants. No age differences were found on the MSF-MPF scores (F[6,116]
= .93; p > .10, n.s.) (eta2 = .02).

Further analysis
Correlation between speech measures and verbal fluency scores were calculated.
There was a positive and significant correlation between speech rate and the two semantic
fluency scores (animals and food categories), but not with any of the three phonemic
fluency tests. Since both speech rate and performance in verbal fluency increased with
age, a partial correlation was studied between those two variables controlling the factor
age (Table 4). Again, significant correlations were found between semantic fluency tasks
and speech rates (food fluency [.26, p = .004] animal fluency [.20, p = .026]), but not
with phonemic fluency tasks. Number of pauses and TLT had no significant associations
with any of the verbal fluency tasks. As expected speech rate was negatively correlated
with locution time and pauses.

DISCUSSION
In this study we addressed the question of the predominance of nonfluent types of
speech in children with different types of acquired brain lesions by the analysis of narra-
tive speech and verbal fluency in healthy children and adolescents of different ages. The
results obtained showed that: a) several variables used to classify speech fluency have
 SPEECH RATE AND FLUENCY IN CHILDREN 327

Table 4 Partial Bivariate Correlations Coefficients Between Fluency Test and Speech Measures Controlling for
Age.

Speech Animal Food

TLT rate Pauses fluency fluency M fluency P fluency R fluency

TLT 1000 −.341** .687** .112 .017 .094 .155 .082

p = .000 p = .000 p = .226 p = .852 p = .310 p = .091 p = .372
Speech rate 1000 −.502** .204* .262** .070 .091 .144
p = .000 p = .026 p = .004 p = .449 p = .323 p = .118
Pauses 1000 −.039 −.099 −.015 −.024 −.138
p = .673 p = .284 p = 870 p = .794 p = .132
Animal fluency 1000 .514** .283** .286** .323**
p = .000 p = .002 p = .002 p = .000
Food fluency 1000 .366** .332** .352**
p = .000 p = .000 p = .000
M fluency 1000 .641** .529**
p = .000 p = .000
P fluency 1000 .643**
p = .000
R fluency 1000

There is a positive and significant correlation between speech rate and the two semantic fluency scores
(animals and food categories) but not with any of the three phonemic fluency tests (partial correlation control-
ling the factor age). Number of pauses and TLT had no significant association with any of the verbal fluency
measures. Speech rate was negatively correlated with locution time and pauses.
*< .05, **< .005.

significant developmental effects, b) scores obtained by healthy children on some vari-

ables, namely speech rate, are quite low when compared to adult standards, and c) speech
rate correlates with performance on semantic (but not phonemic) fluency tasks, suggesting
that, in part, speech rate variation is explained by the increasing efficacy of a semantic
strategy of lexical search and retrieval with age. We shall analyze these points in more
detail relating them to data from acquired language disorders and the development of
executive functions in children.
The average speech rate produced by this sample of healthy children during a
picture description task was 91.2 WPM, but ranged from 64 to125 WPM, from 5 to
17 years of age. Portuguese speaking adults, performing the same picture description task,
analyzed by the same methodology (Afonso, 1994) produced an average speech rate of
147.1 WPM (+ 32.1, ranging from 87 to 204 WPM). Therefore, compared to adult stan-
dards, the speech of many healthy young children can be considered nonfluent. The
present results suggest that adolescents approach adult speech rate by 17 years of age but
development may not be completed by then.
The observed variation of speech rate during childhood is in accordance with the
findings obtained in English speaking children by Purcell & Runyan (1980) in two other
verbal tasks (story retelling and conversational speech elicited by open-ended questions).
However, this age effect was not observed in children between 3 and 5 years of age by
Pindzola, Jenkins, & Lokken (1989), which can be due to the rather small sample and age
ranges of participants or to different methodologies used.
Speech rate is affected by multiple environmental variables, namely situational and
demographic factors like task roles, difficulty of topic domain, relationship between
328 I. P. MARTINS ET AL.

speakers, gender, and ageing (Bortfeld, Leon, Bloom, Schober, & Brennan, 2001; Searl,
Gabel, & Fulks, 2002). Part of this variation may reflect speech planning strategies or dif-
ficulties. The use of a picture description task is useful to control domain difficulty and
task roles and has been widely used to analyze speech fluency in adult individuals with
aphasia (Knopman et al., 1983).
For clinical purposes, the cutoff point that separates fluent from nonfluent aphasias,
in adults, ranges from 50 to 90 WPM according to different authors (Benson, 1967;
Kerschensteiner et al., 1972). In children and adolescents participating in this study, the
cutoff score between normal and pathologically slow speech rates (considered at the 5th
percentile) ranged from 22.7 and 75.8 depending age. When comparison was made to
minimum or low scores (2 standard deviations below mean) obtained by adults undertak-
ing the same task, a high proportion of children, especially young children, had values of
speech fluency that might be considered within the very low range.
Other variables used in the classification of fluency also demonstrated developmental
effects. Pauses, repetitions and the total locution time decreased with age. The number of
pauses, a reliable measure to differentiate between fluent and nonfluent types of speech
(Kerschensteiner et al., 1972; Van Dongen, Paquier, Creten, Borsel, & Catsman-Berrevoets,
2001), declined almost by half between the eldest and youngest children. Concerning total
speaking time (TLT), it was higher in the young child that showed a tendency to describe
the picture in detail often repeating and reformulating sentences, while adolescents made a
more synthetic approach focusing rapidly onto the most evident aspects of the scene. This
change in discourse strategy with age was also described by Bates et al. (2001), during a
biographic interview to children (5 to 8 years old) and adults.
Concerning VFTs we found a steady increase in verbal fluency with age. This was
expected from previously published studies (De Agostini et al., 1998; Riva et al., 2000;
Korkman et al., 1997). While semantic fluency tasks reached a plateau by 13 years of age,
phonemic fluency continued to increase by age 15 (for letter categories P and R). These
findings are fairly similar to the ones described in the D-KEFS (Delis et al., 2001), in show-
ing that semantic fluency tends to stabilize earlier than phonemic fluency. Our results also
replicate those reported by Riva et al. (2000) in the Italian population, in several aspects: a)
the developmental effect observed, b) the lack of gender effect, c) a better score in categor-
ical than phonemic fluency, and d) the correlation of semantic (but not phonemic) fluency
with other language measures (a naming task in the study of Riva, speech rate in our study).
Both studies suggest that the two types of verbal fluency tasks (semantic and phonemic)
evaluate two different subsystems responsible for language processing, and only one of
them (semantic fluency) is associated with measures of narrative speech. The difference
between semantic fluency and phonemic fluency performance remained constant across
age groups, corroborating the idea that phonemic fluency is a more difficult task in differ-
ent ages, probably because it requires the exploration of more subsets of categories. It does,
therefore, support a hierarchical organization of the two categories, despite the progressive
automatization of the orthographic lexicon with age and literacy skills.
Verbal fluency tasks require the synchronized action of different cognitive skills:
lexical search and retrieval, working memory, initiation, monitoring and inhibitory
control. They appear to be sensitive both to frontal (the executive component) and tempo-
ral lobe (lexical store and retrieval) development and their disorders. Lesion studies and
fMRI in healthy subjects suggest that semantic fluency is more sensible to temporal lobe
pathology while phonemic fluency is more vulnerable to frontal lobe dysfunction (Gourovitch
et al., 2000; Pihlajamaki et al., 2000). On a meta-analytic review of verbal fluency
 SPEECH RATE AND FLUENCY IN CHILDREN 329

performance in patients with focal cortical lesions, Henry & Crawford (2004) found that
different cortical regions participate on phonemic and semantic fluency tasks. Both tests
are sensitive to frontal lobe lesions, and therefore can be regarded as executive functions
tests. Phonemic fluency is more sensitive to frontal than nonfrontal and to left, as opposed
to right, cortical lesions. The extent to which semantic fluency should be regarded as a
measure of executive functions remains less clear. Semantic fluency presumably impli-
cates the same executive process, thought to mediate phonemic fluency performance, such
as initiation, efficient organization of verbal retrieval and recall and self-monitoring.
However, in addition to being sensitive to frontal damage, semantic fluency is also very
sensitive to temporal lobe pathology.
Further to syntactic abilities, narrative speech shares many executive functions with
VFTs, yet the relation between spontaneous speech and VFTs is not clear and functional
imaging studies have provided controversial results about their common anatomical basis
(Gourovitch et al., 2000; Kircher, Brammer, Williams, & McGuire, 2000). Lexical retrieval
during spontaneous speech is contextual, semantically based, and this may explain the
association between semantic tasks and narrative speech measures found in our study.
Letter fluency forces the participant to make a phonological/orthographic search strategy. It
may depend upon literacy skills and the acquisition of the orthographic lexicon, which
might explain its lack of correlation with speech measures. In addition performance in these
two tasks may reflect different maturations of their neural basis, as phonemic tasks put
more demands upon the frontal lobes (late to mature) than semantic tasks.
While speech rate was significantly correlated with the two tasks of semantic
fluency, other measures of speech (TLT, pauses, and repetitions) were not. They may
reflect other aspects of speech programming beyond the lexical domain.
Thus, the developmental effects observed both in VFT and measures of narrative
speech may reflect age-related changes in several domains of explicit and implicit knowl-
edge, such as speed of information processing, planning, monitoring, inhibitory control
and goal-oriented behavior, late developing components of the executive functions
(Anderson, 2002). This is one possible explanation for their increase from late childhood
to adolescence. Although language development is subject to several environmental
factors, namely living area, neighborhood poverty and intrafamilial factors (type of stimu-
lation received, mono or bilingualism and the educational level of the caregiver) (McCulloch
& Joshi, 2001) social factors (indirectly assessed in the present study by SES) had no sig-
nificant weight on most studied measures compared to factor age.
We acknowledge limitations to this study. One concerns the selection procedure
used. The identification of children with developmental language disabilities by the
teacher, and/or information from parents, may be subject to error. Although school
achievement and or specific learning disabilities (apart from those involving oral
language) were not sought, the method used excluded children repeating grades, i.e.
those with lowest achievement, and those born during the last semester of the year.
Therefore school grade could not be included as an independent variable, and further
studies are necessary to clarify its specific effect, namely in the phonemic fluency
tasks.
In this study analysis of verbal fluency was limited to children above 9 years of age,
when reading and writing and the ortograhic-phonemic correspondence is becoming
automatized. This limits the generalization of the results to other age group. Future studies
are necessary to corroborate this effect in younger children, by the time phonological,
grammatical, and pragmatic awareness is developing by school education.
330 I. P. MARTINS ET AL.

This study involved children between 5 and 17 years of age. Although it would be
interesting to extend it to younger age groups, we selected this minimum age limit for two
main reasons. Firstly, because we wanted to guarantee that participants had enough vocab-
ulary and syntactic abilities to be able to describe the Cookie Theft picture with some
detail, and to process the central aspects of the picture and, secondly, because executive
functions such as the ones tested by the VFT (speed of information processing, monitor-
ing, and impulse control) and narrative speech (planning and goal setting), develop late in
childhood (Anderson, Anderson, Northam, & Taylor, 2000) and might be just emerging
before that age. The aim of this study was to understand how some measures of speech
fluency change with age, in order to comprehend symptoms of acquired speech and
language disorders in childhood. It adds another explanation for the paucity of fluent
aphasias in children. The lack of norms for speech fluency in children may have led clini-
cians to under-diagnose fluent aphasias, by the use of criteria that are too restrictive for
this age group. Although some reported cases of childhood aphasia describe speech met-
rics in detail (Klein et al., 1992; Van Dongen, Loonen, & Van Dongen, 1985; Van Dongen
et al., 1994, 2001), in others the fluency dimension is not fully quantified.
Present data provides some normative quantitative guidelines for those assessing chil-
dren with aphasia and speech difficulties due to brain lesions. The low tech methodology
used (just requiring a tape recorder and a stopwatch) is adequate for clinical purposes,
namely for the bedside examination of children. Although speech timings can be determined
by more sophisticated techniques, including phonation rates, they are not always available to
clinicians assessing children in the acute stages of illness, and stopwatch measures have been
shown to be reliable when compared with acoustic techniques to determine speech rates
(Pindzola et al., 1989; Van Dongen et al., 1994). Further studies are necessary to obtain
norms for other measures of speech fluency (phrase length, duration of pauses, effort,
melodic line, agrammatism, etc) in different languages, especially in young children where
variability may be wider. It is also necessary to understand contextual influences (Hansson,
Nettelblad, & Nilhom, 2000) on those measures. Children might become less fluent when
talking with unknown adults, when facing verbal difficulties, or in threatening situations, as
during a hospital admission, introducing other bias in the evaluation of aphasic speech.
Original manuscript received October 14, 2005
Revised manuscript accepted May 26, 2006
First published online October 20, 2006

REFERENCES
Afonso, A. M. S. (1994). Avaliação do discurso espontâneo na afasia. Tese de Licenciatura.
Alcoitão: Escola Superior de Saúde de Alcoitão.
Anderson, P. (2002). Assessment and development of executive function (EF) during childhood.
Child Neuropsychology, 8, 71–82.
Anderson, P., Anderson, V., Northam, E., & Taylor, H. (2000). Standardization of the Contingency
Naming Test for school-aged children: A new measure of reactive flexibility. Clinical Neurop-
sychological Assessment, 1, 247–273.
Anderson, V., Anderson, P., Northam, E., Jacobs, R., & Catroppa, C. (2001). Development of exec-
utive functions through late childhood and adolescence in an Australian sample. Developmental
Neuropsychology, 20, 385–406.
Baker, R., Lee, L., Evans, M., Hall, C., Kelley, C., King, A., & Rose, K. (2001). Word fluency mea-
sures: Normative values for younger and older adults. Journal of Medical Speech - Language
Pathology, 9, 63–70.
 SPEECH RATE AND FLUENCY IN CHILDREN 331

Bates, E., Reilly, J., Wulfeck, B., Dronkers, N., Opie, M., Fenson, J., Kriz, S., Jeffries, R., Miller, L., &
Herbs, K. (2001). Differential effects of unilateral lesions on language production in children
and adults. Brain and Language, 79, 223–265.
Benson, D. F. (1967). Fluency in aphasia: Correlation with radioactive scan localization. Cortex, 3,
373–394.
Borkowsky, J. G., Benton, A. L., & Spreen, O. (1967). Word fluency and brain damage. Neuropsy-
chologia, 5, 135–140.
Bortfeld, H., Leon, S. D., Bloom, J. E., Schober, M. F., & Brennan, S. E. (2001). Disfluency rates in con-
versation: effects of age, relationship, topic, role, and gender. Language and Speech, 44, 123–147.
Cranberg, L. D., Filley, C. M., Hart, E. J., & Alexander, M. P. (1987). Acquired aphasia in child-
hood: Clinical and CT investigations. Neurology, 37, 1165–1172.
Dayer, A., Roulet, E., Maeder, P., & Deonna, T. (1998). Post-traumatic mutism in children: Clinical
characteristics, pattern of recovery and clinicopathological correlations. European Journal of
Paediatric Neurology, 2, 109–116.
De Agostini, M., Metz-Lutz, M., Van Hout, A., Chavance, M., Deloche, G., Pavão-Martins, I., &
Dellatolas, G. (1998). Batterie d´ evaluation du langage ora de l`enfant aphasic (ELOLA): stan-
dartisation française (4–12 ans). Revue de Neuropsychologie, 8, 319–367.
Delis, D., Kaplan, E., & Kramer, J. (2001). Delis-K Executive Function System (D-KEFS): Exam-
iner’s manual. San Antonio, TX: The Psychological Corporation.
Deloche, G., Jean-Louis, J., & Seron, X. (1979). Study of the temporal variables in the spontaneous speech
of five aphasic patients in two situations, interview and description. Brain and Language, 8, 241–250.
Feyereisen, P., Pillon, A., & de Partz, M. P. (1991). On the measures of fluency in the assessment of
Spontaneous speech production by aphasic subjects. Aphasiology, 5, 1–21.
Frith, C. D., Friston, K. J., Liddle, P. F., & Frackowiak, R. S. (1991). A PET study of word finding.
Neuropsychologia, 29, 1137–1148.
Gaillard, W., Hertz-Pannier, L., Mott, S., Barnett, A., LeBihan, D., & Theodore, W. (2000). Func-
tional anatomy of cognitive development – fMRI of verbal fluency in children and adults. Neu-
rology, 54, 180–185.
Goldman-Eisler, F. (1968). Psycholinguistics: Experiments in spontaneous speech. London: Aca-
demic Press.
Goodglass, H., & Kaplan, E. (1983). The assessment of aphasia and related disorders. Philadelphia:
Lea and Fabiger.
Gordon, N. (2001). Mutism: Elective or selective, and acquired. Brain Development, 23, 83–87.
Gourovitch, M., Kirkby, B., Goldberg, T., Weinberger, D., Gold, J., Esposito, G., Horn, J., &
Berman, K. (2000). A Comparison of rCBF patterns during letter and semantic fluency. Neu-
ropsychology, 14, 353–360.
Greenwald, M. L., Nadeau, S. E., & Rothi, L. J. G. (2000). Fluency. In S. E. Nadeau, L. J. G. Rothi, &
B. Crosson (Eds), Aphasia and language, theory to practice (pp. 31–39). London: The Guilford Press.
Halperin, J., Healey, J., Zeitchik, E., Ludman, W., & Weinstein, L. (1989). Developmental Aspects
of Linguistic and Mnestic Abilities in Normal Children. Journal of Clinical and Experimental
Psychology, 11, 518–528.
Hansson, K., Nettelblad, U., & Nilhom, C. (2000). Contextual influence on the Language production
of children with speech/language impairment. International Journal of Communication Disor-
ders, 35, 31–47.
Henry, J. D, & Crawford, J. R. (2004). A meta-analytic review of verbal fluency performance
following focal cortical lesions. Neuropsychology, 18, 284–295.
Kerschensteiner, M., Poeck, K., & Brunner, E. (1972). The fluency-non fluency dimension in the
classification of aphasic speech. Cortex, 8, 233–247.
Kircher, T. T., Brammer, M. J., Williams, S. C., & McGuire, P. K. (2000). Lexical retrieval during
fluent speech production: An fMRI study. Neuroreport, 11, 4093–4096.
Klein, S., Masur, D., Farber, K., Shinnar, S., & Rapin, I. (1992). Fluent aphasia in children. Defini-
tion and natural history. Journal of Child Neurology, 7, 50–59.
332 I. P. MARTINS ET AL.

Knopman, D. S., Selnes, A. O., Niccum, N., Rubens, A. B., Yock, D., & Larson, D. (1983). A
longitudinal study of speech fluency in aphasia: CT correlates of recovery and persistent non-
fluency. Neurology, 33, 1170–1178.
Korkman, M., Kirk, U., & Kamp, S. (1997). NEPSY. San Antonio, TX: Psychological Corporation.
Kreindler, A., Mihailescu, L., & Fradis, A. (1980). Speech fluency in aphasics. Brain and Language,
9, 199–205.
Loonen, M. C. B., & Van Dongen, H. R. (1990). Acquired childhood Aphasia. Outcome 1 year after
onset. Archives of Neurology, 47, 1324–1328.
Martins, I. P. (1997). Lesões cerebrais adquiridas na infância: dissolução, recuperação e reorga-
nização das capacidades cognitivas na criança. Doctoral thesis. Lisbon; Faculty of Medicine.
Martins, I. P., & Ferro, J. M. (1993). Acquired childhood aphasia: A clinicoradiological study of 11
stroke patients. Aphasiology, 7, 489–495.
McCulloch, A., & Joshi, H. E. (2001). Neighbourhood and family influences on the cognitive ability of
children in the British National Child Development Study. Social Science & Medicine, 53, 579–591.
McLaughlin, S. F., & Cullinan, W. L. (1989). Disfluencies, utterance length and linguistic complex-
ity in nonstuttering children. Journal of Fluency Disorders, 14, 17–36.
Paquier, P., & Van Dongen, H. R. (2000). Aphasic syndromes and localization of lesions in children.
In D. Riva, & A. Benton (Eds.), Localization of Brain Lesions and Development Functions
(pp. 67–77). Eastleigh: John Libbey Company Ltd.
Pihlajamaki, M., Tanila, H., Hanninen, T., Kononen, M., Laakso, M., Partanen, K., Soininen, H., &
Aronen, H. J. (2000). Verbal fluency activities the left medial temporal lobe: A functional mag-
netic resonance imaging study. Annals of Neurology, 47, 470–476.
Pindzola, R. H., Jenkins, M. M. & Lokken, K. J. (1989). Speaking rates of young children.
Language, Speech, and Hearing Services in Schools, 20,133–138.
Purcell, R. M., & Runyan, C. M. (1980). Normative Study of Speech Rates of Children. Journal of
the Speech and Hearing Association of Virginia, 21, 6 –14.
Riva, D., & Giorgi, C. (2000). The cerebellum contributes to higher functions during development: evi-
dence from a series of children surgically treated for posterior fossa tumours. Brain, 123, 1051–1061.
Riva, D., Nichelli, F., & Devoti, M. (2000). Developmental aspects of verbal fluency and confronta-
tion naming in children. Brain and Language, 71, 267–284.
Satz, P. (1991). Symptom pattern and recovery outcome in childhood aphasia: A methodological
and theoretical critique. In I. P. Martins, A. Castro-Caldas, H. R. Van Dongen, & A. Van Hout
(Eds.), Acquired aphasia in children. (pp. 95–114). Dordrecht: Kluwer Academic Publishers.
Sauzéon, H., Lestage., P., Raboutet, C., N’Kaoua, B., & Claverie, B. (2004). Verbal Fluency output in
children aged 7–16 as a function of the production criterion: Qualitative analysis of clustering,
switching processes, and semantic network exploitation. Brain and Language, 89, 192–202.
Searl, J. P., Gabel, R. M., & Fulks, J. B. (2002). Speech disfluency in centenarians. Journal of Com-
munication Disorders, 35, 383–392.
Shipley, K. G., & McAfee, J. G. (1992). Assessment Procedures common to most communicative
disorders. In K. G. Shipley & J. G. McAfee (Eds.), Assessment in speech-language pathology:
A resource manual (pp. 108–109). San Diego, CA: Singular Publishing Group.
Thatcher, R. (1991). Maturation of the human frontal lobes: Physiological Evidence for staging.
Developmental Neuropsychology, 7, 397–419.
Van Dongen, H. R., Loonen, C. B., & Van Dongen, K. J. (1985). Anatomical basis for acquired
fluent aphasia in children. Annals of Neurology, 17, 306–309.
Van Dongen, H., Paquier, P., Creten, W., Borsel, J., & Catsman-Berrevoets, C. (2001). Clinical
evaluation of conversational speech fluency in the acute phase of acquired childhood aphasia:
Does a fluency/nonfluency dichotomy exist?. Journal of Child Neurology, 16, 345–351.
Van Dongen, H., Paquier, P., Raes, J., & Creten, W. (1994). An analysis of spontaneous conversa-
tional speech fluency in children with acquired aphasia. Cortex, 30, 619–633.
Woods, B. T., & Teuber, H. L. (1978). Changing patterns of childhood aphasia. Annals of Neurol-
ogy, 3, 272–280.