You are on page 1of 11

International Journal of

Behavioral Development
Associations between private speech, 1–11
ª The Author(s) 2014
Reprints and permissions:
behavioral self-regulation, and cognitive sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0165025414556094
abilities ijbd.sagepub.com

Tuija Aro,1 Anna-Maija Poikkeus,2 Marja-Leena Laakso,2


Asko Tolvanen,2 and Timo Ahonen1

Abstract
We examined the associations between 5-year-old children’s private speech, behavioural self-regulation, and cognitive abilities. Beha-
vioural self-regulation was assessed using parental and preschool teacher questionnaires. Cognitive abilities (i.e., language, inhibition, plan-
ning and fluency, and memory) were assessed with neurocognitive tests, and the effectiveness of private speech (i.e., whether the child
performs better when using speech than when not using speech) with the Hammer Task. About 43% of the children used private speech
spontaneously, and about 76% performed better on the Hammer Task when they used speech. Associations between behavioural self-
regulation and speech effectiveness were few, but all cognitive scores were significantly associated with the speech effectiveness score,
i.e., the poorer the child’s neurocognitive test performance was, the more he/she benefited from using speech. The findings lend support
to the relevance of children’s cognitive abilities when seeking understanding on the development, use, and effectiveness of private speech.

Keywords
behavioural self-regulation, children, cognitive abilities, private speech

One of the central developmental tasks for a preschool-age child is speech, as well as their capability to use speech to improve their
the development from other-control to self-control of behaviour performance, appears to vary. For instance, children with deficits
through the process of speech internalization (Vygotsky, 1962; see in language development tend to be delayed in their internalization
also Olson & Lunkenheimer, 2009). During the preschool years, the of speech (Lidstone, Meins, & Fernyhough, 2012). Empirical evi-
child shifts from using language merely for communicative pur- dence suggests also that children with deficits in behavioural con-
poses to using it also as a tool to regulate his/her behaviour trol have delays in speech internalization (Corkum, Humpries,
(Vygotsky, 1978). This process entails a change in locus of control Mullane, & Theriault, 2008; Winsler et al., 2000), and children with
from behavioural regulation by the caregiver to regulation by pri- ADHD often use more overt as well as covert private speech than
vate speech, and finally to regulation by inner speech (Kohlberg, typically-developing children (Berk & Landau, 1993; Diaz, Wins-
Yaeger, & Hjertholm, 1968; Vygotsky, 1962; see also Berk, ler, Atencio, & Harbers, 1992; Winsler, 2009; see also Corkum
1992; Winsler, Diaz, Atencio, McCarthy, & Chabay, 2000). Thus, et al., 2008; Winsler et al., 2007).
it is well understood that private speech forms an intermediate link In addition to use of private speech, another area of interest is its
between the control provided by the caregiver’s speech (i.e., other- effectiveness for task performance. Although emergence of private
control) and that provided by inner speech (i.e., self-control). How- speech has been shown to improve performance among young chil-
ever, our understanding on the factors affecting development and dren, to our knowledge, only a few studies have directly addressed
use as well as effectiveness of private speech among young children effectiveness of private speech, especially among children with
is still vague. developmental difficulties. Recently, private speech has been
Vygotsky (1962) postulated that the function of private speech is shown to improve problem-solving performance (the Tower of
to guide one’s thought processes, allowing for selective attention, London) among children with deficits in language development
voluntary memory functions, planning, and problem-solving to (Lidstone et al., 2012) as well as to improve performance in a
develop. Via its regulatory function, self-directed speech (private speech–action coordination task (the Hammer task) among beha-
or inner) is supposed to improve performance; therefore, among viourally at-risk children and control children (Winsler et al.,
young children it is often used when facing challenging tasks. In the 2007). On the other hand, Meichenbaum and Goodman (1969)
early years, children’s task performance typically improves with the
use of speech (e.g. Balamore & Wozniak, 1984; Berk, 1992; Bivens &
Berk, 1990; Fernyhough & Fradley, 2005; Winsler, Manfra, & Diaz, 1
University of Jyväskylä & Niilo Mäki Institute, Jyväskylä, Finland
2007; see also Al-Namlah, Fernyhough, & Meins, 2006; Kraft & Berk, 2
University of Jyväskylä, Jyväskylä, Finland
1998; Müller, Zelazo, Hood, Leone, & Rohrer, 2004), and the amount
of self-directed speech peaks when the task is neither too easy nor too Corresponding author:
difficult (e.g. Diaz, 1992; Fernyhough & Fradley, 2005). Tuija Aro, University of Jyväskylä & Niilo Mäki Insitute, PO Box 35, 40014
During the preschool years, task success with silence typically University of Jyväkylä, Finland.
increases, but the rate at which children shift from private to inner Email: tuija.i.aro@jyu.fi

Downloaded from jbd.sagepub.com at University of Sydney on August 14, 2015


2 International Journal of Behavioral Development

found that self-verbalizations were less effective in directing beha- Goodman (1969), indirectly suggest that poor inhibition or defi-
viour among impulsive kindergarten-age children than among chil- cient executive skills may influence effectiveness of private speech.
dren with more advanced reflective skills. Similarly, Berk and Potts The development and use of private speech may also be interwo-
(1991) reported that private speech was less effective in channelling ven with the development of short-term memory or the ability to use
motor behaviour among 6–12 year-old children with ADHD than it (see also Barkley, 2012) as thinking out loud may activate neural
among children without attention problems. Berk and Landau networks that are involved in memory functions. Hitch, Halliday,
(1993) discuss the possibility that for a subgroup of children with Schaafstal, and Heffernan (1991) noted that speaking or listening
learning deficits inner speech (with external manifestations) does to speech activates an internal articulatory loop, and this activation
not have the same functional relationship with behaviour as it does is important when the child’s ability to use inner speech in short-
for their peers. It can be speculated that the surplus reliance on term memory has not been fully developed. Private speech may
speech beyond what would be typical for the age normalizes the contribute to activation of the articulatory loop, which in turn helps
performance of children with deficits in behavioural control (see in refreshing the memory trace (see Baddeley, 2001; Repovs &
Corkum et al., 2008; Winsler, 2009; Winsler et al., 2007). Baddeley, 2006; see also Al-Namlah et al., 2006).
Development of private speech has been shown to be linked In sum, research linking private speech and cognitive function
with cognitive maturation (e.g. Berk, 1986; Kohlberg et al., 1968; has mainly focused on the association between the children’s cog-
Vygotsky, 1962; see also Müller, Jacques, Brocki, & Zelazo, nitive maturity and use or amount of different types of private
2009), and also the effectiveness of private speech may be affected speech (from overt to inaudible), and there is a paucity of systema-
by cognitive processing abilities and their deficits. It has been sug- tic research literature on the associations between specific cognitive
gested that private speech emerges and peaks earlier among chil- functions and effectiveness of self-directed speech (private or inner
dren with higher IQ (e.g. Deutsch & Stein, 1972; Kohlberg et al., speech). Furthermore, the earlier studies have mainly used correla-
1968), but the association between IQ and amount of private speech tional approach and have not compared child performances in situ-
is negative among older children (Kohlberg et al., 1968). ation with use of private speech to situation without speech
Neuropsychologically-oriented studies have explored the neurolo- (however, see Lidstone et al., 2012, and Winsler et al., 2007).
gical processes involved in overt and covert speech (e.g. Jones, Therefore they provide only indirect evidence on the effectiveness
2009; Jones & Fernyhough, 2007; Palmer et al., 2001), but empiri- of private speech for child’s performance.
cal evidence on the associations between specific cognitive func- In the present study, we measured the child’s use of private
tions and manifestations or effectiveness of self-directed speech speech (i.e., amount) and its effectiveness (i.e., whether a child per-
is scant. Although language, attentional skills (i.e., executive func- forms better when using private speech) with the Hammer Task, a
tions and inhibition), and memory have theoretical associations speech-action coordination task of tapping sequences created by
with emergence of private speech, little is known about their bear- Winsler and colleagues (e.g. Winsler et al., 2007; Winsler et al.,
ings on the effectiveness of private speech. In the present study, we 2000) based on the work by Vygotsky (1962; and subsequently
were interested in both the association between effectiveness of Balamore & Wozniak, 1984). The task involves an experimental
children’s private speech and their behavioural self-regulation design with conditions which aim to elicit both spontaneous and
skills (i.e., hyperactivity, inattention, and lack of effort) and cogni- prompted private speech and conditions where the child is asked
tive skills (i.e., language, inhibition, planning and fluency, and to perform the task without using private speech. These different
short-term memory). conditions, although with relatively small number of items, allow
Berk and Landau (1993) indicated that language skills measured the comparison of the child’s performance in the task with and
with vocabulary had a small but significant negative correlation without private speech. In this study, we calculated a speech effec-
with incidence of task-relevant externalized private speech among tiveness score indicating whether child performs better in task items
elementary-school children, but they did not study effectiveness of when using private speech compared to task items when he/she
private speech. More recently, Lidstone and colleagues (2012) sug- does not use private speech. The Hammer Task has been previously
gested that at younger-age deficits in language development would used in a similar way by Winsler and colleagues (2007) who exam-
be associated with lower rate of private speech due to its later emer- ined the effects of children’s private speech use on task perfor-
gence, whereas in middle childhood, language difficulties would be mance comparing a group of behaviourally at-risk children and a
associated with a lesser degree of internalization. These findings, in group of control children.
line with those of Kohlberg and colleagues (1968), suggest that Behavioural self-regulation skills were assessed in the present
cognitive maturation is associated with the timing of emergence study using parental and preschool teacher ratings of hyperactivity,
and internalization of private speech so that kindergarteners with inattention, and lack of effort with the Five-to-Fifteen questionnaire
better skills tend to peak earlier in rate of private speech. Evidence (Kadesjö et al., 2004), and cognitive performance was measured
on the effectiveness of private speech and language skills is scant, with individually administered neuropsychological tests.
but the recent study by Lidstone and colleagues (2012) suggests that The specific questions addressed and hypotheses set in our study
children with specific language impairment use private speech were as follows:
effectively in a problem-solving task (the Tower of London). What percentage of 5-year-old children use private speech spon-
The central role of inhibition and executive skills in private taneously during the Hammer Task (i.e., the condition at the begin-
speech is supported by the findings that children who are behaviou- ning phase of the task without any manipulation of child’s use of
rally at-risk are often delayed in their speech internalization process speech with tester instructions)? Is spontaneous use of speech asso-
(e.g. Winsler et al., 2000). Children with behaviour-regulation ciated with item difficulty? What percentage of children perform
problems are also shown to have deficits in underlying cognitive better when using speech, i.e., they show speech effectiveness
functions such as attention and working memory (Rapport et al., (speech effectiveness score higher than 1; the probability of cor-
2009; Willcutt et al., 2001). Studies conducted among children with rectly responding to items with private speech divided by the prob-
ADHD, e.g. by Berk and Potts (1991) as well as Meichenbaum and ability of responding correctly without private speech)? According

Downloaded from jbd.sagepub.com at University of Sydney on August 14, 2015


Aro et al. 3

to previous studies (e.g. Al-Namlah et al., 2006; Winsler et al., Questionnaires. At the fourth phase of the study, at 4 years 7
2000) we expected that some of the 5-year-old children in our sam- months of age, 472 out of the 508 families who had participated
ple may have stopped using overt private speech, but many of them in the first data collection phase (between 6 and 24 months of age)
would still use it spontaneously, at least some of the time. Accord- could be located for contact. Out of these 472 parents, 296 (62.7%)
ing to previous literature, we expected that the use of spontaneous returned the parental questionnaire. The preschool teacher’s ques-
speech would peak on items of medium difficulty (e.g. Fernyhough tionnaire was obtained concerning only a subgroup, 182 (38.6%)
& Fradley, 2005). Based on earlier findings (e.g. Müller et al., children, because many of children did not attend preschool at this
2004; Winsler et al., 2007) indicating that young children’s task young age. The questionnaire data used in the present analyses
performance improves with the use of self-directed speech, we involved scales from the parental and teacher questionnaires asses-
expected children to perform better in the Hammer Task when they sing difficulties in self-regulation at the behavioural level, such as
used speech than when they did not use speech. hyperactive and inattentive behaviour, and lack of effort from the
Are the spontaneous use and effectiveness of private speech Five-to-Fifteen questionnaire (FTF; Kadesjö et al., 2004; see also
associated with behavioural self-regulation assessed using parental Korkman, Jaakkola, Ahlroth, Pesonen, & Turunen, 2004).
and preschool teachers’ ratings of hyperactivity, inattention, and
lack of effort? We expected to find a negative association between
Individual assessments. At the fifth phase of the study, at the age of
behavioural self-regulation skills and the amount of spontaneous
5 years 3 months, a subsample of children (n ¼ 102) participated in
private speech (i.e., children with low regulative skills would speak
the individual assessments of cognitive skills. All the children who
more often; see Corkum et al., 2008; Winsler, 2009; Winsler et al.,
participated in the fifth phase of data collection were included in the
2007), but since the literature concerning effectiveness of the use of
present study, and all of them had also participated in the fourth
private speech and the children’s behavioural self-regulation skills
phase of the longitudinal study. Of these 102 children, 70 had par-
is limited (however, see Berk & Potts, 1991, and Meichenbaum &
ticipated in all earlier data collection phases (early communication
Goodman, 1969), we did not set any specific hypothesis.
screen between 6 and 24 months of age, individual assessments at 2
Are the spontaneous use and effectiveness of private speech
and 3 years of age, and parental questionnaire at 4 years 7 months of
associated with children’s cognitive abilities (i.e., language, inhibi-
age) and 32 had participated in the first phase (early communication
tion, planning and fluency, and short-term memory)? Based on the
screen 6 to 24 months) and the fourth phase (parental questionnaire
earlier suggestions that cognitive skills are positively associated
4 years 7 months). In the questionnaire at the 4 years 7 months, the
with private speech at kindergarten-age (Kohlberg et al., 1968),
parents of these 32 children had reported a concern for language
we expected positive associations between cognitive skills and
difficulties or hyperactivity, or for problems in executive func-
amount of spontaneous private speech (i.e., children with higher
tions (i.e., score at or below 1 SDs in the parental questionnaire).
cognitive skills would use more private speech spontaneously). Evi-
The individual assessments at 5 years 3 months of age consisted of
dence on associations between effectiveness of private speech and
two to three sessions and involved measures of memory, language,
different cognitive abilities is much more limited than that concern-
perception, attention, executive functions, and parent–child
ing the associations between amount of private speech and cogni-
interaction.
tion, therefore we did not set any specific hypotheses.

Sample description. All 102 families completed a written consent


form prior to participating in the individual assessments. The chil-
Materials and method dren were all Caucasian and spoke Finnish as their native language.
The children were assessed with a short version (30 items) of the
Participants and procedure Peabody Picture Vocabulary Test (PPVT; Dunn & Dunn, 1981).
The present study involved data from 102 children. This was a sub- The sample mean (MPPVT ¼ 16.67, SDPPVT ¼ 3.63; min ¼ 5 and
sample of a larger longitudinal community-based sample of 508 max ¼ 24) was close to the mean of the comprehensive
children and their families who had been enrolled in a longitudinal population-based sample of 1,837 Finnish children who had taken
study on early language and communication development. The first the same short version of the test at 6 years of age, i.e., 1 year older
data collection phase consisted of early communication screen than the present sample (MPPVT ¼ 19.83; SDPPVT ¼ 3.36).
(ITC; part of the Communication and Symbolic Behavior Scales Since only a subsample of the original community-based sample
Developmental Profile CSBS-DP, Wetherby & Prizant, 2002) filled participated in the present study, comparisons on maternal educa-
out by the parents when the children were between 6 and 24 months tion were made between the children who were involved in the
of age (see Määttä, Laakso, Tolvanen, Ahonen, & Aro, 2012). The present study (n ¼ 102) and those who had participated in the
families were originally recruited through child health clinics in follow-up at previous ages, but not in the present study, and who
Central Finland. Child health clinics are regularly attended by over had reported maternal education (n ¼ 378). These comparisons
95% of Finnish parents and their children from the child’s birth to revealed no statistically significant differences in the mothers’ edu-
school entry. The longitudinal study involved five assessment cation (n ¼ 378; Meducation ¼ 3.78, SDeducation ¼ 1.94 and n ¼ 102;
points, but only a subgroup of the children participated in all of Meducation ¼ 4.22, SDeducation ¼ 2.04). Education was assessed using
them: 1) early communication screen, 2) individual assessments a 7-point scale ranging from a basic level of education to advanced
at 2 years of age, 3) individual assessments at 3 years of age, 4) par- educational training: 1 ¼ ‘‘comprehensive school education with-
ental questionnaires at 4 years 7 months, and 5) individual assess- out any vocational education’’ and 7 ¼ ‘‘comprehensive school
ment at 5 years 3 months. The present analyses involved or upper secondary general school diploma combined with a higher
questionnaire data collected when the children were 4 years 7 university degree, master’s or doctorate-level degree’’. There were
months of age and data from cognitive skills assessments at 5 years also no significant differences between the children who dropped
3 months of age. off before and those participating in the present study on any of the

Downloaded from jbd.sagepub.com at University of Sydney on August 14, 2015


4 International Journal of Behavioral Development

Infant Toddler Checklist scales (ITC, Whetherby & Prizant, 2002) Wozniak (1984), and its later version by Winsler and colleagues
at 2 years of age. (Winsler et al., 2007), was administered in the individual assess-
Next, we conducted analyses concerning the prevalence of lan- ments when the children were 5 years 3 months old. In this task,
guage and/or behaviour regulation problems within the sample of each child was asked tap a sequence of coloured pegs placed on a
children for whom parental ratings were available at the age of 4 rectangular pegboard (one peg located in each corner, and the fifth
years 7 months (n ¼ 296). Parent-rated risks concerning children’s peg in the centre) with a plastic hammer. The task consisted of two
language development or ADHD-type behaviour were found to be parts: The first involved the tapping of sequences of coloured pegs,
somewhat overrepresented in the present sample (n ¼ 102) com- and the second numerical tapping. Both parts included two condi-
pared to those who were not involved in the present study, but who tions: The first condition involved conducting each task without
had the questionnaire data from the age of 4. 7 years (n ¼ 194). The speech instruction and the second with speech instruction (i.e.,
former children had higher risk scores on the receptive language, asking the child to either use or not use private speech, depending
expressive language, and hyperactivity scales of the FTF parental on whether or not the child spontaneously did so under the first
questionnaire in comparison to children who did not participate condition). Before commencing with these two parts officially,
in the individual assessment (F(1, 294) ¼ 6.88; p  .05; F(1, each child was asked to perform two practice task items to ensure
294) ¼ 22.50; p  .001, and F(1, 294) ¼ 6.14; p  .05, respec- that he/she had understood the task and could name the colours of
tively). About 9% of the children had been given parental ratings the pegs.
indicating problems (i.e., below 2.0 SD, Finnish standardization In the first part of the Hammer Task, the examiner asked the
norms) in hyperactivity, and around 8% in receptive and/or expres- children to tap progressively longer sequences of coloured pegs,
sive language skills. According to parental reports, however, only with two task items for each sequence length. The children moved
four of the children had received a diagnosis of specific language on to the next higher sequence if they correctly hit the pegs in the
impairment, and none were diagnosed as having ADHD at 5 years order stated by the examiner for at least one of the two task items
of age. (It is worth noting that it is rather rare in Finland to diagnose (e.g. sequence of two colours: red – blue; sequence of three colours:
preschool-aged children with ADHD.) blue – yellow – green). In the first condition (without speech
instruction; maximum of eight items), the children were simply
asked to tap the pegs in the requested order. In the second condition
Measures (with speech instruction; maximum of eight items), the children
were given a similar set of task items to execute, but this time they
Hyperactivity, inattention, and lack of effort at 4 years 7 months were told to do the opposite of what they had done spontaneously
of age. The Five-to-Fifteen questionnaire (FTF) (Kadesjö et al., with respect to speech use in the first condition. That is, if they had
2004; Korkman et al., 2004) is a screening questionnaire developed used private speech spontaneously even once during the first task,
in the Nordic countries for assessing symptoms of ADHD and its then they were asked to do the following tasks silently; if, on the
comorbidities among children from 5 to 15 years of age. The FTF other hand, they had not used private speech spontaneously during
can be used by parents as well as personnel in preschool settings. It the first tasks, then they were asked to use private speech for the
contains 181 questions covering eight developmental domains following tasks. The maximum number of task items was 16 in the
(motor skills, attention and impulsivity, perception, memory, lan- first part, but not all of the children completed all tasks.
guage, learning, social skills, emotional and behavioural problems). In the second part of the Hammer Task (numerical tapping), the
Three out of the four scales in the Attention and Impulsivity domain examiner asked the children to tap the centre peg a specific number
were used in the present study. The Hyperactivity scale consists of of times. They moved on to the next higher sequence if they cor-
nine items (e.g. Difficulties in waiting for one’s turn), the Inatten- rectly hit the right number of pegs for at least one of the two task
tion scale consists of nine items (e.g. Difficulties in sustaining items. In the first condition (without speech instruction; maximum
attention in tasks and activities), and the Lack of effort scale (orig- of four items), the examiner asked the children simply to hit the peg
inally the Hypoactivity scale) consists of four items (e.g. Seems list- with the hammer a specific number of times without instructing
less, sluggish, and lacking energy). These three scales tap into the them to use private speech. In the second condition (with speech
behavioural level of self-regulation whereas the fourth subscale instruction; maximum of four items), the children were told to do
assesses executive functions (planning and monitoring one’s the opposite of what they had done spontaneously in regard to
actions). The forth subscale was omitted due to time constraints. speech use in the first condition. The maximum number of task
The respondents (parents and preschool teachers) gave ratings on items was eight in the second part, but not all children completed
a 3-point scale: 0 (does not apply), 1 (applies sometimes or to some all tasks. The Cronbach alpha for all of the Hammer Task task items
extent), or 2 (definitely applies). Korkman and colleagues (2004) (8 þ 8 in part 1 and 4 þ 4 in part 2) in the present data was .71.
reported a reliability coefficient of .89 for the Attention and Impul- It must be noted that although in both parts of the Hammer Task,
sivity domain (consisting of the Hyperactivity, Inattention, Lack of the children were instructed to do the opposite in the second condi-
effort, and Executive functions scales), and reported that the FTF tion of what they had done spontaneously in the first condition
showed good external validity and sensitivity in screening devel- regarding private speech, they were not always able to do so. For
opmental disorder with high (93%) rate of positive hits and low instance, the two children who used private speech as many as six
rate (7%) of misses; specificity was not good (63% false posi- times in the first condition of part 1 (task items 1-8) continued to
tives). The Cronbach alpha in the present study (for the three use it a couple of times after being asked to complete the succeed-
scales) for the parental data was .86 (n ¼ 296) and for the pre- ing items silently during the following task items. Of the 58 chil-
school data .92 (n ¼ 182). dren who did not use private speech spontaneously in the first
condition of part 1 (task items 1–8), nine children used speech
Speech–action coordination task (the Hammer Task) at 5 years 3 only two times or less after being asked to use vocalization (task
months of age. The Hammer Task, adapted from Balamore and items 9–16).

Downloaded from jbd.sagepub.com at University of Sydney on August 14, 2015


Aro et al. 5

The number of items in which the child used private speech min ¼ 2, and max ¼ 23). The Korpilahti Sentence Comprehension
spontaneously (i.e., without being prompted to do so) in part 1 (con- test consists of sentences that are read aloud to the child, and the
dition without speech instruction; items 1–8) was used as the mea- child is then asked to choose the picture from three alternatives
sure of amount of spontaneous speech. In order to calculate a that best matches the meaning of the sentence (in the present data,
speech effectiveness score, we first calculated the total number of M ¼ 21.25, SD ¼ 4.17, min ¼ 6, and max ¼ 29).
tasks completed by the children with private speech (either sponta-
neously or after being requested) and the total number of tasks com- Inhibition. The composite score for inhibition consisted of standar-
pleted without speech. Second, two probability scores were dized scores of the following three tests: the Knock and Tap subtest,
calculated: The probability of a child completing a task correctly the Inhibition subtest, and the Statue subtest of the NEPSY (Kork-
1) while using private speech; number of tasks completed correctly man et al., 1998). The Knock and Tap subtest is a nonverbal
while using private speech divided by the total number of tasks conflict-solving task (of the classic ‘‘go/no-go’’ type) that measures
completed with speech; P(C|S), and 2) while not using speech; num- self-regulation and the inhibition of immediate responses. Children
ber of task items completed correctly without using private speech were instructed to knock on a table with their knuckles when the
divided by the total number of task items completed without speech; examiner tapped with his/her flat palm and vice versa. A total of
P(C|NS) (see Winsler et al., 2007). Finally, the speech effectiveness 15 trials were conducted in pseudorandom order (in the present
score; probability of correct reactions while using private speech data, M ¼ 26.97, SD ¼ 3.63, min ¼ 12, and max ¼ 30). The Inhibi-
divided by probability of correct reactions while not using private tion subtest is a timed task designed to assess a child’s ability to
speech; P(C|S)/P(C|NS), was computed for each child. This speech inhibit automatic responses in favour of novel responses, and the
effectiveness score indicated how probable it would be for a child to ability to switch between response types. In the first part, the child
complete a task item correctly while using private speech, taking looks at a series of black and white shapes (squares or circles) and
into account the total number of task items completed with as well names each one of them, after which he/she is asked to name their
as the total number of task items completed without speech. For opposite shape in another run-through (in the present data, M ¼
example, a child completes 24 items. He/she uses speech in 15 34.93, SD ¼ 5.93, min ¼ 0, and max ¼ 40). In the second part, the
items and performs correctly in 11 of them, and he/she does not use child looks at a series of black and white arrows (upward or down-
speech in 9 items and performs correctly in all of them. His/her ward) and names each arrow’s direction, after which he/she is asked
scores would be as follows: P(C|S) ¼ 11/15 and P(C|NS) ¼ 9/9, and to state their opposite direction in a second run-through (in the pres-
the effectiveness score would be .73/1 ¼ .73 (below 1) indicating ent data, M ¼ 31.80, SD ¼ 6.08, min ¼ 1, and max ¼ 40). The Sta-
that he/she got more items correctly without speech than with tue subtest is designed to assess motor persistence and inhibition.
speech. Similarly, another child also completes 24 items. He/she The child is asked to maintain a body position with his/her eyes
uses speech in 12 items and performs correctly in six of them, and closed for a 75-second period and to inhibit his/her impulse to
he/she does not use speech in 12 items and performs correctly in respond to sound distractions made by the examiner at standard
four of them. His/her scores would be as follows: P(C|S) ¼ 6/12 and intervals (in the present data, M ¼ 11.71, SD ¼ 3.92, min ¼ 0, and
P(C|NS) ¼ 4/12, and the effectiveness score would be .50/.33 ¼ max ¼ 15).
1.50 indicating that he/she got more items correctly with speech
than without speech. Planning and fluency. The composite score for planning and flu-
ency was formed from the standardized scores of the following
Cognitive abilities at 5 years 3 months of age. The raw scores of three tests: the Design Fluency and Word Fluency of the NEPSY
the cognitive tests (tests of language, inhibition, planning and flu- (Korkman et al., 1998), and the Block-Design subtest of the
ency, and memory) were first standardized according to local norms WPPSI-R (Wechsler, 1989, 1995). The Verbal Fluency subtest is
and converted to z scores to ensure that all variables were on the a speed test assessing the ability to generate words according to
same scale (M ¼ 0, SD ¼ 1). Next, cognitive composite scores were semantic categories (animal, food, drink), and at later ages also
formed based on these standardized scores. Each of the composite according to phonemic categories (starting with a requested pho-
scores consisted of three to four tests. In the analyses, these cogni- neme). The child needs to bear in mind the requested category and
tive composites will be referred to by the skill respective domain: to avoid repeating words that he/she has already said (in the present
language, inhibition, planning and fluency, and memory. data, M ¼ 15.56, SD ¼ 6.08, min ¼ 1, and max ¼ 29). The Design
Fluency subtest assesses the child’s ability to generate novel
Language. The composite score for language was formed based on designs as quickly as possible from structured and unstructured
the standardized scores of four language tests: the Sentence Repeti- arrays of dots by connecting two or more dots. In order to attain
tion and the Repetition of Nonsense Words of the NEPSY (A a high score on this task, the child needs to systematically monitor
Developmental Neuropsychological Assessment; Korkman, Kirk, his/her progress and avoid repeating patterns (in the present data,
& Kemp, 1998), the Similarities subtest of the Wechsler Intelli- M ¼ 10.92, SD ¼ 4.51, min ¼ 1, and max ¼ 27). For the Block-
gence Scale for Children (WPPSI-R; Wechsler, 1989, 1995), and Design subtest, which is to be completed within a specified time
the Korpilahti Sentence Comprehension test (Korpilahti, 1998). limit, the child uses coloured blocks to build designs based on
The Sentence Repetition assesses a child’s ability to repeat sen- model pattern. The Block-Design task measures both spatial and
tences of increasing complexity and length (in the present data planning processes, via the ability to analyse and synthesize an
M ¼ 21.25, SD ¼ 4.68, min ¼ 6, and max ¼ 29). The Repetition abstract design (e.g. Shallice, 1982) (in the present data, M ¼
of Nonsense Words assesses phonological encoding and decoding 24.74, SD ¼ 6.09, min ¼ 6, and max ¼ 29).
through a task in which the child is asked to repeat nonsense words
(in the present data, M ¼ 9.53, SD ¼ 2.17, min ¼ 2, and max ¼ 14). Memory. The composite score for memory consisted of the standar-
The Similarities subtest is a measure of verbal reasoning and dized scores of the following three tests assessing short-term mem-
abstract thinking (in the present data, M ¼ 16.42, SD ¼ 4.04, ory: the Digit-Span Forward Recall and the Digit-Span Backward

Downloaded from jbd.sagepub.com at University of Sydney on August 14, 2015


6 International Journal of Behavioral Development

Table 1. Descriptives concerning children’s measures.

Variable M (SD) min max

Peabody Picture Vocabulary Test (raw score) 16.67 (3.63) 5 24


Behavioural self-regulation (Five-to-Fifteen raw score / z score)
Impulsivity, parents 5.14 (3.24) / .45 (.97) 0/ 2.87 12 / 1.08
Inattention, parents 4.96 (3.11) / .59 (.99) 0/ 3.13 13 / 1.00
Lack of effort, parents 1.21 (1.58) / .14 (.32) 0/ 1.11 6 / .11
Impulsivity, preschool 3.04 (3.64) / .18 (1.08) 0/ 3.60 16 / 1.08
Inattention, preschool 3.90 (3.77) / .26 (1.19) 0/ 4.08 16 / 1.00
Lack of effort, preschool 1.71 (1.68) / .18 (.34) 0/ 1.32 7 / .11
Cognitive scores (z score)
Language .15 (.81) 2.38 1.75
Inhibition .02 (.67) 3.00 1.17
Planning and fluency .09 (.68) 1.89 1.67
Memory .06 (.71) 2.29 1.61
Hammer Task scores (raw score)
Items completed 21.3 (3.12) 8 24
Correct responses 13.73 (4.35) 2 24
Responses with spontaneous speech in part 1a 1.22 (1.95) 0 8
Responses with speech upon instruction in part 1b 4.05 (3.18) 0 8
Total responses with speechc 8.65 (3.13) 0 15
Probability of task-items correct with speech .73 (.22) 0 1
Probability of task-items correct without speech .59 (.20) 0 1
Speech effectiveness score 1.41 (.73) 0 5.14

Note. n ¼ 101.
a
Sum of responses with speech in task-items 1–8 on part 1 without speech instruction.
b
Sum of responses with speech in task-items 9–16 on part 1 with speech instruction.
c
Sum of responses with speech in all task-items (part 1 and part 2).

Recall from the WPPSI-R (Wechsler, 1989, 1995), and the Corsi items. It was more common for the children to perform tasks with-
Block-Tapping Test of spatial memory (Grossi, Orsini, Monetti, out than with private speech: only 20.8% of the children did more
& DeMichele, 1979). The Digit-Span tests require the repetition task items with speech than without speech, 60.4% of the children
of number sequences both forward and backwards (in the present did more task items without private speech than with private
data, M ¼ 4.75, SD ¼ 1.40, min ¼ 0, and max ¼ 8; M ¼ 1.84., speech, and the rest of the did as many task items with speech as
SD ¼ 1.27, min ¼ 0, and max ¼5, respectively). The Corsi without speech.
Block-Tapping Test is administered using a set of nine 3.1-cm To examine whether the difficulty of the Hammer task items
block cubes irregularly arranged on a board. The examiner taps the was associated with the amount of spontaneous speech, we looked
blocks in sequences of increasing length (up to a maximum at the percentage of correct answers and the amount of speech used
sequence of six blocks), and the child is required to reproduce each in the first eight task items in part 1. The percentage of children per-
sequence in the same order immediately after the demonstration. A forming correctly decreased as the Hammer task items became lon-
score is given for the longest sequence that the child is able to repro- ger: In the first and second task items (both with a sequence of two
duce correctly, with a possible maximum score of six (in the present colours), 89.1% and 91.1% of the children, respectively, performed
data, M ¼ 3.82, SD ¼ .95, min ¼ 0, and max ¼ 6). correctly. In the third and fourth task items (sequence of three col-
ours), 73.7% and 83.0% performed correctly. In the fifth and sixth
task items (sequence of four colours), 37.1% and 44.4%, and in the
Results last two (sequence of five colours), 18.5% and 24.1% performed
Spontaneous use of private speech, task item difficulty, correctly. These percentages demonstrate that the difficulty of the
items increased as they became longer. Only 54 children reached
and speech effectiveness the last two items. The number of correct answers provided for the
The Hammer Task data were available for 101 children (53 boys first and last task items was significantly different—t(53) ¼ 10.74;
and 48 girls) at the age of 5 years and 3 months. One child of those p  .001; paired samples t test comparing number of correct
participating to the individual assessments at 5 years 3 months of answers in the first and last task items for those who completed all
age had missing data on the Hammer Task. On average, the children the items—indicating that more correct answers were provided by
completed 21 task items and performed 13 of these correctly (see the children in the first task items.
Table 1), the maximum score being 24 (8 þ 8 in part 1, 4 þ 4 in Children’s use of spontaneous speech did not vary much after
part 2). Private speech was used spontaneously by 42.6% of the the very first tasks: although the easiest task items elicited less
children (i.e., they spoke spontaneously at least once in the condi- speech than the longer and more difficult ones, the amount of
tion without speech instruction which consisted of the first eight speech did not increase linearly with the difficulty of the items after
task items in part 1). On average, the children used speech in the third task item. In the first and second task items (sequence of
8.65 (SD ¼ 3.13) of the task items. Only two children (2.0%) did two colours), only 5.5% and 8.9% of the children, respectively,
not use speech at all whereas some children used speech in 15 task used speech; and 10.9% used speech in one or both of the 2-color

Downloaded from jbd.sagepub.com at University of Sydney on August 14, 2015


Aro et al. 7

Table 2. Intercorrelations between behavioural self-regulation, cognitive scores and speech effectiveness score.

1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Behavioural self-regulation, Five-to-Fifteen


1. Impulsivity, parents
2. Inattention, parents .495**
3. Lack of effort, parents .001 .450**
4. Impulsivity, preschool .053 .053 .084
5. Inattention, preschool .025 .275* .183 .536**
6. Lack of effort, preschool .112 .157 .372** .342** .721**
Cognitive scores
7. Language .079 .169 .160 .021 283* .369*
8. Inhibition .141 .142 .077 .168 .244* .296* .387**
9. Planning and fluency .053 .135 .119 .062 .250* .460* .556** .406**
10. Memory .013 .146 .103 .162 .216 .403* .625** .449** .379**
11. Speech effectiveness .081 .005 .175 .010 .008 .236** .312** .320** .200** .322**

Note. n ¼ 101.
*p < 0.05; **p < 0.01; ***p < 0.001.

trials. In the third and fourth task items (sequence of three colours), Behavioural self-regulation skills, speech effectiveness,
28.8% and 17.0% used speech; and 32.0% used speech in one or
both of the 3-color trials. In the fifth and sixth task items (sequence
and the amount of spontaneous private speech
of four colours), the percentages were 18.9% and 23.3%; and 25.6% The associations between private speech and behavioural self-
used speech in one or both of the 4-color trials. In the seventh and regulation skills (as rated by both parents and preschool teachers
eight task items (sequence of five colours), the percentages were using the FTF) were first examined using correlational analyses.
16.4% and 31.5%, respectively; and 31.5% used speech in one or Next, we applied multivariate general linear model (GLM) analysis
both of the 5-color trials. When the number of children using speech separately for parent ratings and preschool teacher ratings with the
in either one or both of the items in each difficulty level were com- measure of private speech in question (speech effectiveness or the
pared, it was found (using paired samples t test comparing number amount of spontaneous private speech) entered as the independent
of children using speech in the two consecutive difficulty levels) variable, and the three continuous FTF scale scores (i.e., impulsiv-
that there was a significant difference in number of children using ity, inattention, and lack of effort) entered as dependent variables.
speech between 2-colour trials (10.9%) and 3-color trials (32.0%), The advantage of using GLM is that multivariate tests of signifi-
t(99) ¼ 4.60; p  .001. No reliable difference was detected between cance can be employed when responses on multiple dependent vari-
3-color trials (32.0%) and 4-color trials (25.6%), t(88) ¼ 1.97; p ¼ ables are correlated. Hypothesis-testing can be made using the
.052, although there was a trend for decreasing amount of speech. independent univariate tests after the multivariate test.
Also, no difference was detected between 4-color trials (25.6%) and Parental questionnaires were available for 101 children, and pre-
5-color trials (31.5%), t(53) ¼ .275; p ¼ .784. The correlation school questionnaire data were available for 71 children. The cor-
between the use of speech and the correctness of the performance relations between the speech effectiveness score and measures of
was similar across that tasks with varying difficulty level; e.g. in the behavioural self-regulation difficulties (i.e., impulsivity, inatten-
first two task items (r ¼ .08–.10) and the last two task items (r ¼ tion, and lack of effort) were non-significant (r ¼ .005 – .174),
.08–.18). except for the significant correlation between speech effective-
To examine the extent to which the children performed better ness and teachers’ ratings of lack of effort (r ¼ .236; see Table
when using speech (either spontaneously or after a request) than 2). The correlations between the amount of spontaneous private
when not using speech, a score indicating speech effectiveness speech and behavioural self-regulation difficulties were also
(i.e., probability of correct responses to task items while using non-significant (r ¼ .14 – .03).
speech divided by probability of correct responses while not using Next, the association between children’s speech effectiveness
speech) based on calculated probabilities of correct responses with and their behavioural self-regulation difficulties were analysed
and without private speech. On average, the children got more items using the GLM technique; the measures of behavioural self-
correct when they used private speech than when they did not: the regulation difficulties (impulsivity, inattention, and lack of effort)
average score indicating the probability of responding to task items were regressed on the speech effectiveness score. The results of the
correctly while using private speech was .73, while the average GLM analysis with parental ratings indicated no association
score when not using private speech was .59. For 76.2% of the chil- between the speech effectiveness score and children’s behavioural
dren, the speech effectiveness score (M ¼ 1.41; SD ¼ .73) was self-regulation difficulties, F(3, 97) ¼ 1.36; p ¼ ns. The results of
higher than 1 (see Table 1) which indicates that for the majority the GLM analysis with preschool teacher ratings, however, showed
of the children speech was effective. To test the statistical signifi- a significant association between the speech effectiveness score and
cance of the probability to perform better with speech than without behavioural self-regulation difficulties, F(3, 67) ¼ 2.78; p  .05.
speech we tested whether the effectiveness score minus 1 is differ- The univariate tests indicated a significant association between the
ent from zero. The statistically-significant finding, t(100) ¼ 5.58; speech effectiveness score and children’s lack of effort, F(1, 69) ¼
p ¼ .000, indicated that children performed more often correctly 4.07; p  .05; partial 2 ¼ .056. The regression coefficients (b ¼
with speech than without speech. The speech effectiveness score .11) indicated that the children rated by their teachers as showing
was used as a continuous variable in the subsequent analyses. less effort (lower z score) had higher speech effectiveness scores; in

Downloaded from jbd.sagepub.com at University of Sydney on August 14, 2015


8 International Journal of Behavioral Development

scores, F(4, 94) ¼ 5.87; p  .001. The subsequent univariate tests


indicated significant associations between the speech effectiveness
score and language skills (F(1, 99) ¼ 14.81; p  .001; partial 2 ¼
.132; b ¼ .406), inhibition skills, F(1, 99) ¼ 11.51; p  .001; par-
tial 2 ¼ .106; b ¼ .302, planning and fluency, F(1, 99) ¼ 5.28;
p  .05; partial 2 ¼ .052; b ¼ .216, and memory, F(1, 99) ¼
17.58; p  .001; partial 2 ¼ .153; b ¼ .376. The regression coef-
ficients indicated that the poorer a child’s performance was on the
cognitive tests, the better he/she performed on the Hammer Task
items when using private speech compared to task items where
he/she did not use speech. The effects of the different cognitive
skills did not differ from each other. There were no associations
between the amount of spontaneous private speech and cognitive
scores, F(4,94) ¼ .21; p ¼ ns.
To illustrate the association between cognitive abilities and
speech effectiveness, the children were divided into subgroups
according to their speech effectiveness score. Children having a
score below or equal to 1 formed the group labelled ‘‘No effect’’
(n ¼ 21), while children having a score above 1 but below or equal
to 2 formed the group labelled ‘‘Some effect’’ (n ¼ 64), and chil-
dren with a score higher than 2 formed the group labelled ‘‘Clear
effect’’ (n ¼ 16). Figure 1 shows that children in the ‘‘Clear effect’’
group had the lowest cognitive scores.
Figure 1. Cognitive scores of the speech effectiveness groups.
Note. No effect group n ¼ 21; Some effect group n ¼ 64; and Clear effect
group n ¼ 16.
Discussion
other words, their performance on the Hammer Task was better The present study examined the associations between 5-year-old
when they used private speech than when they did not use speech. children’s behavioural self-regulation, cognitive abilities, and their
There were no associations between the amount of sponta- private speech during a structured task (the Hammer Task). More
neous private speech and behavioural self-regulation (parent rat- specifically, we analysed the extent to which the children used pri-
ings: F(3, 97) ¼ .47; p ¼ ns; preschool teacher ratings: F(3, 67) vate speech and whether they performed better in the Hammer Task
¼ .49; p ¼ ns). when using speech than when not using it (i.e., speech effective-
ness), and whether the amount of spontaneous speech and speech
effectiveness were associated with the children’s behavioural
Cognitive abilities, speech effectiveness, and the self-regulation (i.e., impulsivity, inattention, and lack of effort) and
cognitive abilities (language, inhibition, planning and fluency, and
amount of spontaneous private speech memory). The results indicated that 42.6% of the children used pri-
To analyse the associations between private speech and cognitive vate speech spontaneously (at least sometimes without being
scores (language, inhibition, planning and fluency, and memory), prompted), and most of the children (76%) performed better (got
we first conducted correlational analyses. Next, we applied multi- more of the items correctly) when they used speech than when they
variate GLM analysis, and the measure of private speech in ques- did not use it. Children used spontaneous speech less in the easiest
tion (speech effectiveness or the amount of spontaneous private task items (i.e., first and second items) than in the longer and more
speech, respectively) was entered as the independent variable, and difficult ones, but the use of speech did not increase linearly with
the four continuous cognitive scores were entered as dependent the task difficulty. Associations were found between speech effec-
variables. tiveness and all cognitive abilities, but associations between speech
Although the Hammer Task data were available for 101 chil- effectiveness and behavioural self-regulatory skills emerged only
dren, two of these children had missing data on some of the cogni- for teachers’ ratings of lack of effort. No associations were found
tive variables; therefore, they were excluded from these analyses. between the amount of spontaneous private speech used during the
The correlations between the speech effectiveness score and cogni- Hammer Task and behavioural self-regulatory skills or cognitive
tive scores were moderate, but statistically significant (p  .05): abilities.
language r ¼ .312, inhibition r ¼ .320, planning and fluency We found, in line with our expectation and earlier literature (e.g.
r ¼ .200, and memory r ¼ .322 (see Table 2). The correlations Winsler et al., 2007), that it is relatively common for 5-year-old
between the amount of spontaneous private speech and cognitive children to use overt private speech in a task situation: 42.6% of the
scores were non-significant (r ¼ .04–.06). children in our study used private speech spontaneously at least
Next, associations between speech effectiveness and cognitive once in the Hammer Task. For most of the children, private speech
abilities were analysed using the multivariate GLM; the child’s was effective, as about 76% had a speech effectiveness ratio above
cognitive abilities (i.e., language, inhibition, planning and fluency, 1, indicating that they performed better when they used speech than
and short-term memory) were regressed on the speech effectiveness when they did not use it. This finding suggests that speech serves as
score. The GLM results indicated a negative significant association a self-regulatory tool (e.g. Vygotsky, 1962; Winsler et al., 2007; for
between the speech effectiveness score and the four cognitive a review, see Winsler, 2009). The children used speech more when

Downloaded from jbd.sagepub.com at University of Sydney on August 14, 2015


Aro et al. 9

prompted to do so than spontaneously. In part 1 under the condition solving strategies and elaboration or including feedback compo-
of no instruction to use speech (first set of 8 task-items), the average nents, might have elicited more variance in use of speech (see
of responses with speech was 1.22 whereas under the condition of Kopecky et al., 2005; Winsler, 2009).
instruction to use speech (second set of 8 task-items) on average The present study focused on the associations between cognitive
4.05 of responses were with speech. This can be interpreted to sug- skills and effectiveness of private speech which has been addressed
gest that 5-year-olds do not typically have the awareness that they less frequently than the emergency and quantity of private speech at
might perform better if they speak out loud during a task. This has different ages. We found strong associations between children’s
been indicated by the findings by Winsler and Naglieri (2003) speech effectiveness and cognitive test performance. Use of private
showing that only about 16% of 5-year-old children were aware speech (spontaneously or after a request) was more beneficial for
of the benefits of using private speech (see also Manfra & Winsler, the children with weaker cognitive skills suggesting that they
2006). needed private speech to control or aid their cognitive performance,
The amount of spontaneous speech during the Hammer task was although they did not differ in the amount of spontaneous speech
not associated with ratings of behavioural regulatory skills, which from other children. On the other hand, children with better cogni-
was contrary to our expectations based on the literature document- tive skills did not seem to need to use private speech to perform well
ing that private speech tends to be more frequent among children in the Hammer Task, i.e., they performed well both with and with-
with self-regulatory problems than without them (for a review, see out speech.
Winsler, 2009; see also Corkum et al., 2008). The present sample It is plausible that children with weaker cognitive skills bene-
involved 9% of children with high scores in hyperactivity (below fitted from private speech more because the Hammer Task was
2.0 SD, Finnish standardization norms), as rated by adults, but more difficult for them. This interpretation is supported by the
none of the children had been diagnosed with ADHD at this early post-hoc analyses indicating rather high correlations between the
age. It is therefore plausible that variation in behavioural self- amount of correct answers in the Hammer Task and cognitive skills
regulatory skills was too small in the present sample for associa- (r varying between .56–.72). Children with less mature cognitive
tions with the amount of spontaneous speech to emerge. skills may have benefited from using private speech because the
Recent studies among children with impairments, e.g. children repetition of the given stimuli aided them in planning their action
diagnosed with ADHD (for a review, see Corkum et al., 2008, and (see Lidstone et al., 2012) and in refreshing memory trace of the
Winsler, 2009) and specific language impairments (Lidstone et al., instruction (see Baddeley, 2001; Repovs & Baddeley, 2006; see
2012), indicate that these children utilize private speech for self- also Winsler, 2009, and Marvel & Desmond, 2012). Recently, Kray
regulation. In the present study, no systematic associations were and colleagues (Kray, Kipp, & Karbach, 2009) found that labelling
found between the effectiveness of private speech and parental rat- the given visual stimulus in a selective inhibitory control task
ings of children’s behavioural self-regulation skills indicating that increased the accuracy of children’s responses, suggesting that
the regulational skills were not strongly associated with speech inner speech helps in the maintenance of task representations. There
effectiveness. However, a correlation emerged between speech is also evidence indicating that speech motor processes aid lan-
effectiveness and preschool teachers’ ratings of children’s lack of guage comprehension (see Pickering & Garrod, 2007; see also Wil-
effort, suggesting that children rated as lacking effort (e.g. listless, son & Knoblich, 2005). On the first part of the Hammer Task,
sluggish, and lacking energy) in the preschool setting benefitted where children need to map names of colours according to their
from using private speech in the Hammer Task. Deficiencies in semantic constituents, use of private speech may serve to link
effort may be more evident in the context of preschool where chil- semantic understanding and speech production. These associations
dren are likely to face challenging tasks that require persistence between self-vocalization, working memory, and semantic process-
than in the children’s homes where similar tasks are less frequently ing warrant further research among different age groups. Among
observed by the parents. The finding of higher efficiency of private younger children, such as 3-year-olds, more association may have
speech among children with poor effort can be understood as an emerged between cognitive skills and both amount of spontaneous
indicator that these children needed private speech to gain control private speech and effectiveness of private speech. Optimally a
of their underlying deficient attentional control system (see also longitudinal design should be used to analyse possible changes in
Winsler, 2009), and therefore they should be encouraged to use pri- the associations between effectiveness of private speech and cogni-
vate speech. Medication has been shown to normalize the use of pri- tive skills from kindergarten to early school years.
vate speech among children with a combined type of ADHD (see Prior studies have indicated that the level of task difficulty is of
Kopecky, Chang, Klorman, Thatcher, & Borgstedt, 2005) suggest- high relevance for the elicitation of private speech, and that private
ing that improved ability to channel effort reduces the need for reg- speech is evoked by optimally challenging tasks (e.g. Diaz, 1992;
ulatory self-speech. Further research is needed to obtain a better Fernyhough & Fradley, 2005). Fernyhough and Fradley (2005)
understanding of the mechanisms through which private speech is showed a quadratic relation between speech and task difficulty with
associated with the successful use of attention and executive func- amount of speech peaking in tasks which were neither too simple
tions in learning situations, as well as which attentional character- nor too difficult. Thus, solving a task in silence can be associated
istics of the child are associated with the effectiveness of private both with a too easy task and a too difficult task (see Winsler, Diaz,
speech. & Montero, 1997). Our data indicated that the amount of sponta-
The literature indicates that cognitive skills are positively asso- neous speech increased after the first two task items, but it did not
ciated with the emergency and internalization of private speech continue to increase although the items became longer, and more
among younger children (Kohlberg et al., 1968; Lidstone et al., difficult. Winsler (2009) claimed that the appearance of private
2012), but later in development the correlation is negative. Contrary speech is ‘‘intimately linked with the individual’s task-specific
to our expectations, we did not find associations between amount of competence’’ (Winsler, 2009, p. 10). It is possible—and also sug-
spontaneous speech and cognitive skills. A more challenging task gested by the rather high correlations between correct answers and
requiring planning and monitoring, metacognitive problem- cognitive skills—that for the 5-year-olds with lower cognitive

Downloaded from jbd.sagepub.com at University of Sydney on August 14, 2015


10 International Journal of Behavioral Development

abilities the Hammer Task was optimal for private speech to be a Balamore, U., & Wozniak, R.H. (1984). Speech–action coordination in
useful tool to improve the performance, whereas for the children young children. Developmental Psychology, 20, 850–858.
with better cognitive abilities the task was so easy that they did not Barkley, R. A. (2012). Executive functions. What they are, how they
need to resort to private speech. This is in line with the earlier find- work, and why they evolved. New York, NY: Guilford Press
ings by Winsler and Naglieri (2003) which show that among chil- Berk, L. E. (1986). Relationship of elementary school children’s private
dren between 5 and 7 years of age, talking to oneself did not speech to behavioral accompaniment to task, attention, and task per-
affect the performance of children with higher-ability levels, formance. Developmental Psychology, 22, 671–680.
whereas the use of speech improved the performance of lower- Berk, L. E. (1992). Children’s private speech: An overview of theory
achieving children. and the status of research. In R. M. Diaz & L. E. Berk (Eds.), Private
Our study contributed to the literature on private speech by speech. From social interaction to self-regulation, (pp. 17–53).
investigating its associations to cognitive abilities and behavioural Hillsdale, NJ: Lawrence Erlbaum Associates.
self-regulation skills. However, there were limitations. First, the Berk, L. E., & Landau, S. (1993). Private speech of learning disabled
study was carried out in a sample of children with a slight overpre- and normally achieving children in classroom academic and labora-
sentation of difficulties in language development or hyperactivity, tory contexts. Child Development, 64, 556–571.
according to parental ratings. Nevertheless, none of the children Berk, L. E., & Potts, M. K. (1991). Development and functional signif-
were diagnosed with ADHD, and only four were diagnosed as hav- icance of private speech among attention-deficit hyperactivity dis-
ing a specific language impairment. Second, a rather simple task ordered and normal boys. Journal of Abnormal Child Psychology,
and only in a single test situation was used to elicitation of private 19, 357–377.
speech. A longitudinal design with multiple assessments would Bivens, J. A., & Berk, L. E. (1990). A longitudinal study of the devel-
have provided opportunities to examine changes across age (see opment of elementary school children’s private speech.
Winsler, 2009). Furthermore, the Hammer Task and the cognitive Merrill-Palmer Quarterly, 36, 443–463.
tests were performed in the same laboratory setting, whereas the Corkum, P., Humpries, K., Mullane, J., & Theriault, F. (2008). Private
self-regulations ratings were based on parent and preschool teacher speech in children with ADHD and their typically developing peers
ratings. The context and its interaction with individual characteris- during problem-solving and inhibition tasks. Contemporary Educa-
tics in measuring self-regulation and executive skills is likely to be tional Psychology, 33, 97–115.
meaningful as pointed out by McClelland, Ponitz, Messersmith, and Diaz, R. M. (1992). Methodological concerns in the study of private
Tominey (2010). A more complex task (requiring planning and speech. In R. M. Diaz & L. E. Berk (Eds.), Private speech. From
monitoring, for example) would probably have elicited more social interaction to self-regulation (pp. 55–81). Hillsdale, NJ:
speech among the children than in the present study where the Lawrence Erlbaum Associates.
majority of the children completed more task items without than Diaz, R. M., Winsler, A., Atencio, D. J., & Harbers, K. (1992). Media-
with private speech. Coding the quantity and type of private speech tion of self-regulation through the use of private speech. Interna-
(i.e., task-relevant, task-irrelevant, or partially internalized; see e.g. tional Journal of Cognitive Education and Mediated Learning, 2,
Berk & Landau, 1993) could have given valuable information 1–13.
regarding the phase of internalization and it’s association to various Deutsch, F., & Stein, A. H. (1972). The effects of personal responsibil-
cognitive skills. The advantage of using the Hammer Task was the ity and task interruption on the private speech of preschoolers.
opportunity to compare conditions with and without speech but a Human Development, 15, 310–324.
different type of test situation would have given possibility analyse Dunn, L. M., & Dunn, L. M. (1981). Peabody Picture Vocabulary Tes-
use of spontaneous private speech. t-Revised. Circle Pines, MN: American Guidance Service.
Our findings lend support to the importance of including chil- Fernyhough, C., & Fradley, E. (2005). Private speech on an executive
dren’s cognitive abilities in studies investigating the development, task: Relations with task difficulty and task performance. Cognitive
use, and effectiveness of private speech. Cognitive tests, particu- Development, 20, 103–120.
larly of planning skills and short-term memory, need to be incorpo- Grossi, D., Orsini, A., Monetti, C., & DeMichele, G. (1979). Sex differ-
rated in future research designs. It is important to raise awareness ences in children’s spatial and verbal memory span. Cortex, 15,
about the effects of using private speech, especially among children 667–528.
with deficits or delays in cognitive abilities, as well as to model its Hitch, G., Halliday, M., Schaafstal, A., & Heffernan, T. (1991). Speech,
use for both teachers and parents. Using private speech effectively ‘‘inner speech,’’ and the development of short-term memory:
in challenging tasks sets the stage for refining one’s metacognitive Effects of picture-labeling on recall. Journal of Experimental Child
strategies and reasoning skills. Psychology, 51, 220–234.
Jones, S., & Fernyhough, C. (2007). Neural correlates of inners speech
Funding and auditory verbal hallucinations. Clinical Psychology Review, 27,
140–154.
This research received no specific grant from any funding agency in
Jones, S. (2009). The neuropsychology of covert and overt speech:
the public, commercial, or not-for-profit sectors.
Implications for the study of private speech in children and adults.
In A. Winsler, C. Fernyhough & I. Montero (Eds.), Private speech,
References executive functioning, and the development of verbal self-regulation
Al-Namlah, A., Fernyhough, C., & Meins, E (2006). Socio-cultural (pp. 69–80). New York, NY: Cambridge University Press.
influences on the development of verbal mediation: Private speech Kadesjö, B., Janols, L.-O., Korkman, M., Mickelsson, K., Strand, G.,
and phonological recording in Saudi Arabian and British samples. Trillingsgaard, A., & Gillberg, C. (2004). FTF (Five to Fifteen): The
Developmental Psychology, 42, 117–131. development of a parent questionnaire for the assessment of AD/HD
Baddeley, A. (2001). Is working memory still working? European Psy- and comorbid conditions. European Child and Adolescent Psychia-
chologist, 7, 85–97. try, 13(suppl. 3), III/3–III/13.

Downloaded from jbd.sagepub.com at University of Sydney on August 14, 2015


Aro et al. 11

Kohlberg, L., Yaeger, J., & Hjertholm, E. (1968). Private speech: Four Palmer, E., Rosen, H., Ojemann, J., Buckner, R., Kelley, W., & Peter-
studies and a review of theories. Child Development, 39, 691–739. sen, S. (2001). An event-related fMRI study of overt and covert
Kopecky, H., Chang, T., Klorman, R., Thatcher, J., & Borgstedt, A. word stem completion. Neuroimage, 14, 182–193.
(2005). Performance and private speech of children with attention- Pickering, M. J., & Garrod, S. (2007). Do people use language produc-
deficit/hyperactivity disorder while taking the tower of Hanoi test: tion to make predictions during comprehension? Trends in Cogni-
Effects of depth of search, diagnostic subtype, and methylphenidate. tive Sciences, 11, 105–110.
Journal of Abnormal Child Psychology, 33, 625–638. Rapport, M. D., Bolden, J., Kofler, M. J., Sarver, D. E., Raiker, J. S., &
Korkman, M., Jaakkola, M., Ahlroth, A., Pesonen, A. E., & Turunen, Alderson, R. M. (2009). Hyperactivity in boys with attention-defi-
M. M. (2004). Screening of developmental disorders in cit/hyperactivity disorder (ADHD): An ubiquitous core symptom
five-year-olds using the FTF (Five-to-Fifteen) questionnaire: A or manifestation of working memory deficits? Journal of Abnormal
validation study. European Child & Adolescent Psychiatry, Child Psychology, 37, 521–534.
13(suppl. 3), III/31–III/38. Repovs, G., & Baddeley, A. (2006). The multi-component model of
Korkman, M., Kirk, U., & Kemp, S. (1998). NEPSY. A developmental working memory: Explorations in experimental cognitive psychol-
neuropsychological assessment. Manual. San Antonio, TX: The ogy. Neuroscience, 139, 5–21.
Psychological Corporation. Shallice, T. (1982). Specific impairments of planning. Philosophical
Korpilahti, P. (1998). Lausetesti: Kuullun ymmärtämisen lausetasoinen Transactions of the Royal Society of London, Series B, 298,
testi [Listening comprehension test]. Helsinki, Finland: Language 199–209.
and Communication Care Oy. Vygotsky, L. (1962). Thought and language. Cambridge, MA: MIT
Kraft, K., & Berk, L. E. (1998). Private speech in two preschools: Sig- Press.
nificance of open-ended activities and make-believe play for verbal Vygotsky, L. (1978). Mind in society. The development of higher psy-
self-regulation. Early Childhood Research Quarterly, 13, 637–658. chological processes. Cambridge, MA: Harvard University Press.
Kray, J., Kipp, K. H., & Karbach, J. (2009). The development of selec- Wechsler, D. (1989). Wechsler Intelligence Scale for Children-III. San
tive inhibitory control: The influence of verbal labeling. Acta Psy- Antonio, TX: The Psychological Corporation.
chologica, 130, 48–57. Wechsler, D. (1995). WISC-III. Wechslerin lasten älykkyysasteikko.
Lidstone, J., Meins, E., & Fernyhough, C. (2012). Verbal mediation of Käsikirja. [Wechsler Intelligence Scale for Children. Manual]. Hel-
cognition in children with specific language impairment. Develop- sinki, Finland: Psykologien kustannus.
ment and Psychopathology, 24, 651–660. Wetherby, A., & Prizant, B. (1992). Profiling young children’s commu-
Manfra, L., & Winsler, A. (2006). Preschool children’s awareness of nicative competence. In S. Warren & J. Reichle (Eds.), Perspective
private speech. International Journal of Behavioral Development, on communication and language intervention: Development,
30, 537–549. assessment, and intervention (pp. 217–251). Baltimore, MD: Paul
Marvel, C. L., & Desmond, J. E. (2012). From storage to manipulation: H. Brookes.
How the neural correlates of verbal working memory reflect varying Willcutt, E. G., Pennington, B. F., Boada, R., Ogline, J. S., Tunick, R.
demands on inner speech. Brain & Language, 120, 42–51. A., Chhabildas, N. A., & Olson, R. K. (2001). A comparison of the
McClelland, M. M., Ponitz, C. C., Messersmith, E. E., & Tominey, S. cognitive deficits in reading disability and attention-deficit/hyper-
(2010). Self-regulation. Integration of cognition and emotion. In W. activity disorder. Journal of Abnormal Psychology, 110, 157–172.
F. Overton (Vol. Ed.) & R. M. Lerner (Ed. in chief), The Handbook Wilson, M., & Knoblich, G. (2005). The case for motor involvement in
of Life-Span Development. Cognition, Biology, and Methods. Vol. 1 perceiving conspecifics. Psychological Bulletin, 131, 460–473.
(pp. 509–553). Hoboken, NJ: Wiley. Winsler, A. (2009). Still talking to ourselves after all these years: A
Meichenbaum, D., & Goodman, J. (1969). Reflection-impulsivity and review of current research on private speech. In A. Winsler, C. Fer-
verbal control of motor behavior. Child Development, 40, 785–797. nyhough & I. Montero (Eds.), Private speech, executive function-
Müller, U., Jacques, S., Brocki, K., & Zelazo, P. (2009). The executive ing, and the development of verbal self-regulation (pp. 3–41).
functions of language in preschool children. In A. Winsler, C. Fer- New York, NY: Cambridge University Press.
nyhough & I. Montero (Eds.), Private speech, executive function- Winsler, A., Diaz, R., & Montero, I. (1997). The role of private speech
ing, and the development of verbal self-regulation (pp. 53–68). in the transition from collaborative to independent task performance
New York, NY: Cambridge University Press. in young children. Early Childhood Research Quarterly, 12, 59–79.
Müller, U., Zelazo, P., Hood, S., Leone, T., & Rohrer, L. (2004). Inter- Winsler, A., Manfra, L., & Diaz, R. M. (2007). ‘‘Should I let them talk?’’
ference control in a new rule use task: Age-related changes, label- Private speech and task performance among preschool children
ing, and attention. Child Development, 75, 1594–1609. with and without behavior problems. Early Childhood Research
Määttä, S., Laakso, M.-L., Tolvanen, A., Ahonen, T., & Aro, T. (2012). Quarterly, 22, 215–231.
Developmental trajectories of early communications skills. Journal Winsler, A., & Naglieri, J. (2003). Overt and covert verbal
of Speech, Language, and Hearing Research, 55, 1083–1096. problem-solving strategies: Developmental trends in use, aware-
Olson, S., & Lunkenheimer, E. (2009). Expanding concepts of ness, and relations with task performance in children aged 5 to
self-regulation to social relationships: Transactional processes in 17. Child Development, 74, 659–678.
the development of early behavioral adjustment. In A. Sameroff Winsler, A., Diaz, R. M., Atencio, D. J., McCarthy, E. M., & Chabay, L.
(Ed), The transactional model of development: How children and A. (2000). Verbal self-regulation over time in preschool children at
contexts shape each other (pp. 55–76). Washington, DC: American risk for attention and behavior problems. Journal of Child Psychol-
Psychological Association. ogy and Psychiatry, 41, 875–886.

Downloaded from jbd.sagepub.com at University of Sydney on August 14, 2015

You might also like