Introduction

Developmental stuttering is a speech disorder characterized by the repetition, prolongation, and blocking of speech sounds. In the International Classification of Diseases 11th by the World Health Organization, it is coded as “6A01.1 Developmental speech fluency disorder.” Typically, the onset of stuttering occurs between 2 and 6 years old1. The incidence of stuttering is estimated to be about 5–8%, but 70–80% of children spontaneously recover within 3 years of onset2, leading to a prevalence of approximately 1%3.

The characteristics of preschool children who stutter (CWS) have been examined in terms of language developmental factors. A meta-analysis4 revealed that although preschool CWS have language abilities within the typical range, they tend to perform lower on overall language measures compared to preschool children who do not stutter (CWNS), although some studies have reported no significant differences5. Regarding phonological and articulatory development, numerous studies reported that CWS show delays, suggesting an association with recovery5,6. However, language developmental factors alone do not clearly distinguish CWS from CWNS.

The characteristics of CWS beyond language development have also been investigated, particularly concerning emotional and attentional features. CWS have been reported to have more difficulty controlling their emotional responses than CWNS7,8. In terms of attentional capacity, CWS have been reported to have difficulty with divided attention9,10 and attention regulation11. The accuracy and efficiency of CWS compared with CWNS have also been highlighted in these areas.

As described above, emotional regulation, along with behavioral regulation, is one of the core components of executive function12. Divided attention, which is a function of higher-order attention, is associated with executive function and working memory13. Thus, the specificity of executive function can be inferred from the background of the emotional regulation and attentional performance characteristics noted in CWS. Executive function is defined as “goal-oriented thought, action, and emotion regulation”14. Executive function development in early childhood has been reported to be a possible predictor of social success and health status15, and it has attracted attention in many fields related to child development. Miyake et al.16 reported that executive function consists of three factors: shifting, updating (working memory), and inhibition. Shifting refers to the flexible switching of tasks, updating (working memory) refers to a system that monitors and updates information held in working memory, and inhibition refers to the suppression of dominant behaviors and thoughts. In recent years, executive functions have also been categorized into “cool” and “hot”12. Cool executive function refers to the ability to function in neutral, noncontextual situations, whereas hot executive function involves controlling instinctive demands and emotions to achieve goals in emotional situations. Furthermore, it is characterized by its rapid development in early childhood. Notably, the present study also aims to investigate these “cool” EF skills in preschool CWS.

The behavioral and neuroscientific characteristics of CWS suggest the need for further investigation into the executive functioning characteristics of CWS. Behaviorally, several studies have reported that people who stutter have a higher rate of comorbid ADHD17,18. ADHD has been identified as one of the main factors contributing to poor executive functioning19, which may be another reason for the increased interest in executive functioning in CWS. Moreover, neurological findings support a relationship with executive function. Chang et al.20 noted that in young children with persistent stuttering, the frontoparietal network, which is largely associated with executive function, has abnormal connections with the default mode network. The executive function and default mode networks are inversely correlated, suggesting that in CWS, weakness in executive function may lead to excessive interference of the default mode network21.

The association between stuttering and executive function has also been explored. Cognitive aspects of executive function have been suggested to influence speech fluency by affecting resource allocation during speech production planning and language production. For example, it has been reported that inhibition and weak working memory may reduce the stability of phonological and/or lexical representations of words in the mental lexicon. Consequently, insufficient attentional resources may be allocated for fluent language production, thereby affecting stuttering22. From the emotional aspect of executive function, the fact that CWS have higher emotional reactivity than CWNS and that stuttering are more severe during sympathetic hyperarousal23 suggests that executive function affects difficulties in emotional regulation. It has also been suggested that stuttering symptoms may occur as a result of inadequate attention allocation to speech motor control due to insufficient working memory, a component of executive function24. The demands and capacities model (DCM)25, a theoretical account of childhood stuttering, and the multifactorial dynamic pathways theory (MDPT)26, as well as the dual diathesis-stressor model of emotional and linguistic contributions to developmental stuttering27, also address emotion as a factor influencing stuttering. Furthermore, in the RESTART-DCM method, an approach to treating CWS based on the DCM25, language development and emotion are considered to be two of the four speech-related factors associated with the occurrence of stuttering. This approach encourages planning for the child’s fluency by considering the child’s language development and ensuring that language production is guided by the length and content of speech that is less stressful for the caregiver. The RESTART-DCM method has been demonstrated to be as effective as the Lidcombe Program, an approach with a high level of evidence in early childhood28, suggesting that emotional regulation may influence stuttering.

The previous studies outlined above suggest a relationship between the characteristics of CWS, the occurrence of stuttering, and executive function. In the following sections, we will review studies that have measured the executive function of CWS and attempt to analyze their characteristics. Ofoe et al.29 conducted a meta-analysis of the executive function of CWS, examining its subcomponents: shifting, verbal short-term memory, and inhibition. The results indicated that verbal short-term memory performance is lower in CWS. However, as this study included subjects aged 3 to 18 years old, it is necessary to separate the analysis into early childhood and later periods, as executive function rapidly develops in early childhood and the pathophysiology of stuttering may differ between these periods. Kakuta and Kawasaki30 conducted a review focusing on the executive function of preschool CWS. The results indicated that, despite some disagreements, several studies reported significantly lower working memory performance in CWS, and a few studies reported lower performance in other subcomponents of executive function. However, most reports on working memory focused primarily on verbal short-term memory, with visual tasks being insufficiently investigated. Therefore, future studies should determine which subcomponents of working memory are characterized by specific features in CWS. In addition, previous studies often focused on individual subcomponents, such as working memory or inhibitory function, and few have provided an integrated examination of executive function. There are studies that have comprehensively examined EF in preschool CWS, notably those by Ntourou et al.31 and Choo et al.32. Ntourou et al.31 examined whether EF differs between preschool CWS and preschool CWNS and found that, for behavioral measures, only three-year-old CWS exhibited significantly lower scores, while parent ratings indicated that CWS performed significantly lower on the Working Memory and Shift scales. Choo et al.32 used data from the U.S. National Health Interview Survey (NHIS) to investigate the EF of preschool CWS and preschool CWNS and its association with comorbidities. The results showed that CWS performed lower on EF than CWNS, and that comorbidities were also related to EF. These findings suggest that preschool CWS may show lower EF performance compared to their typically developing peers.

Based on previous studies, we hypothesize that executive functions are involved in the derivation of fluent speech by interfering with speech motor control. We also propose that environmental factors may act as moderating variables. As a first step, we conducted an age- and gender-matched case–control study to comprehensively examine executive function through behavioral observation, with the aim of characterizing the overall executive function of CWS compared with CWNS. We hypothesize that the decline in executive function observed in previous studies cannot be explained solely by a decline in verbal working memory and that other components of working memory and other subcomponents of executive function may contribute to the observed decline in performance.

Methods

Participants

This study included 40 preschool children (20 CWS, 20 CWNS) aged 5:00 to 6:11 years. The demographic characteristics of the participants are shown in Table 1. All the participants were native monolingual speakers of Japanese, and none had any difficulties with hearing, development, or intellectual functions per parental report and well-child visits.

Table 1 Demographics characteristics of participants.
Full size table

Talker group matching

The participants in the two talker groups were matched using the same criteria as those of Anderson, Wagovich, and Hall33, ensuring that the age difference between the CWS and CWNS was within ± 4 months and that the gender distribution was identical. The mean age was 67.3 months (standard deviation [SD] = 2.94) for the CWS and 68.85 months (SD = 3.27) for the CWNS, with no significant difference between the groups (P > .05)..

Group classification criteria

The rate of stuttering core behaviors during a free conversation with an examiner (involving 50 or more clauses) and during the picture description tasks of the Stuttering Test (preschooler version)34 was calculated to classify the children into groups. The CWS group consisted of children with stuttering core behaviors occurring in 3% or more of their speech, whereas the CWNS group included children with < 3% stuttering core behaviors and no continuous stuttering symptoms reported by their parents. The Stuttering Inventory was developed to evaluate the stuttering core behaviors in Japanese speakers and is available in three versions: infant version, primary school version, and junior high school and above version. The preschool version includes free conversation with the examiner and sentence/picture description task as essential test items. Additional tests include free conversation with family members, question-answering, and word repetition. Stuttering core behaviors are defined as sound, mora, and syllable repetition, part-word repetition, prolongation, and blocking. The rate of stuttering core behaviors is calculated as the number of stuttering core behaviors divided by the total number of clauses uttered. The results of Welch’s t-test for the Stuttering Test indicated significant differences between the talker groups (P < .05)..

Development and Language criteria

The Kinder Infant Development Scale (KIDS) Type C35 and the Picture Vocabulary Test-Revised (PVT-R)36 were administered to confirm that the children were developing typically and that the tasks in this study were appropriate. The KIDS Type C is a parent-completed questionnaire that focuses on a child’s mental and physical development. It consists of 133 questions answered using a two-choice system (○ = clearly able, able in the past, never done but able if given the chance; × = clearly not able, not sure as never done). The scores are calculated for developmental age, developmental index, and specific domains. In this study, the scores of the subscales were matched to the developmental profile, and the developmental age was calculated. The PVT-R is a test that evaluates vocabulary development by asking the child to select from a set of four illustrations corresponding to a vocabulary word presented in an audio recording. Standard scores are calculated based on the number of correct and incorrect responses. A multivariate analysis of variance with age in months as a covariate for the aforementioned tests revealed no differences between the groups (P > .05) (Table 2).

Table 2 Mean standard scores (standard deviation) on development and Language tests for CWS and CWNS.
Full size table

Procedure

Data collection was performed as follows: the child was administered with the PVT-R by the examiner, followed by the Stuttering Test (preschooler version), and then child carried out the executive function tasks. The executive function tasks are described in more detail below. The KIDS scale was completed by the parents during the child’s performance of the tasks or the day before.

Materials

Executive function task

The executive function tasks measured three factors, namely, updating (working memory), inhibition, and cognitive flexibility, as outlined by Miyake et al.16. Working memory was measured separately for verbal and visual working memories. Verbal working memory was evaluated using a digit span test, visual working memory using a visual counting span task, inhibition using a black-and-white task, and cognitive flexibility using a Simon Says task.

(a) Verbal working memory: digit span test.

The digits 1–9 were presented aurally at a rate of one per second, and the participants were required to orally repeat them in reverse order (e.g., stimulus “2 − 1,” correct response “1–2”). A practice trial with two digits was conducted, and the main trial began after two correct responses. The numbers presented ranged from 2 to 5 digits. For each digit length, three trials were conducted, and if the participant provided a correct response in at least one trial, the next digit length was administered. One point was awarded for each correct response, with a maximum score of 12.

(b) Visual working memory: visual counting span test.

This task was classified as a Counting Span Test37. The participants were presented with a sheet displaying two types of objects (e.g., a star and a sun) and were asked to count one type of object. The maximum number of objects to be counted was five. The experimenter then turned over the sheets one by one, and the participants were asked to recall the number of objects on the sheets in the correct order. In the practice trial, the participants practiced counting and remembering the objects on two sheets. This trial involved two to five digits. Three trials were conducted for each digit length, and if the participant correctly answered in at least one trial, the next digit length was administered. One point was awarded for each correct response, with a maximum score of 12.

(c) Inhibition: black-and-white task.

The experiment was conducted based on Moriguchi38. The participants were instructed to respond “white” when shown a black card and “black” when shown a white card. Two practice trials were given. If the participants answered both trials correctly, they proceeded to the main trial. If an incorrect response was given during the practice phase, the task instructions were repeated. The main trial consisted of 16 trials, during which the experimenter randomly presented the white and black cards once per second, and the participants were asked to respond as described. One point was awarded for each correct response, with a maximum of 16 points.

(d) Cognitive flexibility task: Simon says.

The traditional “Simon Say” game was adapted39. The participants were told they would play a game where they must touch different parts of their body. If the instruction began with “Simon says,” they were to touch the part as instructed; otherwise, they were to touch a different part. Two practice trials were conducted: “Simon says, touch your head” and “touch your foot.” If the participant correctly answered both trials, the test proceeded. If an incorrect response was given during the practice phase, the task instructions were repeated until two consecutive correct responses were obtained. The main trial involved 10 trials of “Simon says” (noninhibition) or not (inhibition), presented in a random order. One point was awarded for each correct response, with 5 points each for noninhibition and inhibition, with a maximum of 10 points.

Data analysis

This study used the scores obtained from each task measuring executive function as the dependent variable to examine differences between the groups. The scores for verbal working memory, visual working memory, cognitive flexibility, and inhibition were included in the analysis. Before the data analysis, the scores from the executive function tasks were checked for normality to determine if they met the assumptions of parametric statistics. Normality was evaluated using the Shapiro–Wilk test and histograms. Group differences were examined using the Mann–Whitney U test, a nonparametric method, for tasks where normality could not be confirmed. Levene’s test was employed to examine equal variances for tasks where normality was confirmed. ANCOVA was conducted to examine differences between the groups for tasks where equal variances were confirmed. As it has been reported that executive function performance tends to increase with age40, “age in months” was used as a covariate. Welch’s t-test was conducted for tasks where equal variances were not confirmed. For statistical analysis, HAD41 was used.

Ethical statements

This study was conducted in full compliance with ethical standards. Approval was obtained from the Ethical Review Committee of the National Rehabilitation Center for Persons with Disabilities (2023-006). Informed Consent was obtained from the parents or guardians of the children participating in this study. We ensured that the children understood the tasks by providing examples during implementation. After confirming their willingness to participate, the trials were conducted.

Results

Descriptive statistics and histograms of executive function tasks for each group

The descriptive statistics of the executive function tasks for each group are presented in Table 3. The histograms of the results of each group’s performance on the executive function tasks are shown in Fig. 1. Regarding the results, the verbal working memory task did not violate the assumptions of normality or homogeneity of variances (P > .05). Therefore, an ANCOVA was conducted for this task. On the other hand, normality was violated in the visual working memory and inhibition tasks for the CWS group, and in the cognitive flexibility task for the CWNS group (P < .05). As a result, the visual working memory task, the inhibition task, and the cognitive flexibility task were considered non-normally distributed. For these tasks, the Mann–Whitney U test was used in the subsequent analyses.

Table 3 Mean standard scores (standard deviation) on executive function tasks tests for CWS and CWNS.
Full size table
Fig. 1
figure 1

Histogram of the results of the executive function tasks in each group.

Full size image

Differences in performance on the executive function task

The Analysis of Covariance (ANCOVA) results for the number of correct responses on the verbal working memory task indicated a significant main effect (F(1, 37) = 5.11, P = .03, partial η2 = 0.12) due to talkers (CWS: mean = 2.26, standard error [SE] = 0.34; CWNS: mean = 3.38, SE = 0.34). The chronological age covariate was also significant (F(1, 37) = 16.01, P < .01, partial η2 = 0.30).

In the visual working memory task, CWS scored significantly lower than CWNS, as revealed by a Mann–Whitney U test (Mdn = 3.0 vs. 4.0, U = 93.0, P = .003, r = .48). For the inhibition task, the Mann–Whitney U test showed no significant difference between CWS and CWNS (Mdn = 13.0 vs. 12.0, U = 195.5, P = .91, r = –.02). Regarding the cognitive flexibility task, no statistically significant group difference was found (Mdn = 7.0 vs. 7.5, U = 172.5, P = .46, r = .12), based on the results of the Mann–Whitney U test.

Discussion

The present study comprehensively investigated the characteristics of executive function in CWS. The results indicated that as initially hypothesized, the CWS exhibited lower performance in verbal and visual working memories. In the following, we will discuss the association between executive function and the occurrence of stuttering symptoms based on the results of the working memory task, in which the CWS had significantly lower performance than the CWNS, and cognitive flexibility and inhibition, in which both groups showed no significant differences in terms of performance.

First, although Baddeley’s model was not directly tested in this study, the low performance of the CWS in the working memory tasks could be interpreted in terms of the working memory model proposed by Baddeley42. In this model, under the central executive function, which is responsible for the control of attention, there are three components: the phonological loop, which stores verbal information; the visuospatial sketchpad, which stores visual information; and the episodic buffer, which retrieves information from long-term memory. The central executive function and phonological loop are classified as verbal working memory and the visuospatial sketchpad together as visual working memory. In the present study, the CWS had significantly lower performance in the verbal and visual working memory tasks (P = .04, partial η2 = 0.11). Most of the studies that investigated the executive function of young CWS have examined verbal working memory through a number-singing or nonword-recitation task. As a result, many studies reported that the performance of the CWS was significantly lower than that of the CWNS. However, young CWS have also been found to have delayed phonological development. Delayed phonological development may lead to difficulties in performing linguistic tasks and, consequently, to smaller verbal short-term memory capacity. In other words, the low verbal working memory capacity of CWS may be a confounded result of slow phonological development6. In the present study, visual working memory, which is less affected by phonological development, was also measured, suggesting that the performance in the verbal and visual working memory tasks was low. This result suggests that the observed low performance in working memory might not be solely explained by deficits in specific subcomponents, such as phonological storage or visuospatial abilities alone, but may involve broader attentional control, potentially implicating the central executive system. However, this interpretation must be viewed with caution because the present study found no statistically significant differences between CWS and CWNS in inhibition and cognitive flexibility. According to Miyake’s model, these two components are also closely related to EF. The absence of differences in these tasks raises the possibility that either the central executive deficits in CWS are domain-specific or that the measures used in this study may not fully capture the complexity of EF in CWS. Future research employing more diverse measures of EF is necessary to clarify this issue.

The current findings provide preliminary behavioral evidence that attentional control capacities associated with working memory could be compromised in young CWS, partially aligning with findings in adults who stutter43. Previous studies indicate that many CWS exhibit learning disabilities (LD)44, which have been linked to verbal and visual working memory impairments45,46. It is therefore plausible, though speculative at this stage, that reduced working memory and attentional control reflect inherent cognitive vulnerabilities in stuttering populations rather than consequences of persistent stuttering.

No significant differences in inhibition and cognitive flexibility were observed between the CWS and CWNS. Regarding the results of inhibition, it is possible that the bimodal distribution of the black-and-white task performance of the stuttering group had influenced the results, and thus, caution should be taken in the interpretation of these results. The bimodal distribution was observed only in CWS, suggesting that there may be a group of CWS who are extremely poor at inhibition, even among CWS who are typically developing. However, although no ceiling effect was observed, the black-and-white task used in this study targeted children older than those in previous research (3–4 years old)38, and thus, the absence of group differences may be due to lower task difficulty. Future studies should consider using more challenging inhibition tasks to better capture individual differences in executive function among CWS. Furthermore, it has been suggested that the task of inhibition may be compensated for by the use of self-talk47 in late infancy due to the development of internalization of languages. The subjects in this study were also in late infancy and may have been affected by language. This point has also been suggested by Anderson & Wagovich48, who examined differences between CWS and CWNS using an explicit task involving suppression of explicit verbal stimuli (grass-snow) and an implicit task involving suppression of implicit verbal stimuli (baa-meow). Their results showed that CWS had slower responses than CWNS on both explicit and implicit verbal response inhibition tasks and also demonstrated lower accuracy on the implicit task. These results indicated that the performance was better with explicit verbal stimuli. Therefore, it is possible that the number of correct responses was compensated for by other factors, such as language ability.

Additionally, in the present study, there was also no difference between CWS and CWNS in cognitive flexibility49,50,51. Several studies have examined cognitive flexibility among age-matched children who stutter. Eichorn et al.49 and Ofoe & Anderson50 reported significantly lower performance in CWS. Paphiti & Eggers51 found no differences among preschool-age children; however, among school-age children (above 7 years), CWS exhibited slower responses and more errors in set-shifting tasks compared to CWNS. These previous studies used tasks measuring response time as an outcome. In contrast, the current study employed tasks focusing solely on accuracy, with minimal pressure for rapid responses, potentially influencing the outcomes. Given that results vary depending on task characteristics, further research is necessary to clarify the cognitive flexibility characteristics of CWS.

Hypotheses regarding the association between executive function and the occurrence of stuttering based on the results of this study will be discussed. DCM25 and MDPT26 are representative hypotheses regarding the occurrence of stuttering symptoms in early childhood. These hypotheses postulate that stuttering symptoms occur when the child’s capacities are unable to meet internal and external demands during the rapid development of speech-related factors in early childhood. Speech-related factors include speech motor, language, social–emotional, and cognitive abilities. Social–emotional abilities are assumed to correspond to executive functions. Among these factors are language (e.g., vocabulary52, speech length53 and executive function54 are a period of rapid development. In this context, there is no significant difference in language development between CWS and CWNS5. The present study suggests that subcomponent performance may be significantly lower with respect to executive functioning, which is highly associated with social–emotional competence. Thus, it is possible that CWS have an imbalance between language development and executive function development in early childhood, which may be a factor in the occurrence of stuttering. Further discussion of the interaction of these factors suggests that speech motor development, which is considered to be necessary in DCM and MDPT, is slower in 4- and 5-year-old toddlers who stutter than in CWNS, indicating that they have difficulty with consistent speech motor development55,56. In addition, speech movements are influenced by linguistic and emotional factors. Regarding linguistic factors, speech movements are more unstable and are influenced by increased utterance length and word novelty55. Emotional factors also influence speech motor control57, and the degree of variability is greater in young children with persistent stuttering58. Based on these findings, it is possible that linguistic and emotional factors each influence speech motor control or are related to each other, which results in the development of stuttering symptoms. Moreover, executive function is the basis for the development of language and emotional control, which may indirectly affect these factors.

Finally, we discuss some issues to be addressed in this study.

The first is the method employed to measure executive function. The executive function tasks used in this study include those that require responses in speech. CWS may have felt anxious or stressed when responding with speech, which may have affected their performance on the task. In particular, the tasks used to measure verbal and visual working memories, which showed significant differences in this study, required responses in speech. In the future, it will be necessary to examine the performance in tasks that do not involve verbal responses. Furthermore, in this study, only the performance in the task was the dependent variable, and we were not able to assess the anxiety and effort required to perform the task, as described above. Even if the task performance does not change, an increase in the physiological indices of sympathetic activity that reflect anxiety and effortfulness may be observed. It would be desirable to examine executive function from the perspective of effortfulness in tasks without speech stress.

The second is the executive function model used in this study. In this study, we attempted to measure the executive function of CWS using Miyake’s model. There are other executive function models, such as that of Zelazo et al.59, which focuses on selective attention and planning. Miyake’s model16 was based on a model of children with normal development. Therefore, it is necessary to determine whether the model used in this study is suitable as a framework of executive function for children with neurodevelopmental disorders, such as stuttering and communicative disorders. Indeed, Cirino et al.60 reported that Miyake’s model was not suitable for elementary school students with reading disabilities. This study conducted a factor analysis of the components of executive function and extracted five specific executive function factors: working memory span, manipulation and planning, working memory updating, generative fluency, self-regulated learning, and metacognition. The components of executive function may differ among CWS and those who have communication difficulties due to speech impediments. In this study, we have elucidated the characteristics of executive function in CWS compared with normal children. However, to examine the executive functions required in the lives of young CWS, we believe that it is necessary to reexamine the subcomponents necessary for “goal-directed control of thoughts, actions, and impulses” in young CWS in the future.