A Study on Disfluency Characteristics in Non-Stuttering Children for the Development of a Stuttering Prediction Test

Ⅰ. Introduction

Speech is the most fundamental means of human communication, and fluency refers to the ability to produce intended messages at an appropriate rate and rhythm without disruptions to the flow of speech (Guitar, 1998). Fluency serves as the foundation for effective communication, and disruptions in fluency may negatively affect not only children’s everyday interactions but also their emotional well-being and social relationships. In particular, stuttering is a representative fluency disorder that typically begins in early childhood and, in some cases, persists into adulthood. When early intervention is delayed, stuttering may lead to long-term difficulties across multiple domains, including academic performance, self-esteem, and interpersonal relationships (Guitar, 2014; Yairi & Ambrose, 2005).

Stuttering is defined as a disruption in the smooth flow of speech characterized by repetitions, prolongations, and blocks (Wingate, 1964), and it most commonly emerges between the ages of 2 and 5, a period of rapid language development. In developmental disorders, including stuttering, early identification and intervention are considered the most effective strategies for preventing symptom chronicity and long-term persistence (Curlee, 1999). Although many preschool-age children recover spontaneously within one to two years after onset without professional intervention, approximately 20~25% continue to stutter (Yairi, 1997). Thus, childhood stuttering may represent either a transient developmental phenomenon or an early indicator of persistent stuttering extending into adulthood. Even though 75~80% of preschool children recover spontaneously, appropriate caregiver responses and accurate clinical judgment remain critical regardless of recovery outcomes. In particular, children who do not recover spontaneously are at high risk for chronic stuttering, and longer duration of stuttering is associated with prolonged treatment and poorer prognosis. Therefore, deciding whether to monitor preschool children’s disfluency as a normal developmental phenomenon or to initiate early intervention because of elevated risk for persistence constitutes a critical clinical decision. Such decisions must be grounded in sufficient empirical evidence regarding the characteristics of childhood stuttering and the processes of onset and recovery (Curlee, 1999).

A fundamental issue that must be addressed prior to such diagnostic decisions concerns the definition of early stuttering itself. The concept of early stuttering includes two key factors: the child’s age and the duration since stuttering onset. However, many previous studies have inferred the characteristics of childhood stuttering from data on adults who stutter or have failed to adequately control for children’s age and time since onset during participant selection (Yairi et al., 1993). As a result, information specific to early stuttering has remained limited, and the substantial overlap between disfluencies observed in children who stutter and those in typically developing children has made it difficult to clearly distinguish early stuttering from normal developmental disfluency.

Disfluencies observed in typically developing children are considered a natural part of language development and frequently occur during the preschool years, a period marked by rapid expansion of the language system. During this stage, children experience rapid vocabulary growth, increased mean length of utterance, and the production of increasingly complex sentence structures, all of which place temporary demands on speech planning and language organization processes. Consequently, developmental disfluencies in typically developing children often take the form of interjections, revisions, incomplete phrases, phrase repetitions, and pauses, and are generally produced with minimal speech tension or articulatory effort (Conture, 2001; Guitar, 2014; Yairi & Ambrose, 2005). Although the frequency of disfluency varies across individuals, it typically decreases or stabilizes with age, reflecting developmental improvements in speech organization and fluency during the later preschool years (Byun et al., 2004; Kim et al., 2012). In contrast, pathological disfluencies such as blocks, prolongations, and part-word repetitions are rarely observed in typically developing children and are considered core behaviors more prominently exhibited by children who stutter (Byun et al., 2004; Conture, 2001; Wingate, 1964).

Because typically developing children also experience various disfluencies as part of the language acquisition process, these behaviors may superficially resemble early stuttering. However, early stuttering and normal developmental disfluency substantially overlap in both type and frequency, making it difficult for clinicians to clearly differentiate the two phenomena in clinical settings (Conture, 2001). In particular, typically developing children tend to exhibit higher frequencies of normal disfluencies, whereas children who stutter demonstrate more prominent pathological disfluencies, including sound- or syllable-level repetitions, prolongations, and blocks, indicating a qualitative distinction between the two groups (Byun et al., 2004; Seong & Jeon, 2019).

Against this backdrop, there has been increasing demand for the development of assessment tools for the early identification and prediction of stuttering. To achieve this goal, it is essential to clearly delineate the differences between pathological stuttering and normal developmental disfluency. In particular, it is important to establish foundational data through systematic analyses of age-related and context-specific disfluency characteristics in typically developing children. However, much of the existing domestic and international literature has focused primarily on characterizing children who stutter, resulting in a relative lack of quantitative and qualitative data on natural disfluency patterns in typically developing children. Furthermore, many studies have been limited to simple frequency-based analyses or have failed to adequately consider the relationships among age, speaking context, and disfluency characteristics, thereby limiting their practical applicability as indicators for stuttering prediction in clinical settings.

Recent research on fluency screening and assessment has rapidly expanded beyond the simple measurement of disfluency frequency toward multidimensional screening frameworks that simultaneously consider persistence risk factors, language abilities, communication attitudes, caregiver reports, and co-occurring issues. For example, Walsh et al. (2021) suggested that the prediction of persistence in early stuttering requires examination of the relationships among multiple risk factors rather than reliance on a single indicator. Singer et al. (2022) further reported that a cumulative risk approach is useful in predicting the risk of persistent stuttering in young children. Winters and Byrd (2024) showed that caregiver ratings provide important information for understanding children’s communication attitudes. Most recently, Koenraads et al. (2025) reported associations between early language ability and the onset and persistence of stuttering, emphasizing the need for early screening approaches that incorporate language ability from the initial stage of stuttering emergence.

Similar trends are becoming increasingly evident in Korea. Kim et al. (2025) proposed a parent-response-based screening module embedded within a telepractice platform model for childhood stuttering, including case history, communication attitude, disfluency, and parental perception items. In addition, Choi et al. (2025) standardized the Korean version of the Stuttering Behavior Inhibition Scale (SBIS) for children, demonstrating that the assessment of childhood stuttering is expanding beyond speech behaviors alone to include temperament and emotional characteristics. Considering these recent domestic and international research trends, foundational data on the natural disfluency characteristics of typically developing children can serve as a critical benchmark for multidimensional early screening systems. In other words, clearly identifying the developmental characteristics of normal disfluency represents the starting point for constructing assessment systems that enable early detection of persistence risk and linkage to individualized intervention, underscoring the strong academic and clinical timeliness of the present study.

Accordingly, the present study aims to provide foundational data for the development of a stuttering prediction measure by conducting an in-depth analysis of disfluency characteristics in typically developing children according to age and speaking context.

Ⅱ. Methods

Ⅲ. Results

1. Age-related differences in total disfluency ratios

To examine age-related differences in the total disfluency ratio, Welch’s one-way ANOVA was conducted. Descriptive statistics indicated that the mean total disfluency ratio was highest in the 4-year-old group (M=4.21, SD=2.37), followed by the 5-year-old group (M=2.03, SD=1.18) and the 6-year-old group (M=1.95, SD=1.30). However, the differences among age groups were not statistically significant, F_(2,11.4)=2.16, p=.160 (Table 1).

Table 1.

Descriptive statistics for total disfluency ratios across age groups

2. Comparisons between ND and AD ratios within age groups

Paired samples t-tests were conducted to compare the ratios of normal disfluency (ND) and abnormal disfluency (AD) within each age group.

The results showed that, in the 4-year-old group, the ND ratio (M=4.06, SD=2.41) was significantly higher than the AD ratio (M=0.15, SD=0.16), t(6)=-4.21, p=.006. Similarly, in the 5-year-old group, the ND ratio (M=2.03, SD=1.18) was significantly higher than the AD ratio (M=0.12, SD=0.17), t(6)=-4.15, p=.006.

In contrast, in the 6-year-old group, the difference between the ND ratio (M=3.28, SD=3.81) and the AD ratio (M=0.56, SD=0.86) was not statistically significant, t(6)=-2.12, p=.078 (Table 2).

Table 2.

Comparisons between ND and AD ratios within age groups

3. Differences among disfluency type

Comparisons of disfluency forms across age groups showed distinct developmental changes for each type. First, part-word repetitions gradually increased with age and were highest in the 6-year-old group. In contrast, interjections occurred most frequently in the 4-year-old group and gradually decreased in the 5- and 6-year-old groups.

Revisions and hesitations generally remained at low frequencies across all age groups, although both increased slightly in the 5-year-old group. Incomplete phrases occurred at a very high frequency in the 4-year-old group, but decreased markedly after age 5 and then remained relatively stable. Finally, word and phrase repetitions were highest in the 4-year-old group, decreased sharply in the 5-year-old group, and then increased slightly again in the 6-year-old group. In contrast, no prolongations or blocks were observed in any age group (Table 3).

Table 3.

Age-related differences in disfluency forms

Ⅳ. Discussion and Conclusion

The present study examined disfluency in preschool children from multiple perspectives, including age-related changes in total disfluency ratios, differences between normal disfluency (ND) and abnormal disfluency (AD), and variations across disfluency forms. This multidimensional approach reflects the understanding that fluency development in children is a complex process that cannot be adequately explained by a single index alone. In particular, the preschool years are characterized by rapid changes in linguistic, cognitive, and speech-planning abilities, during which both the frequency and forms of disfluency serve as important indicators of developmental change. Accordingly, the present discussion interprets the findings with a focus on the characteristics of fluency development and their developmental implications.

First, age-related differences in total disfluency ratios were examined among 4-, 5-, and 6-year-old children, and no statistically significant differences were found across age groups. However, descriptive statistics indicated that the 4-year-old group showed relatively higher disfluency levels than the 5- and 6-year-old groups. This finding suggests that disfluency may occur more frequently during the earlier stages of preschool development and is consistent with previous domestic findings on developmental characteristics of normal disfluency in early childhood (Byun et al., 2004). Nevertheless, because the sample size in each age group was small and the 4-year-old group showed relatively greater variability, this descriptive difference should be interpreted with caution. In other words, the present finding may reflect the individual variability commonly observed in normal disfluency rather than a stable developmental group difference. In addition, previous studies with preschool children have reported that disfluency increases as discourse demands or task complexity increase (Kim et al., 2012). In this context, the relatively higher disfluency levels descriptively observed in the 4-year-old group may reflect differences in language processing demands associated with developmental stage. Future studies using larger samples and longitudinal designs are needed to examine developmental changes in preschool disfluency more precisely.

Second, comparisons of ND and AD ratios within each age group revealed statistically significant differences in the 4- and 5-year-old groups, whereas no significant difference was found in the 6-year-old group. These findings indicate that ND and AD are qualitatively distinguishable phenomena throughout much of the preschool period. Specifically, ND levels were significantly higher than AD levels in the younger age groups, reflecting the fact that normal disfluencies such as repetitions, revisions, and interjections commonly emerge during speech adjustment processes, whereas AD consists of core disfluency behaviors that directly disrupt speech flow. These findings are consistent with previous domestic research showing that normal and pathological disfluencies differ not only in frequency but also in qualitative composition (Byun et al., 2004). In contrast, the absence of a statistically significant difference in the 6-year-old group may be related to the relatively large within-group variability at this age, suggesting that individual fluency characteristics become more diverse in later preschool years. Previous studies have emphasized that analyses of disfluency types may serve as more sensitive indicators for early stuttering identification than simple frequency measures alone. Typically developing children and children who stutter have been reported to differ qualitatively in repetition type, utterance position, and duration (Kim et al., 2012). These findings further support the importance of the ND-AD distinction confirmed in the present study.

Third, the analysis of disfluency forms across age groups revealed clear developmental changes in both the composition and frequency of disfluency types. This finding suggests that preschool disfluency is not a unitary phenomenon but varies according to developmental stage. Interjections and incomplete phrases occurred at relatively high frequencies in younger children but gradually decreased with age. The particularly high occurrence of these forms in the 4-year-old group may reflect immaturity in speech planning and sentence formulation abilities. In contrast, part-word repetitions increased with age and were highest in the 6-year-old group. This may suggest that as utterance length and overall linguistic output increase with age, repetition is more frequently used as a speech adjustment strategy. In particular, the large standard deviation observed in part-word repetitions among 6-year-olds suggests that within-group variability increased substantially, possibly influenced by one or two children who produced markedly high repetition frequencies. This result indicates that repetitions may reflect not only developmental changes but also individual characteristics. Revisions and hesitations remained relatively infrequent across ages, and no clear age-related trend was observed. This suggests that these forms may be influenced more by individual speaking style or task characteristics than by developmental factors alone. Word and phrase repetitions were highest in the 4-year-old group, decreased in the 5-year-old group, and then showed a slight increase again in the 6-year-old group. This may reflect a developmental shift from reliance on short-unit repetitions toward more organized speech strategies. In contrast, no prolongations or blocks were observed in any age group, indicating that these core disfluency behaviors are rarely found in typically developing preschool children.

Acknowledgments

This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2022S1A5A2A01046610).

References

• Byun, J. W., Shim, H. S., & Lee, E. J. (2004). Disfluency characteristics of children with early stuttering. Korean Journal of Communication Disorders, 9(1), 1-14. uci:G704-000725.2004.9.1.003
• Choi, D., Kim, S., Shim, H. S., Lee, S. B., Kang, H. W., & Ntourou, K. (2025). Standardizing the short behavioral inhibition scale for Korean children who stutter. Communication Sciences & Disorders, 30(3), 568-578. [https://doi.org/10.12963/csd.250120]
• Conture, E. G. (2001). Stuttering: Its nature, diagnosis, and treatment. Boston, MA: Allyn and Bacon.
• Curlee, R. F. (1999). Stuttering and related disorders of fluency. New York, NY: Thieme.
• Guitar, B. (1998). Stuttering: An integrated approach to its nature and treatment (2nd ed.). Philadelphia, PA: Lippincott Williams & Wilkins.
• Guitar, B. (2014). Stuttering: An integrated approach to its nature and treatment (4th ed.). Philadelphia, PA: Lippincott Williams & Wilkins.
• Kim, H. J., Shin, M. S., & Chang, H. J. (2025). Development of a stuttering tele-practice platform model for stuttering in children. Journal of Speech-Language & Hearing Disorders, 34(3), 87-95. [https://doi.org/10.15724/jslhd.2025.34.3.087]
• Kim, S. H., Jeon, H. S., & Kwon, D. H. (2012). Characteristics of disfluency according to discourse type in preschool children. Journal of Speech-Language & Hearing Disorders, 21(3), 53-68. [https://doi.org/10.15724/jslhd.2012.21.3.004]
• Kim, Y. T., Sung, T. J., & Lee, Y. K. (2003). Preschool Receptive-Expressive Language Scale (PRES). Seoul: Seoul Community Rehabilitation Center.
• Koenraads, S. P. C., Jansen, P. W., Labuschagne, J. E., van der Schroeff, M. P., & Franken, M. C. (2025). Risk of stuttering onset and persistence linked to early language skills: Results from the generation R study. Journal of Fluency Disorders, 85, 106145. [https://doi.org/10.1016/j.jfludis.2025.106145]
• Lee, Y., Lee, H., & Choi, J. (2018). A study of validity and reliability of the CSBS DP behavior sample in Korean toddlers. Communication Sciences & Disorders, 23(2), 539–548. [https://doi.org/10.12963/csd.18514]
• Seong, J. S., & Jeon, H. S. (2019). Comparison of disfluency and speech rates in preschool-age children who stutter according to the linguistic units. Journal of Speech-Language & Hearing Disorders, 28(4), 39-47. [https://doi.org/10.15724/jslhd.2019.28.4.039]
• Shim, H., Shin, M., & Lee, E. (2004). Paradise-Fluency Assessment. Seoul: Paradise Welfare Foundation.
• Singer, C. M., Otieno, S., Chang, S.-E., & Jones, R. M. (2022). Predicting persistent developmental stuttering using a cumulative risk approach. Journal of Speech, Language, and Hearing Research, 65(1), 70-95. [https://doi.org/10.1044/2021_JSLHR-21-00162]
• Walsh, B., Christ, S., & Weber, C. (2021). Exploring relationships among risk factors for persistence in early childhood stuttering. Journal of Speech, Language, and Hearing Research, 64(8), 2909-2927. [https://doi.org/10.1044/2021_JSLHR-21-00034]
• Wingate, M. E. (1964). A standard definition of stuttering. Journal of Speech and Hearing Disorders, 29(4), 484-489. [https://doi.org/10.1044/jshd.2904.484]
• Winters, K. L., & Byrd, C. T. (2024). Caregiver predictions of their 3- to 6-year-old child who stutters’ communication attitude. Journal of Speech, Language, and Hearing Research, 67(7), 2086-2105. [https://doi.org/10.1044/2024_JSLHR-23-00662]
• Yairi, E. (1997). Home environment and parent-child interaction in childhood stuttering. Boston, MA: Allyn & Bacon.
• Yairi, E., & Ambrose, N. G. (2005). Early childhood stuttering for clinicians by clinicians. Austin, TX: Pro-Ed.
• Yairi, E., Ambrose, N., & Niermann, R. (1993). The early months of stuttering: A developmental study. Journal of Speech and Hearing Research, 36(3), 521–528. [https://doi.org/10.1044/jshr.3603.521]
• Yim, D., Park, W., Cheon, S., Lee, Y., & Lee, J. (2015). Interaction skills via spontaneous speech sample analysis: 2- to 5-year-old children with and without language impairment. Communication Sciences & Disorders, 20(3), 364–373. [https://doi.org/10.12963/csd.15240]