Speech disfluencies in bilingual children An inquiry into the differential diagnosis of stuttering

Abstract

Distinguishing between normal disfluencies and stuttering in bilingual children is a crucial and compelling issue, warranting the attention of both clinicians and researchers. Despite the increasing number of bilinguals, research and clinical practice predominantly focus on monolinguals, leading to a scarcity of empirical data on the development of typical fluency patterns and the manifestation of stuttering in bilingual children. Furthermore, assessment methods that consider the unique characteristics of bilingual language and fluency development are limited, increasing the risk of misdiagnosis.

This thesis aimed to gain comprehensive insights into the manifestation of disfluencies in two groups of bilingual participants: children who stutter (CWS) and children who do not stutter (CWNS). Five main research objectives were specified: (1) to investigate the accuracy of identifying bilingual CWS and CWNS by speech-language pathologists (SLPs) in Lebanon, (2) to examine the extent to which stuttering severity ratings of SLPs to bilingual children are influenced by a child’s stuttering status, (3) to chart the disfluency profiles of bilingual CWNS and CWS in both their dominant and non-dominant languages, (4) to determine which types of disfluencies are most indicative for accurate assessment of stuttering in bilingual children, and (5) to assess whether accuracy of assessment is similar between the dominant language and non-dominant language, and whether accuracy improves when both languages are considered.

The research project involved 92 bilingual Lebanese children aged between 4 and 7 years. Of these, 70 were CWNS and 22 were CWS. Bilingual profiles were determined using the Parents of Bilingual Children Questionnaire. Data collection comprised video-recordings of spontaneous conversations and narrative samples (Frog stories) in the dominant (L1) and non-dominant (L2) languages, resulting in four speech samples per child and 368 speech samples in total. All these speech samples were analyzed and coded for stuttering-like disfluencies (SLD) and other disfluencies (OD). Additionally, 32 SLPs were recruited to evaluate speech samples of eight children, including two CWS and six CWNS, to assess the accuracy of stuttering identification and the extent of misdiagnosis.

Study I revealed that bilingual children where frequently misidentified by Lebanese SLPs with CWNS more often misidentified as CWS than CWS being misidentified as CWNS. Misidentification rates varied within the CWS and CWNS categories as well, with a larger proportion of incorrect identification of CWNS correlating with a higher amount of SLD and the misinterpretation of physical concomitants.

Study II demonstrated that SLPs assigned on average higher stuttering severity ratings to CWS compared to CWNS. However, there is great variation in severity ratings within the CWNS group, mainly depending on the number of SLD a nonstuttering child exhibits. Moreover, individual CWNS ratings occasionally match those of CWS.

Study III showed that, on average, bilingual CWNS exhibited a significantly lower percentage of SLD and iterations than CWS in both languages. Nonetheless, the SLD percentages of CWNS typically surpassed monolingual thresholds. Importantly, the study showed that there is considerable overlap between CWS and CWNS: several CWS exhibited fewer SLD than some CWNS, while, conversely, numerous CWNS displayed more SLD than a number of CWS. Language dominance did not influence the overall frequency and types of disfluencies, although there were some differences for some specific categories (e.g., monosyllabic word repetitions). Dysrhythmic phonation and number of repetitions emerged as reliable predictors for participant group classification (CWS vs. CWNS) in both the dominant and non-dominant language, but combining predictors from both languages resulted in more accurate classification than relying on predictors from just one language.

In conclusion, this research project highlighted that assessing the prevalence and the severity of stuttering among children is not the same for bilingual children as for monolingual children. Bilingual CWNS often demonstrate a relatively high number of SLD and are susceptible to being misdiagnosed as CWS, but CWS may also display a relatively small number of SLD, increasing the risk of not being diagnosed as such. By carefully examining the disfluency profiles in relation to their bilingual status, the accuracy of clinical diagnoses of stuttering in bilingual children can be markedly enhanced.