Skip to main content
NCBI home page
Journal of Speech, Language, and Hearing Research : JSLHR logoLink to Journal of Speech, Language, and Hearing Research : JSLHR
. 2026 Mar 12;69(4):1804–1813. doi: 10.1044/2025_JSLHR-24-00771

Racial Identity Perception, Social Affiliation, and Sentence Repetition in Children

HaeJi Lee a,✉, Elizabeth Ancel a, Kerry Ebert a, Benjamin Munson a
PMCID: PMC13086190  PMID: 41820020

Abstract

Purpose:

Speech-language pathologists (SLPs) often use sentence repetition (SR) tasks to screen children for language disorders. The influence of talker-specific sociolinguistic factors on SR performances in children remains unexplored. This study examined the influence of two sociolinguistic variables on SR task performance: whether the children's racial identity matched that of the person being repeated and whether the children perceived themselves to be socially affiliated with the person being repeated.

Method:

A total of 134 four- to 8-year-old children (55 boys and 79 girls) were tested at two public events. Each child completed an SR task with prompts produced by a Black, East Asian, and White talker and answered three questions about their perceived social affiliation with each talker. Generalized logit mixed-effects models were used to examine the relationship among predictor variables of age, racial identity match, and perceived social affiliation and the outcome variable of SR accuracy.

Results:

Overall, 67% of sentences were repeated back accurately. The Black talker and the East Asian talker were associated with more accurate SR compared to the White talker. A match between the racial identity of the child and the talker was associated with lower SR scores. No significant effect was found between SR scores and social affiliation questions. Age was a consistent predictor of SR scores; however, no evidence was found for the interaction terms with either racial identity or social affiliation questions.

Conclusions:

The differences in SR accuracy across the three talkers suggest the potential for clinically significant differences depending on which talkers are used in assessing SR. More diverse child samples and talker samples are needed to rigorously assess the hypothesis that racial match or mismatch affects SR accuracy.

Supplemental Material:

https://doi.org/10.23641/asha.31446301

Children observe, learn, and mimic various skills from adults. In speech-language pathology, research has examined different individual factors among children that influence assessments and treatments, yet relatively few studies have considered how clinician-specific factors influence assessments and treatments. This study investigated the relationship between children's performance on a language assessment task, sentence repetition (SR), and characteristics of the talker presenting the sentences. Sociolinguistic factors related to the talker may influence children's task performance. This study specifically considered the impact of the talker's race as well as children's perceived affiliation with the talker.

SR Tasks and Developmental Language Disorder

SR tasks involve children listening to an adult model a sentence and repeating the sentence verbatim. SR tasks tap into many different abilities, requiring a child to hear and encode a sentence, hold the sentence in memory, and repeat grammatical structures and vocabulary accurately (Pratt et al., 2021). The score is based on the number of errors the child makes in their production of words and grammatical elements. Speech-language pathologists (SLPs) often use SR tasks as a screening tool or as part of a comprehensive assessment of language (e.g., the Clinical Evaluation of Language Fundamentals–Fifth Edition [CELF-5]; Wiig et al., 2013) to identify children with developmental language disorder (DLD). DLD is one of the most common neurodevelopmental disorders, impacting approximately 7% of the population (Norbury et al., 2016; Tomblin et al., 1997). Previous research has shown that performance on SR tasks distinguishes between children with and without DLD with high sensitivity and specificity (Pham & Ebert, 2020; Vang Christensen, 2019; Wang et al., 2022), which indicates that the SR task is an overall promising measure for identifying DLD. Establishing efficient, sensitive, and culturally responsive ways of identifying DLD is critical to ensure that the diverse population of children with DLD receives appropriate intervention services (Ebert et al., 2020; Rujas et al., 2021).

Researchers have demonstrated that the SR task is effective for identifying DLD across a variety of languages (see Rujas et al., 2021, for a review). For example, in a recent study, Wang et al. (2022) found that Mandarin SR accuracy differentiated children with and without DLD with high sensitivity. Furthermore, Friesen et al. (2022) and Pratt et al. (2021) demonstrated that SR tasks are both reliable measures of DLD with both monolingual and bilingual children and are correlated with individual differences in bilingual language exposure, memory, and vocabulary skills.

The Impact of Sociolinguistic Factors on Children

Researchers have investigated children's individual factors but have not accounted for the possible influence of the SLP on SR performance. The potential influences include both the way that SLPs produce SR prompts and the identities they hold. The latter issue is important because previous research has shown that adults' identities might affect a child's perceived social affiliation with them (i.e., Kinzler et al., 2007; Kinzler & Spelke, 2011), and these in turn might affect children's willingness to repeat from them. From a young age, children develop preferences based on the socially relevant characteristics of those around them, such as gender, language spoken, accent, and race (i.e., Kinzler et al., 2007; Shutts, 2015). Preference choices are often measured through visibly looking toward or accepting toys from the preferred individuals (Kinzler et al., 2007; Quinn et al., 2002) or explicitly asking preference questions (e.g., “whom you would like to have as your friend?”; Kinzler et al., 2007). Quinn et al. (2002) stated that infants as early as 3 months old demonstrate gender preference based on the gender of their primary caregiver. Kinzler et al. (2007) completed a number of experiments in which both infants and 5-year-old children demonstrated their social preference toward individuals who spoke the same language when asked to choose between their native language and a foreign language. Additionally, children 2.5–5 years of age show a preference for individuals of their same race (Kinzler & Spelke, 2011).

Children's social preferences are complex and may not be consistent across different languages. DeJesus et al. (2017) conducted studies comparing French–English and Korean–English bilingual children's preferences for others based on language and accent. Regardless of their language dominance, these 5- to 7-year-old children showed their social preferences with native-accented talkers, as compared with non–native-accented talkers of the same language. English-dominant children did not show a specific preference for talkers of English versus other languages; however, children who were dominant in either French or Korean demonstrated their preference for talkers speaking their dominant language over talkers speaking English (DeJesus et al., 2017). These studies demonstrate that children may have strong social preferences based on accent, race, and communication patterns.

Furthermore, previous sociolinguistic and sociocognitive studies suggest that an individual's social characteristics (i.e., race, accent, language spoken) can influence children's social assessments of that individual, in addition to children's social preferences (e.g., Kinzler & Dautel, 2012; Tong et al., 2020). For example, Tong et al. (2020) demonstrated that racial identities and accent influenced children's judgment on whom to trust. They completed a series of meta-analyses with 51 unique studies focusing on children's trust preferences based on epistemic (e.g., expertise, accuracy) and social (e.g., racial identity, accent, niceness) characteristics. The analyses showed that younger children (3-year-olds) more often trusted informants who had similar social characteristics, whereas older children (4- to 6-year-olds) more often trusted knowledgeable informants despite social characteristics. As a result, children's preferences and trust were based on both epistemic and social characteristics including race, accent, and language (Tong et al., 2020). Furthermore, the results of Kinzler and Dautel (2012) indicated that older children aged 9–10 years preferred individuals who were the same race, whereas younger children aged 5–6 years mostly preferred individuals who spoke the same language. Because children's social assessments of individuals are impacted by their social characteristics, it is possible that children's impressions of a person play a role in their performance on various speech and language tasks.

Sociolinguistic Factors and SR

In the United States, approximately 91% of SLPs are White, 96% are female, and only 8% are reported to provide services in more than one language (American Speech-Language-Hearing Association, 2023). This means that the majority of the time a monolingual, White, female SLP administers a child's speech and language assessment. These assessments are often administered without considering how the assessor's speech and language use could convey their social identities (e.g., gender, race, language) and with the assumption that these social factors would not impact the assessment results. However, some researchers have highlighted the ties between social factors and language. Clarke et al. (2018) addressed the importance of SLPs building trust and positive relationships with their young clients to maximize therapy outcomes. Koenig and Harris (2005) and Tripp et al. (2021) demonstrated that young toddlers and children learned novel words better from adults with whom they have built trust and whom they regard as reliable, showing that children's social evaluations affect who they are most likely to learn language from.

Studies of the impact of sociolinguistic variation on measures of language processing in adults suggest that this variation could and should impact performance. Tripp et al. (2022) and Tripp and Munson (2022) reviewed the effects of sociolinguistic factors in SR performance with adults. Their review was focused on studies that use SR to study speech intelligibility rather than to assess language disorders. They summarized a variety of studies that suggest SR accuracy in adults is affected by their perception of the racial identity of the talker. For example, Tripp et al. (2022) showed that the audiovisual intelligibility of Black talkers was approximately 10% less than that of White talkers to a racially diverse group of listeners, despite the fact that both groups of talkers were recruited with the same inclusionary criteria. They argued that this effect reflects a mix of implicit and explicit bias against different racial groups and assumptions about the ways that different racial groups use language (Tripp et al., 2022; Tripp & Munson, 2022). These findings suggest that studies of similarly structured SR tasks with children should consider whether perception of sociolinguistic variation and talker identity affects children's SR accuracy. We might unintentionally disadvantage some children on language assessments, because the talkers, SLPs who provide these task items, are almost always White, monolingual females.

The Current Study

This study investigates the relationship between children's social affiliation with talkers and children's SR performance. It uses a racially diverse set of talkers producing SR stimuli because children's performance may be impacted by the perception of racial identity. We aimed to increase our understanding of how characteristics of both children and SLPs influence children's performance on SR tasks. Therefore, this study expands on prior work by measuring children's perceived affiliation with the person being repeated to understand whether any effects of talker race on SR accuracy can be explained by children's perception of social affiliation. In pursuit of these aims, we addressed the following research questions:

  1. Do children repeat sentences more accurately when the talker's racial identity matches their own racial identity?

  2. Do children repeat sentences more accurately when they have a higher social affiliation with the talker of the sentence?

  3. Do either of the above effects interact with children's age?

We hypothesize that children will repeat more accurately with the talker who held the same racial identity as the talker with whom they have a higher social affiliation. In addition, based on Kinzler and Dautel's (2012) finding that older children preferred individuals of the same race while younger children preferred individuals who spoke the same language, we predict that the effects of talker's racialized identity may be greater for older children.

Method

The procedures used in this study were approved by the University of Minnesota Institutional Review Board (Study S00018968).

Setting

Data collection occurred in summer 2023 at the Minnesota State Fair (held in urban St. Paul, MN) and Beltrami County Fair (held in rural Bemidji, MN). All data were collected in the University of Minnesota's research pavilion at the respective fairgrounds, where fair attendees could volunteer for various research projects. Children completed the task at a table within a semiprivate space in the research pavilion at each fairground. The intensity of the ambient noise in the research pavilion was measured with Decibel X mobile app (Decibel, 2023) every 1–2 hr. The average loudness was 61.0 dBA, with a minimum of 56.5 dBA and maximum of 66.3 dBA. The setting and the environment of this study may differ from typical language assessment environment. However, this loudness range corresponds to the upper range of background noise in unoccupied classrooms reported by Knecht et al. (2002).

Participants

A total of 181 children were recruited. Caregivers provided written consent, and children verbally assented prior to participating in any study procedures. The inclusion criteria for this study were (a) age 3–8 years old; (b) native English speaker from birth; either monolingually or bilingually (c) normal or corrected-to-normal vision and hearing; and (d) no history of speech, language, or cognitive impairment, all as indicated by caregiver report. We excluded participants if caregivers reported any significant conditions that might affect task performance, such as autism, Down syndrome, or seizure disorder.

The unique nature of data collection at fairs precluded an in-depth screening to ensure that subjects met inclusionary and exclusionary criteria. Hence, more data were collected than those used for the analysis. Additional data cleaning was completed to verify inclusion and exclusion criteria. After the data cleaning process, 47 of the 181 recruited children were excluded in data analysis due to the following reasons: (a) age not reported (n = 1), (b) not meeting age criteria (n = 4), (c) not a native English speaker (n = 3), (d) history of speech-language therapy (n = 24), (e) history of neurological disorders (n = 8), (f) history of failed hearing test (n = 1), and (g) having an incomplete SR task (n = 6). Therefore, 134 children were included in the data analysis: 55 boys and 79 girls, between the ages of 3.08 years and 8.92 years (M = 6.47, SD = 1.38). Gender was assigned via parent report and included an option for a gender that is neither exclusively male nor exclusively female. A detailed summary of participant demographics is described in Table 1.

Table 1.

Demographic information for participants.

Characteristics Participants n = 134
Age (years)
 M 6.47
 SD 1.38
 Min–Max 3.08–8.92
Sex
 Male:Female 55:79
Race
 White European–American (non-Hispanic) 116
 Hispanic/Latino/a of any race 1
 African American 1
 Asian/Pacific Islander 3
 American Indian/Alaskan Native 1
 More than one racial identity 12
Language spoken at home in addition to English
 French 2
 Mandarin 1
 Ojibwe 1
 Spanish 9
 Thai 1
 Vietnamese 1
Open in a new tab

Stimuli

We used audiovisual speech stimuli that were created and used in previous studies by Tripp et al. (2022) and Tripp and Munson (2022). The original stimuli included 28 racially diverse talkers speaking sentences adapted from the Harvard Sentences (Institute of Electrical and Electronics Engineers, 1969) and the Basic English Language sentence set (Calandruccio & Smiljanic, 2012) in both audio-only (AO) and audiovisual (AV) formats. The three talkers were chosen because they represented three different racial identities (one Black talker, one East Asian talker, and one White talker) and because they were found to elicit similar intelligibility in noise to adult listeners in Tripp et al. (2022) using a conventional “listen and transcribe” method to assess intelligibility. Tripp et al.'s study examined intelligibility in both AO and AV conditions, as one of the purposes of that study was to determine whether differences in intelligibility across talkers were greatest in AV conditions in which talker identity could be assessed by both visual and auditory cues. The current study followed that design and included both AO and AV stimuli. The three talkers were well matched for both AO and AV intelligibility as determined by the listeners in Tripp et al.: Black talker 77.9% AO, 84.8% AV; East Asian talker 74.2% AO, 80.8% AV; White talker 76.2% AO, 81.3% AV.

For the current study, two research assistants went through the stimuli and selected sentences that they judged to be most appropriate for children. Then, they selected tokens of these from the three talkers, resulting in 10 different sentences (five AO, five AV) from each talker for a total of 30 sentences. Two versions of the experiment were created, such that the assignment of sentence token to modality was counterbalanced across participants (i.e., if Sentence 1 by the Black talker was presented in AO in Version A, it was presented in AV in Version B). These selected sentences were randomized, and the same random order was played to each child. Three extra sentences from three additional talkers were selected for practice items.

Procedure

Caregivers completed background questionnaires after consenting to participation. The background questionnaires included a series of questions regarding race, ethnicity, age, spoken language, medical history related to neurodevelopmental disorders, and history of or ongoing special education services.

Each participant completed an SR task. A trained research assistant presented sentences on either a computer or a tablet. The research assistant asked the participant to repeat each sentence after listening to it carefully. The task started with at least three practice items, which the research assistant could repeat at the participant's request. Once the SR task started, the research assistant reminded participants that they could only listen to each sentence once. The research assistant provided basic verbal encouragement as needed, such as “Remember, repeat it back as soon as you hear it,” “Great job! Keep going,” “We are almost done.” The research assistants immediately scored all participants' repeated responses and also audio-recorded responses for later analysis.

Once the SR task was completed, participants answered three social affiliation questions using pictures of the three talkers from the SR task: (a) Who would you most want to play with?, (b) Who do you think lives closest to where you live?, and (c) Which one do you think looks most like you? These questions were designed to capture three different aspects of social affiliation: likeability (Question 1), proximity (Question 2), and similarity (Question 3). The research assistants presented these pictures using PowerPoint slides on either a computer or a tablet. Participants pointed to the pictures to indicate their response to the question. If they did not provide any response, the question was rephrased with encouragement to choose one (e.g., “just pick one you want to play with” or “imagine they are in your neighborhood”).

Scoring and Reliability

All research assistants completed relevant preprofessional coursework and received study-specific training in scoring SR tasks. Research assistants marked the scoring sheet while the participant completed the SR task. A copy of the scoring sheet is available in Supplemental Material S1. Each participant's response was assigned to one of three categories: correct, incorrect, or no response. Only sentences repeated back exactly as given were considered correct. Sentences repeated verbatim but with distorted sounds or articulation errors were also considered correct. If children substituted, transposed, or omitted any part of the sentence, it was considered incorrect. The no response category includes cases where children either stated that they did not hear or repeated nothing back. Due to the data collection site being at a fairground, we excluded no response items from analysis, under the assumption that the nonresponse was due to distraction and noise rather than to item difficulty. If the participant did not respond to any of the first five items, we discontinued the task and excluded the child from analysis. With unsure or missed items, which were marked with a question mark on the paper scoring sheet, the first author listened to the audio recording to score them. A subset of 25% of the data was rescored by two research assistants. Each research assistant independently scored this subset and documented their results alongside the original scores for comparison. Interrater reliability between the raters and the original scorer was 90% and 89%, respectively, with 92% agreement between the two raters. Given the high reliability, a formal consensus process was not completed among disagreements.

Statistical Analysis

We completed all statistical analyses in RStudio software (RStudio Team, 2024). The analyses included descriptive statistics for demographic information and generalized logit mixed-effects models to examine the relationship among the outcome variable (children's SR score) and the predictor variables. The predictor variables were the child's age in months, the child's racial identity match with the talker, presentation modality (AV or AO), and responses to the three social affiliation questions. Age was transformed to a z score to improve model fit as mixed-effects models using maximum likelihood estimation fit better when using predictor variables that are centered (i.e., whose median is 0) and on the same scale. The child's racial identity match was coded as whether the child's racial identity (as reported by the parent) matched that of the talker (as reported by that talker as part of their participation in Tripp et al., 2022). The responses to the three affiliation questions were coded as 1 if the child answered “yes” to the question for that talker, or 0 if the child did not. Some talkers were never chosen by the children as answers to the three questions and hence had 0 for all three questions.

Results

SR Scores

For the SR scores, 6.2% of data (251 of 4,020) were excluded as they were marked as no response.1 The average SR score across children was 67% after excluding no-response answers, which means 67% of sentences were accurately repeated back. Children overall performed higher with the Black talker (M = 74%) than East Asian (M = 70%) and White (M = 57%) talkers. The large difference between the White talker and the two others was not expected, given that the three talkers were selected because they elicited similar intelligibility scores to adults in Tripp et al. (2022). A box plot of children's accurate responses is shown in Figure 1, which separates responses by each talker and each presentation modality.

Figure 1.

Box plots of the children's proportions of accurate responses by talker and audio visual condition. The talkers are Black, East Asian, and White. The audiovisual conditions are audio only (AO) and audio video (AV). The median accuracy for Black talkers is 0.74 for both AO and AV conditions. The median accuracy for East Asian talkers is 0.72 for both AO and AV conditions. The median accuracy for White talkers is 0.60 for both AO and AV conditions.

Open in a new tab

A box plot for children's proportions of accurate responses, separated by each talker (Black, East Asian, and White) and each presentation modality (audio-only vs. audiovisual). The dots have been jittered to facilitate comparison. AO = audio-only; AV = audiovisual.

SR and Racial Identity

First, we fit a simple generalized mixed-effects model to investigate the relationship between a child's SR score and the racial identity match variable indicating whether the racial identity of children and the talkers matched or not, to answer Question 1. A series of models were built successively to determine whether the inclusion of fixed effects and interactions better predicted the children's SR scores. The baseline model included only random intercepts for each child's individual participant number and SR item number to account for potential clustering effects. The second model had the same random effects structure but included fixed effects for presentation modality (AV = 1, AO = 0) and racial identity match (match = 1, mismatch = 0). This model fit the data significantly better than the baseline model, χ2(df = 2) = 47.35, p < .001. The next model added age and terms for the interactions between age and modality and between age and the racial identity match variable. That model fit the data significantly better than the model without age, χ2(df = 3) = 66.49, p < .001. The random effects analysis indicated significant variability in individual participant numbers (variance = 1.25, SD = 1.12) and SR item numbers (variance = 0.42, SD = 0.65).

Contrary to predictions, the racial identity match variable for the most complex model had a significant negative effect on the SR score (estimate = −0.69, standard error [SE] = 0.11, z = −6.38, p < .001), suggesting that achieving a higher SR score decreases when the racial identity of child and talker match. Unsurprisingly, age affected SR scores significantly (estimate = 0.56, SE = 0.099, z = 5.67, p < .001). The racial identity match variable did not interact significantly with age. The presentation modality variable was not significant (estimate = −0.17, SE = 0.11, z = −1.50, p = .13) but did interact significantly with age (estimate = 0.27, SE = 0.088, z = 3.30, p < .001), indicating that the effect of age on SR scores was stronger for AV stimuli than AO ones.

The finding that racial identity match had a negative influence on SR accuracy was unexpected. We reasoned that it might be due to the unexpectedly low SR accuracy for stimuli produced by the White talker and the preponderance of White participants. To examine this, we conducted an additional analysis. We fit a logit mixed-effects model that included children's racial identity and talker racial identity (using the White talker as the reference) rather than racial identity match. Given how few non-White children were in the sample, children's racial identity was coded as either White or Black, Indigenous, and People of Color (BIPOC). The model also had fixed effects for age and presentation modality, and random effects for participant and item, as in the earlier models. Interaction terms were included for age and talker identity, age and modality, and presentation modality and talker identity. As in the earlier model, there was an interaction between age and presentation modality (estimate = 0.28, SE = 0.083, z = 3.32, p < .001). Moreover, repetition accuracy for the Black and East Asian talkers was significantly higher than that of the White talker (Black talker: estimate = 0.99, SE = 0.30, z = 3.20, p < .001; East Asian talker: estimate = 1.30, SE = 0.30, z = 4.28, p < .001). The interaction between the fixed effect for the East Asian talker and modality interacted significantly (estimate = −0.67, SE = 0.29, z = −2.31, p = .021). This interaction may have been due to the East Asian talker eliciting a slightly higher audiovisual benefit for SR accuracy than either of the other talkers. No other main effects or interactions were significant. The result of this additional analysis suggests that the negative influence of racial identity match on SR accuracy occurred because the White talker's productions were repeated less accurately than the other two talkers. We return to this point in the discussion.

SR and Social Affiliation

Overall, the racial identity of the child did not predict whether the child would choose the talker of their own race. For the first social affiliation question (“Who would you most want to play with?”), 51% of children chose the East Asian talker, 34% chose the White talker, 13% chose the Black talker, and 2% did not make any choice. For the second social affiliation question (“Who do you think lives closest to where you live?”), children chose the East Asian talker (39%), the White talker next (36%), and the Black talker the last (21%). For the last social affiliation question (“Which one do you think looks most like you?”), 57% of children identified the White talker, 28% of them identified the East Asian talker, 12% identified the Black talker, and 3% of children did not answer the question.

To investigate the relationship between SR scores and social affiliation toward three talkers, we conducted another generalized linear mixed-effects analysis. As in the additional analysis described in the previous section, this model included the fixed effects for talker, child's racial identity (again using two categories: White and BIPOC), age, and presentation modality, and interactions between age and modality, age and talker, and talker and modality. This model included three additional fixed effects for each of the social affiliation questions, along with interaction terms between these factors and age. This model did not fit the data better than the model without the social affiliation questions, χ2(df = 6) = 0.69, p = .71.

Discussion

The purpose of this study was to examine the relationships among the racial identity of talkers, children's perceived affiliation toward different talkers, and children's SR performances.

Research Question 1

Our first research question aimed to investigate whether children exhibit more accurate SR scores when the racial identity of the talker matches their own racial identity. Our analysis indeed showed a significant negative effect of racial identity matching, indicating children were more likely to show lower SR accuracy with the talker who shared their racial identity. This finding does not align with previous research studies, which have shown young children around 6 years of age tend to exhibit a preference toward their own racial group (Aboud, 2003; Shutts, 2015; Weisman et al., 2015). However, Tong et al. (2020) found that children aged 4–6 years based their preferences not only on race but also on a combination of other epistemic factors, such as expertise and accuracy. While our study did not directly measure other additional factors, the results highlight the potential impact of talker differences on the SR accuracy of children and underscores its relevance in the context of administering other language assessments.

Surprisingly, when we analyzed the talker differences in our study, we found that children repeated sentences more accurately from both the Black and East Asian talkers than the White talker. Given that 87% of our participants identified as White (non-Hispanic), we initially expected that the White talker would elicit the most accurate SR performance. However, opposite to our hypothesis, the White talker was associated with the least accurate SR performance. These three talkers were selected based on their comparable intelligibility to adults (Tripp et al., 2022; Tripp & Munson, 2022). In the original studies, the adult participants judged these recorded sentences in more favorable listening conditions, whereas the children in our current study had to repeat sentences in a noisier environment. However, the original study also examined speech perception in speech-shaped noise, and hence, it is unlikely that the differences are due entirely to the listening environment. The reason for these differences is unclear, and a full discussion of them is outside the scope of this brief report. However, one generalizable conclusion from this study is that differences in children's SR accuracy across different adult talkers may not just reflect differences in those talkers' intelligibility to adults. This finding has important implications for the design of assessments or experimental tasks that use recorded stimuli, because judgments of the suitability of the recordings from adults alone might not predict the behaviors of children.

Research Question 2

For the second research question, we focused on the relationship between a child's SR accuracy and their social affiliation with talkers. Surprisingly, the results revealed that children did not necessarily perform better with a talker they preferred to play with, lived nearby, or identified as looking like them. This finding indicates that these particular social variables did not influence children's SR performance; however, it does not necessarily rule out potential influence of other social variables. For example, all three talkers spoke English without any perceivable influence of another language. Kinzler et al. (2007) stated that young children overall preferred native-accented talkers regardless of the talker's race, suggesting visual information like race could have less impact on children's social preference when talkers speak the same native non-accented language.

Another potentially important social variable that we could not capture in this study is the talker's interests. Young children tend to develop affiliation with strangers who share similar interests alongside physical features (Sparks et al., 2017). Having a meaningful interaction, like discovering shared interests, can motivate children to develop affiliation with that person (Giner Torréns & Kärtner, 2019; Sparks et al., 2017). Therefore, perhaps, in this current study, children may not have formed any strong affiliation due to limited direct interaction with the talkers. Instead, children's direct social interaction occurred with the research assistants who administered the task. It is possible that children developed social affiliation with the research assistants through in-person interaction rather than with the recorded talkers and that children's responses were affected by their affiliation with the research assistants.

In addition, each talker in our study had distinct background and physical characteristics (e.g., wearing glasses, headphones, different hairstyles), which could have contributed to children's response to social affiliation questions. For example, while answering the third social affiliation question (similarity), a few children commented about a certain talker having similar features like them (e.g., “she has her hair down just like me”). This reaction from children is not surprising because, according to Jordan and Wynn (2022), children as young as 3 years old demonstrated more affiliations toward adult individuals who explicitly comment on resemblances like hair color, eye color, and so forth. Even without explicit comments from adults, older children (6-year-olds) spontaneously showed preferences toward adults who shared the same physical characteristics. Therefore, perhaps asking children the reasons why they selected a certain talker could have enhanced our understanding of their selections in this study. Even though the results of this question were unexpected, they highlight that children's performance varied across the different talkers. This suggests that clinicians should consider not only a child's individual characteristics but also their own, as assessor-related factors may influence assessment outcomes.

Research Question 3

Last, we examined whether these effects interacted with children's age in influencing SR scores. Overall, the effects of age itself on children's SR accuracy were significant, which was not a surprise, as we expected the older children to perform better in SR tasks (Polišenská et al., 2015). We also originally predicted that the influence of racial identity would be greater for children who are older. However, the inclusion of an interaction term showed no significant effect. This indicates that the influence of racial identity and social affiliation on SR scores remains relatively consistent across age and implies that these factors may influence children and their SR performance consistently regardless of age. Alternatively, it is possible that, with this SR task, a greater skill in word recognition with older children made these extraneous variables, like the talker's racial identity, less susceptible.

Limitations and Future Directions

In this study, we recognized some limitations that could be improved in future studies. First, all the talkers in our study were identified as female. Weisman et al. (2015) highlighted that children show their preference for adults based on gender more than race. In addition, within the stimuli's original study (Tripp et al., 2022; Tripp & Munson, 2022), the talkers were recorded at home during the COVID-19 pandemic, resulting in varied backgrounds and settings for the talkers. This could have introduced additional variables, such as glasses, headphones, and backdrop picture, which might have influenced children's responses to social affiliation questions. For future studies, minimizing these potential variables would help isolate the impact of race more precisely.

Our study showed significant differences in children's SR accuracy across different racialized talkers, which suggests potential variations in child's language task performance based on the task administrator. However, we did not find a significant relationship between children's SR accuracy and their perceived social affiliation. Future research should consider alternative methods for measuring children's perceived affiliation toward talkers.

Furthermore, the homogeneity of our participants, with approximately 87% identified as White, limits the generalizability of our findings to a broader population. Future studies should aim to recruit and include participants from more diverse backgrounds to provide a more comprehensive understanding of these relationships. Last, we only selected one talker per racial identity. Perhaps having various talkers per racial identity could have changed children's affiliation as Xiao et al. (2015) suggest a variety of exposure assists in children's perception and preferences of different racial identities.

Although the relationship between children's SR performance was in the opposite direction of our predictions, we nonetheless found evidence that children's SR performance varies in response to the talker's race. Combined with other literature (e.g., Tripp et al., 2022; Tripp & Munson, 2022) on the effects of racial identity and social affiliation on speech and language perception, this study counters the assumption that all talkers delivering language assessments will elicit similar results and highlights the need for investigation into talker characteristics that meaningfully influence children's performance.

Data Availability Statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

Supplementary Material

Supplemental Material S1. Scoring sheet.
JSLHR-69-1804-s001.pdf (654KB, pdf)

Acknowledgments

The funding is from National Institutes of Health Grant R21 DC018070 (awarded to Benjamin Munson). We thank Nora Thompson for the reliability coding as well as Brooke Dammen, Caden Christ, Kait Sisney, Meredith Hogan, and Tara Lhatsang for their assistance in collecting data.

Funding Statement

The funding is from National Institutes of Health Grant R21 DC018070 (awarded to Benjamin Munson).

Footnote

1

The pattern of significant and nonsignificant results was the same regardless of whether we included the full set of participants or only the subset that had fewer than 10 nonresponses.

References

  1. Aboud, F. E. (2003). The formation of in-group favoritism and out-group prejudice in young children: Are they distinct attitudes? Developmental Psychology, 39(1), 48–60. 10.1037/0012-1649.39.1.48 [DOI] [PubMed] [Google Scholar]
  2. American Speech-Language-Hearing Association. (2023). 2023 member & affiliate profile. https://www.asha.org/siteassets/surveys/2023-member-affiliate-profile.pdf [PDF]
  3. Calandruccio, L., & Smiljanic, R. (2012). New sentence recognition materials developed using a basic non-native English lexicon. Journal of Speech, Language, and Hearing Research, 55(5), 1342–1355. 10.1044/1092-4388(2012/11-0260) [DOI] [Google Scholar]
  4. Clarke, A., Meredith, P. J., Rose, T. A., & Daubney, M. (2018). A role for epistemic trust in speech-language pathology: A tutorial paper. Journal of Communication Disorders, 72, 54–63. 10.1016/j.jcomdis.2018.02.004 [DOI] [PubMed] [Google Scholar]
  5. Decibel, X. (2023). Decibel X - Pro sound meter (Version 9.2.6) [Mobile application software]. SkyPaw. [Google Scholar]
  6. DeJesus, J. M., Hwang, H. G., Dautel, J. B., & Kinzler, K. D. (2017). Bilingual children's social preferences hinge on accent. Journal of Experimental Child Psychology, 164, 178–191. 10.1016/j.jecp.2017.07.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Ebert, K. D., Ochoa-Lubinoff, C., & Holmes, M. P. (2020). Screening school-age children for developmental language disorder in primary care. International Journal of Speech-Language Pathology, 22(2), 152–162. 10.1080/17549507.2019.1632931 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Friesen, D. C., Ward, O., & Archibald, L. M. D. (2022). Sentence repetition performance differences in bilingual and monolingual children. Journal of Speech, Language, and Hearing Research, 65(8), 2948–2961. 10.1044/2022_JSLHR-21-00596 [DOI] [Google Scholar]
  9. Giner Torréns, M., & Kärtner, J. (2019). Affiliation motivates children's prosocial behaviors: Relating helping and comforting to imitation. Social Development, 28(3), 501–513. 10.1111/sode.12357 [DOI] [Google Scholar]
  10. Institute of Electrical and Electronics Engineers. (1969). IEEE recommended practice for speech quality measurements. IEEE Transactions on Audio and Electroacoustics, 17(3), 225–246. 10.1109/TAU.1969.1162058 [DOI] [Google Scholar]
  11. Jordan, A. E., & Wynn, K. (2022). Adults' pedagogical messages engender children's preference for self-resembling others. Developmental Science, 25(3), Article e13206. 10.1111/desc.13206 [DOI] [Google Scholar]
  12. Kinzler, K. D., & Dautel, J. B. (2012). Children's essentialist reasoning about language and race. Developmental Science, 15(1), 131–138. 10.1111/j.1467-7687.2011.01101.x [DOI] [PubMed] [Google Scholar]
  13. Kinzler, K. D., Dupoux, E., & Spelke, E. S. (2007). The native language of social cognition. Proceedings of the National Academy of Sciences, 104(30), 12577–12580. 10.1073/pnas.0705345104 [DOI] [Google Scholar]
  14. Kinzler, K. D., & Spelke, E. S. (2011). Do infants show social preferences for people differing in race? Cognition, 119(1), 1–9. 10.1016/j.cognition.2010.10.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Knecht, H. A., Nelson, P. B., Whitelaw, G. M., & Feth, L. L. (2002). Background noise levels and reverberation times in unoccupied classrooms: Predictions and measurements. American Journal of Audiology, 11(2), 65–71. 10.1044/1059-0889(2002/009) [DOI] [PubMed] [Google Scholar]
  16. Koenig, M. A., & Harris, P. L. (2005). Preschoolers mistrust ignorant and inaccurate speakers. Child Development, 76(6), 1261–1277. 10.1111/j.1467-8624.2005.00849.x [DOI] [PubMed] [Google Scholar]
  17. Norbury, C. F., Gooch, D., Wray, C., Baird, G., Charman, T., Simonoff, E., Vamvakas, G., & Pickles, A. (2016). The impact of nonverbal ability on prevalence and clinical presentation of language disorder: Evidence from a population study. The Journal of Child Psychology and Psychiatry, 57(11), 1247–1257. 10.1111/jcpp.12573 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Pham, G., & Ebert, K. D. (2020). Diagnostic accuracy of sentence repetition and nonword repetition for developmental language disorder in Vietnamese. Journal of Speech, Language, and Hearing Research, 63(5), 1521–1536. 10.1044/2020_JSLHR-19-00366 [DOI] [Google Scholar]
  19. Polišenská, K., Chiat, S., & Roy, P. (2015). Sentence repetition: What does the task measure? International Journal of Language & Communication Disorders, 50(1), 106–118. 10.1111/1460-6984.12126 [DOI] [PubMed] [Google Scholar]
  20. Pratt, A. S., Peña, E. D., & Bedore, L. M. (2021). Sentence repetition with bilinguals with and without DLD: Differential effects of memory, vocabulary, and exposure. Bilingualism: Language and Cognition, 24(2), 305–318. 10.1017/S1366728920000498 [DOI] [Google Scholar]
  21. Quinn, P. C., Yahr, J., Kuhn, A., Slater, A. M., & Pascalis, O. (2002). Representation of the gender of human faces by infants: A preference for female. Perception, 31(9), 1109–1121. 10.1068/p3331 [DOI] [PubMed] [Google Scholar]
  22. RStudio Team. (2024). RStudio 2023.12.1. https://www.rstudio.com/
  23. Rujas, I., Mariscal, S., Murillo, E., & Lázaro, M. (2021). Sentence repetition tasks to detect and prevent language difficulties: A scoping review. Children, 8(7), Article 578. 10.3390/children8070578 [DOI] [Google Scholar]
  24. Shutts, K. (2015). Young children's preferences: Gender, race, and social status. Child Development Perspectives, 9(4), 262–266. 10.1111/cdep.12154 [DOI] [Google Scholar]
  25. Sparks, E., Schinkel, M. G., & Moore, C. (2017). Affiliation affects generosity in young children: The roles of minimal group membership and shared interests. Journal of Experimental Child Psychology, 159, 242–262. 10.1016/j.jecp.2017.02.007 [DOI] [PubMed] [Google Scholar]
  26. Tomblin, J. B., Records, N. L., Buckwalter, P., Zhang, X., Smith, E., & O'Brien, M. (1997). Prevalence of specific language impairment in kindergarten children. Journal of Speech, Language, and Hearing Research, 40(6), 1245–1260. 10.1044/jslhr.4006.1245 [DOI] [Google Scholar]
  27. Tong, Y., Wang, F., & Danovitch, J. (2020). The role of epistemic and social characteristics in children's selective trust: Three meta-analyses. Developmental Science, 23(2), Article e12895. 10.1111/desc.12895 [DOI] [Google Scholar]
  28. Tripp, A., Feldman, N. H., & Idsardi, W. J. (2021). Social inference may guide early lexical learning. Frontiers in Psychology, 12, Article 645247. 10.3389/fpsyg.2021.645247 [DOI] [Google Scholar]
  29. Tripp, A., Lyons, T., & Munson, B. (2022). Predicting talker intelligibility: Contrasting talkers' racialized identities and listeners' racializations. The Journal of the Acoustical Society of America, 152(Suppl. 4), Article A177. 10.1121/10.0015951 [DOI] [Google Scholar]
  30. Tripp, A., & Munson, B. (2022). Perceived linguistic marginalization and speech intelligibility in a racially diverse cohort of talkers. The Journal of the Acoustical Society of America, 152(Suppl. 4), Article A177. 10.1121/10.0015952 [DOI] [Google Scholar]
  31. Vang Christensen, R. (2019). Sentence repetition: A clinical marker for developmental language disorder in Danish. Journal of Speech, Language, and Hearing Research, 62(12), 4450–4463. 10.1044/2019_JSLHR-L-18-0327 [DOI] [Google Scholar]
  32. Wang, D., Zheng, L., Lin, Y., Zhang, Y., & Sheng, L. (2022). Sentence repetition as a clinical marker for Mandarin-speaking preschoolers with developmental language disorder. Journal of Speech, Language, and Hearing Research, 65(4), 1543–1560. 10.1044/2021_JSLHR-21-00401 [DOI] [Google Scholar]
  33. Weisman, K., Johnson, M. V., & Shutts, K. (2015). Young children's automatic encoding of social categories. Developmental Science, 18(6), 1036–1043. 10.1111/desc.12269 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Wiig, E. H., Semel, E., & Secord, W. A. (2013). Clinical Evaluation of Language Fundamentals–Fifth Edition (CELF-5). NCS Pearson. [Google Scholar]
  35. Xiao, W. S., Fu, G., Quinn, P. C., Qin, J., Tanaka, J. W., Pascalis, O., & Lee, K. (2015). Individuation training with other-race faces reduces preschoolers' implicit racial bias: A link between perceptual and social representation of faces in children. Developmental Science, 18(4), 655–663. 10.1111/desc.12241 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material S1. Scoring sheet.
JSLHR-69-1804-s001.pdf (654KB, pdf)

Data Availability Statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.

Articles from Journal of Speech, Language, and Hearing Research : JSLHR are provided here courtesy of American Speech-Language-Hearing Association

RESOURCES