Developmental Language Risk and Social and Behavioral Functioning in Bilingual Latinx Adolescents

Abstract

Purpose:

This study examines the relationship between language development risks and social and behavioral functioning, specifically externalizing problems (EPs), internalizing problems (IPs), and social competence (SC), among bilingual Latinx youth.

Method:

We used a community-based sample of 129 bilingual Latinx youth and their caregivers from a large urban city in the U.S. South. Participants completed standardized parent- and child-report measures in Spanish and English to assess developmental language disorder (DLD) risk and social and behavioral concerns (SBCs), and children completed assessments of their narrative skills in English and Spanish. Fisher's exact tests were used to assess differences in SBC identification and bootstrap tests to examine differences in SBC scores between youth with and without DLD risks. Multiple regression analyses were employed to explore the association of children's narrative skills in each language with SBCs, controlling for key demographics.

Results:

Youth with DLD risk were more likely to have clinically significant IP, based on Youth Self-Report. Youth at risk of DLD reported higher levels of IP in both parent- and self-reports, but there were no significant differences for EP and SC. Regression analyses revealed low but statistically significant associations between Spanish narrative skills and lower parent-reported SC and low but statistically significant English narrative skills and higher self-reported SC.

Conclusions:

Bilingual Latinx youth with DLD may experience IP across multiple settings, making self-reports more sensitive to capturing the relations between IP and DLD risks. Oral narration skills showed small but significant associations with SC, suggesting a potential role in social functioning. These findings highlight the importance of including community-based samples to understand better the diverse experiences of youth that might not be represented in clinical settings.

Language and communication skills play an important role in adolescents' social and behavioral functioning. Language is critical for youth to communicate their needs and build relationships with others (Denham et al., 2003; Nelson, 2007). Language skills support the development of youths' ability to regulate attention, behavior, and emotion (e.g., Lupyan, 2016; Schweitzer & Waytz, 2021), as well as to understand their own and others' mental states (Carpendale & Lewis, 2006; Fivush et al., 2006) and relate positively with others (Hebert-Myers et al., 2006).

Language difficulties such as those characterized by developmental language disorder (DLD; difficulties using/learning spoken language; McGregor, 2020) may negatively impact a youth's ability to regulate their emotions and behaviors and interfere with the development of social competencies (McCabe & Meller, 2004), leading to the development of emotional and behavioral concerns (Brinton & Fujiki, 1993). Elevated co-occurrence of DLD with behavioral concerns has been documented in children (Snow & Powell, 2011). Most studies on language development and behavioral concerns have focused on monolingual children (e.g., Conti-Ramsden et al., 2013), although one study showed high rates of language disorders in bilingual youth referred for psychiatric care (Toppelberg et al., 2002). Understanding the link between risk for DLD and social and behavioral concerns (SBCs) in bilingual youth is crucial for developing effective interventions to improve outcomes for youth at risk for both language and behavioral problems in this growing population.

In 2019, Latinx youth represented 26% of the U.S. youth population and was one of the fastest growing groups of youth in the United States (Chen & Guzman, 2021). Furthermore, in the United States, approximately 22% of children ages 5–17 years speak a language other than English at home (Migration Policy Institute, 2023), with 69% of these youth being Latinx and speaking Spanish at home. Bilingual Latinx youth may have access to psychological and social assets and resources that may promote positive development and protect against SBCs (Chen & Padilla, 2019), such as cognitive strengths (e.g., Bialystok & Viswanathan, 2009; Diaz, 1985; Hakuta, 1987), language brokering skills (Borrero, 2015), and the economic benefits of bilingualism (Gándara, 2018). Bilingual Latinx youth may also experience systemic racism, including exclusion from supportive educational environments for bilingual language learning (Gándara, 2018) as well as over-identification of language learning difficulty (Cuba & Tefera, 2024) or under-identification (Restrepo & Gutiérrez-Clellen, 2001) of language learning difficulty, all of which may increase the risk for SBCs (Müller et al., 2020). As such, it is important to understand the associations between DLD risk and SBCs in Latinx bilingual youth.

Language Development in Adolescence

Adolescence, broadly defined as between the ages of 10 and 19 years old, is a developmental period during which youth continue to master skills to use more complex and sophisticated language (Nippold, 2016). Adolescents demonstrate increasing syntactic complexity in their narrative language (Nippold et al., 2015), and pragmatic language with peers becomes increasingly sophisticated (Nippold, 2000). These narrative and pragmatic skills serve as the basis for more sophisticated social communication as well as academic skills such as writing (Gillam et al., 2023). Oral narration provides a valuable window into adolescents' ability to organize and express complex ideas and perspectives across contexts (Hadley, 1998). Because oral narratives ask for the integration of linguistic, cognitive, and social skills, they are particularly useful in identifying language development differences in bilingual youth (Fiestas & Peña, 2004).

Although most of research on adolescent language development has focused on monolingual populations, a limited but growing body of research has begun to explore these processes in Latinx bilingual adolescents. Emerging evidence suggests that bilingual adolescents may follow different language development paths than their monolingual peers, both in narrative macrostructure and syntactic complexity (Gamez et al., 2016; Ordoñez, 2004). One study found that bilingual adolescents increased daily exposure and use of English (rather than Spanish) was associated with lower Spanish narrative skills, but unrelated to their English narrative skills, suggesting that language development may vary depending on the adolescent's context (del Carmen Perez et al., 2023). Thus, it is important to understand specific processes of risk, resilience, and developmental outcomes that shape the language profiles of bilingual Latinx adolescents, especially those at risk of DLD.

DLD: Prevalence and Outcomes

DLD are prevalent in childhood, with approximately 7.5% of monolingual English-speaking children (Law et al., 2000; Tomblin et al., 1997) and Spanish-speaking children (Auza et al., 2024) meeting criteria for a DLD. There are no studies estimating the prevalence in bilingual children; however, Peña et al. (2023) found that children with bilingual exposure had no greater risk for DLD than children who fell on the more monolingual English or Spanish ends of the continuum. However, Latinx children are less likely to be diagnosed with DLD than their non-Latinx peers, potentially due to language barriers in assessment, limited availability of culturally and linguistically appropriate testing tools, or other cultural factors (Restrepo & Gutiérrez-Clellen, 2001). Often, identification is delayed, leading to delays in children receiving the needed services (Hibel & Jasper, 2012). The lower rates of detection, delayed identification for bilingual Latinx youth, and a lack of intervention may also increase risk for social and behavioral difficulties that co-occur with developmental language difficulties in adolescence (Hibel & Jasper, 2012).

SBCs in Adolescence

Adolescence is a developmental period marked by significant physical, cognitive, and social changes, alongside increased vulnerability to social, behavioral, and emotional concerns. Internalizing concerns, such as anxiety, depression, social withdrawal, and difficulties with emotional regulation, commonly emerge or intensify during this period (Chan et al., 2008). Externalizing behaviors, including aggression, rule breaking, and acting out, are also prevalent, with a combined prevalence of internalizing and externalizing concerns between 17.8% and 34.4% (Danielson et al., 2021). These SBCs are linked to a range of adverse outcomes, including academic failure, disciplinary problems, and engagement in risky behaviors (Atherton et al., 2018; Highlander & Jones, 2022). Social competence (SC), the ability to understand, navigate, and interact with others appropriately and effectively, has been identified as a key protective factor in adolescence (Stepp et al., 2011). Higher levels of SC are associated with fewer internalizing and externalizing concerns and a wide variety of positive outcomes, including educational attainment, positive relationships with peers, and well-being (Bornstein et al., 2010; Jensen et al., 2015). In contrast, lower social competency has been associated with increased risk of engagement in risky and antisocial behaviors, internalizing and externalizing problems (EPs), and lower educational outcomes (Bornstein et al., 2010; Caplan et al., 1992; Sørlie et al., 2008).

In addition to the risks associated with DLD under-identification, Latinx adolescents experience increased risk for SBCs. Compared to non-Latinx White youth, Latinx adolescents report higher levels of internalizing concerns, including depressive and anxiety symptoms, and have among the highest rates of suicidal ideation among all youth (Anderson & Mayes, 2010). Although Latinx youth have similar rates of externalizing concerns to those of the general population, the sociocontextual environments in which many Latinx live (e.g., poverty, structural racism, and other systems of oppression) may increase their risk for these concerns. For example, Prelow et al. (2007) found that socioenvironmental stressors, including parental distress and financial strain, were associated with higher externalizing symptoms, a relationship mediated by lower SC. Of note, SC is significantly associated with language skills in children with DLD (Wieczorek et al., 2024). As such, it is important to consider the influence of language development on factors such as SC that may moderate or mediate the effects of socioenvironmental risk factors on Latinx youth's mental and behavioral health. Despite the growing body of literature on risk and resilience processes in Latinx youth, less is known about how SBCs and DLDs are linked, particularly in bilingual language development. Understanding the intersection between SBCs and DLD is critical for informing culturally and linguistically appropriate identification and intervention efforts.

DLD and SBCs in Adolescence

The co-occurrence of DLDs and SBCs has been well documented in monolingual children. Children with language difficulties or delays are at increased risk for developing emotional and behavioral problems later in adolescence (Conti-Ramsden et al., 2013). For example, children with DLD have elevated levels of internalizing and externalizing difficulties later in childhood and adolescence compared to peers without specific language impairments (Conti-Ramsden et al., 2013; Yew & O'Kearney, 2013). Furthermore, peer problems appear to mediate the relation between DLD risks in early childhood and emotional problems in adolescence (Forrest et al., 2018). Despite these risks, DLDs often go unnoticed or undiagnosed until other mental health concerns are identified (Cohen et al., 1993). For example, one study found that approximately 53% of children referred to urban community clinics had a language disorder, with half of those children receiving their first diagnosis only after the psychiatric referral (Cohen et al., 1993). This is particularly concerning, as undetected DLD has been linked to more severe SBCs (Cantwell & Baker, 1991; Cohen & Horodezky, 1998).

In bilingual children, DLD identification presents additional challenges. DLD may be misattributed to normal dual-language development, or conversely, their normal bilingual language development may be inappropriately pathologized (Peña et al., 2023). However, most studies of language development and child psychopathology focus on monolingual and non-Latinx children and have explicitly excluded bilingual children (Toppelberg et al., 2002). Additionally, youth with DLD concerns are more likely to experience peer victimization (Forrest et al., 2018; van den Bedem et al., 2018), and emergent bilingual students also have an increased risk of bullying, which might be exacerbated in the presence of DLD (Aguilar-Mediavilla et al., 2024). These experiences of peer victimization and social challenges also operate as risk factors for internalizing and EPs. Given the potential compounding effects of language difficulties and social marginalization, examining the co-occurrence of DLD risk and SBCs in bilingual youth is a critical area for further investigation.

An Ecological–Sociocultural Developmental Framework

The present study was guided by the integrative risk and resilience model, an ecological framework for understanding how diverse experiences and interconnected factors shape the adaptation of immigrant–origin children and youth (Suárez-Orozco et al., 2018). Recognizing the multiple levels of influence on the developing child (Bronfenbrenner, 1994), the model conceptualizes adaptation as a dynamic process shaped by interconnected factors from the individual (e.g., language, immigration status) and family (e.g., socioeconomic status, parental level of education) to broader sociocultural factors, including schools, community, and sociopolitical structures (e.g., attitudes toward immigrants, systemic racism). In addition to risks, the model addresses the importance of resilience and positive adaptation, highlighting the role of developmental assets and protective factors (Benson, 2007) that promote positive outcomes in Latinx youth (e.g., Bulut et al., 2015). This model provides a culturally relevant framework to examine the co-occurrence of DLDs and SBCs in a broader ecological context and understand how linguistic and behavioral risks may occur in the Latinx youth population.

The Current Study

The present study addresses this critical gap in knowledge by using cross-sectional data to examine the relationship between Latinx youth's internalizing problems (IPs), EPs, and social competency with language development risks. While some research has explored DLD and mental health in bilingual populations, these studies typically rely on clinical samples rather than community samples (Toppelberg et al., 2002). Our study examined the association of DLD risks and child mental health concerns in a community-based sample of 129 bilingual Latinx youth ages 10–15 years. This study addresses the following research questions:

1. Are there differences in SBC identification for youth with and without DLD risk?

2. To what extent do SBC scores differ between youth with and without DLD risk?

3. To what extent are language development factors, such as children's narrative skills in each language, associated with lower levels of SBCs, while controlling for key demographic variables of age, gender, and parental level of education?

Method

Participants and Data Collection

As part of a federally funded and institutional review board–approved study (Protocol 2014-10-0123), the study included 129 bilingual Latinx and their parents/caregivers living in or around a large urban city in the southern United States. Youth participants (47.3% female) ranged in age from 10 to 15 years (M = 12.00, SD = 1.15), with the majority (85.3%) between the ages of 11 and 13 years. Approximately 7.8% of youth were reported to have received speech-language services. Most families (87.6%) qualified for free/reduced lunch. Parents/caregivers were predominantly female (95%) and identified as Latinx, with the majority reporting a Mexican-origin background (88.4%) or a Central American origin (7.8%). Ninety-three percent of parents/caregivers preferred to complete the study measures in Spanish. Table 1 provides additional demographic information. Eligibility criteria for child participants were as follows: (a) age between 10 and 15 years at the start of the study, (b) have some proficiency in both English and Spanish, and (c) identify as Latinx or Hispanic. For parents/caregivers, the sole requirements were being 18 years of age or older and allowing the participation of only one parent/caregiver and one child from each family. Child participants had varying degrees of English and Spanish proficiency and exposure.

Table 1.

Sociodemographic characteristics of participants (N = 129).

Sociodemographics	n	%
Youth
Female	61	47.3
Qualify for free or reduced lunch	112	87.5
Received speech-language services	10	7.8
White	63	50.8
Black/other race	61	49.2
Age (in years)
10	12	9.3
11	36	27.9
12	32	24.8
13	42	32.6
14–15	7	5.4
Parents/caregiversa
Female	123	95.3
First generation immigrant	116	90.6
Mexican origin	114	88.4
Central American origin	10	7.8
Other Latinx origin	5	3.8
Study language preference (Spanish)	120	93
Highest level of education
High school diploma or less	96	74.4
Some college or higher	33	25.6

^aParents/caregiver demographics are based on those who responded to the survey.

The study recruited participants from schools, community events, and local community hubs (e.g., food trucks). Collaborative efforts focused on schools with a high proportion of low-socioeconomic Latinx students. Researchers coordinated with school principals for study dissemination and engaged eligible students in classrooms. Community-based recruitment occurred at various events, such as back-to-school nights, fairs, and Latinx cultural events targeting Latinx families. Interested families completed a permission-to-contact form, and then a bilingual researcher assessed eligibility, provided study details, and scheduled the first assessment for those interested. Both Spanish and English versions of consent and permission forms were available. Additional details of the recruitment and enrollment procedures are described elsewhere (Ocegueda et al., 2024).

Measures

In this study, we employed widely used parent- and child-informant measures of child language development and social and behavioral functioning. The selected language measures included the Bilingual English–Spanish Assessment—Inventory to Assess Language Knowledge (ITALK; Peña et al., 2018), a parent report designed to identify risk for DLD in Spanish–English bilingual children, and the Test of Narrative Language (TNL) in English (Gillam & Pearson, 2004) and Spanish (Gillam et al., 2010) to assess children's narrative skills in each language. Social and behavioral functioning was evaluated using the IPs, EPs, and SC scales from the Child Behavior Checklist (CBCL) and the Youth Self-Report (YSR), which are part of the Achenbach System of Empirically Based Assessment (ASEBA; Achenbach & Rescorla, 2001). The ASEBA instruments are available in English and Spanish and completed in the participant's preferred language.

The ITALK and CBCL measures are completed by parents/caregivers, while the TNL and YSR measures are completed by their children, providing distinctive perspectives on the behaviors being assessed. Incorporating multiple informants is especially critical when assessing psychopathology in children, as reliance on a single source may lead to different conclusions concerning the associations or risk factors associated with the disorder (De Los Reyes & Kazdin, 2005). Prior research suggests low-to-moderate informant agreement of ratings in clinical measures, including those for internalizing and EPs (De Los Reyes & Kazdin, 2005). Our data are consistent with prior research and shows low-to-moderate correlations between caregiver-reported and youth-reported measures, ranging from r = .32 to .37 (see Table 2).

Table 2.

Correlation matrix for the social and behavioral concerns study variables.

Variable	1	2	3	4	5
1. CBCL internalizing problems	1
2. CBCL externalizing problems	.66***	1
3. CBCL social competence	−.26**	−.26**	1
4. YSR internalizing problems	.32**	.17	.04	1
5. YSR externalizing problems	.12	.37***	−.02	.47***	1
6. YSR social competence	−.03	−.06	.35***	.10	−.07

Note. CBCL = Child Behavior Checklist; YSR = Youth Self-Report.

^** p < .01.

^*** p < .001.

ITALK

The ITALK (Peña et al., 2018) is a parent/caregiver questionnaire part of the Bilingual English–Spanish Assessment battery. It is designed to measure the child's relative use of English and Spanish along with their level of speech and language development in vocabulary, grammar, sentence production, comprehension, and phonology in both languages. The parent/caregiver rates the child's use of each language in each domain on a scale of 0–5, and then these ratings are averaged separately for each language. The ITALK has a cut score of 4.18 to identify risk for language disorder (at risk ≤ 4.18; not at risk > 4.18). The cut score has a sensitivity and specificity of .80 for each language, based on a sample of children aged 7–12 years (Peña et al., 2018). The measure was used as designed, without modifications for use with older youth. In a study by Pratt et al. (2022), the performance of the ITALK was further explored for preschool and early elementary-aged children, showing 90% sensitivity and 63% specificity. Similarly, Peña et al. (2023) found the ITALK to be an effective measure for exploring assumptions of language delay in bilingual children, 5–12 years old, with and without DLD. These findings, collectively, support the utility of ITALK as a screener to identify DLD risk for a broad age range, particularly into early adolescence.

In the current study, children with scores at or below the cut score in both languages were flagged as at risk for possible concerns about DLD. Those scoring above the cut score in at least one language were categorized as not at risk, reflecting typical language development. Among the study sample, 11% were flagged as at risk of DLD. From those identified not at risk of DLD (n = 113), 6.2% were not at risk in English only, 31% not at risk in Spanish only, and 62.8% in both languages.

ASEBA

The ASEBA (Achenbach & Rescorla, 2001). The ASEBA is a comprehensive and empirically validated assessment system designed to evaluate social, emotional, and behavioral functioning in children and adolescents. It consists of various assessment tools, such as the CBCL for parents/caregivers and the YSR for adolescents to complete. The CBCL and YSR are norm referenced based on samples representative of the U.S. population and have been translated into Spanish, demonstrating appropriate cross-cultural applicability for Latinx youth (Achenbach & Rescorla, 2001). This study used the CBCL and YSR composite scales of IPs, EPs, and SC. All measures were administered orally by a trained researcher to reduce potential literacy barriers with participants.

IPs (Achenbach & Rescorla, 2001). This composite scale consists of three subscales: Somatic Complaints, Anxious/Depressed, and Withdrawn/Depressed. Items are on a 3-point scale with the options being not true, somewhat or sometimes true, and very true or often true. The internal consistency of IP is .90 in the CBCL and the YSR scales (Achenbach & Rescorla, 2001). We used the normalized T scores of IP, where scores can be interpreted as normal (T < 60), borderline (60 ≤ T ≤ 63), and clinical (T > 63) behavior. For this study, scores in the borderline and clinical range (T ≥ 60) were grouped together and classified as “At Risk,” else normal.

EPs (Achenbach & Rescorla, 2001). This composite scale is the sum of two subscales: Rule-Breaking Behavior and Aggressive Behavior. Items are on a 3-point rating scale. The internal consistency of EP is .94 in the CBCL measure and .90 in the YSR measure (Achenbach & Rescorla, 2001). In normalized T scores, the predefined clinical classifications of EP are normal (T < 60), borderline (60 ≤ T ≤ 63), and clinical (T > 63). For this study, T scores at or above 60 were combined into an “At Risk” group, and T scores under 60 stayed in the same classification.

SC (Achenbach & Rescorla, 2001). This scale consists of 20 items in multiple-choice and open-ended formats that ask about children's involvement in school activities, interactions with family and friends, and school performance. In contrast to the IP and EP scales, higher scores on the SC represent greater SC and therefore indicate fewer concerns. The internal consistency of SC is .79 in the CBCL measure and .75 in the YSR scale (Achenbach & Rescorla, 2001). The clinical interpretation of SC in T scores is clinical (T < 36), borderline (36 ≤ T ≤ 40), and normal (T > 40). For this study, T scores less than or equal to 40 were grouped together and classified as “At Risk,” else normal.

TNL

The TNL (Gillam & Pearson, 2004). The TNL is a norm-referenced standardized measure of narrative comprehension and oral narration. The areas tested include grammar, organization of language, and inference from narratives. The TNL is an untimed measure that takes between 15 and 25 min to complete. The Spanish TNL is an experimental version developed to be parallel to the English TNL, and it is culturally adapted to Spanish-speaking students in the United States (Gillam et al., 2010). This study used the Oral Narration subtest in English (TNL-ONE) and Spanish (TNL-ONS). According to test developers, the internal consistency for the English Oral Narration subtest is good, with an alpha level of .87. Preliminary analyses indicate a coefficient alpha of .93 for the Spanish Oral Narration subtest (Peña et al., 2020).

Analysis Procedure

We employed Fisher's exact tests, bootstrap tests, and multiple regression analyses to answer Research Questions 1, 2, and 3, respectively. Research Question 1 examines the relationship between DLD risk classification (at risk vs. not at risk) and social and behavioral functioning classification (at risk vs. normal) in the overall sample. Fisher's exact test was preferred as the significance test to account for low expected frequencies in the data (Prescott, 2019). Research Question 2 extends this analysis by exploring variations in scores in the SBC measures, conditioned on DLD risk. We used a bootstrap test to evaluate mean score differences on the CBCL and YSR scales between youth with and without DLD risks. For each scale, data were randomly sampled, ensuring the assumption of independence was met for the bootstrap test. Monte Carlo p values were computed using 2,999 bootstrap replicates from the data. Cohen's d (Cohen, 1988) was used as the effect size indicator. Statistically significant mean differences were determined using a significance level of .05 in the bootstrap test.

Last, we used multiple regressions to examine the extent to which social and behavioral functioning is influenced by language development factors while controlling for key demographic variables. Language development was measured using the raw scores from the TNL-ONE and Spanish (TNL-ONS). Control variables included the child's age (measured in years), gender (Female = 0; Male = 1), and parental level of education. Analyses were performed using R software (R Core Team, 2024).

Results

Association of SBC and DLD Risk Identification

Table 3 provides descriptive statistics for the association between each SBC indicator and DLC classification. For the CBCL scales, the Fisher's exact tests detected no statistically significant relationships for IP (p = .57), EP (p = .28), or SC (p = .99) with ITALK risk classification. For the YSR scales, a Fisher's exact test revealed a significant association between IP and ITALK classification, p = .044. However, statistically significant associations were not detected between YSR EP (p = .99) or SC (p = .78) with ITALK risk classification.

Table 3.

Summary of classification in the social and behavioral functioning concerns and risk of developmental language disorders.

SBC	DLD risk
	Not at risk (%)	At risk (%)	p
CBCL internalizing problems			.57
Not at risk	73 (57.9)	8 (6.3)
Risk	39 (31)	6 (4.8)
CBCL externalizing problems			.28
Not at risk	93 (73.8) (48)	10 (7.9)
Risk	19 (15.1)	4 (3.2)
CBCL social competence			.99
Not at risk	59 (47.2)	7 (5.6)
Risk	52 (41.6)	7 (5.6)
YSR internalizing problems			.044*
Not at risk	67 (52.8)	4 (3.1)
Risk	46 (36.2)	10 (7.9)
YSR externalizing problems			.99
Not at risk	90 (70.9)	11 (8.7)
Risk	23 (18.1)	3 (2.4)
YSR social competence			.78
Not at risk	66 (52.8)	9 (7.2)
Risk	45 (36)	5 (4)

Note. Fisher's exact test is statistically significant at the .05 level. SBC (social and behavioral concern) is a subscale of Child Behavior Checklist (CBCL) and the Youth Self-Report (YSR). DLD risk (risk of developmental language disorder) and assessed with Inventory to Assess Language Knowledge. Percentages are based on the total frequency count within each 2 × 2 cross-table.

^* p < .05.

SBC Mean Score Differences by DLD Risk Identification

The bootstrap tests revealed statistically significant mean differences between youth at risk of DLD and those not at risk in both the CBCL and YSR scales for IP. However, no statistically significant mean differences were detected between DLD risk classification with EP and SC in either the CBCL or YSR scale. As shown in Table 4, youth at risk of DLD scored higher on average in CBCL IP (M = 12.57, SD = 10.77) compared to those not at risk of DLD (M = 8.15, SD = 6.64). This difference was statistically significant (Monte Carlo p = .028), with a moderate effect size of d = 0.62. Similarly, youth at risk of DLD scored higher on average in YSR IP (M = 20.07, SD = 9.04) compared to youth not at risk of DLD (M = 14.84, SD = 8.10), with a statistically significant mean difference (Monte Carlo p = .022) and an effect size of d = 0.65. Table 4 provides the mean scores and standard deviations for the CBCL and YSR scales conditioned on DLD-Risk classification.

Table 4.

Mean differences in social and behavioral concerns (SBCs) by risk of developmental language disorder (DLD).

Variable	DLD risk
	Not at risk		At risk
SBC	M	SD	M	SD
CBCL
Internalizing problems*	8.15	6.64	12.57	10.77
Externalizing problems	6.31	6.65	7.71	7.28
Social competence	6.46	2.26	6.32	2.23
YSR
Internalizing problems*	14.84	8.10	20.07	9.04
Externalizing problems	10.92	6.53	9.50	7.50
Social competence	7.19	2.26	7.11	1.88

Note. SBC (social and behavioral concern) was measured by the Child Behavior Checklist (CBCL) and the Youth Self-Report (YSR), while DLD risk (risk of developmental language disorder) was assessed with Inventory to Assess Language Knowledge.

^*Statistically significant mean differences (p < .05) using Monte Carlo p values.

Language Development Factors Associated With Social and Behavioral Functioning

As stated in Research Question 3, we examined the extent to which oral narration ability in each language is associated with fewer SBCs while controlling for key demographic factors. Specifically, the TNL-ONE and Spanish (TNL-ONS) were used as independent variables, with child age, gender, and parental level of education as control variables. In the overall sample, the mean score for TNL-ONE was 53.14 (SD = 10.12) and for TNL-ONS was 44.80 (SD = 13.42), with a moderate correlation between the English and Spanish TNL measures (r = .52, p < .001). When disaggregating by DLD risk status, youth at risk scored lower on average in both measures, 50.21 (SD = 16.45) for TNL-ONE and 43.67 (SD = 18.9) for TNL-ONS, compared to 53.59 (SD = 9.04) for TNL-ONE and 44.85 (SD = 12.78) for TNL-ONS, respectively, for youth not at risk of DLD. Six regression models were applied to estimate each of the CBCL and YSR scales as dependent variables.

Table 5 presents the regression results for the CBCL measures. The omnibus test for the SC model (Model SC) was statistically significant F(5, 113) = 2.63, p = .03 with R² = .10. The TNL-ONS coefficient was negative and statistically significant (β = −0.04, p = .04), indicating that as the TNL-ONS score increases by one unit, the SC score decreases by 0.04 units on average, holding all other variables constant. Regression coefficients were not statistically significant in the IP and EP models (see Table 5).

Table 5.

Regression results for the Child Behavior Checklist (CBCL) indicators.

Variable	Model IP		Model EP		Model SC
	β	SE	β	SE	β	SE
Intercept	1.27	7.54	7.48	7.12	5.67*	2.39
TNL-ONE	0.08	0.08	0.01	0.08	0.04	0.03
TNL-ONS	−0.04	0.06	−0.05	0.05	−0.04*	0.02
Age	0.35	0.55	0.08	0.52	−0.14	0.17
Gender	−2.54	1.28	−1.50	1.20	0.67	0.41
Parental level of education	0.76	0.18	0.05	0.17	0.13*	0.06
R 2	.06		.02		.10

Note. IP = internalizing problem; EP = externalizing problem; SC= social competence; SE = standard error; TNL-ONE = Test of Narrative Language Oral Narration English; TNL-ONS = Test of Narrative Language Oral Narration Spanish.

^*Statistically significant at the .05 level.

Similar to the results obtained for the CBCL measures, the omnibus test for the YSR SC model (Model SC) was statistically significant F(5, 112) = 2.91, p = .02 with R² = .11. However, for this model the TNL-ONE coefficient was statistically significant (β = −0.08, p = .001) after holding all other predictor variables constant. This indicates that for a unit increase in the TNL-ONE score, the SC score increases by .08 units on average, while holding constant all other variables. The coefficient of determination (R²) was .08. TNL-ONE and TNL-ONS were not statistically significant for the IP and EP models (see Table 6).

Table 6.

Regression results for the Youth Self-Report indicators.

Variable	Model IP		Model EP		Model SC
	β	SE	β	SE	β	SE
Intercept	26.27*	9.04	14.48*	7.21	5.83*	2.36
TNL-ONE	0.07	0.10	−0.03	0.08	0.08*	0.02
TNL-ONS	−0.06	0.07	−0.09	0.05	−0.03	0.02
Age	−.95	0.66	0.37	0.53	−0.16	0.17
Gender	−3.82*	1.54	−2.54*	1.22	−0.38	0.40
Parental level of education	0.14	0.22	−0.11	0.17	0.04	0.06
R 2	.08		.08		.11

Note. IP = internalizing problem; EP = externalizing problem; SC= social competence; SE = standard error; TNL-ONE = Test of Narrative Language Oral Narration English; TNL-ONS = Test of Narrative Language Oral Narration Spanish.

^*Statistically significant at the .05 level.

Discussion

General Implications

Language and communication skills are essential for youth to express their needs, build positive relationships, regulate behavior, attention, and emotions, and understand their own and others' mental states (Carpendale & Lewis, 2006; Hebert-Myers et al., 2006; Lupyan, 2016; Nelson, 2007). DLD can hinder these skills, leading to the development of emotional, social, and behavioral concerns (Brinton & Fujiki, 1993). High co-occurrence of DLD and SBCs has been documented in monolingual children (e.g., Snow & Powell, 2011) and bilingual youth (e.g., Toppelberg et al., 2002); however, links between language development and behavioral concerns in bilingual Latinx youth have been vastly understudied (Conti-Ramsden et al., 2013). Understanding the link between DLD risk and SBCs in bilingual Latinx youth is crucial for developing effective interventions to improve outcomes for youth at risk for language and behavioral problems. The main goal of this study was to address several gaps in the literature on DLD risks and SBCs in bilingual Latinx youth.

Using cross-sectional data from a community-based sample, we examined differences in SBC classification and scores between bilingual Latinx youth with and without DLD risks. The study further explored youths' pragmatic language in English and Spanish associated with lower levels of SBCs. The implications of our findings are discussed in more detail below in relation to each research question.

Association of SBC and DLD Risk Identification

Our first research question was whether there were differences in SBC identification for youth with and without DLD risk. Our findings did not reveal statistically significant relationships between DLD risk and SBC identification in the CBCL and YSR scales, except for YSR IP scale. Although youth with DLD risk endorsed higher levels of IP (as seen in mean differences for Research Question 2), these generally did not translate into a difference in clinical classification. The significant difference across DLD groups in YSR IP identification indicates that youth who are self-reporting clinically significant IPs are more likely to have DLD risk; however, we did not observe this for parent report of IPs. This discrepancy aligns with a low-to-moderate correlation between the CBCL and YSR IP scores (r = .32, p = .001), and consistent with previous research on informant discrepancies (De Los Reyes & Kazdin, 2005). In adolescence, youth may have more awareness of the full spectrum of their IPs across settings (e.g., in school, with peers, at home) while parents may focus on the IPs present in a more restricted range of settings (e.g., at home with family), which could partially explain this discrepancy (e.g., Lau et al., 2004). Moreover, Latinx parents/caregivers may also be less attuned to youth SBCs due to lower mental health literacy, public stigma around psychological and emotional difficulties, and having to manage other potential stressors, including immigration and socioeconomic factors (Lau et al., 2004; Pérez-Florez & Cabassa, 2021). Youth with DLD risk may also be more likely to experience IPs across these settings, making the YSR more sensitive to capture relations between IP clinical risk and DLD risk. These findings also imply that youth presenting to clinical care for IPs (especially due to youth-reported problems) may also experience characteristics of DLD risk, and that collaborative care across behavioral health and speech and language services may benefit these youth.

Our second research question was to what extent SBC scores differ between youth with and without DLD risk. Our results showed statistically significant mean differences between youth with and without DLD risk on both the CBCL and YSR IP scales. Similar to our findings for Research Question 1 about clinical risk identification, these results highlight the relation between IPs and DLD risk for bilingual Latinx adolescents. These findings are consistent with a recent meta-analysis that identified a clear relation between language disorders and internalizing, but not externalizing, disorders (Donolato et al., 2022) and extend prior research with monolingual samples to document this within a bilingual Latinx sample that may already experience heightened risk for IPs. Early adolescents with DLD risk may experience language difficulties in childhood that affect the development of their emotional expression and regulation abilities, leading to a decreased ability to manage stress and greater IPs (Van dem Bedem et al., 2018). Clinically, youth with DLD risk may benefit from additional behavioral health supports as they enter adolescence to support positive emotion regulation strategies and reduce or prevent the development of IPs.

Language Development Factors Associated With Social and Behavioral Functioning

Our third research question was to what extent are children's narrative skills in English and Spanish associated with lower levels of SBCs, while controlling for child age, gender, and parental level of education. Our regression analyses revealed a negative but small and statistically significant association between oral narration in Spanish and SC (as measured by the CBCL). Although higher Spanish oral narration skills were a statistically significant predictor of lower SC, this effect was small and did not represent a clinically meaningful effect. For instance, an average increase of 25 points in the TNL-ONS measure would lead to a one-point decrease in the SC score. Regression analyses also showed a positive but small and statistically significant association between oral narration in English and SC (as measured by YSR). Interestingly, the findings were specific to SC, with no significant relations detected with internalizing or EPs. This may be because oral narration abilities, especially in English, may impact peer relationships and social skills in early adolescence, when more complex social relationships arise and more developed oral narration skills allow youth to navigate these social relationships with peers. This is consistent with previous research showing that adolescents with language disorders are less likely to have high-quality friendships (Durkin & Conti-Ramsden, 2007). Clinically, early adolescence may be a critical time for language and behavioral health providers to provide support in social development, such as offering social skills interventions for youth with DLD risk.

Limitations and Future Research Directions

Although this study addresses a critical gap in understanding the associations between language development risks and social and behavioral functioning in bilingual Latinx youth, it has some limitations. First, the cross-sectional nature of the data prevented us from establishing causal relationships or identifying developmental trends over time (Wang & Cheng, 2020). Longitudinal studies are needed to track language and behavioral risk development and better understand how they interact over time. However, given the limited research on the co-occurrence of DLD and SBC in bilingual Latinx youth, the present study provides valuable information to guide future longitudinal studies. Second, the study involved participants recruited from community settings, restricting the generalizability of the findings to higher risk groups. The relatively low frequency of DLD and SBC in the community-based sample may have restricted our ability to detect all associations among the two, and our findings may not align with findings typically observed in clinical samples (e.g., Toppelberg et al., 2002). Third, while the sample consisted of youth between the ages of 10 and 15 years, we did not examine potential age-based differences. Future research is needed to explore how developmental stage influences the association between DLDs and SBCs. Last, this study focused on three specific SBC indicators: IPs, EPs, and SC. Future research could expand the scope and examine other social and behavioral indicators.

Conclusions

Our findings are informative in highlighting the complexity of the interplay between language development and social and behavioral functioning, indicating that other factors may influence these relationships, and encourage researchers to examine additional variables that could influence these outcomes. The statistically significant findings in IP identification when self-reporting and the moderate effect sizes in IP score differences suggest that DLD risk may have a clinically meaningful impact on the development of IPs (Donolato et al., 2022), especially in early adolescence. Overall, our results highlight the association between DLD risk and social and behavioral functioning within a community-based sample of bilingual Latinx youth, extending prior research focusing mainly on clinical and monolingual populations. These findings point toward future research to further examine the links between DLD risk and SBCs for informing more effective interventions for this population.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Funding Statement

This work was supported by the National Institute of Child Health and Human Development Grant R03 HD087648 and by funding from the College of Education at the University of Texas at Austin (Principal Investigator: Erin M. Rodríguez).