Underrepresentation of Females and Racial and Ethnic Minority Groups in Research on Pragmatic Language in Autism: A Scoping Review

Abstract

Differences in language use in social contexts (i.e., pragmatics, such as conversational ability) serve as a defining clinical feature of autism, which can have an important impact on social functioning and well-being in a neurotypical society. Despite decades of research, most studies have not included representative samples, calling into question the generalizability of findings. This scoping review evaluates reporting practices and representation of autistic females and individuals from racially, ethnically (Hispanic/Latine), and linguistically diverse backgrounds. An initial pool of >1,800 articles were individually reviewed using a detailed coding system of exclusionary criteria, resulting in 308 empirical studies of pragmatic language in autism that were subjected to fine-grained analysis. Approximately 80% of studies reported sex, and about 70% documented language spoken. However, race and ethnicity were reported in less than 10% of studies. When reported, females and individuals from racial/ethnic minority backgrounds were starkly underrepresented. Reporting rates increased over time (1977–2022) for sex, ethnicity, and language spoken, and representation improved for racial minority groups and non-English speakers. Evidence suggests a striking bias towards inclusion of White, non-Hispanic/non-Latine males. This lack of representation has critical implications for interpretations of extant literature on pragmatics in autism, relevant to clinical characterization and future research.

Lay Abstract

Participation of Females and People from Racial and Ethnic Minority Groups in Studies of Social Communication in Autism

Social communication, or “pragmatic” language, refers to skills like maintaining topics of conversation and telling stories. People with autism often show differences in social communication, such as going off-topic during conversation. Social communication differences are important in diagnosing autism. These differences can also impact an autistic individual's daily life, such as making it difficult to find friends. For several decades, researchers have studied social communication in people with autism. However, many argue that most of these studies did not include females with autism or autistic individuals from non-White races (e.g., Black, Asian) or ethnicities (e.g., Hispanic/Latine), or who speak languages other than English as their native language. In this study, we sought to assess and document this critique by conducting a wide-ranging search of all published research articles on social communication in autism. Over 1,800 articles were identified, with 308 articles meeting criteria of the study. We discovered that most, but not all, studies reported sex and native language of study participants, while race and ethnicity were rarely reported. When this information was reported, females and people from racial and ethnic minority backgrounds were not represented as often as White, non-Hispanic males. In summary, there is a striking bias towards including White, non-Hispanic males in research on social communication in autism. Future studies need to include more females and more people from diverse backgrounds to support accurate diagnosis of autism, intervention for social communication differences when appropriate, and education for the neurotypical population on communication styles in autism.

Introduction

Autism is a neurodevelopmental condition characterized by challenges in social communication and the presence of restricted or repetitive behaviors, activities, or interests (American Psychiatric Association, 2013). Pragmatics is a highly complex domain of language which includes skills such as initiating and maintaining conversations, appropriately modulating prosody (e.g., rate, rhythm, volume), coordinating communication style with that of an interlocuter, and repairing communication breakdowns (Adams, 2002; Grice, 1975; Martin et al., 2017; Paul et al., 2025). Importantly, pragmatic language differences are a key feature of the social communication profile of autism that broadly and significantly affects daily life for autistic individuals ¹ (Landa, 2000). For instance, pragmatic skills are critical to the development of social relationships as well as practical life achievements, such as obtaining employment, where the fluid use of language in social contexts is important for successful job interviews and the day-to-day social demands of interacting with co-workers (Helland et al., 2014; Ketelaars et al., 2010; Morgan et al., 2014; Smith et al., 2015). This review approaches the topic of pragmatics in autism from a mostly clinical perspective, consistent with those adopted in the vast majority of studies reviewed. However, alternative and more modern viewpoints, such as the Double Empathy Problem (Milton, 2012), point out that pragmatic challenges may reflect mutual misunderstandings between autistic individuals and neurotypical communication partners. From this perspective, for example, social interactions in the workplace could be supported by educating neurotypical employees about the communication styles of their autistic co-workers.

Given its core role in autism diagnostic criteria (American Psychiatric Association, 2013) and potentially robust impact on social functioning, daily living skills, and well-being, pragmatics has been a primary focus of research on communication skills in autism (den Hartog et al., 2023; Liu et al., 2024). A wide body of literature documents significant pragmatic differences in autism, including reduced conversational reciprocity, difficulty with topic and information management, a pedantic or overly formal conversational tone, and differences in narrative skills such as less clear and cohesive stories, limited integration of protagonists’ perspectives, and less explanation of thoughts and feelings to motivate the ongoing narrative (Capps et al., 1998, 2000; Diehl et al., 2006; Ghaziuddin & Gerstein, 1996; Losh & Capps, 2003, 2006; Tager-Flusberg, 1995; Tager-Flusberg & Sullivan, 1995). However, while a rich literature characterizes the pragmatic profile of autism, the samples that comprise this work may not fully represent the autistic population, calling into question the generalizability of findings in this critical clinical-behavioral domain, central to diagnostic and clinical practice.

It is well known that autism is diagnosed more frequently in males than females, with a current prevalence ratio of 3.8:1 (Maenner et al., 2023). This may represent a true difference in prevalence; however, given that the clinical understanding of autism, including the development of widely used diagnostic instruments, has been largely based on autistic males, it is possible that autistic females display a distinct clinical phenotype not adequately captured with current instruments and diagnostic practices, leading to under-diagnosis (Fusar-Poli et al., 2022; Gould & Ashton-Smith, 2011; Hull & Mandy, 2017). In line with this hypothesis, some studies of sex differences in general autism symptomatology have found that social communication is less impacted in autistic females than in males (Beggiato et al., 2017; Lai et al., 2011; Lawson et al., 2018; Rynkiewicz & Łucka, 2018). However, findings are mixed overall, with some work reporting greater pragmatic language skills in males (Frazier et al., 2014; Grove et al., 2017; Ros-Demarize et al., 2020), and other studies finding no notable sex differences (Fulton et al., 2017; Holtmann et al., 2007; Mandy et al., 2012; Tillmann et al., 2018). A recent review of 19 studies specifically examining pragmatics and peer relationships in autism reported mostly mixed findings for sex differences, but some evidence that autistic females demonstrate stronger pragmatic skills than males (den Hartog et al., 2023). Other work, however, has characterized such differences as only “superficially better” in autistic females (Lai et al., 2011), which may help to explain why they are diagnosed later or remain undetected relative to autistic males.

Findings of seemingly stronger pragmatics in autistic females are consistent with recent theories of social camouflaging, in which females with autism display reduced autism-related characteristics through effortful attempts to mask autistic traits (Beck et al., 2020; Cage & Troxell-Whitman, 2019; Hull et al., 2020; Parish-Morris et al., 2017). In a systematic review, Tubío-Fungueiriño et al. (2021) suggested that the discrepancy in autism prevalence amongst males and females may be attributed to camouflaging, leading to a subtler display of autistic traits (from the perspective of neurotypical communication partners and diagnostic professionals, at least) and, potentially, reduced rates of diagnosis in females. Whereas camouflaging has been hypothesized to serve adaptive functions during social interactions (Tubío-Fungueiriño et al., 2021), it has also been associated with mental health outcomes (Cook et al., 2021), suggesting more research is needed to fully understand the impact of masking on pragmatic language phenotypes and overall emotional well-being, especially in autistic females. Given the phenotypic heterogeneity inherent to autism and recent evidence indicating that autistic females are under-diagnosed, perhaps in part due to a male-centric conceptualization of autistic symptomatology and higher rates of camouflaging, characterizing pragmatic language phenotypes in autistic females is vital to identifying relevant treatment targets and ultimately for developing effective interventions generalizable to a wider population on the autism spectrum when intervention is appropriate. To address this concern, the current study aimed to systematically investigate the representation of females across the lifespan in extant literature on pragmatics in autism. An additional focus concerns the reporting and representation of race and ethnicity in the autism pragmatics literature, as described below.

Historically, as is the case in clinical research involving other neurodevelopmental conditions (Riccioni et al., 2024), individuals from racial and ethnic minority backgrounds have been underrepresented in autism research (Girolamo et al., 2023; Kover & Abbeduto, 2023; Malone et al., 2022; Mire et al., 2024; Roman-Urrestarazu et al., 2021). Several explanations have been proposed for this, including disparities in rates of diagnosis (Weitlauf et al., 2024), as well as cultural views toward autism and research (Girolamo et al., 2023). For example, according to the Centers for Disease Control and Prevention (CDC), Hispanic children with autism are diagnosed later than their non-Hispanic White counterparts (Baio et al., 2018), which could lead to their exclusion from research focused on autism in early childhood. Furthermore, the reporting of racial and ethnic data in autism research has been inconsistent at best. In a recent systematic review of over 1,000 autism intervention studies published from 1990–2017, only 25% of the studies reported on participants’ racial and ethnic backgrounds (Steinbrenner et al., 2022), an overwhelming minority of studies. In studies where this information was reported, samples were disproportionately White, and the authors noted little change in participant demographics over time. Similarly, Girolamo et al. (2023) found that only 29% of reviewed studies reported on race and ethnicity, and for those that did, 77% of participants were White. Given the known disparities in autism diagnosis between White and minoritized racial and ethnic groups, the underrepresentation of individuals from racial and ethnic minority backgrounds in autism research is of concern, particularly since early diagnosis and access to treatment can be vital for optimal outcomes for autistic individuals and their families. Additionally, the lack of sample diversity limits the generalizability of findings and the applicability of interventions across diverse populations. Many standardized autism assessments and interventions were developed and validated primarily with White, Western samples, raising concerns about their cultural appropriateness and effectiveness for individuals from varied backgrounds (Kim et al., 2022; Stevanovic et al., 2021; Vanegas et al., 2016).

In sum, although prior autism research has been centered on White male samples, findings have often been generalized to the broader autistic population, including females and individuals from racially and ethnically diverse backgrounds. Thus, characterizing the current state of the literature and prevailing scientific reporting practices for sample demographics is a vital first step towards increasing the representation of females and racial and ethnic minority groups in autism research, especially in studies focused on pragmatics. To this end, the current study addressed two primary aims, both focused on marginalized autistic groups in pragmatic language research: 1) to provide a detailed description of reporting practices and the representation of autistic females in autism pragmatics research, and 2) to provide a detailed description of racial and ethnic (Hispanic/Latine ² ) minority representation, and related reporting practices, in this research.

Method

This investigation utilized a scoping review approach (vs. systematic) as the most apt evidence synthesis strategy for identifying gaps in the research literature, and assessing research methodology approaches (Munn et al., 2018). Procedures followed search strategy guidelines and data extraction procedures required for scoping reviews (Munn et al., 2018), as detailed below.

Search Strategy

Using a combination of targeted search terms, a literature search was conducted to identify studies of pragmatic language in individuals with autism. PubMed and PsycInfo were selected, having the greatest breadth of clinical and psychological research studies in their databases pertaining to autism. The final search string consisted of a combination of terms referring to 1) the condition of interest (i.e., autism spectrum disorder, ASD, autistic, autism, Asperger, PDD-NOS, pervasive developmental disorder-not otherwise specified) and 2) the language domain of interest (i.e., pragmatic, narrative, discourse, communication breakdown). Asperger's syndrome and pervasive developmental disorder-not otherwise specified were included as search terms to increase the breadth of our search results, given they were included as diagnoses in the Diagnostic and Statistical Manual (DSM) 4^th Edition (American Psychiatric Association & Association, 1994) before being subsumed under the umbrella of autism spectrum disorder in the DSM-5 (American Psychiatric Association, 2013). Of note, some terms were truncated and demarcated with an asterisk to capture all potential variants of the search term (e.g., narrat* for narrative and narration). Autism-related terms were required in the article titles, and language-related terms were required in the titles and/or abstracts of articles populated in the search pool. The literature search was completed on November 2^nd, 2022. This approach resulted in a targeted pool of studies related to autism and pragmatic language (n = 2,424). Duplicate articles present in searches of both databases were removed, resulting in a final pool of 1,812 articles.

Data Selection Criteria

Studies in the resultant search pool (n = 1,812) were systematically reviewed by a team of trained and reliable reviewers, who examined the title, abstract, and, when necessary, the full-article text to determine eligibility for analysis. Exclusion codes were refined through multiple rounds of review between coders to address important nuances in coding. Studies were excluded in accordance with the criteria outlined in Table 1.

Table 1.

Study Exclusion Criteria.

Not applicable	Did not measure pragmatic language skills, or measured a skill that supports pragmatic language rather than directly measuring pragmatics (e.g., social cognition)
Written skills	Measured written pragmatic skills, as opposed to verbal pragmatic abilities
No AU group	Did not measure pragmatics in an AU group
Syndromic AU group	Included a sample of AU participants with known genetic causes of AU (e.g., fragile X syndrome, Rett syndrome), but no idiopathic AU group
Non-empirical study	Was not an empirical, primary research article (i.e., reviews, meta-analyses, editorials, errata, textbook chapters)
Pragmatics from diagnostic instrument	Included as primary measure of pragmatics only summary scores derived from a general diagnostic instrument (e.g., Autism Diagnostic Observation Schedule)
Intervention study	Evaluated an intervention or clinical trial targeting pragmatic skills
Case study or series	Employed a case study or case series design
Not peer-reviewed	Was not peer-reviewed (e.g., dissertations, conference papers)
Non-English or full text unavailable	Was unavailable in an English, full-text version

Note. AU = autism.

Consensus Coding and Reliability

When uncertain about any coding category, coders flagged articles for discussion and consensus coding. A total of 248 articles (13.69% of the 1,812 articles) were flagged and consensus coded. In addition, articles were randomly selected for reliability analysis twice during the project, resulting in a sampling of 9.27% of the 1,812 articles. Reliability analyses revealed that coders agreed on a study's eligibility for inclusion 91.39% of the time, and when disagreements arose, studies were consensus coded. Taken together, a total of 261 articles (14.40% of 1,812 articles) were consensus coded.

Data Extraction and Analysis

For the 308 articles that met inclusion criteria, demographic data were extracted as outlined in Table 2. Importantly, if a study did not report key demographic data (e.g., sex composition), “not reported” was entered for that variable. When the reported sex, racial, or ethnic composition of a study sample did not sum to the total number of subjects reported, “reporting concerns” was entered for the variable of interest. “Reporting concerns” was also entered in cases where race and ethnicity appeared to be conflated (e.g., treating Hispanic as a race that was mutually exclusive with White, Black, etc.), limiting the ability of coders to extract accurate information regarding a study's racial and ethnic composition. Articles with “reporting concerns” were excluded from pertinent analyses.

Table 2.

Data Extracted for Included Studies.

Basic study characteristics	Publication year
	Size of autistic sample
Sex composition (subject n)	Males
	Females
Racial composition (subject n)	White
	Black or African American
	Native Americana
	Asian
	Native Hawaiian or Other Pacific Islander
	Multi-racial, biracial, mixed race
	Other race
	Race not reportedb
Ethnic composition (subject n)	Non-Hispanic/Non-Latine
	Hispanic/Latine
	Ethnicity not reported
Language spoken	English
	Non-English
	Cross-linguistic
	Language not reported

Notes. Table depicts the information extracted from studies included in analyses (n = 308). Subject n's refer to autistic groups only. ^a Subsumes the U.S. Census categories American Indian and Alaska Native. ^b Refers to participant not electing to report their race, not that the study did not report race.

The ratio of autistic males to females was extracted from all studies that included both males and females, including studies that collapsed males and females into a single group as well as studies that examined sex effects (i.e., in preliminary analysis, as a covariate, or through planned group comparisons). For studies that analyzed data for males and females separately, we also considered whether they were adequately powered to detect sex effects with an a priori criterion of n ≥ 20 females. To examine the representation of autistic females in pragmatics research over time, we conducted a logistic regression and several Pearson correlations between publication year and 1) whether sex was reported, 2) autistic female sample size, 3) overall male-to-female sex ratio, 4) sex ratio for studies that collapsed males and females with autism into a single group, and 5) sex ratio for studies that examined sex effects. To determine whether the representation of racial, ethnic, and linguistic diversity has changed with time, a series of logistic regressions were performed to examine relationships between publication year and 1) whether race was reported, 2) whether ethnicity was reported, and 3) whether language was reported. Subsequent Pearson correlations and logistic regressions were performed to examine the relationship between publication year and 1) the percentage of autistic participants from racial minority groups, 2) the percentage of Hispanic/Latine subjects with autism, and 3) the percentage of studies that employed a cross-linguistic approach or were conducted in a language other than English.

Results

As noted in Methods, the initial literature search yielded a pool of 2,424 articles related to pragmatic language in autism. Of those, 612 duplicates were identified and removed, leaving a total of 1,812 unique articles. After applying exclusion criteria, 308 articles on pragmatic language in autism remained and were included in analyses. These articles were published between 1977 and 2022. Figure 1 summarizes methodological details, as well as the resulting number of studies per sub-category (e.g., 24 studies that reported race). Figure 1 uses a flow diagram adapted from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework (Page et al., 2021) to illustrate the study selection process. The results of analyses examining the reporting and inclusion practices of studies across sex, race, ethnic, and language backgrounds are presented below.

Figure 1.

PRISMA Flow Chart of Study Selection Process. Notes. PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses (Page et al., 2021). The literature review is scoping, not systematic. AU = autism.

Sex-Related Analyses

Of the 308 included studies, 81.82% (n = 252) reported sex of participants without reporting concerns. Within this pool of studies, 99.60% (n = 251) included males, whereas 80.95% (n = 204) included females. Of the studies that reported their sample's sex composition, 80.56% (n = 203) included both males and females, with 11.82% (n = 24) exploring differences in pragmatic language across sex, either in preliminary analysis, as a covariate, or through planned group comparisons. Notably, when sex differences were explored, only 12.5% (n = 3) of studies were adequately powered to detect sex effects, including a sample size of 20 or more females (i.e., with n's of 34, 41, and 50 female participants, respectively).

On average, the sex ratio (M:F) for studies that included both males and females (n = 203) was 5.72:1 (see Figure 2). Of these studies, those that did not examine sex effects (n = 179) had an average sex ratio of 5.95:1, whereas those that examined sex effects (n = 24) had an average sex ratio of 4.01:1.

Figure 2.

Representation of Autistic Females in Pragmatics Research Over Time. Note. A. Overall number of autistic female participants increased significantly over time. B. Male:Female ratio did not change as a function of time, hovering around 5:1.

A logistic regression revealed that publication date significantly predicted whether sex was reported in the article's sample demographics (β = 0.045; p = 0.004). Pearson correlations were performed to further characterize the relationship between publication year, female sample size, and sex ratio (M:F). Results revealed a significant positive relationship between publication year and female sample size (r = 0.282; p < 0.001; see Figure 2A), whereas the correlation between publication year and sex ratio was nonsignificant (r = -0.060; p = 0.399; see Figure 2B). No significant associations between publication year and sex ratio emerged (ps > 0.100) when separately examining studies that collapsed participants across sex (r = -0.034) or analyzed the effects of sex on pragmatics (r = -0.137).

Race, Ethnicity, and Language-Related Analyses

Only 7.79% (n = 24) of the 308 included studies reported on the racial composition of their autistic sample without reporting concerns. Of these studies, 88.76% of subjects with autism identified as White, 3.54% identified as Black or African American, 0.57% identified as Native American, 2.12% identified as Asian, 0.10% identified as Native Hawaiian or Other Pacific Islander, 4.30% identified as Multiracial, Biracial, or Mixed Race, 2.02% identified as other race, and 1.88% elected not to report their racial background (see Figure 3). Results of a logistic regression assessing the relationship between publication year and whether race was reported were statistically nonsignificant (β = 0.043; p = 0.165). However, for studies that reported on the racial composition of their study samples, the relationship between publication year and the percentage of autistic participants from racial minority groups was marginally significant and positive (r = 0.371; p = 0.074).

Figure 3.

Racial and Ethnic Composition of Autistic Participants in Pragmatic Language Studies. Note.

With regards to ethnicity, 2.27% (n = 7) of studies documented the ethnic composition of their sample, while 91.56% (n = 282) of studies did not, and the remaining 6.17% (n = 19) had reporting concerns. For the studies that clearly reported on ethnicity, 72.78% of participants with autism were non-Hispanic/non-Latine on average (see Figure 3). Results of a logistic regression assessing the relationship between publication year and whether ethnicity was reported were statistically significant and positive (β = 0.260; p = 0.049). No significant relationship was detected for publication year and percentage of Hispanic/Latine participants (r = 0.336; p = 0.461).

The language spoken by participants was documented in 70.78% (n = 218) of studies. Of these, 56.88% included English speakers, 40.37% included non-English speakers, and 2.75% employed a cross-linguistic design. Results of a logistic regression assessing the relationship between publication year and whether language was reported were statistically significant and positive (β = 0.057; p < 0.001). The relationship between publication year and the percentage of studies classified as Non-English or cross-linguistic was statistically significant and positive (β = 0.105; p < 0.001).

Discussion

This study reports findings from a scoping review of the extant literature on pragmatic language in autism, with a focus on the representation of females, racial and ethnic minority groups, and non-English speakers in this body of research. Pragmatics is the language domain most influenced by autism, and pragmatic language ability can have important implications for daily functioning and well-being. Of the 1,812 articles captured in our search, 308 met inclusion criteria for subsequent analyses. Reporting practices for sample demographics and the representation of these marginalized groups were evaluated in the final study pool to inform the extent to which disparities in reporting and representation may impact our current understanding of pragmatic language differences in autism. Importantly, findings suggest that reporting of these critical demographic characteristics is low, especially for race and ethnicity. Further, when studies do report this information, autistic females and individuals from racial and ethnic minority groups are underrepresented relative to White, non-Hispanic/non-Latine males with autism, and studies are significantly underpowered to examine potential differences and similarities across these groups. Over time, some, but not all, of these metrics have improved; the implications of these findings are discussed in detail below.

Representation of Females

Approximately 82% of reviewed studies on pragmatic language in autism reported the sex of participants. This exceeds findings from a previous review of studies examining language impairments more broadly and in autistic children, indicating reporting rates of 54% (Girolamo et al., 2023). Within this context, an 82% reporting rate suggests that studies of pragmatic language, across the lifespan, are more transparent about the sex composition of their samples. Nevertheless, given known sex effects in autism and other neurodevelopmental conditions, current practice should dictate 100% reporting.

For those studies that did report the sex of participants, females were significantly underrepresented, as nearly 100% of studies reporting on sex included males, whereas only ∼80% included females. Strikingly, when a study included both males and females, the average sex ratio (M:F) was about 5.7:1. This ratio is notably more disparate than the current autism prevalence ratio of 3.8:1 (Maenner et al., 2023), providing a quantifiable metric to characterize the underrepresentation of females in autism research, while considering differences in prevalence across sex. Although the sex ratio of study subjects did not change over time, female sample size significantly increased, suggesting that although more recently published studies included larger samples sizes overall, males continue to be significantly overrepresented. When females with autism were included in studies of pragmatic language, they were most often collapsed with males into a single group for analyses, with no attention paid to potential sex differences. Only 12% percent of studies that included females considered sex effects, either during preliminary analysis, as a covariate, or by including a discrete group of females. However, of the studies that included females, only three were adequately powered to detect sex effects.

Pragmatic language profiles characterized through research play a vital role in the diagnosis of autism (American Psychiatric Association, 2013), and in guiding clinical intervention approaches when intervention is appropriate. Extant pragmatic language research over the past several decades has played a pivotal role in the scientific and clinical understanding of key challenges experienced by autistic individuals, including identification of relevant treatment targets and direction of public health resources. However, the underrepresentation of females in the literature may contribute to sex disparities in autism identification, as the female pragmatic phenotype has yet to be fully illuminated. This gap in research is particularly problematic given robust evidence of distinct patterns of pragmatic skill development in males versus females, and across different languages and cultures in the general population (Berghout Austin et al., 1987; Burton et al., 2020; Conlon et al., 2019; Cook et al., 1985; den Hartog et al., 2023; Kauschke et al., 2016; Lai & Szatmari, 2020; Leaper, 1991; Sturrock et al., 2020b).

The few studies focused on autistic females have revealed that they may show poorer global pragmatic skills and less sophisticated narratives than non-autistic females, but stronger overall pragmatic and narrative skills than autistic males (Burton et al., 2020; Conlon et al., 2019; Kauschke et al., 2016; Sturrock et al., 2020a; Sturrock et al., 2020b). This is consistent with findings from a recent scoping review of pragmatics and friendship based on 19 articles (den Hartog et al., 2023), which reported that autistic females tend to demonstrate stronger pragmatic skills than autistic males, notwithstanding some mixed findings, and that females camouflage autistic traits while interacting with peers. Thus, continued research efforts to characterize the more subtle female pragmatic phenotype in autism and increased clinical translation of existing research findings will be critical to the reliable diagnosis of autism in girls and women. Further still, while the mechanisms underlying sex differences in pragmatic language have yet to be fully revealed, evidence suggests that auditory and visual processing differences contribute differentially to autism symptomatology in males and females (Aykan et al., 2020), which may underly downstream differences in social communication skills. As such, future work with a female focus should aim to further disentangle these mechanistic differences to inform more precise targets for pragmatic language interventions, when such treatment is recommended.

Representation of Racial, Ethnic, and Linguistic Diversity

Findings from this review also reveal that studies of pragmatic language in autism require more transparent and consistent reporting of the racial and ethnic compositions of study samples. Strikingly, only about 8% of included studies clearly reported their study sample's racial composition, and approximately 2% clearly reported ethnic composition. Taken together, the vast majority of included studies did not report, or reported unclearly, the racial and ethnic composition of samples, limiting the interpretability and applicability of research findings to specific target groups (e.g., Hispanic females, Black males, etc.). Of note, these findings are consistent with systematic reviews of autism intervention studies (Steinbrenner et al., 2022), and studies of language impairments in autism not restricted to pragmatics (Girolamo et al., 2023), where race or ethnicity was reported in less than 30% of studies. Although findings from this review revealed that ethnic demographic data is being reported more frequently in recent years, the reporting of racial demographic data is not similarly improving. Thus, more complete and transparent reporting of racial and ethnic demographic data is warranted to more directly assess the generalizability of pragmatic language findings to specific cultural groups. When these demographic data are reported in the extant literature, racial and ethnic minority groups are underrepresented relative to their White and non-Hispanic/non-Latine peers. In the present review, studies that reported on racial and ethnic sample compositions were overwhelmingly White (i.e., 89%) and non-Hispanic/non-Latine (i.e., 73%).

Although the reporting of race did not change over time, studies that did report the racial composition of their samples included more participants from racial minority groups over time, a marginally significant result that suggests a subtle increase in racial diversity among autistic study samples. The pattern of results was opposite for ethnicity. Reporting of ethnicity did increase over time; however, for those studies that reported ethnicity, the percentage of Hispanic/Latine participants did not increase, suggesting limited change to the ethnic diversity of autistic research samples over time. Although these results are somewhat more encouraging than those of Steinbrenner et al. (2022), which had indicated little change over time in representation of racial and ethnic diversity, the reporting and representation of racial and ethnic minority groups in pragmatic language research remain an immediate concern.

This scoping review also examined linguistic diversity. Approximately 30% of studies did not report the language spoken by participants, although it is possible that many of these were English language studies that did not explicitly report on language spoken. Of the remaining studies, about 57% were clearly English language studies, 40% were non-English, and 3% examined pragmatics cross-linguistically. Although one third of the studies on average did not report language spoken, analyses revealed that reporting of language has increased over time, as has the number of studies classified as non-English and cross-linguistic. While these trends are encouraging, native language should be reported explicitly in all studies, and further research on pragmatic language in non-English and especially cross-linguistic samples will be critical.

Summary, Limitations, and Directions

This scoping review has identified the underreporting of sample characteristics and the underrepresentation of marginalized autistic populations as overarching concerns that may limit the generalizability of pragmatic language findings and contribute to the underdiagnosis of autism in females, individuals from racial and ethnic minority backgrounds, and non-English speakers. Of course, representation can only fully be assessed when the appropriate demographics are reported. Thus, whenever possible, researchers should report, and journals should require that they report, demographic characteristics pertaining to sex, race, ethnicity, and language spoken. Importantly, race and ethnicity should not be conflated in these efforts, as meaningful information is lost when they are treated as mutually exclusive. Assuming that, in most cases, the underrepresentation that we identified would be the same or more concerning in studies that did not report this demographic information, the problem of underrepresentation is clearly of utmost importance. Given disparities in the diagnosis of autism, recruitment of similarly sized male and female groups, and groups that match or exceed the population estimates regarding race and ethnicity, certainly represent real and significant challenges that will be critical to address in future research in general, and to further investigations of important and still poorly understood pragmatic profiles.

This investigation also has some limitations that should be considered in future research. Our ability to fully characterize the state of the literature, especially as it pertains to Hispanic/Latine representation and linguistic diversity, was limited by excluding studies that were not published in English. Finally, the first aim provides an initial step in evaluating systematically how the field has considered biological sex in studies of pragmatics in autism; however, it can help to inform future studies of gender identity in autism and how gender (which does not necessarily conform with sex) influences pragmatics in potentially unique ways. Given the complexity of gender expression, gender identity (e.g., male, female, or non-binary) should be assessed in addition to considering sex as a dichotomous biological variable.