Skip to main content
NCBI home page
Autism & Developmental Language Impairments logoLink to Autism & Developmental Language Impairments
. 2026 Aug 3;11:23969415261465246. doi: 10.1177/23969415261465246

Adaptive Functioning in Expressively Nonverbal and Minimally Verbal Autistic Children: Nonverbal Intelligence as a Predictive Factor

Ekaine Rodríguez-Armendariz 1,*,✉, Zuriñe Ábalos 1, Elena Castroviejo 1,**, Agustín Vicente 1,2
PMCID: PMC13434895  PMID: 42553549

Abstract

Background and Aims

Adaptive functioning is a strong predictor of long-term outcomes for autistic individuals, yet evidence regarding its association with ADOS-2 comparison scores, nonverbal IQ (NVIQ), and early language development remains inconsistent. Findings on the contributions of IQ and ADOS-2 comparison scores are mixed, and early linguistic abilities have received comparatively less attention. This study examined how receptive language delay, NVIQ, and ADOS-2 comparison scores relate to later adaptivity in nonverbal and minimally verbal autistic children.

Methods

Participants were 4- to 12-year-old autistic children (N = 44) who completed ADOS-2 Module 1 at baseline, indicating limited expressive language but variability in age and developmental profiles. Baseline assessments included ADOS-2 comparison scores, NVIQ (Leiter-3), and receptive vocabulary (PPVT-III). Adaptive functioning was measured 2 to 3 years later using Vineland-3 subscales. To examine the contribution of each variable, we fitted multiple linear regression models with Vineland-3 scores as the dependent variable and standardized test scores as continuous predictors, while controlling for children's age (in months) and the testing interval between the ADOS-2 and the Vineland-3. We first fitted separate models for each predictor, followed by additive models including the three predictors. Additional models included interaction terms between NVIQ and ADOS-2 comparison scores, PPVT-III scores and ADOS-2 scores, NVIQ and PPVT-III scores, and their three-way interaction. Model comparisons were conducted to determine which model provided a better fit for the data.

Results

NVIQ emerged as the only significant predictor of adaptive functioning. Although ADOS-2 comparison scores and receptive vocabulary were significantly associated with adaptivity in the simpler models, these effects were no longer significant once NVIQ was included, suggesting that their predictive value was largely accounted for by NVIQ. Unlike earlier reports, higher NVIQ in this sample was not associated with greater adaptive difficulties, which suggests this relationship is not uniform across subgroups of autistic children. Interaction analyses did not reveal significant moderation effects, although some interactions involving NVIQ or receptive language delay and ADOS-2 approached conventional significance thresholds. These effects require confirmation in larger samples.

Conclusions

Adaptive functioning in autistic children with limited expressive language abilities is predicted more strongly by NVIQ than by ADOS-2 comparison scores, receptive language delay or the interaction of these three variables.

Implications

These findings indicate that early NVIQ remains a meaningful predictor of adaptive functioning in minimally verbal autistic children, underscoring the importance of including nonverbal cognitive assessments in early prognostic evaluations. The absence of meaningful associations with ADOS-2 comparison scores and receptive vocabulary delay when NVIQ was accounted for suggests that adaptive outcomes may depend less on trait profile or linguistic comprehension than may be assumed, and that autistic children may successfully compensate for early marked autistic traits and comprehension difficulties.

Keywords: Nonverbal and minimally verbal autism, adaptivity, nonverbal intelligence, language delay

Introduction

Adaptive functioning skills, defined as the abilities that allow individuals to operate independently in daily life, are a critical predictor of long-term outcomes for autistic individuals. These outcomes range from educational achievement (De Bildt et al., 2005) to everyday independence (Paul et al., 2004). Pinpointing the factors that predict stronger adaptive behavior is therefore essential for anticipating areas of need, tailoring interventions more effectively, and supporting autistic individuals in achieving optimal independence and quality of life.

Autism, however, is marked by substantial heterogeneity across the very domains expected to contribute to adaptive behavior. Intellectual ability, language skills, and autistic traits all vary widely among autistic individuals, and each domain is theoretically linked to important aspects of everyday functioning: cognition supports problem-solving and self-management; language underpins communication, learning, and social participation; and autistic traits reflect the degree of social-communication difficulty and behavioral rigidity a person experiences. Yet these abilities and characteristics do not co-vary in predictable ways, nor, we may suspect, do they combine additively. In many individuals, strengths in one domain coexist with challenges in another domain, and the relationships among cognition, language, symptom expression, and adaptive functioning can turn out to be nonlinear or conditional on developmental stage.

A growing and recent body of research has examined how different developmental domains—including autistic traits, cognitive ability, language, and executive functions—shape adaptive functioning in autistic individuals. Despite considerable attention, findings remain inconsistent, in part because studies vary in sample characteristics, measurement instruments, and developmental stages.

Work focusing on autistic traits has produced particularly mixed results. For example, Failla et al. (2024), studying a diverse group of 60 autistic children aged 35 to 120 months with full-scale intellectual quotients (IQs) ranging from 16 to 118, found that ADOS-2 calibrated scores, hereafter referred to as the comparison score (CS), directly predicted poorer adaptive functioning when assessed using the PsychoEducational Profile (PEP-3). Their findings highlight that higher autistic trait scores were consistently associated with lower adaptivity, even after considering age and language groups. In contrast, large-scale studies such as that by Kanne et al. (2011), involving 1,089 verbally fluent autistic youths between the ages of 4 and 17 with full IQ scores ranging from 19 to 167, detected little to no association between ADOS-2 comparison scores and adaptive outcomes on the Vineland Adaptive Behavior Scales (VABS-II), despite strong IQ–adaptivity correlations. A similarly broad study by Pathak et al. (2019) using data from 2,538 autistic children aged 5 to 17, also found only weak links between ADOS-2 comparison scores and adaptive functioning once age and cognitive ability were accounted for. Yet, other work points to more domain-specific symptom impacts: using the large EU-AIMS LEAP cohort (N = 417, age range = 6–31 years, full-scale IQ = 40–148), Tillmann et al. (2019) demonstrated that social-communication trait scores, rather than repetitive behaviors or sensory features, was the most consistent predictor of reduced adaptive functioning.

Across studies, IQ emerges as one of the strongest and most consistently examined predictors, though its relationship with adaptivity is far from straightforward. Failla et al. (2024) observed a linear, positive effect of composite developmental IQ—measured by Griffiths Scales of Child Development (Griffiths III)—across all adaptive domains in early childhood. In Kanne et al.'s large adolescent sample, verbal IQ (VIQ) explained over half of the variance in adaptive functioning beyond ADOS-2 comparison scores, and older youth showed a widening mismatch between VIQ and daily living skills. However, Pathak et al.'s stratified analyses (<70, 70–85, >85) revealed a nonlinear pattern: while borderline and average IQ groups showed lower adaptive scores than expected, children with low IQ (<70) paradoxically demonstrated better-than-expected adaptivity relative to their cognitive level. Pathak et al.'s measures of IQ were obtained mostly through either the Stanford–Binet or the Wechsler scales, the Stanford–Binet being “less verbally-mediated” than the Wechsler scales. Tillmann and colleagues similarly found that higher IQ was associated with a larger discrepancy between cognitive ability measured by the Wechsler scales and functional skills in older individuals, suggesting that IQ-adaptive gaps may actually widen for those with higher cognitive potential. Most strikingly, Alvares et al.'s (2020) large registry analysis of 2,225 autistic children and teenagers (ages 1–18) showed that IQ (measured with a variety of scales) is often a poor predictor of functional abilities, especially for autistic individuals without intellectual disability, calling into question the clinical utility of terms like “high-functioning autism.”

With respect to language, Miranda et al. (2023) conducted a longitudinal study of 45 autistic children without intellectual disability, aged 7 to 11 at baseline, to examine how early structural and pragmatic language skills predict later functional outcomes. Using parent-completed Children's Communication Checklist-2 (CCC-2; Bishop, 2004) ratings in childhood and follow-up assessments more than four years later, they found that pragmatic language was the strongest and most consistent predictor of adolescent outcomes across social, adaptive, and behavioral domains. Structural language (especially semantics) predicted a more limited set of outcomes, mainly prosocial behavior and socialization skills.

Finally, some evidence also suggests some role of executive functions (EF). A longitudinal study by Fossum et al. (2025), following 66 individuals (26 autistic, 40 non-autistic) from late childhood (mean age ≈ 11.8) to emerging adulthood (mean age ≈ 21.4), found that childhood EF difficulties robustly predicted poorer adaptive functioning a decade later. Importantly, EF predicted adaptive outcomes more strongly than autism diagnosis or co-occurring symptoms.

Taken together, the literature illustrates a complex and often inconsistent picture: while IQ frequently emerges as a relatively robust, though far from straightforward, predictor of adaptive functioning, findings regarding ADOS-2 comparison scores are considerably more variable, and research on language and EF is comparatively limited. Moreover, the available evidence increasingly suggests that these domains do not operate in isolation. Rather, these domains appear to interact in complex, sometimes counter-intuitive ways. In this regard, the patterns reported by Hill et al. (2015) complement findings from larger studies such as those by Tillmann et al. (2019) and Alvares et al. (2020): all three highlight that higher cognitive ability does not necessarily translate into stronger daily living skills and may even magnify discrepancies between potential and actual functioning. Hill et al. (2015) demonstrated that the relationship between ADOS-2 comparison scores and adaptive functioning can shift direction depending on a child's cognitive level and developmental stage, revealing patterns in which greater ADOS-2 comparison scores were associated with better adaptive outcomes under certain combinations of age and IQ. Such results challenge assumptions that either IQ or ADOS-2 comparison scores exert uniform, linear effects on adaptivity and highlight the need to consider the conditional and potentially interactive nature of these influences.

As Petrolini and Vicente (2022) argue in their broader theoretical work on non-additive interactions, developmental variables often combine in ways that cannot be inferred simply by summing their individual contributions. A somewhat similar view is held by Borsboom's (2017) proposal to understand mental conditions as networks of symptoms. From a network perspective, ADOS-2 comparison scores, IQ, and language ability can be conceptualized not as independent predictors of adaptivity, but as mutually interacting nodes within a system that shapes adaptive functioning. ADOS-2 scores relate to the intensity of autistic behavioral patterns (e.g., social-communication differences, restricted/repetitive behaviors), which can directly constrain opportunities for adaptive learning (e.g., reduced social engagement or flexibility). However, their impact on adaptivity is likely mediated and moderated by cognitive and linguistic resources. For instance, higher IQ may support compensatory strategies (e.g., problem-solving, rule-based coping), while stronger language abilities can facilitate social understanding, help-seeking, and environmental negotiation.

Crucially, these factors likely reinforce each other over time. Higher ADOS-2 scores may limit language-rich interactions, which in turn can slow language development, further constraining adaptive communication. Conversely, stronger language skills may attenuate the functional impact of ADOS scores by enabling alternative routes to adaptation. IQ and language may be also interlinked, such that differences in reasoning abilities can support or hinder the acquisition and flexible deployment of language in real-world contexts. In this sense, we may think, adaptivity may reflect not the additive contribution of ADOS comparison scores, IQ, and language, but the structure and strength of their connections.

In our study we focused on children who had completed ADOS-2 Module 1, a group characterized by very limited expressive language at the time of assessment. This provides a meaningful baseline while accommodating variability in age and developmental profiles. Despite their relevance, children with this profile remain an understudied population. Among autistic children classified as nonverbal or minimally verbal, substantial heterogeneity has been documented, both in receptive language (Chen et al., 2024; Saul et al., 2026) and nonverbal intelligence (Slusna et al., 2021; though see Guerrera et al., 2025), with a possible association with ADOS-2 comparison scores (Guerrera et al., 2025). In our paper, we examined the relations of receptive language delay, nonssverbal IQ (NVIQ), ADOS-2 comparison scores, and their interactions to adaptive outcomes (assessed using the Vineland-3 subscales) after a period of 2 to 3 years.

While previous studies (e.g., Miranda et al., 2023) suggest that pragmatics plays a more important role than semantics or structural language in adaptivity scores, because our sample consisted of nonverbal or minimally verbal children at least production-wise, measures of pragmatic language could not be collected. We therefore relied on receptive language scores, which are both obtainable and informative in this population. However, our primary interest was not children's absolute language level, but rather the degree of language understanding delay; that is, how far their linguistic comprehension skills diverged from what would be expected for their chronological age (CA). Our rationale was that receptive vocabulary delay as a proxy for linguistic difficulties may relate more strongly to adaptivity than raw language scores alone: for example, it is not equivalent for a 7-year-old and a 3-year-old to have a receptive vocabulary of a typically developing 2-year-olds. Although the level of linguistic comprehension difficulties itself has not been systematically examined in relation to adaptive functioning in autism, we hypothesized that such difficulties could contribute to how each child adapts to their environment, as mentioned in previous lines. For instance, it has been suggested that children with Developmental Language Disorders exhibit issues in adaptivity related to their level of language difficulties (Fujiki et al., 2004; Maggio et al., 2014).

In sum, our research questions were:

  1. Do receptive language delay, NVIQ and ADOS-2 comparison scores predict adaptivity in different areas controlling for the time elapsed since the initial assessment?

  2. Are interactions between these variables more predictive of adaptivity scores than the main effects of each variable?

Although this investigation was largely exploratory, several predictions could be drawn based on previous research. Concerning variable predictability, we expected to find some relation between NVIQ and general adaptivity. It was unclear whether such a relation would also emerge for the other two variables, namely ADOS-2 comparison scores and receptive vocabulary delay. As explained above, previous studies report more conflicting results regarding the relation between ADOS-2 comparison scores and adaptivity than between IQ and adaptivity. In contrast, receptive language delay has not been studied in this context, and although research on children with language disorders suggests some connection between language difficulties and adaptivity, it was difficult to predict whether linguistic difficulties would show a similar relation in autism. It can be expected that language delay will affect communication abilities, but it remains unclear whether it will also predict adaptivity in other areas.

Building on the literature reviewed above, we expected adaptive scores to be shaped by the joint influence of these variables. To repeat the point made earlier, we hypothesize that, adaptivity is likely shaped by the combined and mutually constraining influence of cognitive, linguistic and clinical factors rather than by their independent effects (see also, Hill et al., 2015).

Method

Participants

This study included 44 children diagnosed with autism (6 females and 38 males) who belong to the database of the Language in Neurodiversity Lab (Lindy Lab) at the University of the Basque Country (UPV/EHU). All these children had been administered the ADOS-2 (Lord et al., 2012) and the Leiter-3 NVIQ (Koch et al., 2019; Roid & Miller, 2013) tests, and all but 9 completed the PPVT-III (Dunn et al., 2010) measure of receptive vocabulary. The 9 children who did not complete the PPVT-III did not do it due to insufficient pointing skills; their data were therefore treated as missing values.

The inclusion criteria were: (a) having an autism diagnosis according to the ADOS-2 scores, (b) having completed the ADOS-2 Module 1 (i.e., being productively nonverbal or minimally verbal at the time the ADOS-2 was administered), and (c) not being diagnosed with any co-occurring condition at the time the Vineland scales were administered. As said, these criteria restricted the sample to 44 children, who were between 4 and 12 years of age at the time of the Vineland assessment. All children attended the local Early Attention Service to receive intervention, although there may be differences among them in whether they accessed additional intervention services outside of this setting and in the degree of parental involvement. Families of participants meeting these criteria were contacted to explain the study objectives and obtain consent for participation. This study was approved by the Ethics Committee for Research Involving Human Beings (CEISH) at the University of the Basque Country (UPV/EHU).

Measurement Variables and Instruments

ADOS-2 Comparison Scores

We used the Calibrated Severity Score proposed by Gotham et al. (2009), based on ADOS-2 Modules 1, 2, and 3 for participants aged 2 to 16 years (Lord et al., 2012). This score, referred to as the CS, derived from the combination of the administered module, participant age, and ADOS-2 raw score, provides a standardized range from 1 to 10, where values of 1 to 3 correspond to NS (Nonspectrum), 4 to 5 to ASD (Autism Spectrum Disorder), and 6 to 10 to AUT (Autistic). Scores for Social Affect (SA) and Restricted and Repetitive Behaviors (RRB) were calibrated following Hus et al. (2012).

Nonverbal Intelligence

Nonverbal intelligence was assessed using the Leiter-3 scale (Koch et al., 2019; Roid & Miller, 2013), which does not require verbal instructions or oral responses from participants. The instrument is organized into three blocks: Fluid Intelligence subtests, Attention and Memory subtests, and the Socioemotional scale. Fluid Intelligence subtests yield a nonverbal IQ (NVIQ). All measures provided are measures of Fluid Intelligence.

Receptive Vocabulary (VQ)

Receptive vocabulary skills were assessed using the Spanish version of the PPVT-III (Dunn et al., 2010). Since the task involves pointing to the corresponding stimulus, the test does not require reading, writing, or expressive verbal responses, making it suitable for nonreaders or individuals without fluent oral language. Based on raw scores and CA, an equivalent IQ score is obtained, which we used as a proxy for language delay (we refer to this IQ as verbal quotient—VQ—to avoid misunderstandings). We thought that language delay was a more reliable measure of linguistic difficulties than pure verbal mental age (VMA), especially since participants’ CAs were varied. As explained above, if the aim is to relate language measures with adaptivity, we thought that a child whose CA is 7 and whose VMA is 3 is unlikely to adapt as well as a child with CA = 3.5 and VMA = 3 to their corresponding environments.

Adaptive Behavior

Adaptive behavior was assessed using the Vineland Adaptive Behavior Scales, Third Edition (Vineland-3) (Sparrow et al., 2016), a widely used instrument for measuring adaptive skills. The Comprehensive Form, administered via the Parent/Caregiver Form, provides the Adaptive Behavior Composite (ABC) score and detailed information on three main domains: Communication, Daily Living Skills, and Socialization. This approach provided both global and specific indicators for accurately assessing participants’ adaptive functioning.

Table 1 summarizes the age-related characteristics of the sample at the time of administering the different instruments.

Table 1.

Age of Participants at the Time of Instrument Administration.

Mean (SD) Min Max
ADOS-2 57.30 (16.54) 33 107
PPVT-III 59.25 (16.54) 34 107
Leiter-3 59.48 (16.09) 36 103
Vineland-3 85.41 (22.67) 52 144
Open in a new tab

Note. The values shown indicate months.

The results of the administered tests are detailed in Table 2.

Table 2.

Description of Results for CS Values, NVIQ, and PPVT-VQ Scores and Vineland-3 Questionnaire.

Mean (SD) Min Max
ADOS-2 5.95 (1.61) 2 10
SA (ADOS-2) 6.32 (1.84) 3 10
RRB (ADOS-2) 6.18 (1.86) 1 10
Leiter-3 92.45 (13.35) 71 117
PPVT-III 69.57 (18.93) 55 107
Vineland-3 Adaptative Behavior Composite (ABC) 62.61 (13.08) 39 90
Vineland-3 Communication (COM) domain 59.07 (22.05) 20 103
Vineland-3 Daily Living Skills (DLS) domain 65.48 (11.23) 42 95
Vineland-3 Socialization (SOC) domain 57.84 (17.63) 24 94
Open in a new tab

Note. ADOS-2 scores (including Social Affect [SA] and Restricted and Repetitive Behaviors [RRB]) are presented as Comparison Scores (CS), Leiter-3 scores as Nonverbal IQ (NVIQ), PPVT-III scores as VQ scores, and Vineland-3 scores as Standard Scores.

Analysis

All statistical analyses were conducted in R (R Core Team, 2023). To examine the contribution of each predictor (ADOS-2 comparison scores, nonverbal cognitive skills, and receptive vocabulary) to adaptivity, we fitted multiple linear regression models using the lm function, with Vineland-3 scores as the outcome variable, children's scores on the standardized tests (ADOS-2, Leiter-3, PPVT-III) as continuous predictors, and age (in months) as covariate. The time elapsed between the administration of the ADOS-2 and the Vineland-3 was included as a control variable in all models to ensure that this interval did not influence the results.

First, we estimated models including each predictor (ADOS-2, Leiter-3, PPVT-III) separately, to examine their individual associations with adaptive functioning. Second, we fitted additive models including all predictors simultaneously to estimate the contribution of each variable while accounting for shared variance among predictors. Finally, we fitted models including interaction terms between predictors to test whether adaptive functioning was better explained by nonadditive effects. Model comparisons were conducted using the anova function from the “stats” package to assess whether additive models provided a better fit than models including each predictor separately and whether models including interaction terms improved fit beyond additive models.

All models were evaluated using standard diagnostic procedures, including visual inspection of residuals to assess normality, linearity and homoscedasticity. In addition, residual normality was further assessed using the Shapiro–Wilk test. No substantial violations of model assumptions were detected. For purposes of clarity, only coefficients from the best-fitting models are reported in the main text. The complete R code with all model outputs is provided in Supplemental Material.

Results

Correlation Analyses

Descriptive statistics for all variables are reported in Tables 1 and 2. To examine preliminary associations among the three variables of interest, bivariate correlations were computed using Spearman's rank-order correlations. The correlation matrix, including ADOS-2 comparison scores (global CS, CS-SA, CS-RRB), nonverbal cognitive skills, receptive vocabulary delay and age, is presented in Table 3.

Table 3.

Bivariate Correlations Between CS, NVIQ, PPVT-VQ, and Age.

CS CS-SA CS-RRB NVIQ VQ Age
CS –
CS-SA −0.04 –
CS-RRB 0.14 0.23 –
NVIQ −0.49 −0.12 0.02 –
VQ −0.54 −0.13 0.03 0.77 –
Age 0.15 0.10 −0.20 −0.28 −0.43 –
Open in a new tab

Note. Spearman's rank-order correlations (ρ) are reported. Correlations with p < .05 are in bold.

As shown in Table 3, CS scores were negatively correlated with both NVIQ and VQ, while NVIQ and VQ showed a strong positive correlation. In addition, age was negatively correlated with VQ. No significant correlations were observed among the remaining variables.

Regression Analyses

The initial set of linear regression models assessed the individual associations between ADOS-2 comparison scores, nonverbal intelligence and receptive vocabulary, and adaptive functioning, while controlling for children's age (in months) and the testing interval between the ADOS-2 and the Vineland-3. 1 When examined separately, both Leiter-3 NVIQ and PPVT-III VQ scores significantly predicted Vineland-3 scores, with higher NVIQ and VQ being associated with higher levels of adaptive functioning across all domains of the Vineland-3. ADOS-2 CS were found to significantly predict Vineland global (ABC) scores, as well as scores in Communication (COM) and Socialization (SOC), with higher CS scores corresponding to lower adaptive functioning levels in all these subdomains. Daily Living Skills (DLS) scores were not found to be significantly associated with CS. These results are illustrated in Figure 1 (full model outputs are provided in Supplemental Material).

Figure 1.

Figure 1.

Open in a new tab

Linear predictions from separate adjusted models for Vineland-3 domains as a function of the three assessments.

Note. ADOS-2 scores are reported using Comparison Scores (CS), Leiter-3 scores are reported as NVIQ scores, and PPVT-III scores are reported as VQ scores. Lines represent model-fitted values, and points represent observed (raw) data.

We then fitted additive multiple regression models including CS, NVIQ and VQ as main effects. In this combined model, only NVIQ scores emerged as a significant predictor of Vineland-3 ABC scores (β = 0.88, SE = 0.20, t = 4.52, p < .001), with higher NVIQ corresponding to higher levels of adaptivity (see Figure 2). ADOS-2 (β = −1.12, SE = 1.42, t = −0.79, p = 0.44) and VQ scores (β = −0.09, SE = 0.15, t = −0.59, p = 0.56) were no longer found to be significantly associated with Vineland-3 global scores. This model accounted for a substantial proportion of variance (R2 = 0.63, adjusted R2 = 0.57) and showed improved fit relative to the previous models including the predictors of interest separately.

Figure 2.

Figure 2.

Open in a new tab

Linear predictions from additive models for Vineland-3 domains as a function of the three assessments.

Note. ADOS-2 scores are reported using Comparison Scores (CS), Leiter-3 scores are reported as NVIQ scores, and PPVT-III scores are reported as VQ scores. Lines represent model-fitted values, and points represent observed (raw) data.

A similar pattern was observed across the three main domains of the Vineland-3. NVIQ emerged as the only significant predictor of Communication (β = 1.23, SE = 0.33, t = 3.76, p < .001), Daily Living Skills (β = 0.90, SE = 0.19, t = 4.86, p < .001), and Socialization (β = 1.00, SE = 0.26, t = 3.84, p < .001) scores. In contrast, VQ did not significantly predict scores in the Communication (β = 0.08, SE = 0.24, t = 0.33, p = 0.74), Daily Living Skills (β = −0.19, SE = 0.14, t = −1.39, p = 0.18) and Socialization (β = −0.13, SE = 0.19, t = −0.69, p = 0.49) domains. Similarly, ADOS-2 comparison scores were not found to significantly predict scores in Communication (β = −0.23, SE = 2.38, t = −0.10, p = 0.92), Daily Living Skills (β = −0.02, SE = 1.35, t = −0.011, p = 0.99) and Socialization, although the effect approached significance in the Socialization domain (β = −3.37, SE = 1.90, t = −1.78, p = 0.086). Model fit indices were comparable across domains (Communication: R2 = 0.58, adjusted R2 = 0.51; Daily Living Skills: R2 = 0.55, adjusted R2 = 0.48; Socialization: R2 = 0.65, adjusted R2 = 0.59). These results are illustrated in Figure 2.

To further examine potential interaction effects among predictors, we performed a series of models including interactions between (i) NVIQ and ADOS-2, (ii) VQ and ADOS-2, (iii) NVIQ and VQ, and (iv) a three-way interaction between NVIQ, ADOS-2, and VQ. Each interaction model was then compared to the corresponding additive (main-effects) model including the same set of predictors but no interaction terms, to assess whether the inclusion of interactions improved model fit.

Across all domains, models including interaction terms did not result in a substantial improvement in model fit over the corresponding main-effects models (see Supplemental Material). Several interaction effects showed p-values close to conventional significance thresholds. Specifically, the interaction between NVIQ and ADOS-2 for ABC (β = 0.18, SE = 0.10, t = 1.83, p = 0.078) and Communication (β = 0.31, SE = 0.17, t = 1.84, p = 0.076) scores, as well as the interaction between VQ and ADOS-2 in the Socialization domain (β = 0.15, SE = 0.08, t = 1.95, p = 0.06), suggested that the association between NVIQ or VQ and adaptive functioning may be stronger at higher levels of autism traits. However, none of the tested interaction effects reached conventional levels of statistical significance. So, overall, these results do not provide clear evidence for interaction effects between NVIQ, VQ and CS in predicting adaptive functioning.

Discussion

We examined how autistic children with different cognitive and linguistic profiles who had been administered the ADOS-2 Module 1 adapted to their environment after a brief lapse of time (2–3 years), using the Vineland Adaptive Scales. Our research questions were, first, whether ADOS-2 comparison scores, language delay, and NVIQ would predict adaptivity; second, whether interactions between these variables would outperform their main effects in predictive power.

Although the study was largely exploratory, we expected that adaptivity would be better predicted by the joint contribution of the variables rather than by their independent effects. This expectation was based on previous research as well as on theoretical considerations, which suggest that factors such as NVIQ, language delay, and ADOS scores or autistic traits are likely to interact in complex ways. As explained in the introduction, we expected that these variables could form a network (Borsboom, 2017) of interrelated traits whose mutual relationships would have a significant impact on adaptive functioning. Among the individual variables, drawing on prior literature, we anticipated that NVIQ could emerge as the strongest predictor. However, we also expected that the effect of NVIQ would be moderated by the other two variables. Based on prior research, it was unclear whether a similar predictive association would also emerge for the other two variables, ADOS-2 comparison scores and receptive vocabulary delay, especially this latter variable, as it has not been examined in the context of autism.

Of the three variables examined, only NVIQ proved to be a significant predictor of adaptivity scores. In contrast, VQ and CS did not remain significant when included simultaneously in the same model alongside NVIQ, suggesting that their effects were largely explained by shared variance with NVIQ (note that NVIQ was significantly correlated with both CS and VQ; see Table 3). In principle, the pattern we observed aligns with some studies in the literature (Kanne et al., 2011; Failla et al., 2024), but contradicts others that do not find IQ to be a significant predictor of adaptive functioning (e.g., Alvares et al., 2020). In addition, we did not find that autistic children with higher NVIQs in our population show relatively more adaptive difficulties than children with somewhat lower IQs, unlike the findings reported by Pathak et al. (2019). However, it is important to note that those studies primarily used VIQ as their measure of intelligence, whereas we used NVIQ. It remains to be seen whether different types of intelligence relate differently to adaptive functioning. In this respect, the clear association we observe between NVIQ and adaptive functioning can be seen as a relatively underexplored aspect in the literature, since, as noted, NVIQ has comparatively rarely been isolated as an independent factor in these studies. Finally, it is also possible that the relationship between IQ and adaptive functioning is not uniform across subgroups of autistic children, and that the range of profiles captured in the present sample may not include those in which high intelligence may work against adaptivity.

On the other hand, unlike some previous studies, we did not find a relation between ADOS-2 comparison scores and adaptive behavior scores (Failla et al., 2024) once NVIQ was taken into account. Similarly, linguistic comprehension delay also did not appear to relate to adaptivity either globally or in any particular area. We did not have clear predictions regarding this relation, given that the existing evidence is weak and is largely based on studies of children with language disorders. Nonetheless, we measured receptive comprehension because we reasoned that greater delays might lead to more difficulties in adapting to the environment, if only because this environment is, to a large extent, linguistic. Yet, we observed no such association when NVIQ was included in the model, which might suggest that autistic children may, other things being equal, manage to compensate for their comprehension difficulties.

Concerning our second research question, the interaction models suggested a possible moderation of NVIQ by ADOS-2 comparison scores in both the global adaptivity score and the Communication domain. Such moderation effects were not observed in the case of the Daily Living Skills domain or the Socialization area. In the Socialization domain, the interaction between VQ and ADOS-2 approached significance, suggesting a potential moderating role of vocabulary comprehension delay in the association between ADOS-2 and Socialization outcomes. Nevertheless, these effects were not significant, and it remains unclear whether they would reach conventional levels of significance with a larger sample.

Taken together, these findings have several implications for both research and practice. First, they suggest that early NVIQ is a meaningful predictor of adaptive functioning in minimally verbal autistic children, underscoring the importance of including nonverbal cognitive assessments in early prognostic evaluations. Second, the absence of strong associations with either kind of ADOS-2 comparison scores and receptive vocabulary delay suggests that adaptive outcomes may depend less on trait profile or linguistic comprehension than it may be assumed, and that autistic children may successfully compensate for early marked autistic traits and comprehension difficulties. Finally, these findings motivate future longitudinal research to examine whether different forms of intelligence relate differently to adaptive functioning, and whether the relation between NVIQ and adaptivity observed here remains stable over time or evolves as children develop.

Several limitations of the present study should be noted. First of all, vocabulary comprehension data were unavailable for nine participants due to difficulties with pointing. This reduced the sample size for analysis involving VQ, and therefore results involving this variable should be interpreted with caution. Second, it must be borne in mind that all participants received ADOS-2 Module 1, indicating that the sample was largely nonverbal or minimally verbal at baseline; as a result, findings may not generalize to autistic children with more advanced language or to older age groups. Third, our study relied on a single measure for each domain (NVIQ, ADOS-2 comparison scores, receptive vocabulary), and different instruments or more fine-grained assessments (e.g., pragmatic language, executive functioning, nonverbal reasoning subcomponents) might reveal different predictive patterns. Fourth, adaptive functioning was assessed only at one follow-up point, 2 to 3 years after baseline, which prevents us from examining developmental trajectories. Longitudinal analyses with multiple time points would provide a richer picture of how early cognitive and linguistic profiles relate to evolving adaptive skills. External factors not measured here—such as intervention type and intensity, family resources, or environmental supports—may also contribute to adaptive functioning and could interact with the profiles identified. Finally, future studies with larger samples are needed to more reliably test interaction effects, given that these were complex models fitted to a modest sample size.

Supplemental Material

sj-html-1-dli-10.1177_23969415261465246 - Supplemental material for Adaptive Functioning in Expressively Nonverbal and Minimally Verbal Autistic Children: Nonverbal Intelligence as a Predictive Factor
sj-html-1-dli-10.1177_23969415261465246.html (3.4MB, html)

Supplemental material, sj-html-1-dli-10.1177_23969415261465246 for Adaptive Functioning in Expressively Nonverbal and Minimally Verbal Autistic Children: Nonverbal Intelligence as a Predictive Factor by Ekaine Rodríguez-Armendariz, Zuriñe Ábalos, Elena Castroviejo and Agustín Vicente in Autism & Developmental Language Impairments

Acknowledgments

The authors are very grateful to all the families and children who have worked with them during the last 6 years. Thanks also to therapists from the early attention service in Vitoria for their trust and continuous collaboration.

1.

Additional models examining the predictive contribution of the ADOS-2 domain-specific (SA and RRB) CS to adaptivity were fitted. However, since these models did not yield significant effects and did not improve model fit relative to the model including global CS (see Supplemental Material), all analyses are reported using the global CS.

Footnotes

Ethical Approval and Informed Consent Statement: Families of participants meeting study criteria were contacted to explain the study objectives and obtain consent for participation. This study was approved by the Ethics Committee for Research Involving Human Beings (CEISH) at the University of the Basque Country (UPV/EHU).

Author Contributions: ER-A did conceptualization, methodology, data curation, formal analysis, investigation, visualization, writing—original draft, writing—review and editing. ZA performed conceptualization, visualization, writing—original draft, writing—review and editing. EC performed conceptualization, methodology, funding acquisition, project administration, supervision. AV did conceptualization, funding acquisition, investigation, project administration, supervision, writing—original draft, writing—review and editing.

Funding: The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research has been partially supported by project FUNLAT (PID2021-122233OB-I00), funded by the Ministry of Science and Innovation (MCI) / Spanish Research Agency (AEI) and the European Regional Development Fund (FEDER, EU), and by the IT1537-22 Research Group (Basque Government).

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Supplemental Material: Supplemental material for this article is available online.

References

  1. Alvares G. A., Bebbington K., Cleary D., Evans K., Glasson E. J., Maybery M. T., Pillar S., Uljarević M., Varcin K., Wray J., Whitehouse A. J. (2020). The misnomer of ‘high functioning autism’: Intelligence is an imprecise predictor of functional abilities at diagnosis. Autism, 24(1), 221–232. 10.1177/1362361319852831 [DOI] [PubMed] [Google Scholar]
  2. Bishop D. (2004). The children’s communication checklist: CCC-2. Harcourt Assessment. https://sapere-aude.group/read/yangxingyu/Source_Files/2023-10-08-YXY.pdf . [Google Scholar]
  3. Borsboom D. (2017). A network theory of mental disorders. World Psychiatry, 16(1), 5–13. 10.1002/wps.20375 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Chen Y., Siles B., Tager-Flusberg H. (2024). Receptive language and receptive-expressive discrepancy in minimally verbal autistic children and adolescents. Autism Research, 17(2), 381–394. 10.1002/aur.3079 [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. De Bildt A., Sytema S., Kraijer D., Minderaa R. (2005). Prevalence of pervasive developmental disorders in children and adolescents with mental retardation. Journal of Child Psychology and Psychiatry, 46(3), 275–286. 10.1111/j.1469-7610.2004.00346.x [DOI] [PubMed] [Google Scholar]
  6. Dunn L. M., Dunn L. M., Arribas D. (2010). PPVT-III. PEABODY. Test de Vocabulario en Imágenes. Tea Ediciones. [Google Scholar]
  7. Failla C., Scarcella I., Vetrano N., Previti S., Mangano R. M., Tartarisco G., Vagni D., Pioggia G., Marino F. (2024). Autism, intelligence, language, and adaptive behavior, disentangling a complex relationship. Frontiers in Psychiatry, 15, 1411783. 10.3389/fpsyt.2024.1411783 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Fossum I. N., Øie M. G., Orm S., Andersen P. N., Skogli E. W. (2025). Longitudinal predictors of adaptive functioning in emerging adults with and without autism spectrum disorder. Research on Child and Adolescent Psychopathology, 53(2), 137–149. 10.1007/s10802-024-01265-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Fujiki M., Spackman M. P., Brinton B., Hall A. (2004). The relationship of language and emotion regulation skills to reticence in children with specific language impairment. Journal of Speech, Language, and Hearing Research, 47(3), 637–646. 10.1044/1092-4388(2004/049) [DOI] [PubMed] [Google Scholar]
  10. Gotham K., Pickles A., Lord C. (2009). Standardizing ADOS scores for a measure of severity in autism spectrum disorders. Journal of Autism and Developmental Disorders, 39(5), 693–705. 10.1007/s10803-008-0674-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Guerrera S., Fucà E., Petrolo E., De Stefano A., Casula L., Logrieco M. G., Valeri G., Vicari S. (2025). Exploring the clinical features of minimally verbal autistic children. Frontiers in Psychiatry, 16, 1549092. 10.3389/fpsyt.2025.1549092 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Hill T. L., Gray S. A. O., Kamps J. L., Varela R. E. (2015). Age and adaptive functioning in children and adolescents with ASD: The effects of intellectual functioning and ASD symptom severity. Journal of Autism and Developmental Disorders, 45(12), 4074–4083. 10.1007/s10803-015-2522-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Hus V., Gotham K., Lord C. (2012). Standardizing ADOS domain scores: Separating severity of social affect and restricted and repetitive behaviors. Journal of Autism and Developmental Disorders, 44(10), 2400–2412. 10.1007/s10803-012-1719-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Kanne S. M., Gerber A. J., Quirmbach L. M., Sparrow S. S., Cicchetti D. V., Saulnier C. A. (2011). The role of adaptive behavior in autism spectrum disorders: Implications for functional outcome. Journal of Autism and Developmental Disorders, 41(8), 1007–1018. 10.1007/s10803-010-1126-4 [DOI] [PubMed] [Google Scholar]
  15. Koch C., Miller L. J., Pomplun M., Roid G. H. (2019). LEITER-3. Escala de Inteligencia No Verbal. Giunti Psychometrics. [Google Scholar]
  16. Lord C., Rutter M., DiLavore P. S., Risi S., Gotham K., Bishop S. (2012). (ADOS®-2) Autism Diagnostic Observation Schedule™, Second Edition. Western Psychological Services. https://www.wpspublish.com/ADOS-2-autism-diagnostic-observation-schedule-second-edition . [Google Scholar]
  17. Maggio V., Grañana N. E., Richaudeau A., Torres S., Giannotti A., Suburo A. M. (2014). Behavior problems in children with specific language impairment. Journal of Child Neurology, 29(2), 194–202. 10.1177/0883073813509886 [DOI] [PubMed] [Google Scholar]
  18. Miranda A., Berenguer C., Baixauli I., Roselló B. (2023). Childhood language skills as predictors of social, adaptive and behavior outcomes of adolescents with autism spectrum disorder. Research in Autism Spectrum Disorders, 103, 102143. 10.1016/j.rasd.2023.102143 [DOI] [Google Scholar]
  19. Pathak M., Bennett A., Shui A. M. (2019). Correlates of adaptive behavior profiles in a large cohort of children with autism: The autism speaks Autism Treatment Network registry data. Autism, 23(1), 87–99. 10.1177/1362361317733113 [DOI] [PubMed] [Google Scholar]
  20. Paul R., Miles S., Cicchetti D., Sparrow S., Klin A., Volkmar F., Coflin M., Booker S. (2004). Adaptive behavior in autism and pervasive developmental disorder-not otherwise specified: Microanalysis of scores on the vineland adaptive behavior scales. Journal of Autism and Developmental Disorders, 34(2), 223–228. 10.1023/B:JADD.0000022612.18116.46 [DOI] [PubMed] [Google Scholar]
  21. Petrolini V., Vicente A. (2022). The challenges raised by comorbidity in psychiatric research: The case of autism. Philosophical Psychology, 35(8), 1234–1263. 10.1080/09515089.2022.2052829 [DOI] [Google Scholar]
  22. R Core Team. (2023). R: A language and environment for statistical computing. [Computer software]. R Foundation for Statistical Computing. ISBN 3-900051-07-0. http://www.R-project.org/. [Google Scholar]
  23. Roid G. H., Miller L. J. (2013). (Leiter-3) Leiter International Performance Scale, Third Edition. WPS. [Google Scholar]
  24. Saul J., Cooke M., Munaweera S., Matthews D. (2026). Sub-groups of spoken language and broader communication skills in a large heterogenous cohort of minimally verbal school-age children: Evidence of discrepant profiles. Molecular Autism, 17(1), 8. 10.1186/s13229-026-00701-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Slusna D., Rodríguez A., Salvadó B., Vicente A., Hinzen W. (2021). Relations between language, non-verbal cognition, and conceptualization in non- or minimally verbal individuals with ASD across the lifespan. Autism & Developmental Language Impairments, 6, 239694152110532. 10.1177/23969415211053264 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Sparrow S. S., Cicchetti D. V., Saulnier C. A. (2016). Vineland-3: Vineland Adaptive Behavior Scales. Manual (3rd ed.). Pearson. [Google Scholar]
  27. Tillmann J., San José Cáceres A., Chatham C. H., Crawley D., Holt R., Oakley B., Zwiers M. P. (2019). Investigating the factors underlying adaptive functioning in autism in the EU-AIMS longitudinal European autism project. Autism Research, 12(4), 645–657. 10.1002/aur.2081 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

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

Supplementary Materials

sj-html-1-dli-10.1177_23969415261465246 - Supplemental material for Adaptive Functioning in Expressively Nonverbal and Minimally Verbal Autistic Children: Nonverbal Intelligence as a Predictive Factor
sj-html-1-dli-10.1177_23969415261465246.html (3.4MB, html)

Supplemental material, sj-html-1-dli-10.1177_23969415261465246 for Adaptive Functioning in Expressively Nonverbal and Minimally Verbal Autistic Children: Nonverbal Intelligence as a Predictive Factor by Ekaine Rodríguez-Armendariz, Zuriñe Ábalos, Elena Castroviejo and Agustín Vicente in Autism & Developmental Language Impairments

Articles from Autism & Developmental Language Impairments are provided here courtesy of SAGE Publications

RESOURCES