Predictors of High and Low Language Skills in Young Children On the Autism Spectrum

Background:   A primary goal of behavioral research on autism spectrum disorder (ASD) has been to identify predictors of language outcomes (Paul, Chawarska, Cicchetti, & Volkmar, 2008).  Variability in language skills among children with ASD is considerable, with some children remaining nonverbal throughout childhood and others attaining age-appropriate language ability.  Findings have been mixed with respect to the roles of nonverbal cognition, joint attention, socioeconomic status (SES), socialization skills, and autism severity for language growth (Bopp & Mirenda, 2011; Paul et al., 2008; Thurm et al., 2007).

Objectives:   The purpose of the current study was to further characterize the range of variability in language skills in young children with ASD by identifying factors at approximately 2½ years of age that serve as predictors of language outcomes (i.e., high or low language ability) at 5½ years of age. 

Methods:   Participants (n = 103) were drawn from a longitudinal investigation of language development in toddlers with ASD.  Best estimate clinical ASD diagnoses were based in part on the Autism Diagnostic Observation Schedule (ADOS) or Toddler Module (ADOS-T).  Initial assessments at approximately 2½ years of age yielded nonverbal cognition scores (Bayley-III cognitive composite), response to and initiation of joint attention (Early Social Communication Scales), SES (years of maternal education), socialization skills (Vineland Socialization standard scores), and autism severity (calibrated ADOS scores).  These child and environmental characteristics were considered as predictors in discriminant function analyses between participants with high and low language ability.  Participants with high and low language ability were those with Preschool Language Scale-4 (PLS-4) total standard scores that fell within the highest or lowest 15% of the sample of the 103 children at 5½ years of age.  Children with high language ability (n = 15; PLS-4 standard score M = 120.93; SD = 5.55) and children with low language ability (n = 16; PLS-4 standard score M = 50.25; SD = 1.00) did not differ in age, p= .674.  All participants with low language ability were minimally verbal (i.e., scored 3 or 8 on the ADOS item for Overall Level of Non-Echoed Language, indicating fewer than 5 spoken words). 

Results:   Discriminant functions were limited to three predictors due to sample size.  The best 3-predictor model included nonverbal cognition, SES, and response to joint attention and correctly classified 92.3% of valid cases (n = 14 high language; n = 12 low language): 85.7% of high language cases and 100% of low language cases.  Nonverbal cognition alone correctly classified 85.7% of cases: 92.3% for low language and 80.0% for high language. 

Conclusions:   Early nonverbal cognition is a critical predictor of language outcome; however, the ability to respond to others’ bids for attention and environmental factors, such as SES, also discriminated between high and low language ability.  These findings have implications for predicting childhood outcomes for individuals with ASD and for identifying potential targets for intervention.