Reading Comprehension, Language Disturbance, and the Social Communication Phenotype of ASD

Background:   Previous research provided preliminary evidence that reading comprehension impairment may be part of the social communication phenotype of higher functioning school-aged children with ASD (Norbury & Nation, 2011; Ricketts et al., 2013).  The core features of autism spectrum disorders (ASDs) place these children at risk for oral language difficulties. Proficient structural and higher-order language abilities develop within frequent and high-quality social interactions encompassing a wide variety of topics and activities (e.g. Tomasello, 2003) and have been shown to significantly support reading accuracy and comprehension in children with ASD (e.g. Brown et al., 2013). Therefore a finer-grained analysis of ASD symptomatology, oral language, reading accuracy, and reading comprehension is required to model this complex interactive process.

Objectives:   This study was designed provide a detailed model of reading comprehension, reading accuracy, and oral language, and their relation to core symptoms of autism, in a large sample of high functioning children with autism (HFASD). 

Methods:  This sample consisted of 80 (66 males) 8- to16-year-old higher functioning students with ASD (HFASD; FSIQ≥75; ASD diagnosis confirmed with ADOS-2).    Mplus, version 7 (Muthen & Muthen,1998-2012) was used to generate structural equation models (SEM) that tested relations between ADOStot (ADOS-2 total scores) and three latent variables that were each comprised of multiple indicators (see Figures 1&2 notes for complete list):  ReadAcc (reading accuracy), OralLang (structural + higher order language), and ReadComp (reading comprehension).

Results:  Previous research on this sample indicated approximately 50% of the students displayed reading comprehension and oral language impairments (McIntyre, IMFAR 2014). Figure 1 displays the SEM model testing the hypothesis that ADOStot and StrcLang explain unique variance in ReadComp, but that ReadAcc does not.   The model fits the data well given the relatively small sample size: χ²=67.353(50df),p=.0513;RMSEA =.066,90% C.I.=.000-.104;CFI=.947;SRMR=.070.  ADOS scores predicted unique variance in ReadComp even after their significant relations with ReadAcc and StrcLang were taken into account.  Figure 2 displays the SEM model testing the hypothesis that the addition of three higher-order language indicators to the oral language latent variable would fully mediate the relation between ADOStot and ReadComp.  This model provided a better fit to the data:  χ²=102.662(86df),p=.1063;RMSEA =.050,90% C.I.=.000-.083;CFI=.961;SRMR=.076. ADOStot no longer directly explained variance in ReadComp. 

Conclusions:   Results indicated that core features of ASD play a major role in reading comprehension impairments. However, this role appears to be fully mediated by the relation of ASD symptomatology to oral language abilities in these children, particularly with regard to higher-order language processes (inference generation, gist processing, and narrative understanding). Surprisingly, ADOS scores were also significantly related to reading accuracy abilities in this sample, which provided evidence that in a subset of students the social communication disturbance extends to deficits in phonology, word reading, and rapid naming.  However, consistent with prior research (e.g. Newman et al., 2007)  a pattern of dissociation between reading accuracy and reading comprehension was displayed.  This study provided direct evidence that language and reading impairments are integral to the social communication disturbance in ASD.