Sentence Repetition and Nonword Repetition As Markers of Structural Language Impairment in ASD

Background: The nature of linguistic profiles in ASD remains unclear: many children show structural language abilities similar to typically developing age-peers (TD), others display deficits like those in Specific Language Impairment (SLI) (Leyfer et al., 2008; Lloyd et al., 2006; Loucas et al., 2012; Tuller et al. 2016; Zebib et al., 2013). There is consensus from work on SLI that the most sensitive tools for detecting structural language impairment are Sentence Repetition (SR) and Nonword Repetition (NWR) (Conti-Ramsden et al., 2001). Previous studies using these tools in ASD suggested that low performance is due to low nonverbal level and/or severity of ASD symptoms (Harper-Hill et al., 2013; Williams et al., 2013, a.o.). These studies focused on adolescents and high-functioning individuals, and the specific NWR/SR tasks used contained few structures, and had predominant memory components (Riches et al. 2010, 2011; Whitehouse et al., 2008).

Objectives: With the aim of investigating structural language abilities across the spectrum, we used two experimental tasks, SR and NWR, both linguistically based, targeting structures of increasing complexity that are difficult for children with SLI. We sought to determine whether children with ASD and impaired language (ASD-LI), perform analogously to children with SLI, and whether children with ASD and normal language (ASD-LN), perform like TD children, independently from nonverbal level and ASD severity.

Methods: 15 French-speaking children with ASD (data processing for 10 additional children in progress) aged 7–11;8 (M 9;1, SD 1;9) were administrated the SR and NWR tasks. Children varied in ASD severity (ADOS range: 4–10) and nonverbal cognitive level (WISC IRP range: 38 - 119). Standardized expressive and receptive language scores revealed children with ASD-LI (7) and children with ASD-LN (8). Comparison groups included 15 children with SLI (data for 10 additional children to be processed) aged 6;1–8;5 (M 7;6; SD 0;4) and 73 TD children aged 4;0–8;9 (to be completed).

Results: The ASD-LI group performed analogously to the SLI group both, on SR (p=.972) and NWR (p=.084), while the ASD-LN group performed similarly to TD children on SR (when compared to TD 7-8, p=.126) and NWR (when compared to TD 5-6, p=.156). ASD-LI performed significantly worse than ASD-LN on both SR (p=.002) and NWR (p=.019). In the whole ASD group, a significant correlation was found between performance on SR and NWR (r_s=.575; p=.025) and between each repetition task and a composite standardized language score: SR (r_s=.890; p= .000) NWR (r_s=.643; p=.010). No significant correlations were found between SR and ADOS (r_s=-.186; p=.543) or SR and IRP scores (r_s=.513; p=.061) and between NWR and ADOS (r_s=-.450; p=.122) or NWR and IRP scores (r_s=.189; p=.517). No significant correlation was found between either task and age (SR, r_s=.176; p=.531; NWR, r_s=.477; p=.072).

Conclusions: Our study shows the validity of linguistically based SR and NWR tasks as markers of structural language deficits in children with ASD. Furthermore, it found that the LI/LN distinction in ASD cuts across both nonverbal abilities and ASD severity, suggesting that language can be independent in verbal ASD.