Overall Communication Ability
Communication ability was assessed on the basis of GCC scores, with GCC scores < 55 (≤ 10th percentile) indicating that a child has significant communicative difficulties. For participants with Sotos syndrome, GCC scores ranged from 1 to 57 (< 1st percentile–13th percentile), demonstrating that the majority of children with Sotos syndrome had communicative difficulties of varying severity, with only one participant scoring in the normal range. GCC scores for participants with WS ranged from 4 to 60 (< 1st percentile–15th percentile), indicating that the majority of children with WS had communicative difficulties and once again, only one participant scored in the normal range. An independent samples t-test was used to compare overall communicative ability of children with Sotos syndrome and children with WS. The analysis revealed no significant difference between GCC scores for the Sotos syndrome participants (M = 27.55, SD = 14.81) and GCC scores for the WS participants (M = 26.82, SD = 10.60); t(63) = 0.23, p = .820, d = 0.06, indicating that both syndromes are associated with a similar level of overall communicative ability.
Language Structure and Pragmatic Language
In order to determine whether children with Sotos syndrome had an uneven profile of greater difficulty with either language structure or pragmatic language, scores for these abilities were compared. The Sotos syndrome profile was also compared to the WS profile to establish any syndrome-specific differences. Language structure scores were calculated as the sum of scaled scores for subscales A, B, C and D for participants with Sotos syndrome (M = 12.68, SD = 8.38) and participants with WS (M = 13.97, SD = 7.11). Pragmatic language scores were calculated as the sum of scaled scores for subscales E, F, G and H for participants with Sotos syndrome (M = 14.87, SD = 8.16) and participants with WS (M = 12.85, SD = 5.85). A 2 × 2 (Group: Sotos/WS x Language skill: structure/pragmatic) mixed measures ANOVA revealed no main effect of skill; F(1, 63) = 0.34, p = .564, ηρ² = .005, indicating no discrepancy between language structure skills and pragmatic language skills overall within this cohort. There was no effect of group; F(1, 63) = 0.05, p = .820, ηρ² = .001 and no skill x group interaction; F(1, 63) = 3.18, p = .079, ηρ² = .048. This suggests that both children with Sotos syndrome and children with WS had a similar degree of impairment with language structure skills and pragmatic language skills.
A Chi square test of independence was used to assess the association between group (Sotos/WS) and category of language ability (impaired/borderline/typical). There was no significant association between group and language structure ability category, X2(2, N = 65) = 0.44, p = .801, and no significant association between group and pragmatic language ability category, X2(2, N = 65) = 3.40, p = .183. This indicates that, for both Sotos syndrome and WS, a similar proportion of individuals had impaired, borderline and typical language structure and pragmatic language abilities (see Table 1).
Table 1 Proportion of participants in the impaired functioning (scores < 17), borderline functioning (scores 17–24) and typical functioning (scores > 24) categories for language structure scores and pragmatic language scores
Although children with Sotos syndrome had similar overall difficulty with both language structure and pragmatic language abilities, there was significant variability in level of functioning of these abilities within each group. In order to investigate this variability, multiple regression was used to establish the extent to which variance in pragmatic language ability is explained by language structure skills, age and sex. The regression equation was significant; R2 = 0.64, F(3, 27) = 16.27, p < .001, and inspection of the beta weights revealed that both language structure skills (β = 0.61, p < .001) and age (β = − 0.53, p < .001) were significant predictors of pragmatic language ability. The same analysis was used to assess variance in pragmatic language ability for participants with WS. The regression equation was significant; R2 = 0.25, F(3, 30) = 3.33, p = .033, and inspection of the beta weights revealed that both language structure skills (β = 0.39, p = .023) and age (β = − 0.37, p = .032) were significant predictors of pragmatic language ability. This suggests that language structure skills and age are significant predictors of pragmatic language ability for both children with Sotos syndrome and children with WS. Although the predictors were the same in both groups, language structure skills and age explained more variance in pragmatic language ability for children with Sotos syndrome (64%) than children with WS (25%). This indicates that other factors may account for variability in pragmatic language ability within the WS population.
Language Structure Subscale Profile
In order to determine whether children with Sotos syndrome had particular difficulty with specific aspects of language structure, a repeated measures ANOVA was used to compare scaled scores for the four language structure subscales: (A) speech (e.g. pronunciation), (B) syntax (e.g. use of pronouns), (C) semantics (e.g. differentiation of words which sound similar) and (D) coherence (e.g. description of a sequence of events). A scaled subscale score ≤ 5 is indicative of significant communicative problems. The analysis identified no effect of subscale; F(3, 90) = 2.16, p = .099, ηρ² = .067, indicating that children with Sotos syndrome had similar ability in all four language structure skills assessed by the CCC-2.
The language structure skills of children with Sotos syndrome were compared with those of children with WS in order to investigate syndrome-specific differences in these abilities. A 2 × 4 (Group: Sotos/WS x Language structure subscales: A, B, C, D) mixed measures ANOVA was used to compare scaled scores for the four language structure subscales. The analysis identified a significant main effect of subscale; F(3, 189) = 3.61, p = .014, ηρ² = .054 but no significant subscale x group interaction; F(3, 189) = 1.23, p = .302, ηρ² = .019 and no significant effect of group; F(1, 63) = 0.45, p = .504, ηρ² = .007. As there was no significant subscale x group interaction, this suggests that children with these syndromes had similar difficulty with the language structure skills assessed by the CCC-2. Figure 1 shows mean scaled scores for the language structure subscales.
Pragmatic Language Subscale Profile
In order to determine whether children with Sotos syndrome had particular difficulty with specific aspects of pragmatic language, a repeated measures ANOVA was used to compare scaled scores for the four pragmatic language subscales: (E) inappropriate initiation (e.g. telling people things they already know), (F) stereotyped language (e.g. repeating back things other people have said), (G) use of context (e.g. consistency of communication skills across different situations) and (H) nonverbal communication (e.g. use of facial expressions and eye contact). The analysis identified a significant effect of subscale; F(3, 90) = 11.32, p < .001, ηρ² = .274, indicating that children with Sotos syndrome had an uneven profile of pragmatic language skills. Post-hoc paired samples t-tests (using a Bonferroni correction, p < .008 required for significance) were used to compare scaled scores for the four pragmatic language subscales. The significant comparisons were stereotyped language > use of context, t(30) = 5.67, p < .001, d = 1.02; stereotyped language > nonverbal communication, t(30) = 4.68, p < .001, d = 0.84; inappropriate initiation > use of context, t(30) = 4.60, p < .001, d = 0.83. There was a trend for inappropriate initiation > nonverbal communication, t(30) = 2.76, p = .01, d = 0.49. This suggests that, in terms of the pragmatic language profile, children with Sotos syndrome have greater difficulty with use of context and nonverbal communication, whereas inappropriate initiation and stereotyped language are less problematic.
The pragmatic language abilities of children with Sotos syndrome were compared with those of children with WS in order to determine the specificity of the Sotos syndrome pragmatic language profile. A 2 × 4 (Group: Sotos/WS x Pragmatic language subscales: E, F, G, H) mixed measures ANOVA was used to compare scores for the four pragmatic language subscales. The analysis identified no significant effect of group; F(1, 63) = 1.33, p = .253, ηρ² = .021. There was a significant main effect of subscale; F(3, 189) = 17.58, p < .001, ηρ² = .218, and a significant subscale x group interaction; F(3, 189) = 4.86, p = .003, ηρ² = .072, indicating that both groups had uneven pragmatic language profiles but the nature of these profiles differed.
As there was a significant subscale x group interaction, post-hoc comparisons (using a Bonferroni correction p < .013 required for significance) were used to compare scores for the four pragmatic language subscales between the groups. The comparisons revealed a trend for the Sotos syndrome participants to have better stereotyped language scores, compared with the WS participants, t(63) = 2.19, p = .032, d = 0.55. There were no significant differences between groups for inappropriate initiation scores, t(63) = 1.27, p = .210, d = 0.32; use of context scores, t(63) = 1.46, p = .148, d = 0.37; or nonverbal communication scores, t(63) = -1.05, p = .298, d = 0.26. This suggests that the Sotos syndrome pragmatic language profile is characterised by less difficulty with stereotyped language. Figure 2 shows mean scaled scores for the pragmatic language subscales.
Autism Subscale Profile
The CCC-2 has two subscales which assess behaviours typically associated with ASD; social relations (e.g. interaction with other children) and interests (e.g. preference for the same favourite activity). In order to determine whether children with Sotos syndrome had relative differences in these behaviours, a paired samples t-test was used to compare these subscale scores. The analysis revealed a significant difference between the social relations subscale and the interests subscale, t(30) = − 5.99, p < .001, d = 1.08, indicating that children with Sotos syndrome had greater difficulty with social relations, compared with restricted interests. In order to establish whether this profile of behaviours was syndrome-specific, a 2 × 2 (Group: Sotos/WS x ASD subscales: I, J) mixed measures ANOVA was used to compare scaled scores for the ASD subscales between participants with Sotos syndrome and participants with WS. The analysis revealed a significant main effect of subscale; F(1, 63) = 27.74, p < .001, ηρ² = .306 and a significant subscale x group interaction; F(1, 63) = 8.79, p = .004, ηρ² = .122, indicating that both groups had uneven ASD subscale profiles but the nature of these profiles differed. There was no effect of group; F(1, 63) = 0.04, p = .835, ηρ² = .001.
As there was a significant subscale x group interaction, post-hoc comparisons (using a Bonferroni correction p < .025 required for significance) were used to compare scores for the two ASD subscales within the Sotos syndrome and WS participants. As reported above, participants with Sotos syndrome had significantly greater difficulty with social relations, compared with restricted interests. However, the same analysis with participants with WS revealed no significant difference between the social relations subscale and the interests subscale, t(33) = − 1.60, p = .120, d = 0.27. Overall, this suggests that children with Sotos syndrome tend to have less difficulty with restricted interests compared with social relations, and this profile was specific to the Sotos syndrome participants. Figure 3 shows mean scaled scores for these subscales.