Finally, the proportion of variance in language outcomes accounted for in the present study (up to 69%) is higher than in most previous studies.
The present analysis suggests, in agreement with Bishop and Snowling (2004), that non-phonological and phonological language skills are the two main sources of variance in language abilities among SLI, dyslexia and control children.
This suggests that knowledge of formulaic sequences helps to explain 67% of the variance in language learners' language proficiency.
In short, the higher explained variance, previous research on the separability of language tasks, and the content validity study of the EPE, all make the three-factor solution more interpretable than the two-factor solution.
Fifty-one percent of the variance in vocabulary outcomes and between 59 to 69% of the variance in language outcomes was predicted by the regression models.
A significant proportion of the variance in language outcome may therefore be explained by additional changes in tracks not examined here.
Tomblin and Zhang reported a confirmatory factor analysis of their measures which provided support for a unidimensional language factor which accounted for 51% of the variance in language test scores.
Despite the large reported variability in language outcomes, the present study was able to account for a significant proportion of the variance in language outcomes (up to 69%), which was higher than that in many studies.
Indeed, it may be that across the normal range, a single factor explains most of the variance in language abilities (as suggested by Hayiou-Thomas et al., 2006; Dale et al., 2010).
We hypothesised that language impairments would be found in children with TBI regardless of dysarthria, with variance in language ability being partially explained by cumulative effects of tract abnormalities.
