Modified COSMIC 23
regulation, children with better language showed greater rates of communication for the
purposes of social interaction and joint attention. This latter finding is consistent with literature
suggesting concurrent association between joint attention skills and language development (e.g.,
Loveland & Landry, 1986; Mundy, Sigman, Ungerer, & Sherman, 1987; Mundy, Sigman, &
Kasari, 1994), and suggests that such an association holds even when evaluating social
communication behaviours across different settings (e.g., the school and the clinic).
M-COSMIC rates of initiation were strongly associated with scores on the standardised
language measures. In contrast, associations between language ability and M-COSMIC rates of
response were significant only with respect to parent and teacher reports of ability, but not the
direct clinician assessment. Such a pattern may be explained by the fact that the M-COSMIC and
parent and teacher reports of language evaluate functional and naturally-occurring
communication behaviours, while a direct assessment by a clinician evaluates the language
generated by specific presses in a contrived setting.
This broad pattern of results therefore indicates aspects of communication form, function
and role that show robust associations with language ability, along with other aspects of
communication that show more limited language association. This suggests a degree of
independence of verbal and non-verbal communication forms, of functions behaviour
regulation, social interaction and joint attention, and of roles of initiation and response, from
one another (Drew et al., 2007; Wetherby et al., 1989). In this way, results yielded by the M-
COSMIC are consistent with the literature on social communication in these children, thereby
evidencing validity of this naturalistic observational tool. Furthermore, specificity of the new
measure is also demonstrated, with a predictable pattern of associations evident among the