Modified COSMIC 9
only limited opportunity for higher-level functions of social interaction (i.e., joint
attention and shared enjoyment). This setting was therefore omitted.
Standardisation was sought for the additional activities sampled. As group
teaching and free play situations were considered likely to elicit different types of child
communication behaviour (e.g., with responses more likely during a structured teaching
period and initiations more likely during unstructured play), both a structured teacher-led
activity and an unstructured free-play session were included within the modified
behaviour sampling. A standard set of highly motivating toys was used so that each
participant child with ASD would have similar opportunities and motivators for
communication. Despite this standardisation, the group interaction remained naturalistic
in that no specific instructions or presses were introduced by the teacher or researcher.
Event recording was retained as the method of coding behavior from session videotapes.
Revisions of the original COSMIC coding structure were also undertaken. Modifications were
made to the coding of the communication forms and functions, so as to reflect the potential
abilities of higher-functioning and more verbally-able children. New codes evolved from those
of the original COSMIC, review of the literature on social communication skills in ASD (e.g.,
Drew et al., 2007; Wetherby, 1986; Wetherby et al., 1989), and the results of pilot work.
Communication functions were classified into three super-ordinate categories; Behaviour
Regulation, Dyadic Social Interaction, and Joint Attention (Wetherby et al., 1989). Forms were
extended to include non-verbal Showing and Giving. Actions directed toward an examiner (e.g.,
bringing and showing or ‘dumping’ an object, and proximal pointing) were considered forms of
proto-joint attention (see Drew et al., 2007). Verbal forms were extended to include