Picture Exchange Communication System: Moderators of Collateral Speech Gains

Background:

For some with autism spectrum disorder (ASD) the Picture Exchange Communication System (PECS) results in collateral speech gains (e.g.Yoder & Stone, 2006). However, speech gains associated with PECS are not universally observed (e.g., Carr & Felce, 2007). Consequently, it is important to identify child variables that may influence the likelihood of collateral speech gains that occur concurrent with or subsequent to the implementation of PECS. Previous research has demonstrated that several variables appear to be associated with collateral speech production including levels of speech prior to intervention (Carr & Felce, 2007); vocal imitation, joint attention, and play (e.g., Toth, Munson, Meltzoff, & Dawson, 2006); and treatment intensity (Warren, Fey, & Yoder, 2007).

Objectives:

The purpose of this paper was to examine whether factors (listed above) associated with collateral speech gains were addressed in experimental investigations involving PECS and whether these factors may have contributed to collateral speech outcomes.   

Methods:

Studies were identified by searching PyscINFO, ERIC, and Academic Search Premier using the terms autism and Picture Exchange Communication System or PECS between 1994 and 2011. To expand on previous reviews (e.g., Flippin et al., 2010), we included adults and children with disabilities (e.g., autism, Down syndrome) and coded participant variables (e.g., level of speech) and treatment intensity.

Additionally, the efficacy of PECS was examined by scrutinizing dependent measures reported in studies reviewed, including independent requests and speech outcomes. We computed effect size as the percent of all non-overlapping data (PAND; Parker, Hagan-Burke, & Vannest, 2007) for studies using a single case design (SCD) and Hedge’s g for group design studies.

Results:

Twenty-five studies were reviewed in this study, including 21 SCD studies and 4 group design studies. Many of the potential moderators described above were rarely reported, and participants were described to varying levels of detail. To date, it appears that the focus of this literature thus far has been to examine the effectiveness of PECS rather than identify potential moderators of collateral speech gains.

Of the 64 participants in the 21 SCD studies, speech-related outcomes were reported for 17. Results showed that participants who had some speech prior to the PECS intervention had greater collateral speech gains (n = 9, average PAND effect size of .80) than participants who were preverbal (n = 7, average PAND effect size .63). Very few studies reported information about participant imitation, play, joint attention, or included comprehensive information about treatment intensity, so these potential moderators were not examined further.

Conclusions:

            Based on this meta-analysis, it appears that children who have some speech prior to intervention may show better speech outcomes than children at a preverbal stage. However, other child characteristics were not investigated as potential moderators because they are rarely reported (e.g., initiating and response to joint attention, verbal and motor imitation). Future studies should adopt a unified assessment procedure to assess and report these variables so that studies can be synthesized and compared.