Poster #115 - The Conversation Compass© Communication Screener – Revised

• In Event: 3-106 - Poster Session 11
  In Poster Session: PS 11 Section - Language, Communication

Integrative Statement

Developing oral language skills, particularly conversational discourse skills, is an important developmental task of early childhood. Despite the importance of conversation skills during preschool, few tools exist for teachers to assess children’s conversation skills specifically within the classroom context. The Conversation Compass© Communication Screener (CCCS) was developed to address this void. The teacher-reported CCCS acknowledges the multifaceted nature of conversation skills by identifying children’s ability to use discourse at school via language-learning conversations in the structured context of the preschool classroom. Prior work assessing the psychometric properties of the CCCS provided initial support for its reliability and validity using a small sample of Head Start teachers and children (Authors, 2017). However, at the conclusion of that study, the authors determined that an adapted, rescaled version of the tool should be utilized in future work (i.e., the Conversation Compass© Communication Screener-Revised; CCCS-R).
This study extends this work by refining and validating the CCCS-R with a larger, more ethnically and linguistically diverse sample. Specifically, we consider the factor structure of the CCCS-R, and then examine the convergent and predictive validity of the tool. Two primary research questions guided our work:
1. Do the CCCS-R subscales adapted from Authors (2017) map onto a statistically significant factor structure with reliable subscales?
2. Is there convergent validity between CCCS-R scores with the CELF-Preschool’s Descriptive Pragmatics Profile subtest and predictive validity with the CELF-Preschool’s Pre-Literacy Rating Scale subtest?

Participants: Children (N=255) from diverse were recruited from 22 Head Start classrooms in the Northeast. Classroom teachers (N=35) identified 28% (n=71) of the children as being dual language learners (DLLs). Children ranged in age from 3 to 5 years (M=4.00, SD=0.69), and 49% were girls.

Measures: Teachers completed the CCCS-R, a 37-item teacher-report measure of preschooler’s conversational language skills, and two subtests of the Clinical Evaluation of Language Fundamentals—Preschool (CELF-P; Semel, Wiig, & Secord, 2004): the Descriptive Pragmatics Profile (DPP), and the Pre-Literacy Rating Scale (PLRS).

Results: Exploratory factor analysis using an oblique rotation yielded four factors; confirmatory factor analyses indicated this fit the data (CFI=.860, RMSEA=.084; SRMR=.091). These factors included: pre-academic language, negative communication behaviors, social communication, and narrative talk and vocabulary knowledge. Nonparametric Mann-Whitney U tests and independent sample t-tests revealed subgroup differences in CCCS-R scores: girls received lower scores on negative communication behaviors, but higher scores on social communication, than boys. Additionally, non-DLL students received higher scores on pre-academic language and narrative talk and vocabulary knowledge than their DLL peers. Results from HLM regression analyses provided evidence of the CCCS-R’s convergent and predictive validity with the CELF-P subtests. For convergent validity, all scores were significantly associated with DPP scores in the directions to be expected. For predictive validity, all scores were significantly associated with PLRS scores in the directions to be expected.

Conclusions: This work suggests that the CCCS-R reliably and validly measures aspects of classroom conversation skills needed for success at school. The CCCS-R can be used in both research and practice to supplement language and literacy batteries used in preschools.

Authors

• Stephanie Michelle Curenton, Boston University
  Non-Presenting Author

• Jacqueline Prince Sims, Boston University
  Presenting Author

• Shana Rochester, Boston University
  Non-Presenting Author

• Shari L Gardner, SRI International
  Non-Presenting Author