Examining the Constructs of Narrative and Component Language: A Confirmatory Factor Analysis

• In Event: Poster Session 07
  In Poster Session: PS 07 Topic - Developmental Disabilities

Abstract

Purpose: Narrative language is often thought of as a specialized skill under the oral language umbrella. However, narrative language may provide unique information about the “use” of language above and beyond other component language measures (e.g., Clinical Evaluation of Language Fundamentals, CELF). In fact, some studies have suggested that narrative language may be a separate skill that bridges the gap between oral language and reading abilities (Snow, 1991; Westby, 1991).
Methods: Participants in this study included 139 children in 3rd and 4th grade between the ages of 7 and 11 years (Mean = 9.30 years). Children in this study were originally recruited to participate in a reading intervention study and received an assessment battery that included standardized and unstandardized language assessments prior to intervention. Raw scores from CELF-4 subtests (i.e., Concepts and Following Directions, Recalling Sentences, and Formulated Sentences), TNL subtests (i.e., Oral Narrative and Narrative comprehension), WASI vocabulary subtest, PPVT-4, and mean length of utterance (MLU) from SALT transcripts of the TNL were used in this analysis. To test our hypothesis about the measurement of component language and narrative language, we constructed two CFA Models: a two-factor model, Model 1 (Figure 1), and a single factor model, Model 2 (Figure 2). Both models fit very well (Model 1: CFI = 1.00, RMSEA = .00, and SRMR = .03, AIC = 6891.66; Model 2: CFI = .99, RMSEA = .02, and SRMR = .04, AIC = 6894.78). Since the two-factor model fit slightly better, we will continue with discussion of these results.
Results: Significant factor loadings ranged from .31 to .77. Individual loadings on component language suggested that these measures were good indicators of the factor: CELF Following Directions (.64), CELF Recalling Sentences (.65), and CELF Formulated Sentences (.59), PPVT (.71), and WASI Vocabulary (.69). Individual loadings for narrative language included: TNL Narrative Comprehension (.77), TNL Oral Narrative (.68), and MLU (.31). Similarly, these measures represented strong indicators of the narrative language factor. The latent factors of narrative and component language correlated highly (r = .86).
Conclusions: Our aim was to examine if narrative and component language abilities were separate constructs. The two-factor model exhibited excellent fit with strong factor loadings however the covariance between the two factors was high. Like prior studies that examined the dimensionality of language, these findings supported that although the two-factor model fit best, the more parsimonious, single-factor model may best characterize narrative and component language skills. Just as clinicians and researchers of language continue to examine receptive and expressive language skills to produce a more fine-grained understanding of language ability, these results support including both component language and functional, narrative language skills to provide more robust descriptions of language ability broadly. Together these findings support the theory that narrative language is not totally separate from component language skills but contributes to a wholistic representation of language skills in children with dyslexia.

Authors

• Casy Walters, Georgia State University
  Presenting Author

• Rose A Sevcik, Georgia State University
  Non-Presenting Author

• Robin Morris, Georgia State University
  Non-Presenting Author