Modeling Confrontation Naming and Discourse Informativeness Using Structural Equation Modeling
We are especially grateful to the study participants and AphasiaBank. We also thank the volunteers in the Aging and Adult Language Disorders Lab at Portland State University and especially Jessica Jensen for assistance with language analyses.
Background: People with aphasia (PWA) and their families identify as their priority the ability to use language at the discourse level in order to meet their daily communicative needs. However, measuring connected speech can be a challenging task due to the complex and multidimensional nature of discourse. As a result, professionals often depend on confrontation naming tests to identify and measure impaired underlying cognitive mechanisms that are also hypothesized to be important for discourse production.
Aims: In the current study, we investigated the validity of making inferences about discourse performance based on scores from confrontation naming tests. Specifically, we investigated the strength of the relationship between word retrieval abilities, and the ability to convey information during discourse production.
Method & Procedures: Data from 118 monolingual PWA were retrieved from AphasiaBank and analyzed using structural equation modeling. Performance in confrontation naming tests was modeled as a latent variable based on the Boston Naming Test, the Western Aphasia Battery – R Naming Subtest, and the Verb Naming Test. Performance at the discourse level was modeled based on indices of informativeness in three discourse tasks (free speech, eventcasts, and story re-tell). Informativeness was quantified using the percentage of Correct Information Units.
Outcomes and Results: Based on the fit statistics, the model exhibited adequate fit, indicating that the relationship between confrontation picture naming and informativeness was adequately reflected in the model. We found a strong relationship between confrontation naming test performance and discourse informativeness (standardized regression coefficient between the two latent factors = .79).
Conclusions: Performance on confrontation naming tests was a strong predictor of the amount of information PWA communicated during discourse production. However, our results also highlight that performance on the latter cannot be predicted solely from the former, as evidenced by the large proportion of unexplained variance in the informativeness latent variable.
Locate the Document
Gerasimos Fergadiotis, Maria Kapantzoglou, Stephen Kintz & Heather Harris Wright (2019) Modeling confrontation naming and discourse informativeness using structural equation modeling, Aphasiology, 33:5, 544-560, DOI: 10.1080/02687038.2018.1482404