Related Experiment Videos
Curvilinear Relationship Between Confrontation Naming and Monologic Narrative Discourse in People With Aphasia: A
Christie Carroll-Duhigg1, Sarah Grace Dalton2, Yu-Yu Hsiao3
1Department of Speech and Hearing Sciences, The University of New Mexico, Albuquerque.
Purpose:
The purpose of this study was to examine the relationship between latent naming and narrative discourse ability in people with aphasia, extending prior research with a revised discourse measurement model focused on narrative production.
Method:
Data from 223 participants from AphasiaBank were analyzed (final n = 220). Latent naming ability was indicated by the Boston Naming Test-Short Form, Verb Naming Test, and Western Aphasia Battery-Revised Naming Subtest (WABN). Latent narrative discourse ability, in the context of AphasiaBank's Cinderella storytelling task, was indicated by main concept analysis (MCA) subscores Accurate and Complete (AC), and Attempts (ATT), Sequencing (SEQ), Story Grammar Episodic Complexity (SGEC), and Core Lexicon (CoreLex). Confirmatory factor analysis (CFA) and structural equation modeling were used to evaluate the narrative discourse measurement model and its association with latent naming ability, respectively. A scatter plot with locally estimated scatterplot smoothing revealed a curvilinear relationship between latent constructs, showing a weaker relationship at lower ranges of naming ability, and a disproportionately stronger relationship at higher levels. Linear (a priori) and quadratic (post hoc) paths were tested using robust maximum likelihood estimation to account for nonnormal distribution of AC and WABN.
Results:
Narrative discourse CFA demonstrated a good fit (e.g., comparative fit index [CFI] = .998, root-mean-square error of approximation [RMSEA] = .031, standardized root-mean-square residual [SRMR] = .004), with high loadings across indicators. The linear model also fit the data well (e.g., CFI = .990, RMSEA = .054, SRMR = .024), showing a strong positive association (β = .66; R2 = .44). However, the curvilinear model improved fit (Δ Akaike Information Criterion = -12) and explained more variance (R2 = .61).
Conclusions:
Differences in elicited narrative discourse performance appear to be minimal among those with lower naming ability. However, larger differences in narrative discourse for those with higher level naming abilities suggest that even mild naming impairment may correspond with noticeably impaired narrative discourse.
Supplemental Material:
https://doi.org/10.23641/asha.34113318.