Construct validation of the verb naming test (Casilio et al., 2025)

Purpose: Although there is widespread agreement pertaining to the cognitive processes underlying spoken word production, more generally in aphasia, multiple competing accounts exist regarding the processes involved for verb production, specifically. Some have speculated that suboptimal control of certain item properties (e.g., imageability) may be partially responsible for conflicting reports in the literature, yet there remains a dearth of research on the psychometric validation of verb production tests for aphasia. The purpose of the present study was to investigate the cognitive constructs underlying the Verb Naming Test (VNT), a relatively commonly used verb production test, by expanding upon an item response theory (IRT) modeling framework we previously described.

Method: Using an archival data set of 107 individuals with aphasia, we specified a series of IRT models to investigate whether item covariates (argument structure, imageability), person covariates (aphasia subtype, severity), and their interactions were predictive of VNT item response patterns.

Results: Across all models, covariates that were most strongly associated with lexical-semantic processing (imageability, aphasia severity) were significant predictors. In contrast, covariates that were most strongly associated with morphosyntactic processing (argument structure, aphasia subtype) were minimally predictive.

Conclusions: VNT item response patterns appear to be primarily explained by covariates representing lexical-semantic processing. In particular, we identified an important role of imageability, a covariate not controlled for in the VNT’s item design, which both aligns with a body of prior research and further illustrates the challenge of differentiating morphosyntactic processing from lexical and semantic processes during word production.

Supplemental Material S1. VNT dataset.

Supplemental Material S2. Analysis code.

Supplemental Material S3. Full results from the Descriptive IRT Model (1-PL IRT with random item effect).

Supplemental Material S4. Preliminary explanatory IRT model and results using common item covariates.

Casilio, M., Fergadiotis, G., Cho, S.-J., Steel, S., Fleegle, M., Dickey, M. W., & Hula, W. (2025). Construct validation of the verb naming test for aphasia. Journal of Speech, Language, and Hearing Research, 68(4), 1932–1949. https://doi.org/10.1044/2024_JSLHR-24-00236