AbstractBackgroundThe semantic fluency task involves naming as many items as possible in a given category (e.g., animals) in one minute. It is often interpreted as a measure of word retrieval, but also draws on executive function abilities to organize responses and develop task‐specific strategies. Indeed, both language skills and dementia severity ratings have recently been found to correlate with semantic fluency performance in patients with primary progressive aphasia (Riello et al., 2022; Scheffel et al., 2021). Here, we focused instead on how performance on language and executive function measures predicts longitudinal change in semantic fluency performance in patients with progressive agrammatic aphasia and/or apraxia of speech (AOS).MethodPatients with isolated agrammatic aphasia (n = 10), isolated AOS (n = 42), or a combination of the two (n = 59) with 1‐7 visits (mean = 2.19) were included. We used Bayesian hierarchical linear models to investigate whether baseline measures of executive function or language were associated with the rate at which semantic fluency performance changes over time while accounting for baseline diagnosis group, sex, education, age at baseline, and baseline disease duration.ResultSemantic fluency performance declined over time, at a similar rate irrespective of diagnostic group (isolated or combined aphasia or AOS; Figure 1). Across all patients, higher executive function abilities at baseline, as indicated by faster Trails Making Test B completion times, predicted higher baseline semantic fluency performance and also a steeper annual rate of decline (Figure 2). Similarly, higher fluency ratings during the Western Aphasia Battery‐Revised picture description at baseline were associated with higher baseline semantic fluency performance and faster decline. Baseline performance on the Boston Naming Test and Token Test were also associated with baseline semantic fluency, but not with longitudinal change.ConclusionThese results better delineate how language and executive function abilities support performance in the semantic fluency task. Higher abilities in those areas at baseline do not appear to be protective over time. Rather, the differences in semantic fluency that are present at baseline disappear as the disease progresses.