Presentation

Marianne Casilio¹, Jillian L Entrup¹, Sarah M Schneck¹, Caitlin Onuscheck¹, Deborah F Levy¹, Maysaa Rahman¹, Michael de Riesthal¹, Stephen M Wilson¹; ¹Vanderbilt University Medical Center

Despite a rich tradition of research investigating the brain-behavior relations of language processing in aphasia, relatively few studies [1–3] have focused on connected speech, although the ability to produce language in naturalistic contexts is a critical functional outcome. The aim of our study in progress is to characterize the neural bases of various connected speech disruptions in aphasia using a novel and validated approach. We extracted audiovisual connected speech samples produced by 51 participants with acute post-stroke aphasia, constituting a subset of participants from a larger study [4]. Each sample was then scored by a speech-language pathologist blinded to all other data using the Auditory-Perceptual Rating of Connected Speech in Aphasia (APROCSA) [5], a 27-item measure with strong psychometric properties that, per a factor analysis, yields scores on four underlying constructs: paraphasia (mis-selection of words and sounds), logopenia (paucity of words), agrammatism (omission of morphosyntactic structures), and motor speech (impaired speech motor programming or execution). We created a new scoring system, where items with >|.4| factor loadings from our original analysis are retained and then summed. Lesions were manually delineated on acute DWI or CT images, following previously described methods [4], and voxel-based lesion-symptom mapping was performed [6]. Lesion coverage spanned the left middle cerebral artery territory, as expected in a stroke cohort, and APROCSA scores were broadly distributed, suggesting there was sufficient variability in both neural and behavioral dimensions. We observed a clear anterior-posterior division among the neural correlates of APROCSA scores that paralleled neoclassical [7] and contemporary [8] views of aphasia. Paraphasia and logopenia, both of which reflect impaired lexical-semantic and phonological processing, had predominantly posterior correlates. Paraphasia was associated with the posterior superior temporal sulcus while logopenia was associated with white matter underlying the middle temporal gyrus, a critical bottleneck for language processing [9]. In contrast, agrammatism and motor speech, which reflect morphosyntactic and motoric processing deficits respectively, had anterior correlates. Both were associated with the ventral precentral gyrus, although agrammatism was additionally associated with the inferior frontal gyrus, and motor speech was associated with the basal ganglia and underlying white matter. Taken together, our preliminary findings suggest that the neural correlates of connected speech in aphasia, as measured using an auditory-perceptual approach, are well-aligned with the extant literature on language processing in the brain. Our next steps will be to score available samples from an additional ~100 participants, to continue to recruit participants, and, when our sample size is larger, to employ multivariate rather than univariate lesion-symptom mapping, correcting for multiple comparisons with permutation testing. [1] Wilson et al., Brain. 2010;133(7):2069-2088. [2] Mirman et al., CABN. 2019;19(5):1286-1298. [3] Ding et al., Brain 2020;143(3):862-876. [4] Wilson et al. Brain. 2022; online. [5] Casilio et al., AJSLP. 2019;28(2):550-568. [6] Bates et al. Nat Neurosci. 2003;6(5):448-450. [7] Geschwind. Brain. 1965;88(2):237-294. [8] Fridriksson et al. Proc Natl Acad Sci USA. 2016;113(52):15108-15113. [9] Turken, Dronkers. Front Syst Neurosci. 2011;5:1-20.

Topic Areas: Disorders: Acquired, Language Production