Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Lesions associated with reduced lexical-semantic contributions to visual word recognition in left-hemisphere stroke survivors
Poster Session C, Thursday, October 1, 10:45 am - 12:45 pm, Wangari Maathai
Yeonwoo Kim1, Ryan Staples1, Elizabeth Chang1, Peter Turkeltaub1,2,3; 1Center for Brain Plasticity and Recovery, Georgetown University, 2Center for Aphasia Research and Rehabilitation, Georgetown University, 3Research Division, MedStar National Rehabilitation Hospital
Left-hemisphere strokes often lead to reading deficits, or alexia. Examining how lesions affect specific reading subprocesses may improve diagnosis and treatment of alexia. We have previously used imageability effects to demonstrate that superior temporal sulcus lesions disrupt semantic-to-phonology mappings, leading to reduced semantic contributions to oral reading (Staples et al, 2025). We have also demonstrated that oral reading and visual word recognition, measured by lexical decision, share neural substrates in the middle temporal gyrus and angular gyrus (Chang et al, 2026), likely because both tasks rely on orthography-to-semantics mappings. Here, we examine imageability and frequency effects in lexical decision to assess orthography-to-semantics mappings without explicit phonological or articulatory demand in a large group of left-hemisphere stroke survivors. Participants include 135 left-hemisphere stroke survivors in the chronic phase (>6 months) (Age:59.20(13.05); Gender:62F/73M; Education:16.19(3.00); Median chronicity:17.54 months) and 128 matched controls (Age:60.29(12.81); Gender:68F/60M; Education:16.84(2.36)). Participants completed a written lexical decision task, in which they decided whether a stimulus was a real word or pseudoword. Performance was measured by accuracy and reaction time (RT). We used efficiency (accuracy/RT) which is sensitive to milder deficits that may not be reflected in accuracy, due to ceiling effects. Word items were counterbalanced in frequency and imageability, resulting in four word-types: high frequency/high imageability (HF/HI), high frequency/low imageability (HF/LI), low frequency/high imageability (LF/HI), and low frequency/low imageability (LF/LI). We used an ANCOVA to test for differences in efficiency scores by frequency, imageability, and group (stroke, control), controlling for age and education. We then used voxel-based support-vector regression lesion-symptom mapping (SVR-LSM) to assess semantic contributions to word recognition measured as a reduction in the imageability effect. We focused this analysis on low frequency words, which rely more on semantic content for recognition compared to high frequency words. Specifically, we examined regions associated with reduced LF/HI word efficiency controlling for LF/LI efficiency, as well as age, education, and lesion size. The ANCOVA revealed a significant interaction of imageability by frequency across groups (F(1,259)=58.085, p=4.75 x 10-13). There were also significant group by imageability (F(1,259)=8.101, p=0.005) and group by frequency (F(1,259)=7.266, p=0.007) interactions in efficiency scores, such that stroke survivors showed exaggerated effects compared to controls. SVR-LSM identified a large cluster related to reduced advantage of high imageability for recognizing low frequency words spanning the left angular gyrus (AG), left posterior middle temporal gyrus, extending to the left middle occipital gyrus (MOG). As a follow-up, we performed a parallel analysis to examine lesions that reduce the advantage of high frequency for low imageability words, and found a highly overlapping cluster in the left MOG and AG. Together these findings suggest that posterior parietal-occipital lesions are associated with loss of lexical-semantic support in visual word recognition. The identified regions are distinct from previous LSM results associated with reduced imageability effects in oral reading, which were associated with semantic-to-phonological mapping. However, they are highly overlapping with areas previously identified to be involved in both oral reading and visual word recognition efficiency overall. This suggests the identified regions are involved in orthography-to-semantic mapping.
Topic Areas: Disorders: Acquired, Meaning: Lexical Semantics