Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Superior temporal lobe lesions are associated with typed spelling deficits in chronic left hemisphere stroke
Poster Session E, Sunday, September 14, 11:00 am - 12:30 pm, Field House
Zoe White1,3, Ryan Staples2,3, Peter Turkeltaub1,2,3,4, Andrew DeMarco1,2,3; 1Department of Rehabilitation Medicine, Georgetown University Medical Center, 2Department of Neurology, Georgetown University Medical Center, 3Center for Brain Plasticity and Recovery, Georgetown University Medical Center, 4MedStar National Rehabilitation Hospital
Acquired agraphia is common after left-hemisphere stroke. Recent work has focused on localizing specific components of a brain network associated with written language production. During spelling-to-dictation, auditory phonological input may be converted to graphemes, or matched to a long-term store of phonological/orthographic word-forms either directly or indirectly via semantics. Lesion studies of agraphia have largely focused on distinguishing components of orthographic long-term vs working memory through lexicality manipulations (real vs pseudoword). These studies have implicated the fusiform gyrus and frontal lobe in orthographic long-term memory, and phoneme-to-grapheme conversion with the angular and inferior frontal gyri. These processes are influenced by psycholinguistic characteristics of individual words, which are well-studied in reading but less in spelling. A large-scale study of psycholinguistic influences on agraphic errors has not been conducted. Here we examine psycholinguistic characteristics influencing spelling errors and their lesion correlates in a large cohort of chronic left-hemisphere stroke survivors. We studied 73 chronic left-hemisphere stroke survivors and 69 healthy controls. Anatomical lesions were traced on MPRAGE/FLAIR scans. Participants completed a typing-to-dictation task consisting of 32 real words presented auditorily (balanced for frequency, imageability, regularity). Logistic mixed-effects models examined whole-word accuracy as a function of group (patient vs control) and psycholinguistic properties (e.g. high vs low frequency) plus interactions, controlling for education, with random effects of participant and item. Multivariate lesion-symptom mapping examined lesion locations associated with accuracy, plus tails of the psycholinguistic effects (e.g. high covarying for low), controlling for lesion size and education (voxelwise p<.005, clusterwise p<.05, 10000 permutations). Accuracy was lower in patients (controls: 90%, patients: 58%; β=-2.52, p<.001). There were main effects of both frequency (high: 79%; low: 68%; β=-1.12, p<.001) and imageability (high: 78%; low: 69%; β=-1.02, p=.017) but no interaction by group. There was a trend of overall regularity (regular: 75% irregular: 72%; β=-0.788, p=.089), with an interaction of group (β=0.639, p=.004). There was no frequency×regularity or frequency×imageability interaction in either group. Overall accuracy was associated with a lesion cluster encompassing mid/posterior superior temporal gyrus, including supramarginal gyrus, and underlying white matter (MNI=-54, -24, 1, volume=21.7cc). Imageability (low, controlling for high) was associated with a cluster encompassing supramarginal and superior temporal gyrus (MNI=-51, -40, 18, volume=3.4cc). We found superior temporal lesions associated with spelling errors, potentially due to disruption of semantic and phonological contributions to spelling. This contrasts with prior findings of agraphia associated with parietal lesions, which may relate to differing task demands (here, typing-to-dictation). Psycholinguistic characteristics influenced errors similarly between patients and controls, with patients more sensitive to irregular words. We found no interaction effects of psycholinguistic features, however there were few words per category. An exaggerated imageability effect was associated with lesions to STG/SMG, suggesting low imageability words place extra emphasis on auditory-phonological representation. In the future we will examine more sensitive measures like Levenshtein-distance, examine error types and self-corrections, and apply connectome-based lesion-symptom mapping.
Topic Areas: Disorders: Acquired, Writing and Spelling