Poster B21, Thursday, August 16, 3:05 – 4:50 pm, Room 2000AB

Localization of dual stream contributions to sublexical and lexical-semantic reading: a multivariate lesion-symptom mapping study of left hemisphere stroke survivors

J. Vivian Dickens1, Mackenzie E. Fama1, Andrew T. DeMarco1, Elizabeth H. Lacey1,2, Rhonda B. Friedman1, Peter E. Turkeltaub1,2;1Georgetown University Medical Center, Washington, DC, USA, 2MedStar National Rehabilitation Hospital, Washington, DC, USA

Introduction: Over the past half century, studies of alexia, an acquired disorder of reading, have enriched our understanding of the neurocognitive architecture subserving reading and language. The reading deficits observed in alexia reflect impaired sublexical and lexical-semantic processing and are frequently characterized along the dimensions of lexicality (word versus nonword), orthographic regularity (regular versus irregular spelling-sound correspondences), and concreteness (abstract versus concrete meaning). Localization of dorsal and ventral stream areas essential to reading aloud words that vary along these dimensions remains a matter of investigation. Here, we apply multivariate lesion-symptom mapping (LSM) in left hemisphere stroke survivors to identify brain areas critical for effects of lexicality, orthographic regularity, and concreteness in reading aloud. Methods: Participants were 73 native English speakers at least 6 months post unilateral left hemisphere stroke with no other significant neurological or psychiatric disorders. The reading battery consisted of three pairs of matched lists targeting effects of lexicality, orthographic regularity, and concreteness, respectively: words/pseudowords, regular/irregular, and concrete/abstract. The latter two tasks were completed by a subset of participants (N = 48). Accuracies in reading aloud items in the six categories served as the behavioral outcomes, with accuracy on the corresponding matched items being covaried in each analysis (e.g., for concrete word reading, abstract word reading was included as a covariate). To account for voxelwise lesion covariance, we applied support vector regression lesion-symptom mapping (SVR-LSM), a multivariate LSM method that considers all voxels in one regression, using a MATLAB toolbox developed in our lab ( To further minimize lesion autocorrelation effects, lesion size was regressed out of the analyses, and significance was determined through a permutation approach at a voxelwise threshold of p < .005, with cluster-level family-wise error controlled at .05. Results: Inaccurate concrete word reading was associated with lesions to pars orbitalis and pars triangularis of left inferior frontal gyrus. Inaccurate regular word reading was associated with lesions spanning left planum temporale, parietal operculum, and ventral precentral gyrus. Inaccurate pseudoword reading was related to lesioned left ventral precentral gyrus, overlapping with the region implicated in reading regular words, although this cluster was just below the cluster-level threshold (p= .0509). Conclusions: Our results demonstrate specific cortical regions important for effects of concreteness, regularity, and lexicality in reading aloud. Lesions to left anterior inferior frontal gyrus may diminish the normal concrete word advantage by degrading semantic control and selection. Left planum temporale, parietal operculum, and ventral precentral gyrus underlie auditory, somatosensory, and articulatory representations, respectively. That lesions to these regions resulted in diminished reading of regular words points to a critical role for sound-motor integration in sublexical decoding of print. The association of poorer pseudoword reading with lesioned ventral precentral gyrus further supports the role of articulatory-based representations in phonological decoding. Overall, our application of multivariate LSM provides evidence for a specialized role for dorsal stream areas in accessing articulatory-based representations in sublexical reading and for ventral stream areas, particularly left anterior inferior frontal cortex, in lexical-semantic contributions to reading.

Topic Area: Perception: Orthographic and Other Visual Processes