Poster A33, Thursday, August 16, 10:15 am – 12:00 pm, Room 2000AB
Neural Correlates of Sentence Processing in Acute Stroke Patients
Sigfus Kristinsson1, Brielle Stark1, Grigori Yourganov1, Alexandra Basilakos1, Helga Thors1, Julious Fridriksson1;1Center for the Study of Aphasia Recovery at the University of South Carolina
Introduction Syntactic processing is a fundamental aspect of language and is often compromised in aphasia. The ventral inferior frontal gyrus has long been suggested as an important area for sentence processing, particularly Broca’s area.1-7 Other areas commonly implicated in syntactic processing include the posterior temporal areas,8-10 and anterior superior and middle temporal gyrus.11-13 Contrasting previous studies, recent voxel-based lesion-symptom (VLSM) studies have consistently implicated a temporo-parietal area, including the posterior middle temporal gyrus (pMTG), as being important for sentence processing.13-18 The current study examined the association between left hemisphere stroke and sentence processing in persons with acute aphasia. We hypothesized that damage to the pMTG, and not to Broca’s area, would predict performance on the syntactic processing task. Method This was a prospective study. A total of 56 participants were recruited from Iceland and 56 from a local hospital. All participants had incurred a single stroke to the left hemisphere (Table 1). Participants completed a 45-item sentence-picture-matching task which includes nine types of sentence structures (Table 2).19 Performance on sentences with canonical word order (C: Types 1, 4, and 8) was compared to that on non-canonical sentence types (NC: 2, 5, and 9). All participants underwent MRI scanning. Brain lesions were demarcated on DWI images by a trained neurologist. Our main analysis was a stepwise regression analysis using proportional damage in all predefined language ROIs within The Dual Stream Model of speech/language processing20-21 as independent variables and total score (TS), C scores and NC scores as dependent variables, in three separate analyses. Lesion volume was controlled for in all analyses. Results Participants’ mean TS was 35.8 points (SD=9.1, range: 12-45). Mean score on C sentences was 13.0 points (SD=2.96, range: 4-15), and mean score on NC sentences was 11.3 points (SD=3.6, range: 3-15). Patients scored significantly higher on C sentence types than NC on average (t=7.44, p<.001; Figure 1). Ceiling effects were significantly greater on C sentence types than NC sentence types (52 vs. 31, X2=7.24, p=.007). Figure 2 shows a lesion overlay map. Region-wise analysis revealed that impaired SP was best predicted by proportional damage to pMTG, angular gyrus (AG) and the inferior frontal gyrus pars triangularis (F=31.0, R2=.482, p<.001; Table 3). Performance on C sentence types was best predicted by damage to AG, the middle temporal gyrus (MTG) and the superior temporal gyrus (STG) (F=21.6, R2=.393, p<.001). Damage to pMTG, AG and the inferior frontal gyrus pars opercularis predicted performance on NC sentences (F=31.4, R2=.485, p<.001). Conclusion These results indicate that damage to temporo-parietal cortex predicts performance on an auditory-visual sentence processing task in acute stroke patients. Damage to the pMTG is the strongest predictor of TS and NC scores, explaining 33.7% and 37.1% of the variance in scores, respectively. Damage to a temporo-parietal area predicted C scores as well. These results suggest that the pMTG may play a crucial role in complex sentence processing. Importantly, the results furthermore show that Broca’s area does not seem to be critically involved. These aspects will be studied further.
Topic Area: Grammar: Syntax