The role of LIFG-based executive control in recovery from garden-paths during sentence comprehension

• ¹ Rice University, United States

Introduction Individuals with damage to the left inferior frontal gyrus (LIFG) have difficulties when sentence comprehension depends on accessing a dispreferred word meaning or sentence structure, and these comprehension difficulties have been attributed to co-occurring deficits in executive control (e.g., Novick, Kan, Trueswell, & Thompson-Schill, 2009; Vuong & Martin, 2011). The present study further investigated the consequences of LIFG-based executive control deficits on comprehension by examining the role of verb bias on the ability to recover from an object/subject garden-path. We predicted that verb bias toward the inappropriate structure would magnify the syntactic garden-path effect. However, it was possible that verb bias alone could create a garden-path effect for patients with LIFG damage. Methods Two patients had lesions that included the LIFG while one patient had a non-LIFG (parietal) lesion. All were screened for good comprehension of single words and of active and passive reversible sentences. Executive control abilities were examined by three tasks: Stroop, picture-word interference, and recent negatives recognition memory (see Vuong & Martin, 2011). The sentence materials were drawn from Vuong and Martin (2014). Sentence structure was manipulated by including object/subject garden-path sentences and closely matched syntactically unambiguous control sentences (see examples). Prepotency of the garden-path interpretation was manipulated by including verbs biased towards the contextually inappropriate interpretation (i.e., transitive bias – e.g., “visit”) and neutral verbs (i.e., transitive and intransitive interpretations equally likely – e.g., “coach”). Garden-path: While the man/ coached/ the woman/ attended/ the party by herself. Control: While the man/ coached,/ the woman/ attended/ the party by herself. The three patients, and a group of eight healthy controls, were tested on two self-paced sentence processing tasks, using sentence regions as above, involving meaning interpretation (Experiment 1) and grammaticality judgments (Experiment 2). Results In contrast to healthy controls and the non-LIFG patient, one of the LIFG patients showed a deficit in meaning interpretation that was worse with biased verbs, across garden-path as well as syntactically unambiguous sentences (mean verb bias effect = 32.5%, which was 7 standard deviations above the control mean of -3%, SD = 4.6). She showed spared performance on grammaticality judgments. The other LIFG patient presented with a more severe executive control deficit, and showed spared performance only for grammaticality judgments in the easiest condition, namely the condition involving syntactically unambiguous sentences with neutral verbs (but not in the remaining conditions, including the corresponding condition with biased verbs). Conclusion The results add to prior studies in arguing for a role of executive control across diverse sentence processing situations, including overcoming verb bias preferences.