Two dissociable semantic mechanisms predict naming errors and their responsive brain sites in awake surgery. DO80 revisited

How do we choose words, and what affects the selection of a specific term? Naming tests such as the DO80 are frequently used to assess language function during brain mapping in awake surgery. The present study aimed to explore whether specific semantic errors become more probable under the stimulation of specific brain areas. Moreover, it meant to determine whether specific semantic characteristics of the items may evoke specific types of error. A corpus-based qualitative semantic analysis of the DO80 items, and the emitted naming errors to those items during direct cortical electrostimulation (DCE) revealed that the number of hyperonyms (i.e. 'vehicle' for car') of an item predicted the emission of a synonym ('automobile' for 'car'). This association occurred mainly in frontal tumor patients, which was corroborated by behavior to lesion analyses. In contrast, the emission of co-hyponyms was associated with tumors located in temporal areas. These two behavior-lesion associations thus dissociated, and were also dependent on item semantic characteristics. Co-hyponym errors might generate from the disruption in a temporal semantic-to-lexical process, and the production of synonyms could be the result of an impairment in a frontal lexical-selection mechanism. A hypothesis on the lexical selection mechanisms exerted by the inferior frontal gyrus is proposed. Crucially, the present data suggest the need for more restrictive naming tasks, with items conditioned by tumor location.

Automated summary

This study investigates the impact of brain surgery on language function, identifying a link between specific brain stimulation and semantic errors, which are influenced by the semantic characteristics of items. Patients with frontal tumors are more likely to produce synonym errors, while those with temporal tumors are more likely to produce co-hyponym errors. Therefore, stricter naming tasks conditioned by tumor location are necessary to understand the effects of brain lesions on language selection mechanisms.