Abstract
In the present study, we combine linguistic annotation of oral texts in Russian with the registration of BOLD signal in functional MRI experiments to determine how and where semantic categories are represented in the human brain. Using the same stimuli material, we also analyze the differences in cortical activation in three thematic domains: description of nature, description of working principles of technical devices and more self-referential texts, addressing the question of human identity in conflict situations. We discuss methodological problems within the two approaches (microanalysis and macroanalysis) to study brain activation in natural conditions, i.e. under a continuous speech flow. Within the thematic domain studies, only minimally significant differences in brain activation were registered during the listening to texts from the three thematic groups. This outcome leads to the conclusion that the approach of thematic group contrasts (cognitive subtraction methodology) is not sufficient to study the mechanisms of text comprehension, and should be replaced by the modeling of multidimensional representations of semantic categories in time. Within the semantic category approach, we describe the neurolinguistic process of text understanding as the activation of 15 clusters responsible for semantic categories (e.g. "Conflict", "Mental", "Social"). Our data demonstrate that the clusters are widely distributed across the human brain. In contrast to the previous studies, we suggest that deep subcortical structures are involved in the processing of certain categories as well. The observed lateralization of category processing underlines the involvement of the right hemisphere in the processing of meaning.