Analysis of the Discourse Macrostructure in the Narratives of People With Aphasia: a Comparison of Different Methods and Approaches

In modern clinical linguistics, there is still no generally accepted approach to the analysis of the discourse macrostructure. The macro level encompasses many linguistic competencies, the consideration and evaluation of which require additional sensitive criteria that allow to compare both micro-level elements that influence the holistic perception of speech and global tools that allow the speaker to link discourse elements (for example, sentences or larger parts in the narrative organisation). This study examined the common approaches to the discourse analysis of narratives in patients with aphasia at the macro level: Quality of Content (QoC) scale, global and local coherence, depth of the rhetorical trees (according to the Rhetoric Structure Theory), number of elementary discursive units (EDU) in the narrative, as well as criteria of the Russian Aphasia Test (RAT) for the speech evaluation at the macrolevel, and four aspects of discourse coherence. Each method was used on a subcategory of 16 narratives of patients with aphasia and 16 participants without speech disorders selected from the sub-tests norming of the RAT. As a result of the study, all methods except global coherence and RST methods have shown sensitivity to differences in the discourse between people with aphasia group and normal population. Subsequently, the correlation analysis revealed that four aspects of coherence demonstrate a strong correlation with local connectivity, as well as with all parameters of the RAT scale, especially with micro-linguistic parameters included, affecting both the local coherence and the understandability, informativeness, general connectedness and clarity of discourse. The relationship between the content quality criterion and the four aspects of discourse was also proved and interpreted as a result of the same overall concept of discourse evaluation, which was the final perception of the discourse as a unified whole, and not its constituent parts. The paper also discusses in detail the relationship of key parameters from each method in order to explain their conceptual similarity, and hence the more accurate nature of the phenomena under study.