Discourse complexity in the light of eye-tracking: a pilot Russian language study

Abstract

The paper explores the influence of discourse structure on text complexity. We assume that certain types of discourse units are easier to read than others, due to their explicit discourse structure, which makes their informational input more accessible. As a data source, we use the dataset from the MECO corpus, which contains eye movement data for 12 Russian texts read by 35 native speakers. We demonstrate that the approach relying on elementary discourse units (EDUs) can be felicitously used in the analysis of eye movement data, since fixation patterns on EDUs are similar to those on whole sentences. Our analysis has identified EDU outliers, which show shorter time of first fixation than estimated. We arranged these outliers into several groups associated with different discourse structures. First, these are statements with nominal predicates that set exposition of the text or macroproposition and, following those, EDUs that elaborate on the previous statement and signal the beginning of the narrative. Second, they are EDUs that serve as the middle component of a listing or a group of coordinated clauses or phrases. The final group represents EDUs that are part of an opposition, contrast or comparison. Discourse analysis based on EDUs has never been applied to eye movement data, so our project opens many avenues for further research of complexity of discourse structure.