We argue in this paper that deep properties of discourse structure can be revealed by mapping discourse representations to networks and analyzing them using tools from network theory. Two talk clarified corpora, C58 and STAC, that have a place with various talk types and dialects were looked at and examined. The discourse representations of both corpora were based on a variety of key network indices, which demonstrate the distinct network profiles of the two kinds of discourse. In addition, strong tendencies in building or avoiding the construction of discourse relations between utterances for permissible three-node discourse subgraphs were illuminated by the discovery of both network motifs and antimotifs for the discourse networks in the two corpora. New discourse structure rules based on the properties of the networks that underpin discourse representation may emerge as a result of these findings.
HTML PDFShare this article
