Discourcenotation

Discourcenotation is a proposed framework for annotating and representing the structure of discourse in both spoken and written language. It seeks to provide a unified notation for delineating discourse units, their hierarchical relations, and interactional functions such as questioning, answering, elaboration, and contrast.

In this scheme, discourse units are labeled (for example U1, U2) and connected by relation tags such as Rel=Question, Rel=Answer, Rel=Elaboration, and Rel=Contrast. Optional annotations may record prosodic cues, stance, or discourse markers (Tone=Neutral, Marker=Well). A simple depiction might be: U1: 'Where are you going?' Rel=Question → U2: 'To the library.' Rel=Answer. More complex sequences can include multiple relations and cross-references, reflecting shifts in topic or stance.

Discourcenotation overlaps with established theories such as Rhetorical Structure Theory (RST) and Segmented Discourse Representation Theory (SDRT) but emphasizes interactional dynamics and explicit cross-unit relations in dialogue. It aims to be compatible with existing corpora and transcription guidelines while adding a compact, machine-readable layer of discourse structure.

Because discourcenotation is not widely standardized, its use is mostly experimental. Proponents argue it can support corpus annotation, improve dialogue systems, and aid second-language learning by making discourse relations more explicit. Critics note that the approach can be complex and require extensive training to annotate reliably.

See also: RST, SDRT, Discourse Analysis, Conversation Analysis, Natural Language Processing.