Fables-DTR: A Corpus of Fables Annotated for Discourse and Temporal Relations

Abstract

This paper presents Fables-DTR, a corpus of Aesop’s fables annotated for discourse and temporal relations, designed to explore how event sequencing and aspectual features and discourse relations interact. Building on the ISO 24617 Semantic Annotation Framework, integrating Part 1 (Time and Events) and Part 8 (Discourse Relations), the resource provides a unified representation of discourse structure and temporal and aspectual features. The corpus comprises 15 fables in English, automatically translated into European Portuguese and Polish (45 texts in total), with all translations manually validated by native linguists to preserve semantic and discourse features. Each fable is annotated in two layers: (i) for discourse relations, argument roles, and signals; (ii) for temporal relations, and event attributes, such as Tense, Aspect, Polarity. The resulting dataset provides relevant information about the association between discourse relations and their temporal and aspectual features. Fables-DTR contributes both a valuable resource for cross-linguistic and narrative discourse analysis and empirical evidence for integrating ISO standards in multilayer annotation. It also provides a foundation for computational applications in discourse parsing, event ordering, and implicit relation detection.

Resources

Details