Annotating Similes in Literary Texts

Annotated corpora are invaluable resources for researchers in the humanities: on the one hand, for natural processing tasks, they can serve as standards against which results from new automatic methods can be measured; on the other hand, in corpus-based studies, they enable either to answer existing research questions or to explore original ones. In this respect, some annotation frameworks such as the Text Encoding Initiative (TEI) attempt to standardise annotation practices in order to facilitate data reuse and exchange. However, despite the crucial role played by figurative language in general and similes in particular in language, no consensus has been reached so far on how to comprehensively annotate them in literary texts. The present paper proposes a framework for annotating similes in literary texts which takes into consideration their semantic and syntactic characteristics as well as the challenges inherent to the automatic detection of similes.