RiQuA: A Corpus of Rich Quotation Annotation for English Literary Text

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 835–841 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 835 RiQuA: A Corpus of Rich Quotation Annotation for English Literary Text Sean Papay and Sebastian Padó University of Stuttgart, Institute for Natural Language Processing Pfaffenwaldring 5b, 70569 Stuttgart, Germany {sean.papay, sebastian.pado}@ims.uni-stuttgart.de Abstract We introduce RiQuA (RIch QUotation Annotations), a corpus that provides quotations, including their interpersonal structure (speakers and addressees) for English literary text. The corpus comprises 11 works of 19th-century literature that were manually doubly annotated for direct and indirect quotations. For each quotation, its span, speaker, addressee, and cue are identified (if present). This provides a rich view of dialogue structures not available from other available corpora. We detail the process of creating this dataset, discuss the annotation guidelines, and analyze the resulting corpus in terms of inter-annotator agreement and its prop- erties. RiQuA, along with its annotations guidelines and associated scripts, are publicly available for use, modification, and experimentation. Keywords: quotations, interpersonal structure, literary text 1. Introduction In literature, spoken interactions between characters are of- ten of central importance to the narrative. Reported speech, or quotations, provide readers a direct window into the world of the narrative, where the author’s own presence within the work becomes minimally intrusive (Page, 1988). Authors can use quoted speech for storytelling in order to describe complex verbal interactions between characters without ex- plicit exposition. In dialogue-heavy works, quoted speech can often exceed 50% of a novel’s text (Elson and McKeown, 2010). Therefore, quotations are an important structural component of literary texts. When analyzing the information contained in quotations, both content and context are of fundamental importance. A quotation conveys information about what was said, how it was said, who said it, and to whom. Of these data, only the first one is generally provided by a quotation’s content, while broader context is needed for the other three. As a concrete example, take the following sentence from Charles Dickens’ A Christmas Carol: [“What’s to-day?”]direct quotation [cried]Cue [Scrooge]Speaker, calling downward to [a boy in Sunday clothes]Addressee, who perhaps had loitered in to look about him. (1) The quote itself ([“What’s to-day?”]direct quotation) provides the content of the utterance, and the type of the quotation (di- rect speech, as opposed to indirect speech) tells us that this is a verbatim rendering of the utterance. However, the quota- tion tells us relatively little about the interpersonal structure. The reader requires knowledge of the full sentence to under- stand which character is speaking ([Scrooge]Speaker), whom he is addressing ([a boy in Sunday clothes]Addressee), and the manner of speech (he [cried]Cue), indicating anxiety. Often, even a full sentence is not enough, and the interpersonal structure must be reconstructed from the larger context. There exist a substantial number of studies which have shown that knowledge about quotations can be useful in tasks such as extracting social networks (Elson et al., 2010), modeling discourse structure (Redeker and Egg, 2006), for- mality …

Content truncated to 500 words