An Annotated Corpus of Direct Speech

An Annotated Corpus of Direct Speech John Lee, Chak Yan Yeung Halliday Centre for Intelligent Applications of Language Studies Department of Linguistics and Translation City University of Hong Kong E-mail: jsylee@cityu.edu.hk, chak.yeung@my.cityu.edu.hk Abstract We propose a scheme for annotating direct speech in literary texts, based on the Text Encoding Initiative (TEI) and the coreference annotation guidelines from the Message Understanding Conference (MUC). The scheme encodes the speakers and listeners of utterances in a text, as well as the quotative verbs that reports the utterances. We measure inter-annotator agreement on this annotation task. We then present statistics on a manually annotated corpus that consists of books from the New Testament. Finally, we visualize the corpus as a conversational network. Keywords: direct speech, coreference, corpus annotation 1. Introduction There is increasing interest in analyzing social networks in literary texts. Such networks have been constructed for Alice in Wonderland (Agarwal et al., 2012), Les Misérables (Newman and Girvan, 2004), Biographies of Eminent Monks (Bingenheimer et al., 2011), a set of British novels (Elson et al., 2010), Hamlet (Moretti, 2011) and various Classical Greek tragedies (Rydberg-Cox, 2011), to name just a few examples. Social networks can visualize the protagonists, closely related characters and communities in the text. They also support the analysis of text structures and properties, such as the perspective holder (Agarwal et al., 2012) and the density of dialog interactions (Elson et al., 2010). While characters can socially interact in a variety of ways, their conversations — who talked to whom, and what they talked about — are key indicators of their relationships. This paper proposes a scheme for annotating direct speech in literary texts, based on the Text Encoding Initiative (TEI) and the coreference annotation guidelines from the Message Understanding Conference (MUC). The scheme encodes the speakers and listeners of utterances in a text, as well as the quotative verbs that reports the utterances. To the best of our knowledge, this is the first reported attempt of such an annotation task on a corpus of literary texts. The rest of the paper is organized as follows. The next section presents the proposed annotation scheme. Section 3 reports inter-annotator agreement. Section 4 presents statistics on a manually annotated corpus that consists of books from the New Testament. 2. Annotation Scheme We propose an XML annotation scheme that marks direct speech, the speakers and listeners, as well as the associated quotative verbs. The scheme aims to adopt the best practices from coreference annotation, and conforms to the standards of the Text Encoding Initiative (TEI) where possible. A summary of the elements and their attributes are provided in Table 1. An utterance is a span of spoken words, typically within a pair of quotation marks. We enclose utterances with the TEI element <said>. The first <said> element in Figure 1, for example, marks the sentence “We have come from … with us” as an utterance. Our annotation excludes written communications, but includes ambiguous cases where the words might have been delivered verbally or in …

Content truncated to 500 words