Modelling the Usage of Discourse Connectives as Rational Speech Acts

Proceedings of the 20th SIGNLL Conference on Computational Natural Language Learning (CoNLL), pages 302–313, Berlin, Germany, August 7-12, 2016. c⃝2016 Association for Computational Linguistics Modeling the Usage of Discourse Connectives as Rational Speech Acts Frances Yung Nara Institute of Science and Technology 8916-5 Takayama, Ikoma, Nara, 630-0192, Japan pikyufrances-y@is.naist.jp Kevin Duh John Hopkins University 810 Wyman Park Drive, Baltimore, MD 21211-2840, USA kevinduh@cs.jhu.edu Taku Komura University of Edinburgh 10 Crichton Street, Edinburgh, EH8 9AB, United Kingdom tkomura@inf.ed.ac.uk Yuji Matsumoto Nara Institute of Science and Technology 8916-5 Takayama, Ikoma, Nara, 630-0192, Japan matsu@is.naist.jp Abstract Discourse relations can either be implicit or explicitly expressed by markers, such as ’therefore’ and ’but’. How a speaker makes this choice is a question that is not well understood. We propose a psy- cholinguistic model that predicts whether a speaker will produce an explicit marker given the discourse relation s/he wishes to express. Based on the framework of the Rational Speech Acts model, we quantify the utility of producing a marker based on the information-theoretic measure of surprisal, the cost of production, and a bias to maintain uniform information den- sity throughout the utterance. Experiments based on the Penn Discourse Treebank show that our approach outperforms state- of-the-art approaches, while giving an ex- planatory account of the speaker’s choice. 1 Introduction Speakers or authors1 produce informative utter- ances, such that the listeners or readers can under- stand his/her message. Grice’s Maxim of Quantity states that human speakers communicate by being as informative as required, but no more (Grice, 1975). If a speaker always tries to provide as much information as possible, the resulting utter- ance could become excessively long and tedious. Such utterance is not only effort consuming for the speaker to produce, but also contains redundant in- formation that is not necessary for the listener. 1‘Speakers’ and ‘listeners’ are interchangeably used with ‘authors’ and ‘readers’ in this article In this work, we model how speakers plan the presentation of discourse structure optimally in terms of informativeness. Specifically, we propose a model that predicts whether the speaker will use or omit a discourse connective, given the sense of discourse relation s/he wants to convey. Discourse relations are relations between unit of texts (known as arguments) that make a document coherent. These relations can be marked in the sur- face text or inferred by the readers, as shown in the below examples. 1. It was a great movie, but I did not like it. 2. It was a great movie, therefore I liked it. 3. It was a great movie. I liked it. The word ‘but’ indicates a Concession relation in Example (1), and ‘therefore’ indicates a Result re- lation in Example (2). We call ‘but’ and ‘there- fore’ explicit discourse connectives (DCs). In Ex- ample (3), DCs are absent but a Result relation can be inferred. We say the DC is implicit in this case. Explicit DCs are highly informative cues to identify discourse relations (Pitler et al., 2008) while implicit DCs are more ambiguous. For …

Content truncated to 500 words