An Evaluation Dataset for Identifying Communicative Functions of Sentences in English Scholarly Papers

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 1712–1720 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 1712 An Evaluation Dataset for Identifying Communicative Functions of Sentences in English Scholarly Papers Kenichi Iwatsuki1, Florian Boudin2, Akiko Aizawa31 1The University of Tokyo, 7-3-1, Hongo, Bunkyo, 113-8656 Tokyo, Japan 2Universit´e de Nantes, 2 rue de la Houssini`ere, 44322 Nantes, France 3National Institute of Informatics, 2-1-2, Hitotsubashi, Chiyoda, 101-8430 Tokyo, Japan iwatsuki@nii.ac.jp, florian.boudin@univ-nantes.fr, aizawa@nii.ac.jp Abstract Formulaic expressions, such as ‘in this paper we propose’, are used by authors of scholarly papers to perform communicative functions; the communicative function of the present example is ‘stating the aim of the paper’. Collecting such expressions and pairing them with their communicative functions would be highly valuable for various tasks, particularly for writing assistance. However, such collection and paring in a principled and automated manner would require high-quality annotated data, which are not available. In this study, we address this shortcoming by creating a manually annotated dataset for detecting communicative functions in sentences. Starting from a seed list of labelled formulaic expressions, we retrieved new sentences from scholarly papers in the ACL Anthology and asked multiple human evaluators to label communicative functions. To show the usefulness of our dataset, we conducted a series of experiments that determined to what extent sentence representations acquired by recent models, such as word2vec and BERT, can be employed to detect communicative functions in sentences. Keywords: multi-word expression, formulaic expression, communicative function, rhetorical structure, sentence representation 1. Introduction Formulaic expressions are defined as ‘multi-word expres- sions that speakers remember and process as wholes, rather than constructing them “online” with each use’ (Durrant and Mathews-Aydınlı, 2011). They are frequently used in scientific papers because they convey specific commu- nicative functions in the rhetorical structure of papers, which indicate the author’s purpose or intention (Swales, 1990; Swales, 2004). For example, the formulaic expres- sion ‘in this paper we propose’ has the communicative function ‘stating the aim of the paper’. There is, how- ever, no consensus as to what constitutes the minimal text span that realises a communicative function. For exam- ple, to convey the function ‘describing the limitations of current research’, some may regard ‘beyond the scope’ as the minimal formulaic expression, while others may con- sider a larger span such as ‘is beyond the scope of this pa- per’. Here, we follow past research (Hirohata et al., 2008; Dayrell et al., 2012; Fiacco et al., 2019) and deal with this issue by regarding the whole sentence as the minimal unit of a communicative function. Formulaic expressions and their communicative functions have been investigated mainly in academic writing re- search to help people write papers more rapidly and flu- ently (Cortes, 2013; Mizumoto et al., 2017; Omidian et al., 2018). There even exist some computer systems for academic-writing assistance12 that rely on these commu- nicative functions to improve the user’s writing skills by suggesting commonly-used, alternative formulaic expres- sions. This is especially helpful for …

Content truncated to 500 words