The OpenCourseWare Metadiscourse (OCWMD) Corpus

The OpenCourseWare Metadiscourse (OCWMD) Corpus Ghada Alharbi and Thomas Hain Department of Computer Science, University of Sheffield Regent Court, 211 Portobello, Sheffield, S1 4DP, United Kingdom galharbi1@sheffiled.ac.uk, t.hain@sheffield.ac.uk Abstract This study describes a new corpus of over 60,000 hand-annotated metadiscourse acts from 106 OpenCourseWare lectures, from two different disciplines: Physics and Economics. Metadiscourse is a set of linguistic expressions that signal different functions in the discourse. This type of language is hypothesised to be helpful in finding a structure in unstructured text, such as lectures discourse. A brief summary is provided about the annotation scheme and labelling procedures, inter-annotator reliability statistics, overall distributional statistics, a description of auxiliary data that will be distributed with the corpus, and information relating to how to obtain the data. The results provide a deeper understanding of lecture structure and confirm the reliable coding of metadiscursive acts in academic lectures across different disciplines. The next stage of our research will be to build a classification model to automate the tagging process, instead of manual annotation, which take time and efforts. This is in addition to the use of these tags as indicators of the higher level structure of lecture discourse. Keywords: metadiscourse, disciplines, OpenCourseWare lectures 1. Introduction Academic lectures offer rich opportunities for studying a variety of complex discourse phenomena that help in build- ing system to interpret the information in lectures dis- course. Lectures contain regions where speakers intro- duce some concepts, emphasise others, interact with stu- dents, and engage in other interesting discourse phenom- ena. Thus, it is critical to locate these regions in order for a system to interpret lecture discourse. Moreover, lectures of- ten involve strategies used to indicate these regions, which in turn reflect both topical and functional structure, as well as other high-level discourse dynamics. These strategies are known as metadiscourse (MD), which are linguistic ex- pressions that are often referred to as discourse about dis- course that has just occurred or is about to occur (Schiffrin, 1980; Crismore et al., 1993). Some examples of metadis- course expressions include the Introduction (“Today I want to talk; Now moving on to”), Conclusion (“To conclude”), or Previewing (“We’ll be coming to that”). Metadiscourse in spoken lectures poses interesting chal- lenges to descriptive and theoretical models of discourse, as well as to downstream applications research, including the summarising of a meeting according to its activities (Niekrasz, 2012). Most recently, this has involved build- ing presentation skills tools using Ted Talks (Correia et al., 2014). Previous work by Alharbi et al. (2015) considered only five metadiscourse tags of the selected schema for about 47 university lectures. However, the present study describes a new corpus of hand-annotated metadiscourse, consider- ing all the tags in the schema for roughly 106 lectures from different disciplines, namely Physics and Economics. The lectures were recorded at both Yale University and the Mas- sachusetts Institute of Technology; this constituted as part of the OpenCourseWare initiative. In this paper, a description of the OpenCourseWare Metadiscourse (OCWMD) is provided, in addition …

Content truncated to 500 words