Advanced Corpus Solutions for Humanities Researchers

Advanced corpus solutions for humanities researchers James Wilson, Anthony Hartley, Serge Sharoff, Paul Stephenson Centre for Translation Studies University of Leeds, UK Abstract This paper describes the design and implementation of an interface to corpora in 12 languages, stemming from the analysis of the needs of a diverse group of users: lan- guage teachers and language students, (non-computational) linguists, researchers in history and translation studies. We identified a set of requirements shared across the disciplines, as well as more specific requirements from the targeted user groups. The interface is designed to handle large-scale corpora of 20-500 million words. Keywords: corpus interface, corpora, language teaching, history, translation studies 1 Introduction This paper showcases the Intelligent Tools for Creating and Analysing Electronic Text Corpora for Humanities Research (hereafter, IntelliText) project that is being conducted by the Centre for Translation Studies (CTS) at the University of Leeds, UK. It describes how we have developed and extended our existing Leeds-based electronic corpora and tools 1 to meet the research demands of academics in various areas of the humanities. After briefly outlining some general modifications to the interface and its documentation, we describe in detail how we have implemented the following functions: • searches using metadata and statistics for metadata; • automatic genre identification; • advanced definition of shallow patterns; • operations with frequency lists, including affix-based searches; • classification of concordance lines according to their level of difficulty and appropriateness for language learners. We have tested the tools with a range of corpora and languages, including representative web- derived corpora for Arabic, Chinese, English, French, German, Greek, Italian, Japanese, Pol- ish, Portuguese, Russian and Spanish (Sharoff, 2006a), as well more specific collections, such as newswire or business corpora. In this paper, we present examples from Chinese, Japanese and Russian. 2 Outline of the project IntelliText was conceived in order to allow humanities researchers, including those with little or no experience of working with electronic corpora, to make use of advanced methods of text collection and analysis. Rather than producing a new product we are developing our existing tools by enhancing the range of their functions and their usability. We are doing so by liaising with humanities researchers who represent several disciplines in order to cater for teaching and research needs across a spectrum of subjects. Our aim is to facilitate the use of corpora for a wide range of users, including students and non-specialists in corpus linguistics. Therefore, rather than determining on their behalf which new functions to add, we consultated corpus-active colleagues – IntelliText board members – from areas such as dialectology, language teaching, history and 1 http://corpus.leeds.ac.uk/it/ PACLIC 24 Proceedings 769 linguistics in order to ascertain their needs at first hand. Given the limited lifetime of the IntelliText project, we adopted the principle that all modifications and enhancements should be applicable to more than one corpus language offered at CTS and should benefit users from more than discipline. Much humanities research relies on or would benefit from analysis of electronic corpora. How- …

Content truncated to 500 words