Visualising Linguistic Evolution in Academic Discourse

Proceedings of the EACL 2012 Joint Workshop of LINGVIS & UNCLH, pages 44–48, Avignon, France, April 23 - 24 2012. c⃝2012 Association for Computational Linguistics Visualising Linguistic Evolution in Academic Discourse Verena Lyding European Academy of Bolzano-Bozen verena.lyding@eurac.edu Ekaterina Lapshinova-Koltunski Saarland University e.lapshinova@mx.uni-saarland.de Stefania Degaetano-Ortlieb Saarland University s.degaetano@mx.uni-saarland.de Henrik Dittmann European Academy of Bolzano-Bozen henrik.dittmann@eurac.edu Christopher Culy The University of T¨ubingen christopher.culy@uni-tuebingen.de Abstract The present paper describes procedures to visualise diachronic language changes in academic discourse to support analysis. These changes are reflected in the distri- bution of different lexico-grammatical fea- tures according to register. Findings about register differences are relevant for both lin- guistic applications (e.g., discourse analysis and translation studies) and NLP tasks (no- tably automatic text classification). 1 Introduction The present paper describes procedures to visu- alise diachronic language changes in academic discourse with the aim to facilitate analysis and interpretation of complex data. Diachronic changes are reflected by linguistic features of reg- isters under analysis. Registers are patterns of lan- guage according to use in context, cf. (Halliday and Hasan, 1989). To analyse register change, we extract lexico- grammatical features from a diachronic corpus of academic English, and visualise our extraction re- sults with Structured Parallel Coordinates (SPC), a tool for the visualisation of structured multidi- mensional data, cf. (Culy et al., 2011). Our approach is based on the inspection and comparison of how different features change over time and registers. The major aim is to deter- mine and describe tendencies of features, which might become rarer, more frequent or cluster in new ways. The amount and complexity of the in- terrelated data, which is obtained for nine disci- plines in two time periods (see section 2) makes the analysis more difficult. Structured Parallel Coordinates provide a tool for the compact visual presentation of complex data. The visualisation of statistical values for different linguistic features laid out over time and register supports data analysis as tendencies be- come apparent. Furthermore, interactive features allow for taking different views on the data and focussing on interesting aspects. 2 Data to Analyse 2.1 Features and theoretical background When defining lexico-grammatical features, we refer to Systemic Functional Linguistics (SFL) and register theory, e.g., (Quirk, 1985), (Halliday and Hasan, 1989) and (Biber, 1995), which are concerned with linguistic variation according to contexts of use, typically distinguishing the three contextual variables of field, tenor and mode of discourse. Particular settings of these variables are associated with the co-occurrences of certain lexico-grammatical features, creating distinctive registers (e.g., the language of linguistics in aca- demic discourse). We also consider investiga- tions of recent language change, observed, e.g., by (Mair, 2006), who analyses changes in prefer- ences of lexico-grammatical selection in English in the 1960s vs. the 1990s. As a case study, we show an analysis of modal verbs (falling into the contextual variable of tenor), which we group according to (Biber, 1999) into three categories of meaning that rep- resent three features: obligation, permission and volition (see Table 1). 2.2 Resources The selected features …

Content truncated to 500 words