The representativeness threshold for the CETA subcorpus of the Coruña Corpus

Abstract

The concept of representativeness is the main distinguishing characteristic of specialised corpora in comparison to other sets of texts. The Coruña Corpus of English Scientific Writing currently comprises four published subcorpora (astronomy, life sciences, history, and philosophy) plus three others under compilation (physics, chemistry and linguistics). In this paper we aim to assess the lexical density of the text samples in CETA, the Corpus of English Texts on Astronomy, by means of the ReCor tool, a posteriori. The study is motivated by the following question: does quantitative representativeness analysis using ReCor provide, in the form of a cross-check, further validation of previous research on the representativeness of CETA? Previous work (Crespo and Moskowich, 2010) has indicated that the CETA corpus is well designed and valid for the purposes for which it was intended. We will here suggest metrics to measure these findings. The most important contribution of this study is to offer quantitative data collection results using the ReCor tool, which allows data triangulation and consequently ensures overall data quality. Results show that data analysis with the ReCor tool supports previous findings, and thus we are able to verify that CETA is indeed representative of the language of its time and register.