Swiss-AL: A Multilingual Swiss Web Corpus for Applied Linguistics

Proceedings of the 12th Conference on Language Resources and Evaluation (LREC 2020), pages 4145–4151 Marseille, 11–16 May 2020 c⃝ European Language Resources Association (ELRA), licensed under CC-BY-NC 4145 Swiss-AL: A Multilingual Swiss Web Corpus for Applied Linguistics Julia Krasselt, Philipp Dreesen, Matthias Fluor, Cerstin Mahlow, Klaus Rothenh¨ausler, Maren Runte ZHAW Zurich University of Applied Sciences, School of Applied Linguistics Theaterstrasse 17, 8400 Winterthur {krss, dree, fluo, maho, rotk, runm}@zhaw.ch Abstract The Swiss Web Corpus for Applied Linguistics (Swiss-AL) is a multilingual (German, French, Italian) collection of texts from selected web sources. Unlike most other web corpora it is not intended for NLP purposes, but rather designed to support data-based and data-driven research on societal and political discourses in Switzerland. It currently contains 8 million texts (approx. 1.55 billion tokens), including news and specialist publications, governmental opinions, and parliamentary records, web sites of political parties, companies, and universities, statements from industry associations and NGOs, etc. A flexible processing pipeline using state-of-the-art components allows researchers in applied linguistics to create tailor-made subcorpora for studying discourse in a wide range of domains. So far, Swiss-AL has been used successfully in research on Swiss public discourses on energy and on antibiotic resistance. Keywords: web corpus, applied linguistics, annotation, discourse analysis 1. Introduction Applied Linguistics, such as discourse analysis, relies on corpora. Corpus-based methods, e.g., collocation or key- word analysis, have become standard tools of analysis (Bubenhofer, 2009; Baker, 2006; Gabrielatos and Baker, 2008). Until recently, there was no Swiss corpus that met the requirements of applied discourse analysis. In this pa- per, we introduce Swiss-AL, a multilingual Swiss web cor- pus for Applied Linguistics, that fills this research gap. With Swiss-AL we address four major challenges: a) multi- lingual discourse (with the major national Swiss languages German, French, and Italian); b) the representation of the Swiss federal structure in the selection of corpus sources; c) following the perspective of applied discourse analysis, it is necessary to maintain a flexible pipeline and corpus structure that can take into account not only mass media but also, for example, texts from political parties, selected business associations, and NGOs; d) in the context of the increasing range of machine and deep learning algorithms (e.g., topic modeling and neural word embeddings), it is important to have a linguistically annotated corpus at hand that allows for basic corpus linguistic methods as well as for the adaption of such innovative NLP methods. In the following sections, we first present the current state of the art regarding web corpora (Section 2.) and then introduce Swiss-AL and our corpus linguistic processing pipeline (Section 3.) in more detail. In Section 4., we present research based on Swiss-AL as a corpus linguis- tic resource. In Section 5., we describe future plans for extending the processing pipeline. 2. Related Work on Web Corpora Today, the web is used as a linguistic resource quite nat- urally. There are several very large and multilingual cor- pus collections which gather their texts from the web, e.g., WaCky Wide Web …

Content truncated to 500 words