Exploring discourse and ideology through corpora

Miguel Fuster-Márquez, José Santaemilia, Carmen Gregori-Signes and Paula Rodríguez-Abruñeiras

It is increasingly evident that corpus methodologies have already greatly contributed to discourse analysis and shown their potential to unveil political or ideological attitudes, values, or social inequalities. The implementation of corpus techniques, which are increasingly refined every day, allow us access to document social attitudes, to expose injustice or sexism and –hopefully – to fight against discrimination by obtaining results which are more generalisable. Until quite recently, the study of social issues, identity politics or ideological values has favoured qualitative analyses of small collections of texts. Today, however, more and more researchers are convinced that the use of corpus linguistics for social analysis constitutes a very powerful instrument of social inquiry, especially if used in combination with other (qualitative) approaches which require close reading and the interpretation of particularly salient textual features. Corpus-Assisted Discourse Studies (CADS) enjoys today great popularity and academic appeal, partly because of its ability to move back and forth, in a complementary way, between qualitative and quantitative techniques in order to generate new hypotheses and to test existing ones.

Tony McEnery, Helen Baker and Carmen Dayrell are in charge of the second chapter, “Analysing the impacts of 19th-century drought: A corpus-based study”, where they turn their attention to the past in order to access information gaps through textual data about the weather that could be useful to historians and meteorologists, thus proving the usefulness and convenience of carrying out textual analyses of large amounts of historical media texts (in this case, a large sample of 19th century British newspapers reporting on climate conditions in different parts of the country) by means of corpus methodologies. The rigorous examination of discourses can become an indispensible contribution of corpus linguistics to non-linguistic fields such as the one dealt with here in relation to weather conditions in earlier centuries which cannot be accounted for directly through sophisticated modern standard methods. Nevertheless, the authors deem most appropriate to make use of triangulation, hence they use corpus methods in concert with other techniques (concordance geo-parsing and close reading analyses) with a view to reconstructing, as faithfully as possible, media narratives of the droughts and water shortages in different regions of the UK.

Political Discourse, one of the favourite targets of corpus-based analysis, is illustrated in this volume by Salvador Enguix-Oliver and Beatriz Gallardo-Paúls’s study of recent political developments related to the extremist far-right Spanish party Vox during 2018 in their contribution “Coverage of the far-right in the Spanish written press: The case of Vox”. With this aim in mind, the authors have selected a number of widely read Spanish quality written newspapers (El País, El Mundo, La Vanguardia and ABC) and digital dailies (El Español, eldiario.es, Público and Infolibre). According to the authors, these newspaper items belong to their own sub-corpus PRODISNET-201, which is part of the larger PRODISNET (Discursive Processes on the Internet corpus). The hypothesis explored in this chapter is that far-right parties like the Spanish party Vox have been, and are still, receiving largely undeserved media coverage in Western democracies, and this media attitude enhances their public visibility which results in unprecedented electoral success. The authors also resort to the use of quantitative methods such as sentiment analysis (using the software Lingmotif, which reveals a basically negative appraisal) and complementing their study with close pragmatic analyses of all texts, since the authors find, for example, that political evaluation in the media is often expressed implicitly. In their conclusion they highlight that negative political evaluations are most often carried out through presuppositions and anomalous implicatures of manner and quality.

Gender issues –and, specifically, violence against women (VAW) – feature prominently in the growing research that makes combined use of corpus linguistics and other methodologies, as shown in Leanne Victoria Bartley’s chapter “ ‘Nobody is guilty in football. That’s the first thing to understand’: A corpus-assisted critical discourse analysis of the UK press coverage of the Ched Evans case.” Bartley focuses on judicial events which took place in 2012 in the UK, when the famous footballer Ched Evans was convicted of raping a young woman in a hotel room in Rhyl, North Wales, after a night out together. In 2016, Evans’ case was reviewed, his conviction overturned, and he also received a compensation of £800,000. Bartley explores the approaches of the British press (The Daily Mail, The Guardian, The Mirror, The Sun) during this period by means of a CADS approach, based on Martin and White’s (2005) appraisal theory slightly modified by Bednarek’s (2008) recommendations concerning the subcategory of affect. She makes use of UAM Corpus Tool (O’Donnell 2019), which allows users to employ this in-built paradigm for quantitative and qualitative analysis. Bartley investigates how both the perpetrator and his victim were represented in the British press at different stages of this controversial case. She finds that the media discourses changed over time – the view of the alleged perpetrator went from harsh criticism to a positive image. The focus on the victim, by contrast, went from understanding her inability to consent to sexual intercourse to criticising her inappropriate sexual activities.

Alfonso Sánchez-Moya also explores the gender issue of VAW in the chapter “How does violence-motivated online discourse differ from its non-violent counterpart? Insights from a CADS approach.” He makes use LIWC, a sentiment software tool to analyse the discourse of female survivors of Intimate Partner Violence (IPV) in a corpus of online forum messages by women who are undergoing such violence, containing a total of 120,000 words. Insights into this phenomenon with a discursive approach are, in the author’s words, scarce. Some of his conclusions are that, when compared to another neutral/non-violent corpus containing forum messages by women, the discourse in this IPV-related forum appears to be characterized by a narrative-oriented and personal style, focusing more on the here-and-now. Particularly, the sub-corpus dealing with abuse within the IPV corpus shows a severely pessimistic type of discourse. Also, the claims about IPV-related negative emotions and their discursive characterization are validated by this analysis.

Stefania M. Maci presents the chapter “The narrative of the anti-vax campaign on Twitter”, on medical knowledge as disseminated by digital media. Maci describes the discourse strategies employed by US anti-vaccine activists in a small Twitter corpus of 16,768 tweets (75,960 running words) gathered in October 2018 after some children died from measles in New York City, their aim being spreading fake news about vaccination. The corpus has been semantically annotated by USAS to identify key semantic domains and WMatrix has been used for the automatic processing of the most relevant meanings. Maci also carried out a sentiment analysis, employing Socialbearing.com as its software. The results obtained reveal an overall strong, articulate anti-vaccine discourse displayed by these activists of the anti-vaccine movement in their tweets. Maci claims that, although apparently grounded on scientific truths, these activists favour a conspiracy theory between the government and the pharmacological industry.

Readers should notice that authors in this volume have relied on data contained in either small or large corpora. It is convenient to bear in mind that, in the same way as there is no specifically recommended methodology attached to the analysis of (critical) discourses, there is no agreed criterion either about the size of corpora suitable in research. Perhaps ‘the bigger the better’ is a good guiding principle to be seen in many published papers, since large corpora are more likely to yield generalisable findings and insights; however, small targeted corpora, as shown here, are perfectly suitable for many kinds of corpus research into discourse.

Finally, corpus linguistics has been enriched by the incorporation of larger social, historial, ideological and ethical considerations, brought about by critical approaches to discourse. Certainly, this volume is not the first one to be published around these issues. However, it attempts to provide readers with updated research from leading international scholars in Discourse Analysis. All CADS researchers found in this volume are quite aware of the strength which come from applying a methodological synergy that combines different kinds of quantitative and/or qualitative research. A critical attitude towards the choice of methods and the author’s stance undoubtedly contribute to more fine-tuned analyses and to a more responsible and reflexive practice.

We believe this edited volume is an excellent example of this growing symbiosis between methods, types of corpora, disciplines, software programmes, discourse perspectives, and social and ethical concerns.