This thesis investigates the language of risk reporting, more precisely how ten British newspapers use language to discursively construct four “risks to life”: terror attacks, earthquakes, road accidents, and heart attacks. The media do not allot space to these risks equally or in accordance with risk level (i.e., based on number of deaths and likelihood of occurrence). In addition to the media, public perception of risk in the second half of the 20th century was noteworthy for accepting serious everyday risks (i.e., those causing a high number of deaths) while rejecting new low-level technological risks (Zinn & Taylor-Gooby, 2006).
The approach of this thesis comprises theories and methodologies from Critical Discourse Studies (CDS) and corpus linguistics. The interdisciplinarity nature of CDS facilitates drawing upon theoretical frameworks from media and risk studies. News values, which play a role in news story selection, identify the newsworthiness of the risks, while risk characteristics, qualities found to influence public risk perception by the psychometric paradigm, enhance the interpretation of risk construction and how that might influence the public perception.
The data comprise approximately 14 million tokens of news articles distributed unevenly between four corpora but from identical sources and timeframes (January 2017 to January 2020). The analysis focuses on three aspects: risks to life (events), related social actors, and consequences (i.e., death). It also identifies news values and risk characteristics around these aspects. The findings highlight discursive strategies in risk reporting: dramatisation and naturalisation, (im)personalisation, blame / responsibility, and risk management. These contribute differently to the construction of the risks to life and can potentially be linked to how media amplify or attenuate risks in society, a consequence of media language use. I conclude with their manifestation in reporting the risks to life and how they might be linguistically and discursively realised.