Analysing Representations of Obesity in the Daily Mail via Corpus and Down-Sampling Methods

Media reporting of obesity has been criticised in academic research as alarmist. Some researchers describe how such reporting is perceived by obese people as portraying them as freaks and enemies of society who are rarely given a voice unless successfully losing weight, which the authors argue is a form of ‘synoptical’ social control. In reporting on obesity, researchers claim that newspapers can influence perceptions, having implications for public policy such as their diachronic study of U.S. newspapers from 1990 to 2007 indicated a shift from a deterministic view of obesity (e.g. genetic factors) toward one based on personal responsibility (e.g. diet and exercise). This study examines a corpus of articles from the British newspaper The Daily Mail about obesity (published between 2012 and 2016), collected from LexisNexis, with the aim of (i) identifying how language is used to represent obese people and (ii) comparing traditional methods of critical discourse analysis with a corpus-based approach. First, close readings were carried out on four samples of the corpus, using four sampling techniques. Ten articles from each of the sampling conditions was collected, considering phenomena commonly focussed on in critical discourse studies approaches to text analysis, including quotation patterns, narrative structure, and argumentation strategies as well as lexical choice, grammatical relationships, and metaphor. Second, collocation patterns were used with concordance analyses in order to identify salient and consistent ways that obese people are represented across the whole corpus. Having carried out the analyses, a meta-analysis compared the findings elicited by different techniques in order to identify the extent that they overlapped or gave dissonant results. I found that a combination of a corpus approach with one or more sampling methods produced a slightly wider range of complementary findings, especially when corpus techniques were triangulated with samples of articles containing highly frequent mentions of the topic under discussion. The corpus approach as used in this chapter was better at identifying representations around obese people but performed slightly worse than qualitative analyses of samples in terms of revealing a range of causes of obesity.