Roughly one in a hundred people experience symptoms of schizophrenia during their lifetime, symptoms which include experiencing delusions and hallucinations, such as hearing voices (Johnstone and Frith, 2004). The frequency and intensity of symptoms is exacerbated by widespread negative attitudes and increases the likelihood that an individual will comply with command hallucinations (Harrison and Gill, 2010; Goldstone et al., 2012). In spite of this, the mass media continues to represent people with schizophrenia in an inaccurate and stereotypical way (Clement and Foster, 2008; Chopra and Doody, 2007). This fosters widespread misunderstandings and negative assumptions around the disorder. These misunderstandings gain traction because schizophrenia is widely misunderstood by the public at large (Jensen et al., 2016), and because people are unlikely to have first-hand experiences with people with schizophrenia. Despite the harmful nature of media representations, no study to date has seriously considered how misconceptions of schizophrenia and people with schizophrenia are mediated linguistically in the media. This is curious given that scholars in fields outside of Linguistics are increasingly recognising that the manner in which mental illness is represented plays an important role in reproducing stereotypical and prejudiced attitudes (Goulden et al., 2011, Kalucy et al., 2011). With these considerations in mind, this thesis draws on theories and methods from the field of Corpus Assisted Discourse Studies (CADS) to examine representations of people with schizophrenia in articles published in the British press between 2000 and 2015. This thesis uses a combination of corpus toolkits (Wordswmith 5.0, Sketch Engine) to examine repetitive lexicogrammatical patterns in articles published in the press that refer to schizophrenia and people with schizophrenia. It takes a particular interest in ‘non-obvious meaning’ (Partington, 2012:11), meanings that are only visible when examining how lexicogrammatical patterns converge around broader semantic and evaluative structures in large repositories of text. Do these patterns, working cumulatively over hundreds and thousands of texts suggest certain ways of understanding or viewing schizophrenia that would not be discernible to the naked eye?

The root of the problem revealed in the analysis was a tension between reporting
schizophrenia accurately and the press’ interest in reporting on schizophrenia in a way that is newsworthy in accordance with news values (Galtung and Ruge, 1965; Jewkes, 2015). In particular, there was a tendency to report on exceptional cases of people with schizophrenia (e.g. worst cases where people experience florid symptoms, people with schizophrenia who have succeeded creatively) that do not represent the majority of people diagnosed with the disorder. In the same vein, there was also a tendency for the press to repackage hard news as entertainment, for instance, by interdiscursively drawing on language and tropes associated with horror fiction. Salient language patterns converged around two main discourses: (1) that people with schizophrenia pose a risk to others, and (2) that people with schizophrenia who kill are intentional immoral agents. The analysis also identified a problematic metaphorical usage, which potentially reproduced the widespread misassumption that schizophrenia refers to a ‘split personality disorder’ (e.g. Jensen et al., 2015). I conclude by supporting a suggestion made in the academic literature (e.g. Ellison et al., 2018) that the diagnostic term ‘schizophrenia’ should be relabelled so that individuals diagnosed with the disorder do not carry the additional burden of negative stereotypes and misassumptions associated
with the label.