‘Lose weight, save the NHS’: Discourses of obesity in press coverage of COVID-19

ABSTRACT This article examines the discourses that are used by the British press to represent obesity in its coverage of coronavirus disease 2019 (COVID-19). Obesity is understood to be a risk factor for COVID-19, with people with obesity being more likely to die from the virus. This study adopts a corpus-based approach to Critical Discourse Studies and utilises a novel approach to keyword analysis, based on comparing analysis corpora against two reference corpora in order to yield keywords that are, in this case, characteristic compared to general coverage of both COVID-19 and obesity. It is argued that the context of the pandemic has produced a range of discourses around obesity that are more stigmatising than usual in both the tabloids and the broadsheets, with people with obesity being construed, for example, in especially fatalistic terms and as being responsible for the problems facing the country’s healthcare system. At the same time, the pandemic context also seems to have given rise to some more positive changes to press coverage of obesity, with race-related health disparities receiving more focus than usual and the right-leaning press being more likely to critique the Conservative Government than it usually is, resulting in an arguably more balanced style of reporting.


Introduction
Coronavirus disease 2019 (henceforth COVID-19) is caused by severe acute respiratory syndrome coronavirus 2. In March 2020, the spread of COVID-19 was declared a global pandemic by the World Health Organisation (2020). People with obesity are at increased risk of serious illness and death from COVID-19. This paper examines the discourses used by the British press to represent obesity and people with it in the context of the COVID-19 pandemic. The study takes a corpus-based approach to Critical Discourse Studies, examining obesity-related discourses in two corpora: one containing articles about obesity and COVID-19 from UK broadsheets and the other from tabloids. Discourses are identified through keyword analysisa technique in corpus linguistics which highlights words that occur with a statistically marked frequency in one corpus compared to another The report also highlights possible interaction between weight-related comorbidities, such as type 2 diabetes and cardiovascular and respiratory diseases, which are linked to more severe cases of COVID-19, as well as socio-economic and demographic factors. The report acknowledged some limitations of the evidence, such as small and unrepresentative sample sizes. These limitations notwithstanding, the report states that 'the evidence consistently suggests that people with COVID-19 who are living with overweight or obesity, compared with those of a healthy weight, are at an increased risk of serious COVID-19 complications and death' (ibid.). In July 2020, the British Government responded to the link between obesity and severe cases of COVID-19 by publishing a policy paper, titled Tackling obesity: empowering adults and children to live healthier lives (Department of Health & Social Care, 2020). This describes the Government's 'new obesity strategy to get the nation fit and healthy, protect themselves against COVID-19 and protect the NHS [National Health Service]' (ibid.: online). Proposed measures include banning pre-watershed TV and online adverts for high-fat, -sugar and -salt food products, curbing supermarket promotions on 'unhealthy' products, displaying calorie information in restaurants and on alcohol products, and launching a public health campaign intended to encourage people to lose weight by getting active and eating 'better'.
Obesity has been a prominent topic in UK news coverage of the pandemic. Outside of the pandemic context, news representations of obesity have been widely studied (see Atanasova et al., 2012), with the most recent studies carried out as part of a wider programme of research within which the present article is situated (see: Baker at al., 2020;Brookes & Baker, 2021a, p. 2021b. This project takes a corpus-based approach to critically analysing the discourses in a 36-million-word (approx.) corpus of UK news articles about obesity published between 2008 and 2017. This research found that the press largely relies upon discourses of personal responsibility which frame obesity as a moral failing by individuals, who are accordingly responsibilised into eradicating their excess weight through modifications to their diet and by undertaking exercise. This discourse was particularly prominent in right-leaning newspapers, and especially tabloids, which were also more likely to stigmatise and shame people with obesity through euphemistic and 'humorous' language. We have argued such depictions to be likely to contribute to weight stigma (Puhl & Heuer, 2009), which can adversely impact the lives of people with obesity, causing them to be evaluated negatively by others, which can in turn lead to internalised shame (Obesity UK, 2020). The broadsheets (particularly those on the political left) were more likely to frame the causes of obesity in terms of wider socio-political factors and to attribute responsibility for it to institutions such as the Government, food manufacturers and supermarkets.
This article expands on this recent work by examining the discourse representations of obesity in press coverage of COVID-19. Understanding this is important, since such representations contribute towards the messages that the public receives on these public health issues. Indeed, experimental evidence indicates that media representations of obesity can lend support for different ways of assigning responsibility for it, as well as garnering support for different Government policies on the issue (Liu et al., 2019). Given that previous research has found that representations of obesity can be influenced by newspapers' formats (i.e. broadsheets vs. tabloids -discussed above), this study will distinguish newspapers according to these variables and compare the representational discourses that characterise the coverage of obesity and COVID-19 provided by each.

Methodology
The data analysed in this study comprises two purpose-built corpora which respectively contain British broadsheet and tabloid articles mentioning both COVID-19 and obesity. Articles were sourced from the online news repository, LexisNexis. All UK national newspapers (including their Sunday and online editions) were searched for articles containing at least one mention of covid*, coronavirus or corona virus and at least one mention of obese or obesity. The asterisk acts as a wildcard to include cases where covid is optionally followed by other characters (e.g. to capture instances of 'COVID' but also etc.). I used obese and obesity as search-terms as these reflect medical terminology, and so arguably constitute the most 'neutral' labels available for this health issue (terms such as fat have been found to have a negative evaluative prosody in the presssee Brookes and Baker, 2021b). This search yielded only a handful of results in January and February of 2020. I therefore collected articles meeting the above criteria published between 1st March and 31st July 2020 (the start-point of analysis). All national newspapers were considered, though the Morning Star, Daily Star and Metro were excluded as they returned a small number of results (fewer than one article per week on average). This left four newspapers in the broadsheet corpus (785 articles, 1,220,849 words) and four newspapers in the tabloid corpus (959 articles, 1,011,735 words). Results were grouped by 'moderate' similarity (an option within LexisNexis), which helped to reduce the number of duplicate results. Remaining duplicate articles were removed using the 'Corpus Checker' function of WordSmith Tools (Scott, 2020). (Table 1).
The data were analysed using a corpus-based approach to critical discourse studies. Corpus linguistics broadly refers to a set of methods, but also a field of research, which analyses linguistic patterns in large collections of naturally occurring language (i.e. corpora, singular corpus). The appeal of employing corpora in the study of discourse is that that the use of specialist analytical software that is characteristic of contemporary corpus linguistic research affords the (critical) discourse analyst the opportunity to carry out quantitative and qualitative analyses on larger and more representative datasets than would be feasible without computational assistance (Brookes & Harvey 2016).
In this article, I explore the discourses around obesity using the corpus linguistic technique of keywords. Keywords are words that occur with a significantly higher frequency in one corpus compared to another, with the corpus against which the analysis corpus is compared referred to as the reference corpus. When selecting a reference corpus, analysts usually want one that is at least similar in size to the corpus being analysed (if not larger) and which contains texts that are comparable, for instance in terms of their register and genre (unless, of course, these are variables being investigated). This latter consideration helps to ensure that the keywords produced are characteristic of the language use and discourses in the corpus, as opposed to arising due to other variables.
One approach I could have taken to obtaining keywords for my two corpora would involve comparing them against each other directly, with each corpus serving as the reference for the other (i.e. comparing my broadsheet corpus against my tabloid one, and vice versa). However, a disadvantage of this approach is that the resultant keywords would indicate differences between my two corpora while concealing their similarities. I therefore decided to adopt a different approach; I analysed the two corpora separately, focusing on words that were key when I compared each corpus against two purpose-built reference corpora. For each of my analysis corpora, I constructed two reference corpora: one representing general coverage of obesity and the other representing general coverage of COVID-19. This helped to ensure that the resultant keywords were characteristic of the articles in my data relative to general coverage of both obesity and COVID-19. Because I also wanted to match my reference corpora to my analysis corpora in terms of newspaper format (i.e. to compare broadsheet with broadsheet and tabloid with tabloid), this necessitated constructing four reference corpora in total (two for each analysis corpus; see Table 2).
The reference corpora each contain the same number of articles, with each providing a balanced representation of the newspapers and months covered. Articles were sampled using the same search-terms as the analysis corpora and were taken from the same newspapers (i.e. the broadsheet reference corpora contain articles from the Guardian, Independent, Telegraph and Times, while the tabloid reference corpora contain articles from the Express, Mail, Mirror and Sun). While the COVID-19 reference corpora were timematched to the analysis corpora (i.e. 1st March-31st July 2020), the obesity reference corpora contained articles published between 1st August and 31st December 2019 (i.e. the five months prior to the first UK case of . This ensured that the obesity reference corpora represented general, contemporary obesity representation but not in a COVID-19 news context. One-hundred articles were sampled from each newspaper for each month covered by the reference corpora. Where more than 100 articles were available, 100 were taken at random. The decision to sample 100 articles from each newspaper in each month ensured that the reference corpora were larger than my analysis corpora, while also providing a balanced representation of their constituent newspapers. Another important decision regarding the use of keywords concerns the choice of statistical measure(s). For this study, I combined two measures of keyness: log-likelihood and log ratio. Log-likelihood indicates the confidence level that keywords are in fact key and have not arisen due to a sampling error (Dunning, 1993). Log ratio is an effect size measure which indicates the strength of the observed difference in the relative frequencies of a given word between the analysis corpus and the reference corpus (Hardie, 2014). Combining Log-likelihood with log ratio thus helps to ensure that keywords are both reliable and based on sizeable frequency differences between the analysis and reference corpora. I focus here on keywords with a minimum log-likelihood score of 15.13, which indicates a confidence level of 99.99% that keywords are reliable. I also stipulated that keywords should receive a minimum log ratio score of 1, which indicates that keywords are at least twice as common in the analysis corpus relative to the reference corpus. To ensure that keywords were broadly characteristic of the texts in my corporaand not simply used densely by a small number of articles -I also stipulated that keywords should appear in at least 5% of the texts in the analysis corpus. Using the parameters described above, I used WordSmith Tools to compare my corpora against both of their corresponding reference corpora and filtered out keywords that were not present in both lists. For example, for the analysis of the broadsheet corpus, I generated two sets of keywords: one by comparing it against the broadsheet general COVID-19 reference corpus and another by comparing it against the broadsheet general obesity reference corpus. I then filtered out all words that were not present in both keyword lists. This left a set of words that were therefore key relative to the broadsheets' more general coverage of both COVID-19 and obesity. The same procedure was then carried out for the tabloid corpus using its corresponding reference corpora.
I analysed all keywords qualitatively, using concordance lines as a starting point but usually analysing whole texts, to identify discourses that were used to represent obesity. For this purpose, I adopted a broadly social constructionist view of discourse, following Burr (1995, p. 48) who views a discourse as 'a set of meanings, metaphors, representations, images, stories, statements and so on that in some way together produce a particular version of events […] surrounding any one object, event, person etc.'. On a practical level, I traced discourses around obesity through recurring lexical and grammatical choices surrounding the use of keywords, focusing in particular on those which relate to obesity and people with it. The analysis focuses on the majority discourse patterns identified through close analysis of 100 randomly selected uses of each keyword. This sample size was large enough to reveal a range of uses of each keyword, including indicating majority patterns, but small enough to keep the analysis manageable. On a theoretical level, my analysis is broadly situated within the Dialectal-Relational Approach to Critical Discourses Studies (Fairclough, 2015). This involved analysing the dialectal relationship between discourses and social practices, where social practices are viewed as being both constitutive of and constituted by discourse. My interpretation of the discourses in my corpora is thus concerned not just with how these relate to news text producers' immediate objectives, but also considers the broader ideological structures which both enable and are propagated by these discourses.

Analysis
Using the reference corpora and parameters described in the previous section, I generated a list of keywords for each corpus. This gave 22 keywords for the broadsheets (Table 3) and 27 keywords for the tabloids (Table 4). Ten keywords appeared in both lists (backgrounds, BAME, black, die, disproportionately, dying, ethnic, ethnicity, minority, underlying) and so can be considered to be characteristic of both the broadsheets' and tabloids' coverage of obesity in relation to COVID-19.
On the basis of analysis of each of the keywords in both lists, I identified a series of recurring representational discourses which characterised the coverage in either or both of the corpora. These can be grouped into four broad themes which are explored in the following subsections: (i) Obesity as increasing risk of death from COVID-19; (ii)  Ethnicity and race-based discrimination; (iii) Politicians and political processes; and (iv) Medicine and healthcare.

Obesity as increasing risk of death from COVID-19
The first set of keywords I explore constitute a discourse which represents obesity as something which increases people's risk of dying from COVID-19. Such fatalistic depictions of obesity were evident in uses of keywords in both the broadsheets and the tabloids, accounting for the majority patterns of the use of the keywords die and dying.
As well as being linked to obesity directly, heightened risk of COVID-19-related death was also attributed to obesity indirectly in cases where it was linked to diabetes which, in turn, was connected to obesity. In these contexts, diabetes and obesity are conceived as being underlying factors associated with heightened risk of dying from COVID-19. Note that these extracts and those used henceforth were selected on the basis that they were deemed to be representative of the particular discourses and patterns being discussed.
Patients with Type 2 diabetes are twice as likely to die from coronavirusand being obese increases the risk even further, research has revealed. (Mail, 13.06.20) People with type 1 diabetesthe autoimmune form of the diseaseare three-and-a-half times more likely to die if they catch Covid-19 than non-diabetics, while type 2 diabeticsthose with the form closely linked to being overweightare twice as likely to die as non-diabetics. (Guardian,20.05.20) In the broadsheets, the heightened risk of dying from COVID-19 that is caused by obesity is also conceptualised in terms of hospital admissions, which explains why admitted is a keyword for this corpus.
Obesity a deadly risk factor for hospitalised coronavirus patients, study finds; When patients admitted to hospital were obese, their risk of death rose by 37 per cent with heart disease and dementia also risk factors. (Telegraph,29.04.20) The keyness of admitted in the broadsheets (and not the tabloids) betrays another feature of the broadsheet representation which is also evident from this headline; these articles are frequently informed by studies and statistics. Citing studies and statistics helps to legitimate the discursive linking of obesity and COVID-19-related death by invoking the authority of expertise (van Leeuwen, 2008). This helps to account for the prominence of the keyword ONS (short for, 'Office for National Statistics') in the broadsheets. Given the connection between COVID-19 and mortality, it could be expected that die and dying would feature as keywords in the obesity-related COVID-19 articles when compared to a reference corpus of general obesity coverage. Yet it is notable that these deathrelated words were also key in my data compared to the general COVID-19 coverage too. The use of keywords denoting death and dying, and the more fatalistic representations of COVID-19 and obesity (and diabetes) that these signal, indicates that such fatalistic depictions are more pronounced in articles which report on COVID-19 and obesity together, compared against articles reporting on COVID-19 in a more general sense. Such fatalistic representations are found not just in the tabloids, which in general tend towards more fatalistic depictions of obesity (Brookes and Baker, 2021b) but can also be seen in the broadsheets. Similarly, the keyness of admitted in the broadsheets arises because these newspapers are more likely to discuss hospitalisations from COVID-19 when the people concerned also have obesity. It could be argued, then, that the pandemic has engendered more fatalistic representations of obesity from both the tabloids and the broadsheets.

Ethnicity and race-based discrimination
Both the tabloids and the broadsheets contain keywords denoting ethnicity. This is the most substantial theme in both corpora in terms of the number of keywords that can be linked to it, accounting for a third (9/27) of the keywords in the tabloids and over half (13/22) of those in the broadsheets. The majority of the keywords in both cases (n = 7) overlap: BAME, backgrounds, black, disproportionately, ethnic, ethnicity and minority. In addition, the tabloids contained the keywords asian and groups, while the keywords that were unique to the broadsheets include african and minorities, as well as terms which indicate a more explicit focus on racial inequalities: disparities, disproportionate, racial and racism.
The majority pattern in the use of these terms across both corpora is to list ethnicity, and being from a BAME background specifically, alongside obesity as a risk factor that makes people more likely to die from COVID-19 if they contract it.
The Guardian view on BAME death rates: inequality and injustice; Coronavirus is much more likely to claim the lives of black people than white. (Guardian, 07.05.20) Brits from BAME backgrounds ARE more likely to die of Covid-19, Public Health England review confirms. (Mail,02.06.20) The relatively higher frequency of these terms in both analysis corpora compared to general COVID-19 coverage is perhaps unexpected and could indicate that obesity is particularly likely to be mentioned in articles which address COVID-19 risk factors in general (of which ethnicity is one). However, there is a difference between the broadsheets and tabloids here, as the broadsheets also provide a more critical perspective, with both obesity and COVID-19 being framed as outcomes of broader health inequalities.
Then there is the concern regarding the disproportionate impact of the virus on local black and Asian communities.
[…] A separate report by Public Health England (PHE), published earlier this month, said such disparities may be explained by social and economic inequalities, differing risks at work and the prevalence of conditions such as obesity and diabetes among minority groups. (Independent,16.04.20) Racial disparities are regarded more critically in the broadsheets than in the tabloids, then, and this is particularly the case for broadsheets to the left of the political spectrum, which present such racial disparities as part of a wider system of discrimination within society. This is indicated in particular in uses of the broadsheet keywords disparities and racism.
Racism contributing to coronavirus deaths among ethnic minorities, leaked official report says; Government said to have previously not published report recommendations because of 'current global events'. (Independent,13.06.20) This trend is consistent with more general representations of obesity in the press, where the (particularly left-leaning) broadsheets are more likely than the tabloids to attribute the cause of obesity to forms of social inequality (Brookes and Baker, 2021b). Yet it is notable that words explicitly denoting forms of inequality, such as disparities and racism, are key in the broadsheets' coverage of obesity and COVID-19 together compared to when they report on either topic on its own. It seems to be the case, therefore, that the intersection of obesity and COVID-19 has engendered a more explicitly politicised explanation of both of these health issues relative to the broadsheets' more general coverage of them. While this trend was more characteristic of the left-leaning broadsheets, it can also be seen in right-leaning publications too.

Politicians and political processes
The next set of keywords I want to consider relate to politicians and political processes. While this theme is present in both the broadsheets and tabloids, the individual keywords involved are unique to each list and are suggestive of differences between the specific focus of the newspapers representing each format. The broadsheets exhibit a characteristic focus on political social actors, namely the Government (Downing), Prime Minister Boris Johnson (PM's) and Leader of the Opposition, Keir Starmer (Starmer). The picture offered here is largely one of unity, with Labour tending to be represented and quoted as approving of Prime Minister Boris Johnson's renewed focus on reversing the country's obesity rates.
Labour welcomes PM's 'conversion' on obesity after coronavirus scare; No 10 says Boris Johnson to lead health drive, blaming his intensive care stint on excess weight. (Guardian,17.05.20) On the other hand, the tabloids' political keywords indicate a focus on actions that are attributed to Johnson and his Government, particularly relating to the aforementioned anti-obesity policies. These policies are conceptualised as a drive and a strategy, words which are both key in the tabloids, as well as being a metaphorical 'war' or 'crusade'. In both the tabloids and the broadsheets, the strategy is framed as constituting a change in Johnson's and the Government's approach to obesity, and one which has resulted from his own experiences of COVID-19 and in particular his view that his condition was exacerbated by his weight.
Boris Johnson to launch anti-obesity drive after coronavirus scare; BORIS Johnson will launch an anti-obesity strategy after telling friends he thinks his weight was partly the reason he ended up in intensive care. (Express,16.05.20) The more detailed accounts of the measures to be introduced as part of the 'drive' provided by the tabloids also help to explain the keyness of the terms prescribe, cycling and gym in this corpus. While an ostensibly medical term, the keyword prescribe tends to be used to frame Johnson's anti-obesity measures, which include recommending that doctors prescribe exercise in the form of cycling and attending the gym. While the introduction of measures to reverse the country's obesity trends was generally welcomed across the press, newspapers also adopted critical stances towards particular aspects of the proposals, and in this respect the newspapers can also be distinguished in terms of format and political leaning. For example, one of the measures of the antiobesity 'drive' discussed in the press is higher taxation of products high in fat and sugar contentotherwise known as a 'sugar' or 'sin' tax. In 2016, the UK Government introduced higher taxation on the production and importation of drinks containing added sugar in the form of the Soft Drinks Industry Levy (HMRC, 2016). While the leftleaning press (and particularly the broadsheets) were supportive of this legislation, the initiative received more a hostile receptionfrom right-leaning newspapers (Brookes & Baker, 2021b), who criticised it for harming business interests and presumably opposed it because it runs counter to the traditional Conservative values of economic liberalisation and minimal taxation. At the time of writing, the Government has not (yet) implemented an extended form of the Soft Drinks Industry Levy. One reason for this could be that proposals of this measure received criticism from the right-leaning broadsheets, who warned Johnson against introducing such a tax for the reasons mentioned above.
PM's obesity crusade must not abandon Conservative principles; The shift towards statism bodes ill when the Treasury is abandoning fiscal prudence elsewhere. (Telegraph,17.05.20) Another area of criticism directed towards the Government comes from the tabloids, particularly in relation to the 'Eat Out to Help Out' scheme, indexed through uses of the tabloid keyword scheme, which gives discounts to encourage people to eat out at restaurants and fast-food outlets in an effort to bring economic benefit to those industries which have suffered financially due to enforced closure as a part of lockdown restrictions. This scheme is presented as hypocriticalframed as potentially increasing obesity rates and is thus juxtaposed against the aforementioned anti-obesity 'drive'. This criticism was a feature of both the left-and right-leaning tabloids, with the critical stance of the latter being somewhat uncharacteristic, given their otherwise supportive stance towards Johnson and the Conservative Party.
Anti-obesity campaigners say Boris Johnson's plan to ban junk food deals in anti-obesity drive 'defies belief' as it comes just DAYS after Rishi Sunak's 'Eat Out' scheme that lets the whole country gorge on unlimited cut-price meals. (Mail,10.07.20) The left-leaning broadsheets provided a specific kind of criticism of the Government's anti-obesity policy, indicated through uses of the keyword underlying. This keywordencountered earliertends to be used in both the broadsheets and the tabloids in reference to obesity and other factors underlying the development of obesity. However, in a small number of cases it features in criticisms of the new obesity policies for failing to address the underlying socio-political causes of obesity.
New UK obesity plan fails to address underlying problems; Despite some good news for campaigners, bad health's roots in poverty and poor food education are not addressed. (Guardian,27.07.20) Another alternative to Johnson's anti-obesity 'drive' emerges in uses of the broadsheet keyword, parks, with articles from the left-and right-leaning broadsheets arguing that the benefits of parks have been highlighted by the lockdown, granting people the opportunity to exercise and leave their homes, and that these should be invested in as a way to improve the nation's health and, in turn, to save money for the NHS and taxpayers.
By diverting millions into improving and promoting parks, you would save the health service billions, especially in treating obesity-related illnesses. I think we would all be happier about the future if we felt that the huge disruption caused by coronavirus brought a few benefits as well. Rediscovering the therapeutic effects of green space could be one of them. Parks bring tranquillity and natural beauty into our increasingly troubled urban lives. Let's find ways to make them flourish again. (Times,24.07.20) Overall, then, all sections of the press share the view that the link between obesity and COVID-19 warrants political action to reduce national obesity rates. However, there are differences in the kinds of actions that are advocated by the tabloids and broadsheets, and the discourses they draw on to support these positions. The tabloids support measures which encourage individual lifestyle change, drawing on a discourse of personal responsibility. It is for this reason that they are critical of the Government's 'Eat Out to Help Out' scheme, which is framed as encouraging individuals to eat unhealthily. However, the left-leaning broadsheets are critical of the personally responsibilising perspective espousded by the tabloids, and with this the Government, for failing to address social and structural causes of ill-health (including obesity and . This focus on the social and structural factors underlying obesity in the leftleaning broadsheets is also consistent with these newspapers' representations of BAME groups, seen earlier. The right-leaning broadsheets, meanwhile, are somewhere inbetween. They advocate the personal responsibility discourse extolled by the tabloids but also recognise the role that Government decisions (e.g. pertaining to urban planning) play in enabling or constraining the extent to which people can lower their risk of obesity.

Medicine and healthcare
The final set of keywords relate to the theme of medicine and healthcare and were all unique to the tabloids: fitter, hospitalised, ICU [Intensive Care Unit], intensive, NHS, oxygen, proven, supplement and supplements. NHS is the most frequent of these keywords. The NHS is a seemingly ever-topical issue for the British press and previous analysis has shown that obesity is often construed as a severe and avoidable strain on the NHS (Brookes & Baker, 2021b), including posing a 'risk' to it (Brookes & Baker, 2021a). It is telling, then, that the NHS is mentioned relatively more often in obesity-related COVID-19 tabloid articles compared to broader coverage of both COVID-19 and obesity in general. NHS is used in a range of ways in the tabloids. In relation to obesity, it tends to be mentioned either as a source of statistics, as the source of definitions of obesity, or within a discourse which represents the health service as being placed under increased strain by people with a dual diagnosis of obesity and COVID-19.
We must win obesity war for the NHS: Third of COVID-19 deaths linked. (Express,21.05.20) This discourse, of people with obesity placing extra strain on hospitals during the pandemic also manifested in the majority of uses of the keywords hospitalised, ICU and intensive. These keywords tended to occur in relatively straightforward depictions of people with obesity as accounting for a larger proportion of those being treated for COVID-19 in hospital wards and intensive care units, typically with recourse to published statistics.
And the UK has the highest levels of obesity in Western Europe, with research showing 75 per cent of Covid-19 patients in intensive care are overweight or obese. (The Sun,26.05.20) The tabloid keyword oxygen tended to feature in reports about oxygen shortages at hospitals throughout the UK as a result of the need for the gas to be used in the care of people with COVID-19. When we consider that Johnson's purported obesity has been widely reported in the press, including in quotes in which he described himself as being 'too fat', we might consider the potential for reports about oxygen shortages in hospitals to lead readers to conclude that people with obesity are to blame for such shortages.
Boris Johnson has said he struggles with his own weight as he urged people to take action to improve their health after he said he was "way overweight" when he was admitted to intensive care in April as he battled COVID-19 and was put on oxygen. The Prime Minister, who is set to announce the Government's obesity strategy, has been praised by Ms Whately for already losing weight during lockdown. (Express,27.07.20) The solution to the strain on the NHS that is put forward by the tabloids is therefore that people with obesity should lose weight in order to 'save', 'protect' and 'take pressure off' the NHS, including in quotes from Johnson and other members of the Government.
Lose weight, save the NHS: Boris Johnson reveals his fight to diet as government launches anti-obesity drive. (Mail,27.07.20) The PM, who has lost a "stone and a bit" in his own brush with the deadly disease, said: "Losing weight is hard but with some small changes we can all feel fitter and healthier. If we all do our bit, we can reduce our health risks and protect ourselves against coronavirus as well as taking pressure off the NHS." (Mirror,27.07.20) This also helps to account for the keyness of the adjective fitter in the tabloids, with the nation increasing its physical fitness being presented as something that will benefit the NHS and the economy.
Mr Hancock explained being overweight put enormous pressure on the NHS, and getting fitter could boost the nation's health and finances. (The Sun,27.07.20) Related to this, the tabloid keywords supplement and supplements tended to be construed as means for weight loss, including in passages which address readers directly with guidance on how they can lose weight. As the example below shows, the use of supplements was presented as being proven to be effective for weight loss (another tabloid keyword), as well as being presented as a means for individuals to lose weight gained during lockdown, which helps to construct its relevance and newsworthiness.
Best supplements for weight loss: The supplement proven to reduce fatty tissue; BEST supplements for weight loss: The likelihood of putting on weight is greater than ever right now but evidence suggests supplements can help to keep fatty tissue at bay while in lockdown. To that end, a fibre-based supplement has been shown to have a pronounced effect on weight loss. (Express,05.07.20)

Discussion
This article has presented a study of the discourses that are used by the UK broadsheets and tabloids to represent obesity in the context of coverage of COVID-19. Using the corpus linguistic technique of keywords, the analysis has focused on words (and their attendant discourses) that are characteristic of obesity coverage during the pandemic relative to general reportage of both obesity and COVID-19. The analysis has shown that, compared to more general coverage, both broadsheet and tabloid reporting of obesity in the COVID-19 context is more fatalistic, with people with obesity being particularly likely to be construed as dying, or at least as being at heightened risk of dying, from COVID-19. The keyness of lexis surrounding death in the corpora under analysis suggests that people with obesity are more likely than other people to be depicted as being at risk of dying from COVID-19, and that this focus on death and dying is also more pronounced than in more general obesity coverage. For the broadsheets, this represents a marked change in tone, with the pandemic ushering in a more pronounced focus on the connection between obesity and mortality. Such fatalistic discourses are more characteristic of the tabloids in general (Brookes & Baker, 2021a), so their marked frequency in the tabloid corpus assembled for this study indicates that this way of framing obesity has gained even more prominence in the tabloids during the pandemic compared to usual.
The connection between obesity, COVID-19 and death arguably facilitates another discourse observed in the analysis, particularly prominent in the tabloids, wherein people with obesity were constructed as a strain on an already overburdened NHS, including by taking up hospital beds and requiring oxygen therapy to the extent that there are shortages for the treatment of others. Accordingly, the solution to this burden put forward by the tabloids, and to a lesser extent the right-leaning broadsheets, is for people with obesity to lose weight, including through exercise and taking supplements, to 'save' the NHS. Both discourses can be linked to a kind of responsibilisation of people with obesity, not only for their own health but also for harming or, conversely, 'saving' the country's healthcare services. This is consistent with more general tabloid coverage of obesity, though the keyness of certain words indicating these discoursesincluding NHSagain suggests that this has become more pronounced in tabloid articles during the pandemic compared to usual.
In critical discourse studies, it can be useful to ask, 'who benefits from a discourse?'. On a commercial level, by drawing links between obesity and COVID-19, the press can help to maintain the newsworthiness of obesity in the context of what is, in the pandemic, a news story of global relevance. Meanwhile, the fatalistic and responsibilising discourses around obesity observed in the foregoing analysis could be interpreted as helping the news agencies which employ them to key into the news value of 'negativity' as part of the construction of this newsworthiness (Galtung & Ruge, 1965).
The discourse of personal responsibility, which blames people with obesity for problems with the NHS brought to the fore by the pandemic, is not only consistent with a prevailing neoliberal ideology in UK society, but also reflects the alignment of the tabloids and the right-leaning press as a whole with such individualistic values. In fact, so strong is this ideological commitment that it even motivated otherwise uncharacteristic criticism from these newspapers of Boris Johnson and his Government for contemplating expanding taxation on sugary food/drink and for implementing the 'Eat Out to Help Out' scheme whilst at the same time trying to encourage people to eat more healthily in order to lose weight and 'save the NHS'. The former was criticised for running up against the free market values of the political right, while the latter was presented as a counterproductive step in encouraging individuals to practice self-care and to take responsibility for their obesity risk.
In the main, though, most of the newspapers in the UK (and the data studied here) are generally sympathetic towards Johnson's Conservative Government (evidenced through their support of the Party during the 2019 General Election which preceded the startpoint of the data). Imploring people with obesity to lose weight to 'save the NHS' could thus also be viewed as a way for much of the press to articulate its support for Johnson by reproducing his political slogans. Yet the blaming of people with obesity for the strain felt by the NHS during the pandemic may also constitute a more insidious method through which this support is articulated. It could be argued that any problems facing the UK's healthcare system are much more likely to be the result of under-funding by successive (and latterly, Conservative) British Governments (see Baker et al., 2019 for a discussion), rather than arising simply due to the demands placed on it by the pandemic and the care requirements of people with obesity. Thus, blaming people with obesity for the problems faced by the NHS at this time, as the tabloids and some articles in the rightleaning broadsheets do, conveniently shifts focus away from the political decisions that have likely brought the NHS to this point, including a decade-long programme of austerity policies. The view of obesity as a preventable lifestyle disease, propagated by the types of responsibilising discourses described above, is also likely to be relevant here. It is difficult to imagine people with cancer, for example, being blamed for putting the NHS under strain in the way that people with obesity have been.
Whatever their precise motivation may be, discourses of personal responsibility can be viewed as problematic from a number of perspectives. A long-standing criticism of neoliberal political systems, and the responsibilising discourses to which they give rise, is that these often fail to recognise that obesity, like other so-called 'lifestyle' conditions, is not simply the culmination of individuals' lifestyle choices but, rather, is likely to result from a variety of factors (both individual and socio-political), over which individuals often have very little control. To offer a public figure as privileged and as powerful as the Prime Minister as a 'role model' to encourage readers to lose weight, as the tabloids and right-leaning broadsheets have done, risks overlooking precisely the role that factors like social privilege play in the development of obesity and other forms of ill-health, as well as individuals' capacity to remedy these.
When blame for purported problems in society is placed on individuals belonging to a particular group, this can result in the creation of stigma. In the case of obesity, this results in the creation of weight stigma. Stigmatising depictions of people with obesity in the British press are certainly not particular to the context of the pandemic (see, for example, Brookes & Baker, 2021b). However, I would argue that the way much of the press has represented obesity during the pandemic constitutes a 'ramping up' of such stigmatising discourse, as people with obesity have come to be blamed not only for their own health problems but have, in the pages of most British newspapers, shouldered responsibility for a failing healthcare system against the backdrop of the most severe public health crisis of modern times. The weight stigma that is likely to result from such blame-loading may engender further negative attitudes towards people with obesity and result in them experiencing internalised shame (Obesity UK, 2020). Yet I would also argue that the consequences of weight stigma may in fact be intensified by the context of the pandemic, which is widely reported to have had an adverse effect on the population's mental health. A survey of over 16,000 people living in the UK carried out by the mental health charity, Mind (2020) indicated that 65% of adults with a pre-existing mental health problem self-reported that these had worsened during lockdown. This figure was even higher for people aged 13-24 (75%). Meanwhile, 22% of respondents aged 13 or above without prior experience of mental health problems described their mental health during the pandemic as being 'poor' or 'very poor'. A recent report by Public Health England (2020b, p. 7) stated that '[s]tigma experienced by people living with obesity may delay interaction with health care and may also contribute to increased risk of severe complications arising from COVID-19'. In this context, then, we could question how beneficial such stigmatising discourses are for the mental health and wellbeing not only of people with obesity, but all of society.
The intensification of stigmatising discourses is not the only effect that the pandemic seems to have had on the press's coverage of obesity. Both the broadsheets and tabloids, but particularly the former, exhibited a heightened focus on the racial disparities in rates of, and deaths from, COVID-19. For the broadsheets in particular, obesity, like COVID-19, could be constructed as the result of race-related social disparities. Social explanations have been found to be most characteristic of the left-leaning broadsheets in general coverage of obesity (Baker et al., 2020), though the representation of race-related health inequalities as causing obesity has not been noted in previous studies of the British press. We could thus infer that for the broadsheets, and particularly those on the left of the political spectrum, the pandemic has at least provided an opportunity to report more, and more critically, on the role of race-related health inequalities in contributing to obesity and other health problems. In terms of raising audiences' critical awareness of the social determinants of ill-health, this could be considered a potentially effective counter-discourse to that which foregrounds personal responsibility. However, for the tabloids (and, to a lesser extent, the right-leaning broadsheets), this opportunity does not appear to have been seized in the same way, as race-related differences in the prevalence of obesity and death from COVID-19 were rarely if ever reported in relation to the wider structural inequalities of which they are likely a product.

Conclusion
The pandemic context has given rise to a range of obesity press discourses which could be viewed positively or negatively, depending on our perspective. I have argued some of these changes to have potentially negative consequences for people with obesityparticularly the more fatalistic, responsibilising and ultimately stigmatising portrayals that were characteristic of the tabloids. Such representations, it seems, have become more pronounced than usual during the pandemic, where the stakes have been raised. In this context, people with obesity are presented as being at even greater risk of death than usual and as purging healthcare resources at a time when the NHS was widely reported to be in crisis. Yet other changes, such as the greater reporting of government policy from all newspapers, and the broadsheets' heightened focus on race-related health disparities, could be viewed as something of a counter-discourse to the focus on personal responsibility.
In a sense, the ways that newspapers have reported on obesity during the pandemic seem to represent more exaggerated forms of their regular practices. On one end of the spectrum, the right-leaning tabloids are even more concerned with the risks that obesity poses to people's health and have widened the scope of responsibility such that people with obesity can be viewed as being responsible for the NHS's difficulty in handling the number of people infected with COVID-19. On the other end, the left-leaning broadsheets focalised their discussions of structural health inequalities around the issue of race in response to the heightened prevalence of COVID-19 among people from BAME backgrounds, as well as perhaps the global Black Lives Matter movement which was gathering momentum at this time and is mentioned in some of the articles. The right-leaning broadsheets, meanwhile, exhibited some change, for example by discussing the role of issues like health inequality and urban planning more than they usually would. At the same time, these newspapers maintained their firm commitment to traditional conservative values like individual autonomy and economic liberalism.
Following the analysis, I considered the possible commercial, ideological and political motivations underlying the discourses identified in the corpora, as well as their potential ramifications for the health and wellbeing of people with obesity and the public more generally. Given that news reporting in the UK has been found to contribute to stigma and internalised shame in people with obesity, the even more fatalising and responsibilising discourses employed by much of the press during the pandemic is, I have argued, only likely to exacerbate these effects on people with obesity, particularly given that the pandemic context has itself been shown to have a detrimental effect on mental health (Mind, 2020). The counter discourses provided mostly by the left-leaning broadsheets could offer something of a reprieve, though it is worth bearing in mind that these newspapers make up a small minority of the UK press landscape.
The articles examined in this study all represent obesity coverage in a unique news context, and it remains to be seen whether if discourses observed here are particular to this context or whether they have the potential to herald longer-term changes to the press's treatment of what is, in obesity, a persistently newsworthy topic. This is a question for future research, which could adopt a diachronic perspective to track changes to obesity representation within and beyond the timeframe of the pandemic. Another topic for future research pertains to the similarities and differences between newspapers' coverage of obesity and COVID-19 with respect to their political leanings, which I have not had space to analyse systematically in this article. Finally, images can function as powerful connotators of ideologies relating to health and illness (Harvey & Brookes, 2019). The corpus approach I took in this study did not facilitate the analysis of the images featured in the articles in my data. An interesting possibility for future research, then, would be to compare the types of images used in articles about obesity before the pandemic against those used during it, in order to ascertain the extent to which such images may contribute to the more stigmatising linguistic portrayals of obesity observed in the present study.
To conclude on a methodological note, in this study I generated keywords by comparing each of my corpora against two separate reference corpora, respectively representing general COVID-19 and obesity coverage. This usefully allowed me to hone my analytical gaze on those keywords and attendant discourses that are truly characteristic of COVID-19-related obesity coverage, rather than merely representing more general trends in the coverage of either condition. A similar approach could be utilised in future to examine texts which likewise represent an intersection of topics or aspects of identity, with the various keyword comparisons offering insight into different layers of distinctiveness.