9.2.1 Anxiety and Related Terms

Sketch Engine contains a function called Thesaurus, which is based on the theory of distributional semantics to identify synonyms. Thus, words that have similar meanings will share similar collocates. Entering anxiety into the Thesaurus produced a list of lexical items, in order of their proportion of shared collocates with anxiety. Sketch Engine has a large number of general reference corpora that can be useful to get an idea of how linguistic items typically behave. As such, the analysis began with an examination of anxiety in the English Web 2020 Corpus, which contains around a billion words of general English collected from the internet. In this corpus, the 10 words which had the most similar collocates to anxiety were depression, worry, fear, emotion, stress, anger, confusion, frustration, grief, and illness. Intuitively, this makes sense – anxiety is treated in similar ways as negative feelings or mental health conditions – and some of the words on this list (particularly worry and fear) are similar to anxiety and could be subjected to more detailed analyses themselves.

Having obtained an idea of the kind of words that are closest to anxiety in general English, it is now worth focussing in on how it is used in the corpus of forum posts. Here, the top-10 most similar items were symptom, thing, feeling, attack, I, fear, pain, thought, time, and problem. These words are somewhat different and perhaps less obvious candidates, compared to those elicited through the general corpus, and accordingly they indicate that people with anxiety might have different understandings of their condition. It is interesting, for example, that the word thing has similar collocates to anxiety in the forum, and concordance analyses found that about 10 per cent of uses of thing refer anaphorically to anxiety. For example,

The word thing is somewhat vague and general, in addition to being typical of informal ways of communicating. As we will see, there are other ways of conceptualising anxiety, which link to a range of discourses or ways of seeing the world. Therefore, in order to focus more on the collocates of anxiety itself, the Word Sketch tool in Sketch Engine was employed. This essentially identifies collocates of anxiety and then groups them into grammatical categories. Figure 9.1 shows a (partial) screenshot of a Word Sketch of the word anxiety.

Figure 9.1 Word Sketch of anxiety.

The Word Sketch contains several lists of words that collocate with anxiety, arranged in order of their collocational strength (with collocational scores hidden as the default), so the numbers in the table relate to the number of times a word collocates with the word anxiety. The first column shows nouns that are modified by the word anxiety, indicating noun phrases like anxiety disorder, anxiety attack, and anxiety symptoms. The second column shows words (usually adjectives) which modify anxiety (e.g., severe anxiety). The third column shows adjectives which occur after ‘anxiety is’. Due to mis-tagging, there are sometimes cases that appear in the wrong column (e.g., the word most should not be in the adjective predicates list). Finally, the fourth and fifth columns show verbs which collocate with anxiety, in the object or subject position, respectively. The fourth column indicates actions which are done to or acted upon anxiety (e.g., makes my anxiety), whereas the fifth column shows actions where anxiety is the doer (e.g., anxiety takes). The Word Sketch also contains additional columns under these ones (not shown in the figure) which denote further grammatical relationships such as ‘anxiety + and/or’ or ‘pronominal possessors of anxiety’.

This method of grouping collocates into grammatical categories can be helpful for analysts, as it enables them to spot words with similar meanings or functions more easily than a simple list of collocates. In order to identify the various ways that anxiety is represented, it is still necessary for humans to spend time analysing Word Sketch, exploring hypotheses about the ways that collocates position anxiety, through the examination of relevant concordance lines.

For example, in the third column, we can see a number of similar kinds of adjectives that are used to modify anxiety: bad, horrible, awful, hard, severe, terrible. Taken together, these words would suggest a discourse prosody (Stubbs, Reference Stubbs2001) which represents anxiety extremely negatively. Words in other columns also contribute towards this negative depiction (e.g., severe and bad also appear in the second column), while words like sufferers and problems in the first column contribute towards the same kind of picture. The word attack, in the first column, could also be interpreted as contributing to a negative representation of anxiety.

However, another set of words in the first column seem to represent anxiety somewhat differently: disorder, symptoms, meds, medication appear to function in a way which represents anxiety as a medical condition. It is most likely the case that the two representations (anxiety as bad, anxiety as a medical condition) are not incompatible. It is not surprising to see a medical condition also described as negative. But we should still bear in mind that these are separate (if related) representations, and not everything that is bad is a medical condition; some medical conditions can be seen as positive (e.g., genetic resistance to certain diseases, neurodiversity, or high pain tolerance).

9.2.2 Identifying Representations around Anxiety

The task of the researcher, then, is to consider the collocates in the Word Sketch and try to group them into ways that contribute towards discrete representations. Some words will be easier to categorise than others; some words might potentially contribute to more than one representation, depending on the context that they are used in; and for some words it might not be possible to identify a clear representation. The process of identifying categories and the words that go in them is therefore subjective, and it is unlikely that there will be a perfect way of doing it. It is important to try to apply a set of consistent and transparent guidelines when carrying out this kind of categorisation, and it is a good idea to work collaboratively, to help resolve difficult or ambiguous cases. While it is possible to categorise some words almost immediately (e.g., bad, horrible, and awful are fairly unambiguously negative), for other words, a detailed concordance analysis would be required. To be thorough, it is sensible to carry out concordance scans of all the words in the Word Sketch, just to confirm that a word is functioning in its expected way and something strange is not happening with it. Bear in mind, for example, that the Word Sketch shows all words in lowercase, but they might actually be realised with upper-case letters in the corpus. For instance, a word like bad could actually be an acronym (e.g., BAD can stand for a therapeutic approach known as behavioural activation for depression). In addition, negators may switch a collocate’s meaning to its opposite – so in some cases people may write ‘I don’t have bad anxiety’.

As a result of grouping collocates together, the researchers identified four pairs of representations around anxiety, each pair of representations being related to one another. These pairs were (1) medicalising versus non-medicalising, (2) catastrophising versus minimising, (3) anthropomorphising versus abstracting, and (4) owning versus distancing. We will take the third pair as an example, in order to illustrate how these representations were realised through language.

Table 9.1 shows a range of ways that anxiety is metaphorically represented through language as a living being, often a human being but sometimes as an animal or a supernatural entity. This kind of representation casts anxiety as having agency, carrying out actions and having its own thoughts and goals.

Table 9.1Anthropomorphising representation of anxiety

Structure	Examples
anxiety NOUN	bully (87), monster (36), demon (27), beast (13)
anxiety is a NOUN	beast (31), devil (30), bitch (22), culprit (17), enemy (17), liar (16), bully (15), monster (13), trickster (9), demon (9), tiger (8), fraud (9)
anxiety is not a NOUN	friend (10)
TITLE anxiety	Mr (34), Miss (1), Mrs (1)
anxiety VERB	cause (1,710), make (1,246), affect (273), try (249), hit (213), give (179), stop (114), create (109), bring (97), control (94), work (85), want (83), run (67), build (61), put (53), drive (45), like (48), let (45), produce (42), rule (43), provoke (40), need (46), love (40), lead (39), change (35), effect (29), mean (25), decide (23), send (21), prevent (21), rear (58), follow (36), take (641), suck (196), ruin (102), attack (91), feed (79), throw (47), hold (33), cripple (28), wake (26), rob (24), push (22), destroy (21), play (363), tell (167), think (120), act (72), talk (71), know (64), trick (46), mess (39), say (54), scare (29), bother (25), wait (24), exaggerate (23), convince (21), win (61), wear (29), beat (28), overwhelm (25), thrive (22)
VERB + anxiety	fight (316), beat (166), battle (80), conquer (60), tackle (51), combat (48), attack (32), challenge (24)

In some cases anxiety was personified, through the use of prenominal titles like Mr, which occurs 34 times as a collocate of anxiety.

A more common way that anxiety is anthropomorphised is by referring to it as a fantasy entity (beast, devil, monster, demon) or a malign human (bitch, culprit, enemy, liar, bully, trickster, fraud).

Table 9.1 also shows range of verb actions caried out by anxiety (verbs in their base form refer to all forms (e.g., try refers to try, tries, tried, and trying).

Anxiety is also described as having conscious thoughts and desires, deciding, wanting, and loving things. Although love appears to be a positive word, it is used to describe anxiety as desiring negative outcomes for the person experiencing it.

Anxiety is similarly described as playing tricks, scaring, and exaggerating.

In addition, anxiety is assigned as the experiencer of the negative thought processes, as opposed to the person who experiences anxiety.

A less common set of verbs involve descriptions of anxiety as beating the poster. These verbs often occur with negativisers (such as no, not, or never), where posters exhort one another or themselves not to let anxiety beat them.

Other verbs in this representation describe anxiety as moving (creep, follow).

Finally, there is a set of verbs which position anxiety as the patient of an action. Posters talk about challenging, combatting, battling, and beating anxiety. These kinds of verbs cast resolution of anxiety in terms of beating an opponent. In the following excerpt, the poster acknowledges the metaphorical nature of the representation by putting the word weapons in quote marks.

Viewing anxiety as a living being can be seen as management strategy for some posters. Chen, Chen, and Yang (Reference Chen, Chen and Yang2019) have described a study where individuals who were instructed to anthropomorphise sadness or happiness reported less experience of that emotion afterwards. They argue that the reduction of emotion occurs because anthropomorphic thinking increases the perceived distance between the self and the emotion, which results in a sense of detachment.

At other times, posters cast anxiety as an abstract state or entity (e.g., something which has no concrete state), as shown by the collocates in Table 9.2.

Table 9.2Anxiety as an abstract entity

Structure	Examples
anxiety NOUN	disorder (2,913), issue (757), problem (398), thing (195), state (153), condition (56), stuff (40), side (33), part (33), journey (27), diagnosis (28), illness (23), experience (22), situation (21), bout (15), story (15), crap (14), struggle (14), relapse (13), shit (11), cycle (131), loop (38), spiral (25), trap (10)
anxiety is (a) NOUN	illness (76), problem (46), condition (32), issue (32), paradox (25), disorder (25), disease (18), habit (14), pain (28), hell (22), nightmare (19), game (15), circle (12), battle (10), trigger (11), bluff (10), trick (8), cycle (8), feeling (71), fear (71), thought (21), stress (13), reaction (12), emotion (12), response (10), thing (225), something (81), part (64), way (35), step (19), state (15), form (10)
anxiety VERB	kick (246), flare (66), increase (55), worsen (50), strike (33), kill (33), rise (29), grow (28), drain (28), heighten (27), reduce (24), decrease (22)
VERB anxiety	experience (536), feel (817), spike (23), lessen (72), alleviate (32), decrease (24)

Rather than being a living or human entity, the words in this table represent anxiety through a range of different perspectives. For example, some posters metaphorically represent anxiety as an experience, journey, or story, which frames it as part of a person’s life narrative.

A less frequent category of words describe anxiety in terms of a negative and repetitive experience, using terms like cycle, loop, and spiral, a set of characterisations which are linked to acknowledging that negative representations can result in increased anxiety.

In this category, words which involve increasing anxiety are also included, like heighten, worsen, raise, escalate, add, exacerbate, and spike. Again, these verbs are often used in explanatory contexts.

It can be useful to obtain a more general sense of a word’s overall meaning by consulting a larger reference corpus. For example, escalate is used in the English Web 2020 Corpus to refer to abstract phenomena like tension, violence, conflict, war, crisis, confrontation, and dispute – it clearly has a semantic prosody for negative things, as well as abstract concepts.

Subside also has a semantic prosody for abstract entities. In the English Web 2020 Corpus, things that subside include laughter, pain, swelling, anger, fever, storm, flood, and fighting.

Additionally, anxiety is represented as abstract negative phenomena – for example, as a repetitive process (circle, cycle), a place (hell), a bad dream (nightmare), or a contest (battle, game).

Coming somewhere between the abstracting and anthropomorphising representations of anxiety is a much smaller third set which frames anxiety as a non-living object. For example, there are verb metaphors like fix and fuel which cast anxiety as something akin to a machine.

Finally, there are another set of verbs, often denoting physical violence, which characterise anxiety even more negatively, as an entity which physically abuses the sufferer. It is difficult to categorise these verbs as referring to an abstract, living, or non-living entity, as they tend to occur in general language use in a wide range of contexts. Some of these verbs are used in metaphorical ways to describe anxiety as appearing or worsening (e.g., kick in, flare, strike). These verbs can also index natural phenomena. For example, in the English Web 2020 Corpus, strike tends to be associated with lightning or earthquakes, although there are also references to human-made objects (bullet, missile, car) or abstract concepts (tragedy, disaster) striking.

It is important to note that a single lexical item can invoke more than one kind of representation. For example, referring to anxiety as a disorder is a way of representing it as both an abstract concept and a medical condition, while calling anxiety a nightmare will represent it as an abstract concept and an extremely negative phenomena (a form of catastrophisation).

Identifying these kinds of representations is one stage of the analysis. We might want to consider the kinds of contexts that the different representations occur in. For example, the representations which describe anxiety as a living entity often tend to be used when posters provide one another with emotional support or advice on coping, while cases of catastrophisation of anxiety in the corpus tend to occur when people are seeking help. We may also want to consider which kinds of posters use certain types of representations (in terms of, say, demographic characteristics or the length of time they have spent on the forum). And we could also consider the ways that other posters respond to these representations: do they take them up themselves, reject them, or explicitly talk about their effectiveness in helping them resolve or manage their anxiety?

For medical practitioners, it could be argued that it is useful to have awareness of the ways that people use language to refer to anxiety. This could help practitioners reflect on the ways their clients represent anxiety and whether such representations are likely to help or hinder them. For example, there is some evidence that catastrophising is a positive predictor of anxiety (Chan et al., Reference Chan, Chan and Kwok2015). Practitioners could thus support clients in critically considering a range of possible representations of anxiety, helping them to shift their thinking about anxiety towards more constructive means. It should be borne in mind, however, that people are likely to respond differently to the same representation, so there will be no single ideal way of talking about anxiety that works for everyone. Instead, the analysis from the corpus of forum posts provides information about a range of possible representations and encourages awareness relating to the way that those representations impact our experience of health conditions. This sense of gaining a better understanding of the possible set of representations around illness, and their effects, continues in the following section.

9.3.1 Creating a Cancer-Related Corpus and Identifying Metaphors

The MELC team collected a 1.5-million-word corpus consisting of interviews with and online writing by members of three stakeholder groups in cancer care: patients with cancer, unpaid family carers, and healthcare professionals. An overview of the MELC corpus is provided in Table 9.3 (Semino et al., Reference Semino, Demjén, Hardie, Rayson and Payne2018: 50).

Table 9.3Word counts in the MELC corpus

	Patients	Unpaid family carers	Healthcare professionals	Totals
Semi-structured interviews	100,859	81,564	89,943	272,366
Contributions to online forums	500,134	500,256	253,168	1,253,558
Totals	600,993	581,820	343,111	1,525,924

More specifically, the interviews involved

• 29 patients who had been diagnosed with advanced cancer (Payne et al., Reference Payne, Chapman, Froggatt, Gott and Chung2008);

• 17 unpaid carers who looked after a family member with a diagnosis of advanced cancer (Payne et al., Reference Payne, Ingleton, Nolan and O’Brien2009); and

• 16 senior healthcare professionals working in hospice or palliative care.

The online writing was produced in the period from 2007 to 2012 by

• 56 patients and 56 family carers writing on one particular UK-based online forum dedicated to cancer; and

• 307 healthcare professionals, writing on online forums or blogs.

The MELC corpus was too large to be analysed manually for the use of metaphor. On the other hand, no automated system for metaphor identification was deemed appropriate for the purposes of the project (e.g., Dunn, Reference Dunn and Gelbukh2013). The team thus adopted a combination of manual and computational analysis. They created a sample corpus consisting of approximately 90,000 words (i.e., approximately 15,000 words from each of the six sections of the corpus outlined in Table 9.3) and employed existing metaphor identification procedures (Pragglejaz Group, 2007; Steen et al., Reference Steen, Dorst, Herrmann, Kaal, Krennmayr and Pasma2010) to manually code this corpus for all instances of linguistic metaphors relevant to the experience of cancer. The linguistic metaphors were also tagged for the semantic domains that corresponded to their literal meanings – for example, Violence for ‘battle’, Journey for ‘path’ and Sports for ‘marathon’. A tailor-made database was then employed to match the expressions included under each semantic grouping to the semantic domains in the USAS semantic tagger, which is implemented in the online corpus analysis tool Wmatrix (Rayson, Reference Rayson2008), as explained in Chapter 7.

More specifically, during the manual analysis of the sample corpus, the team classified as Violence metaphors

The linguistic metaphors that fit this definition were then matched with the corresponding USAS semantic tags in Wmatrix by means of the database mentioned earlier. This revealed that the following USAS semantic domains included expressions that could be used metaphorically to describe the experience of cancer in terms of various kinds of violent encounters (Semino et al., Reference Semino, Demjén, Hardie, Rayson and Payne2018: 100):

• G3 (Warfare): for example, fight as a verb; battle

• E3– (Violent/angry): for example, hit; attack

• S8+ (Helping): for example, defend; protect

• S8– (Hindering): for example, fight as a noun

• X8+ (Trying hard): for example, struggle

• A1.1.1 (General actions, making): for example, blast and confront

• A1.1.2 (Damaging and destroying): for example, destroy; shatter

The team then analysed all concordance lines for each of these semantic domains in the complete corpus and identified all instances where the relevant lexical item was used metaphorically to capture some aspect of the experience of cancer. The use of semantic domain concordances meant that it was possible to identify metaphorical expressions that did not occur in the sample corpus. This does not of course mean that all Violence metaphors in the analysis would have been captured. However, as described in the following sections, a substantial number of instances were identified, and the fact that the same approach was adopted for all sections of the corpus also means that comparisons were possible.

9.3.2 Violence Metaphors and Different Stakeholder Groups

In this section we report the most relevant findings with regard to Violence metaphors. We begin with quantitative findings and then move on to qualitative observations.

The MELC team found that Violence and Journey metaphors were the most frequently used types of metaphors in all sections of the corpus (Semino et al., Reference Semino, Demjén, Hardie, Rayson and Payne2018: 84) and that Violence metaphors were the most frequently used by patients in particular.

Table 9.4 and Figure 9.2 (from Semino et al., Reference Semino, Demjén, Hardie, Rayson and Payne2018: 101–2) provide raw and normalised frequencies for Violence metaphors in the three groups of people and two genres included in the MELC corpus.

Figure 9.2 Frequencies of Violence metaphors in the corpus per 1,000 words.

Figure 9.2Long description

The y-axis denotes frequency points from 0 to 2.25 in increments of 0.45, and the x-axis denotes the three groups: Patients, Carers, and Professionals. It reads as follows. Patients: 0.71 for Interviews and 1.8 for Online. Crers: 0/98 for Interviews and 1.61 for Online. 0.81 for Interviews and 1.33 for Online.

Violence metaphors are used by members of all three groups in both genres. However, in the online data, patients use Violence metaphors more frequently than family carers and healthcare professionals, with 1.8 instances per 1,000 words versus 1.61 and 1.33, respectively, for carers and professionals. Semino and colleagues (Reference Semino, Demjén, Hardie, Rayson and Payne2018: 102) report that this difference across the three groups is statistically significant (using a log likelihood test for uniformity: p < 0.05, 2 d.f., log likelihood = 23.22).

The MELC team considered all instances of Violence metaphors in order to establish how they are used in context. This led to two main findings. First, there is no single Violence or War metaphor where the disease is an enemy that needs to fought by the patient or by healthcare professionals, as implicitly suggested by previous literature. On the contrary, patients in particular use Violence metaphors to talk about different aspects of their experience of illness, including the relationship with cancer, the effects of treatment, and the patients’ relationship with the healthcare system. Second, while there is evidence in the MELC corpus of the potentially harmful consequences of Violence metaphors, there is also evidence that they can be meaningful and motivating for some patients, at least some of the time. The MELC team expressed this contrast in terms of the concept of ‘(dis)empowerment’, defined as ‘the process through which linguistic choices reflect, facilitate and/or undermine different kinds and degrees of agency, validation, evaluation and control, with implications for identities, emotions and relationships’ (Semino et al., Reference Semino, Demjén, Hardie, Rayson and Payne2018: 7).

Whether a Violence metaphor or, in fact, any metaphor is empowering or disempowering depends on who uses it and how in a specific context (Semino et al., Reference Semino, Demjén and Hampe2017).

In the next section we consider the different aspects of the cancer experience that patients use Violence metaphors for in the MELC data, with special emphasis on the contrast between empowering and disempowering uses (Semino et al., Reference Semino, Demjén, Hardie, Rayson and Payne2018: 105ff.).

9.3.3 Violence Metaphors and Different Aspects of Patients’ Experiences with Cancer

Violence metaphors are used by some patients in the MELC corpus to portray their relationship with cancer as antagonistic and competitive, and to express their determination to recover. These are the kinds of Violence metaphors that had been the focus of previous studies:

In both examples, the patient places themselves in an empowered position; they present themselves as active and focussed, and as hopeful about their outlook.

The potential for Violence metaphors to be used in empowering ways is particularly evident when patients describe themselves as fighters:

This metaphorical use of fighter does not so much evoke an oppositional scenario with the illness as the enemy but rather presents the person as optimistic and ready to do everything they can to get better (cf. determination in the provided extract).

Semino and colleagues (Reference Semino, Demjén, Hardie, Rayson and Payne2018) also notice that some empowering uses of Violence metaphors involve humour, as mentioned in Chapter 5:

Here the cancer is menacingly described as a ‘Demon’, but the use of ‘wahoola’ as part of a Violence metaphor involving physical struggle makes light of what is otherwise a serious situation. The cancer is personified as the butt of the joke, thereby placing the patient in the empowered position of making fun of the illness. The MELC team showed more broadly how this kind of humour was employed by patients in the online data to release tension, demystify the illness, and strengthen the social bonds among contributors to the forum (Semino et al., Reference Semino, Demjén, Hardie, Rayson and Payne2018: 233ff.; see also Semino and Demjén, Reference Semino, Demjén and Hampe2017).

In contrast, when Violence metaphors are used to express the fear or prospect of not recovering, the patient is placed in a disempowered position:

In both examples provided, not getting better (and therefore potentially dying of cancer), is metaphorically described as ‘not winning’, as part of Violence metaphors (cf. battle and fight in the extracts provided). In addition, in both cases, the person suggests that they feel responsible for not recovering or anticipates that others will see them as responsible. This is where Violence metaphors can do harm, as suggested by their critics. In a violent confrontation – whether a physical struggle or a battle between armies – the party that is defeated is typically perceived to weaker, less determined, or less well organised. Within a violence framing for illness, these associations are attributed to the person who does not recover, resulting in the person feeling ‘a failure’ or anticipating embarrassment for not getting better, when in fact they bear no responsibility for the course of their illness.

In the following example, in contrast, the patient is implicitly disempowered in relation to the healthcare system, as they are not being provided with the treatment (armour) that they believe they need to get better:

The MELC team noticed that the term battle, as a noun or as a verb, had a tendency to be used to suggest disempowerment, either because of extreme difficulties experienced by patients or the unsuccessful outcome of treatment (Semino et al., Reference Semino, Demjén, Hardie, Rayson and Payne2018: 109).

While it was previously known that Violence metaphors were frequently used to talk about having cancer and trying to get better, the analysis of the MELC corpus also revealed the use of Violence metaphors for the effects that cancer treatment has on the patient. In such cases the violence-related metaphorical expressions evoke scenarios dealing with personal physical aggression, such as ‘a battering from chemo’, ‘a “hammering” from medication’, or various uses of the verb ‘hit’:

Here the process of being treated with radiotherapy and/or chemotherapy is presented as being on the receiving end of a violent attack. The emphasis is not so much on ‘losing’ against the aggressor but on the damage that results from the attack, especially where the metaphorical expressions suggest extreme violence, as in the case of ‘hit by a sledgehammer’. Cancer treatment is well known to have major side effects, especially in the case of chemotherapy. The use of Violence metaphors for treatment suggests the seriousness of these side effects while also conveying a sense of helplessness and disempowerment in the context of medical interventions that are intended to help the person recover or live longer.

The analysis of the MELC corpus also revealed instances of Violence metaphors where the two opponents were the patient and the healthcare system:

These uses of Violence metaphors reflect the perception of difficulty and effort involved in obtaining the treatment that the patients feel they need. With regard to (dis)empowerment, on the one hand the patients place themselves in the active position of agents advocating for themselves. However, they are not in fact empowered, partly for having to argue for their own treatment at a point of extreme vulnerability, and partly because the decision-making power with regard to that treatment ultimately lies with health professionals and policy makers.

As we showed in Chapter 5, a different use of Violence metaphors for the patient’s relationship with the health system involves empowerment through humour (Semino et al., Reference Semino, Demjén, Hardie, Rayson and Payne2018: 252–5). A group of contributors to a particular thread dedicated to cancer-related humour describe themselves as a small army aiming to ‘liberate’ specific individuals from the hospital and even give each other military titles and promotions to higher ranks. This long-running collaborative joke shows the power of the combination of metaphor and humour to create distance, at least temporarily, from a situation of hardship, and to foster emotional closeness with others in the same situation.

Overall, the use of corpus methods to systematically study the Violence metaphors used by patients revealed a greater variety of uses than had been previously noticed. Among other things, this shows how patients may feel in a situation where they are ‘fighting’ on several fronts – against the illness, against the treatment, and against the healthcare system. On the other hand, while Violence metaphors can be disempowering, they can also be used in empowering ways, especially when they are combined with humour.