‘Good’ and ‘bad’ deaths: Narratives and professional identities in interviews with hospice managers

Abstract

This article explores the formal and functional characteristics of narratives of ‘good’ and ‘bad’ deaths as they were told by 13 UK-based hospice managers in the course of semi-structured interviews. The interviewees’ responses include a variety of remarkably consistent ‘narratives of successful/frustrated intervention’, which exhibit distinctive formal characteristics in terms of the starting point and core of the action, the choice of personal pronouns and metaphors, and the ways in which positive and negative evaluation is expressed. In functional terms, the hospice managers’ narratives play an important role in representing and constructing their professional views, challenges and identities. Overall, the narratives argue for the role of hospices and professional hospice staff in facilitating a ‘good’ death, and, by presenting a relatively unified view, may potentially preclude alternative perspectives.

Introduction

In this article we discuss the formal and functional characteristics of the stories told by 13 UK-based hospice managers ¹ to illustrate what they see as a ‘good’ or a ‘bad’ death in the course of semi-structured interviews. We relate these characteristics to the interviewees’ professional approach to death and dying, and particularly to the professional identities that the hospice managers construct for themselves and their colleagues. Following is a preliminary example of a ‘good’ death narrative from one of our interviews: ²

somebody that comes to mind is somebody that was very very anxious, came into the hospice and they couldn’t sit down, they were so anxious they were just pacing and pacing and gradually we started to have discussions about what their worries are, what they frightened about and we started to relieve that anxiety. And then the patient will eventually sat down and we could have those conversation with them, they wouldn’t go to bed at night time cos they were so frightened so they slept in the chair, and so gradually with the work we were doing they they got into bed and they became less frightened, they talked to us about their symptoms and things that they were really frightened about happening at end of life. Erm that they were gonna die with a lot of pain erm . . . and erm and they were frightened, just very very frightened about what was going to happen. So gradually we worked through those things with them. Erm and to the point where they went to bed at night and had a good night’s sleep and then got up and then had quality of life for a few days where their family members were coming in. And they were talking openly with the family members about what was happening, and then their condition just deteriorated and we were able to put all the things in place that we’d agreed with that patient would make their end of life care and the last few days of the life better. And the patient they died peacefully. In the hospice. And the family were with them. So that was that to me was a good death.

The interviews were conducted in late 2012 as part of the project ‘Metaphor in End-of-Life Care’, funded by the UK’s Economic and Social Research Council and based at Lancaster University. ³ This study, and the interviews with hospice managers in particular, need to be set against the background of contemporary practices and discourses around end-of-life care, dying and ‘quality’ of death in the United Kingdom.

Modern hospice care in the UK developed from the pioneering work of Dame Cicely Saunders in London’s St Christopher’s Hospice in the 1960s. It involves a holistic approach to the needs of patients with life-limiting or terminal illnesses: within this approach, the social, emotional and spiritual aspects of the patient’s needs are taken into account alongside the physical dimension. In 2008, the first ‘End-of-Life Care Strategy’ for England and Wales was published by the UK’s Department of Health, outlining a 10-year strategy for improving care at the end of life. In 2012, one particular approach to the care of the dying, the ‘Liverpool Care Pathway for the Dying Patient’ (LCP), became the focus of public and media attention. The LCP was intended to ensure the comfort and dignity of dying patients. It involved, among other things, the cessation of non-essential medication and treatment for patients who were deemed to be close enough to death not to benefit from further intervention. The pathway became publicly controversial after numerous media reports that it was not always correctly implemented. In 2013, an independent review recommended that it should be phased out. ⁴ Also in 2013, the European Association for Palliative Care (EAPC) launched ‘The Prague Charter’, which ‘urg[es] governments to relieve suffering and recognize palliative care as a human right’. ⁵ These initiatives, among others, demonstrate that the importance of care at the end of life is increasingly being acknowledged.

Generally speaking, however, death and dying are still taboo topics: most people find them embarrassing, uncomfortable and potentially distressing to talk about. Yet there is increasing consensus in the public sphere, including in the UK, that talking explicitly about death and dying is necessary in order to make death as positive an experience as possible, both for the person who is dying and for those close to them. In 2009, the UK’s National Council for Palliative Care created the ‘Dying Matters’ coalition – a wide network of organizations that aim ‘to promote public awareness of dying, death and bereavement’. ⁶ The coalition’s mission statement explicitly links ‘talking about dying’ with the experience of a ‘good’ death. ⁷

The notion of a ‘good’ death, and communication around the topic of death, have been studied from several different perspectives. Payne et al. (1996) conducted interviews with patients and staff in a palliative care unit in order to compare their views of a ‘good’ death. The patients’ descriptions were found to be rather heterogeneous, and included ‘dying in one’s sleep, dying quietly, with dignity, being pain free and dying suddenly’ (Payne et al., 1996: 307). The staff’s descriptions showed greater uniformity and emphasized ‘adequate symptom control, family involvement, peacefulness and lack of distress’ (Payne et al., 1996: 307). In addition, the staff characterized a ‘bad’ death in terms of ‘uncontrolled symptoms, lack of acceptance and being young’ (Payne et al., 1996: 307). Similar views were found by Low and Payne (1996) among nurses and social workers. A ‘good’ death was associated with being able to control the patient’s symptoms and help them prepare for and accept death. A ‘bad’ death was associated with the inability to deal with the patient’s negative emotions, resulting in lack of acceptance of death, and distress for family members.

Taking a discourse analysis perspective, Carpentier and van Brussel (2012) point out how the notions of a ‘good’ or ‘bad’ death have changed over time, and stress the political nature of the ‘discourse of death’. Van Brussel and Carpentier (2012) discuss more specifically the discursive construction of a ‘good’ death in the north Belgian press coverage of three euthanasia cases in 2008. They suggest that the notion of a ‘good’ death has been politicized, particularly in the discourses associated with palliative care on the one hand and the right to die on the other. While the palliative care movement associates a ‘good’ death with control, autonomy and dignity, the right-to-die movement associates it with awareness and heroism. This politicization of death, whether ‘good’ or ‘bad’, can affect the range of ‘acceptable’ choices that patients and families are presented with.

A rather different approach to talking about death features in William Labov’s classic work on oral narratives of personal experience in Harlem in the 1960s (Labov, 1972; Labov and Waletzky, 1967). Narratives were elicited from participants by asking them whether they had ever been in danger of dying. This question arguably exploits a corollary of the taboo status of death, namely the fact that death and potentially fatal incidents are inherently highly ‘reportable’ (Labov, 1972: 370; Labov, 2010) – a claim that has been repeatedly echoed in the narratological literature (e.g. Ryan, 2005). In Labov’s data, the focus was primarily on the risks and exhilaration involved in narrowly avoiding death. This involves a different kind of heroic dimension to (nearly) dying than the one that applies to the right-to-die discourse discussed by van Brussel and Carpentier (2012). More recently, Labov (2013) has restated that death is one of ‘three universal centers of interest’, alongside sex and moral indignation (Labov, 2013: 4).

In contrast, the details of actual deaths, especially when they result from protracted illnesses, can be too personal, sensitive and distressing to be told in the majority of conversational contexts. In this sense, death can be included among the experiences that Norrick (2005) describes as ‘untellable’, not because they are too mundane to reach the minimum threshold for tellability, but because they are ‘too personal, too embarrassing or obscene’, and therefore exceed the maximum threshold for tellability in most contexts (Norrick, 2005: 323). According to Norrick, such topics include illness and medical procedures, and taboo topics such as sex and dying.

In this article we show how hospice managers’ perspectives on ‘good’ and ‘bad’ deaths, expressed in the narratives they told in response to interview questions, revolve around the success or failure of professional intervention in difficult circumstances. We also show that these narratives have some distinctive formal characteristics, especially in terms of the starting point and core of the action, the use of personal pronouns and metaphors, and the ways in which positive or negative evaluation is expressed. Additionally, we point out that, as noted by Payne et al. (1996), the view of what constitutes a ‘good’ death is remarkably consistent across our interviewees. We suggest that these characteristics reflect and support a strong sense of professional worth and identity, which is undoubtedly helpful for hospice staff performing their difficult roles. However, these unified views on ways of dying could potentially limit the range of perspectives and attitudes that are deemed ‘acceptable’ in a hospice context.

In the next section, we present our approach to identifying and analysing ‘narratives’ in our data, and briefly discuss previous relevant work on narratives in relation to evaluation, argumentation and (professional) identities. We then consider examples of ‘good’ and ‘bad’ death narratives in turn, before concluding with reflections on the implications of our findings.

Narratives, professional identities and our data

Three main aspects of narratives are particularly relevant to the discussion of our data. First, stories are an important tool for making sense of the world, of ourselves and of our experiences (e.g. de Fina and Georgakopoulou, 2012: 18; Herman, 2007: 3). Second, story-telling can function, among other things, as an argumentation device: stories can be used to provide ‘evidence’ from one’s own or others’ experiences to support one’s claims, views and judgements (e.g. de Fina and Georgakopoulou, 2012: 97; Schubert, 2010). Third, within a social constructionist approach, narratives are one of the means through which different aspects of tellers’ identities can be constructed and negotiated (e.g. Bamberg and Georgakopoulou, 2008; de Fina and Georgakopoulou, 2012: 105–190; Georgakopoulou, 2007). Within this approach, identities are seen as multiple, transient and negotiable. Discourse – and story-telling in particular – provides the means to project particular images of oneself and of the group(s) one belongs to in relation to situations, events and other people or groups: ‘stories offer their narrators a way of re-imagining their lives and (re)creating identities’ (Harvey and Koteyko, 2013: 91). Even the way the ‘story world’ is constructed ‘points to how the teller wants to be understood, what sense of self they index’ (Bamberg and Georgakopoulou, 2008: 380). A particularly relevant aspect of identities for our purposes is that of ‘professional’ identities, which are linked to the teller’s workplace and their role within it. A number of studies have shown how different types of narratives can be used in workplace settings to construct different kinds of personal and group identities (see Georgakopoulou, 2011; Holmes, 2005; Metz, 2003; Sarangi and Roberts, 1999).

Defining ‘narrative’ is a complex and controversial issue, both theoretically and methodologically (for overviews, see de Fina and Georgakopoulou, 2012: 1–25; Ryan, 2007). Generally speaking, ‘story’ or ‘narrative’ is best seen as a fuzzy category whose members vary in their degree of ‘narrativity’ (e.g. Herman, 2007). Within narratology and sociolinguistics, narrativity has been associated with a set of central characteristics, including the presence of conscious participants engaging in intentional actions, some kind of change or disruption, a sequence of temporally connected events, and ‘meaningfulness’ or ‘tellability’ for a particular audience (e.g. Georgakopoulou, 2007: 37–40; Ryan, 2007).

For the purposes of our study, a ‘story’ or a ‘narrative’ is a stretch of talk within our interviewees’ answers that concerns a patient experiencing a ‘good’ or ‘bad’ death, and includes a set of actions and reactions preceding that death, and some indication of why and how that death is regarded as ‘good’ or ‘bad’. Given that our narratives were elicited in the course of semi-structured interviews, it is not surprising that, in some cases and to varying degrees, they exhibit a similar structure to the ones that were told by the interviewees in Labov’s classic studies. As is well-known, Labov defines narrative as ‘[o]ne method of recapitulating past experience by matching a verbal sequence of clauses to the sequence of events which (it is inferred) actually occurred’ (Labov, 1972: 359–360). While a minimal narrative consists of two temporally oriented clauses, Labov’s (1972) classic framework for ‘narratives of personal experience’ includes five main elements or stages:

• Abstract: an indication that the speaker has a story to tell and/or a brief summary.

• Orientation: a description of the setting, including time, place, people, situation.

• Evaluation: devices that indicate the point of the story, that is, why it is worth telling.

• Complicating action: a series of clauses that present the events that are the core of the story.

• Resolution: an indication of the final event.

• Coda: an indication that the story is finished and potentially some general observations on the effects of the event on the narrator.

The evaluation component of Labov’s model is linked with the notion of tellability: evaluative devices are used to indicate or emphasize why it is appropriate to narrate a particular event in a given context. Within our interviews, the topic of death is of course central to the interviewees’ professional role and institutional context. In addition, the interview questions we consider in this article explicitly concerned death, and encouraged the telling of stories about death and dying. In some respects, this reduces the tellability stakes, as the interviewee is not responsible for the topic and is implicitly allowed space for extended story-telling. At the same time, the setting and questions raise the threshold above which particular details may be ‘untellable’ in Norrick’s (2005) terms. Indeed, several of the responses include detailed descriptions of bodily functions and severe physical symptoms which would be inappropriate to mention in most other contexts. Hence, the tellability of each story hinges on what makes each death ‘good’ or ‘bad’. In this context, it is important to distinguish between valence – which is to do with whether and how a death is seen as ‘good’ or ‘bad’ in itself, and evaluation – which signals tellability.

Of course, as has been found in much recent work (e.g. Georgakopoulou, 2007), a strict application of Labov’s definition and framework does not do justice to the different ways in which our interviewees engaged in story-telling in response to our questions. We therefore also classified as narratives a variety of stretches of text that depart in different ways from the prototypical cases described by Labov, particularly by not including a section that straightforwardly corresponds to the definition of ‘complicating action’. Some of these narratives could be subsumed under the general label ‘small stories’, which Georgakopoulou (2007: 148) uses to capture ‘a whole range of under-represented narrative activities ranging from literally small and fragmented tellings to refusals to tell and deferrals of telling’ (see also Bamberg, 2004). Nonetheless, Labov’s framework proved consistently useful in our analyses in order to account for the distinctive structural characteristics of the stories told by our interviewees.

The status of narratives told in the course of research interviews, like that of research interviews more generally, has been hotly debated (e.g. see Schegloff, 1997, and Wolfson, 1976, for critiques of story-telling in interviews; de Fina and Perrino, 2011, for an overview). For the purposes of our study, we regard narratives as a situated practice (Potter and Hepburn, 2008: 284), the situation here being the semi-structured interview. The questions about ‘good’ and ‘bad’ deaths came towards the end of each interview and were expressed as follows: ‘How would you describe a good death? How would you describe a bad death? Can you provide any examples from your experience?’ Our interviewees were recruited because of their professional role as hospice managers, and were told that our research was concerned with the language used to talk about end-of-life care. The interviews lasted between 40 minutes and an hour. They mostly took place at the interviewees’ place of work, while a few were recorded at a university where the interviewees were attending classes related to their profession. All interviews were conducted by one of two of the authors, and both interviewers were working from the same interview guide. Interviewees were aware that both interviewers were linguists, rather than specialists in end-of-life care, and therefore ‘outsiders’ whom they may have felt in need of convincing of the importance of hospices in end-of-life care and of the value of their profession. As the interviews were part of the ‘Metaphor in End-of-Life Care’ project, the interviewers tried to avoid using metaphor in their questions. Back-channelling responses were offered and follow-up questions asked where appropriate, especially to elicit examples.

As a result of the semi-structured nature of the interviews, interviewees were allowed enough flexibility to express themselves freely in response to open-ended questions in a relaxed atmosphere (Dörnyei, 2007). However, the distinction between ‘good’ and ‘bad’ deaths was always introduced by the interviewers, who also encouraged the interviewee to provide some examples from their experience. We therefore acknowledge the interviewer’s active involvement in the genesis and production of the narratives (Mishler, 1986), and are not in a position to claim that the stories in our data resemble those that the interviewees might tell in naturally occurring contexts. However, recent research has undermined the notion of a stark dichotomy between ‘elicited’ and ‘spontaneous’ story-telling (Koven, 2011). Moreover, the notion of the ‘good’ death in particular is one that hospice managers regularly encounter and discuss in their professional practice.

In our interviews the hospice managers express their views about ‘good’ and ‘bad’ deaths in three main ways: a) general statements about the characteristics of a ‘good’ or a ‘bad’ death, often including metaphorical expressions; b) generic narratives about types of people and experiences; and c) narratives about specific (but unnamed) individuals. As we show in the rest of this article, the specific narratives told in response to our questions tend to include the following elements: a description of the patient’s background, often including family circumstances and relationships; an account of the intervention of hospice staff in the care of the patient; a reference to the patient’s death; and some reflections on what makes the particular patient’s death ‘good’ or ‘bad’. We suggest that the interviewees use these narratives to construct particular professional roles and identities for themselves, that is, to define, elaborate and justify their role in end-of-life care. This is obviously cued by the situational context, as we interviewed people in their role as hospice managers, and mostly conducted the interviews in their workplaces. We will see that professional face and identity are mostly enhanced by emphasizing the role of hospice staff in deaths valenced as ‘good’, while downplaying it in those seen as ‘bad’.

A ‘good’ death and narratives of successful intervention

In response to our question about what constitutes a ‘good’ death, the hospice managers pointed out that a ‘good’ death is a matter of perspective, and emphasized that it is essential to give patients options, and to try to fulfil their wishes. Their own descriptions of ‘good’ deaths included a number of recurring metaphors, such as being ‘at peace/peaceful’, being ‘symptom/pain free’, having ‘open’ conversations with family members, and accepting death as the ‘end’ of one’s ‘journey’. Importantly, several interviewees recognized that most people wish to die at home, without professional intervention. However, they also claimed that hospice care can actually provide patients and their families with a better overall experience, thus making a case for the importance of end-of-life care and their profession. In this section we consider two typical examples of specific narratives from the interviewees’ responses. We discuss their content, structure and linguistic characteristics, and relate these to the argumentative function of these narratives in the interviewees’ accounts of what they see as a ‘good’ death.

Example 1 is typical of narratives that were told to exemplify a ‘good’ death. The interviewee introduces a difficult situation that has the potential to result in a ‘bad’ death. This situation is addressed by professional intervention in the hospice, and changed to such an extent that the patient has a ‘good’ death. We have laid out the story according to what we see as its main component elements. In the discussion that follows, we reflect on how these relate to the elements in Labov’s (1972) framework.

Example 1

[‘Opening’]

Erm I think of another gentleman who came to the hospice,

[‘The patient and his family’]

he was a Portuguese speaker, had pretty much no English at all. And he’d had recurrent hiccoughs for about five months. And the medical team put in a referral for him to have some acupuncture. And looking at his case history, he was getting so depressed not just with the hiccoughs but with his diagnosis of stomach cancer. And he’d been suicidal at one stage, he was so depressed that he couldn’t enjoy his wife’s cooking. He was in one of the wards and was desperate to get better to go home.

[‘Core: Successful professional intervention’]

And we went to see him as a team and did some acupuncture, and the recurrent hiccoughs erm reduced considerably, in the first instance and then and then stopped and he was able to go home.

[‘Reflections on death as “good”’]

And so I think that was a good piece of collective collaborative work. To actually fulfil the wishes of you know he wanted to get home spend a bit of time and be able enjoy his wife’s cooking.

[‘The death’]

Erm he died a few months later,

[‘Reflections on death as “good”’]

but I think that was an illustration of you know an immediate sort of response to a request that worked quite well.

The story opens with a reference to the patient and to the main event that transforms his experience of end of life: came to the hospice. In Labov’s terms, this utterance can be seen as a minimal abstract. This is followed by a detailed description of the initial situation that hospice staff were presented with: the focus is on the patient, and specifically his personal circumstances and the challenges posed by his terminal illness. While this part of the story corresponds to Labov’s ‘orientation’, it is not a prototypical instance in that it is highly valenced: the emphasis is on the negative aspects of the patient’s initial situation (recurrent hiccoughs, depressed, stomach cancer, suicidal, desperate) and on the absence of positive aspects (no English, couldn’t enjoy his wife’s cooking) (see Labov, 2013: 38, on non-prototypical instances of orientation in ‘highly polarized’ narratives). These negative assessments function as signals of evaluation and set up a contrast with the positive outcome of the staff’s intervention, as they suggest that the patient was in a situation that was likely to result in a ‘bad’ death.

In the following stretch of text, the focus switches to the hospice team, who are referred to via the first-person plural pronoun we. The speaker employs a series of temporally ordered clauses to outline the professional intervention in the hospice and its effects. This is the core of the narrative and, in this case, corresponds to Labov’s definition of ‘complicating action’. What is characteristic of this kind of narrative, however, is that the main action provides an example of successful professional intervention, which involves removing some of the undesirable aspects introduced in the description of the patient: the hospice staff intervene to solve one of the main problems that affected the patient’s quality of life, and that would potentially prevent a ‘good’ death. As the person’s main problem (stomach cancer) cannot be eliminated, success here involves dealing with the issues that can be addressed, thereby re-establishing as much ‘normality’ as possible: the hiccoughs stop and the patient returns home.

As with other narratives in our data, in Example 1 the patient’s death is mentioned briefly and almost as an afterthought, as it is an obligatory and predictable part of the story. In Labov’s terms, however, this brief reference to the timing of the patient’s death in relation to the core action arguably counts as the story’s resolution. In the extract above, this reference occurs within a series of reflections on this patient’s death, which, in Labov’s terms, function as ‘external’ evaluation (Labov, 1972: 371). From his current vantage point (explicitly signalled by the repetition of I think), the speaker praises the team’s intervention using adjectives and adverbs (good, quite well) and points out what he perceives to be the main strengths of the professionals’ actions within the institutional setting – collective collaborative work, immediate response. In addition, he makes references to the fulfilment of the patient’s wishes, partly by repeating those elements that were presented as absent in the orientation section: fulfil the wishes, he wanted to get home spend a bit of time and be able enjoy his wife’s cooking.

He presents these as reasons why the patient’s death was ‘good’, and hence why this particular story is tellable in response to a question concerning a ‘good’ death.

This positive evaluation is closely linked to the sense of professional achievement experienced by the interviewee and, arguably, his colleagues, when they dealt successfully with the patient’s problems on his first visit to the hospice. The collaborative nature of end-of-life care is reinforced in the use of we and as a team to refer to the agents who bring about the successful intervention: the interviewee presents himself as part of a group of professionals that acts as one. The team as social actor is further represented as engaging in actions that are expressed by means of verbs (And we went to see him as a team and did some acupuncture), which contrasts with the otherwise nominalized style (collective collaborative work, immediate sort of response) and foregrounds the staff as active professionals. Although the interviewee mitigates his team’s beneficial action to minimize self-praise (I think, sort of, quite; cf. Leech’s, 1983, ‘Modesty Maxim’), it is clear that he here constructs his profession as both necessary and positive.

Some of the ‘good’ death narratives in our interviews are less prototypically structured than Example 1, but perform the same function in terms of argumentation and identity construction. The interviewee we quote in Example 2 begins his response by pointing out that there is no ‘black and white’ distinction between ‘good’ and ‘bad’ deaths. However, he also adds that, although people tend to favour the idea of dying at home, a hospice death can be far nicer [. . .] than struggling in a crisis at home or in the back of the ambulance coming to the hospice. The following narrative supports this general point:

Example 2

[‘The patient and his family’]

it a was a young patient who in his thirties, his wife was there he got two young children. His wife had been told,

[‘Reflections on death as “good”: Alternative hypothetical scenario’]

and if he hadn’t have come to us,

[‘The death’]

and he died the next morning, and it was quick we weren’t expecting it to be that quick

[‘Reflections on death as “good”: Alternative hypothetical scenario’]

but if he hadn’t have been, if his wife hadn’t have been, if he’d have stayed in the hospital acute setting, his wife probably wouldn’t have been told wouldn’t have realized and he might have just died on his own behind some curtains. Erm you know without you know the level of sort of comfort dignity, attention to his needs and his family.

The story begins with a description of the patient and his family, including some details indicative of a difficult situation: young patient, young children. In terms of positioning and linguistic features, this description corresponds to Labov’s ‘orientation’ phase (e.g. the use of the past perfect had been). However, it also includes two references to positive aspects of the patient’s death that resulted from the patient’s admission to the hospice and from the staff’s intervention (his wife was there, His wife had been told). As in Example 1, the reference to the patient’s death can be seen as the story’s resolution, but is described rather briefly in order to point out its timing in relation to admission to the hospice and the hospice staff’s surprise at the speed with which it happened.

The rest of Example 2 consists of reflections on what makes this particular death ‘good’, which, in Labov’s terms, constitute a stretch of evaluation: the speaker spells out in detail why this situation, involving a last-minute professional intervention in the hospice, was still better than an alternative reality that would have come to pass if the person had remained in hospital. This is achieved through a series of what Labov (2010) calls ‘irrealis clauses’, including several conditional clauses (e.g. his wife wouldn’t have been told, might have just died on his own) and negation by means of the preposition ‘without’ (without you know the level of sort of comfort . . .). These devices implicitly evaluate the events that actually did occur in the hospice by comparing them with potential undesirable events or outcomes that the hospice staff were able to prevent. As in the previous example, the interviewee uses the first-person plural pronoun to refer to himself and his colleagues, although they are represented as less directly involved by the use of passive clauses (wouldn’t have been told) and nominalization (attention) instead of active clauses.

In contrast with the previous example, however, Example 2 does not include a separately identifiable core or ‘complicating action’ that would provide an account of the main actions and events that the story is about: the patient was moved from elsewhere (a hospital) to the hospice just in time for his wife to be told what was happening and for him to die with comfort and dignity. Nonetheless, the various aspects of the hospice staff’s successful intervention are suggested by clauses that occur as part of the introduction to the patient (e.g. His wife had been told) and of reflections on what did not happen thanks to the patient’s admission to the hospice (might have just died on his own behind some curtains).

Overall, our interviews show a considerable amount of agreement concerning the characteristics of a ‘good’ death (see Payne et al., 1996). In particular, the narratives that the interviewees told to exemplify their views share several distinctive characteristics. They tend to start at a point when many potentially tellable events have already happened to the patient: they have received a terminal diagnosis and their circumstances have deteriorated to such an extent that they are likely to suffer a ‘bad’ death. It is at this point that hospice staff become involved in ways that provide tellable material for our interviewees in response to a question about ‘good’ deaths. The actions that constitute a successful professional intervention tend to be attributed to a collective agent (the speaker and his/her colleagues), primarily by means of the first-person plural pronoun ‘we’. The patient’s actual death is usually mentioned only briefly, since its occurrence was assumed by the interviewer’s question and the end-of-life/hospice context.

Not surprisingly, a fairly large proportion of these stories consists of reflections that, in Labov’s terms, function as evaluation. Since death is typically unwelcome and associated with physical and emotional suffering, a positive assessment of particular deaths is a delicate matter. However, the idea that some deaths can be described as ‘good’ is central to the hospice managers’ profession, and was, of course, assumed by our interview questions. As we have shown, the evaluative devices used by the interviewees include positive assessments of the hospice team’s intervention and hypothetical references to the possible negative things that were avoided by that intervention. As a result, the events and actions that the narrative is about are not always told directly and chronologically, but are sometimes assumed or presupposed, particularly by means of negatives and conditionals that also function as evaluation devices.

In this way, the narratives are often more or less explicitly used to support the argument that dying in a hospice is not only a better experience than dying in a hospital, but also often better than dying at home without any hospice care. The latter point in particular counters a widely held view that the interviewees explicitly refer to, that is, that it is best to die at home. Presenting dying at a hospice as the best option is a particular construction of the story world that supports the professional identity the speaker wishes to project (Bamberg and Georgakopoulou, 2008). The ‘lay view’ about dying at home is at times directly questioned by the interviewees. In most cases, however, it is implicitly challenged via the stories that the hospice managers tell. This unified focus on intervention seems to suggest the persistence of a ‘medical model’ (see Payne et al., 1996) that potentially leaves little space for alternative views.

A ‘bad’ death and narratives of frustrated intervention

Our interview questions concerning what constitutes a ‘bad’ death were arguably even more challenging for the hospice managers to answer: as the goal of hospice care is to provide the best possible death, a ‘bad’ death in a hospice context is a challenge to the professional ‘face’ of the staff (Goffman, 1967). Indeed, several interviewees commented that ‘bad’ deaths are relatively rare in their experience, and some used different terms in their responses (e.g. ‘difficult’ or ‘hard’ deaths). As with ‘good’ deaths, some interviewees also suggested that what counts as a ‘bad’ death is a matter of perspective.

In the vast majority of cases, the responses to our ‘bad’ death questions can be seen as the converse of responses to the ‘good’ death questions: hospice managers described as ‘bad’ those deaths where professional intervention was unsuccessful. This lack of success was consistently explained in terms of problems that were outside the hospice staff’s control, thereby backgrounding their role in negatively valenced deaths. In a few cases, the interviewees mentioned lack of adequate resources for the provision of optimal care, or unfortunate circumstances such as patients being admitted at particularly busy times. By and large, however, the failure of professional intervention was explained as the result of the attitude of the people involved, namely, the patient’s close relatives or, in most cases, the patient him- or herself, who either rejected help or was unable to benefit from it due to the way they had lived up to that point. This kind of explanation for a ‘bad’ death is rhetorically delicate, as it involves implicit or explicit criticisms of vulnerable people whose care is the main task of the hospice team. The ‘bad’ death narratives in our interviews help to achieve a difficult balancing act between mitigating the hospice team’s responsibility for ‘bad’ deaths and avoiding overly explicit criticisms of dying patients and their relatives.

Example 3 below is representative of our interviewees’ responses to the ‘bad’ death questions. It involves a hospice manager who begins her answer by suggesting that people’s approach to death tends to be consistent with their approach to life. She then uses a series of ‘battle’ metaphors to describe the attitudes and feelings on the part of patients that make it impossible for hospice staff to intervene successfully: for some people their life is a struggle or a battle or a series of conflicts; there’s [. . .] a lot of battling in the background. According to the interviewee, this approach to life leads to intense feelings of fear when the person is faced with imminent death. In addition, she suggests that these patients tend to not communicate their feelings and do not allow the hospice staff to help with these feelings: ‘fear’ is described, again metaphorically, as being in the background; pretending that there are no negative feelings is described as a horrible overlay; and the negative emotions that the patient experiences are described as being located in a scared place that the hospice team are not allowed to reach into. All of this results in the hospice staff not being able to provide support.

These general reflections form part of a lengthy preface to the telling of the story about a specific patient who experienced a death that the interviewee found ‘difficult’.

Example 3

[‘Opening’]

I have one lady who comes to mind right from . . . my early weeks of working in palliative care,

[‘The patient and her family’]

she’s a lady who’d suffered a lot of anxiety in her life. And had a difficult relationship with her husband. And did battle with everyone.

[‘Core: frustrated professional intervention’]

Nothing was ever right, nothing was good enough, nothing. You couldn’t physically get her comfortable. You couldn’t kind of get in a good place to meet her psychologically erm . . .

[‘Reflections on death as “bad”’]

I think she felt that being comfortable was being patronized, and that made it really difficult to offer her any support.

[‘The death’]

And she had quite a traumatic death in that she . . . not only died vomiting. She had a bowel obstruction. Erm but died very fearful and very anxious and wouldn’t accept any pain relief. . . .

[‘Reflections on death as “bad”’]

So I think her life how she’d lived it played out right into the very last moment. And that was difficult for staff team. Cos we couldn’t do the things we each felt we could have done. It was difficult for her family because I think they ended up feeling . . . almost in a sense that she’d actively punished them. And then it was difficult in their bereavement care . . . to sort of . . . make sense of it. And make any meaning out of it.

In the story’s opening, which can be described as a minimal abstract, the speaker uses the singular first-person pronoun to refer to herself as narrator and to locate the events she is about to tell in the context of her career. This is followed by a description of the particular patient that corresponds to Labov’s ‘orientation’: the interviewee outlines a problematic situation, both in terms of emotions (a lot of anxiety) and in terms of family relationships (a difficult relationship with her husband). Here another ‘battle’ metaphor is used to describe in pejorative terms the patient’s approach to others (did battle with everyone).

The following four independent clauses arguably constitute the core of the narrative: the setting is now the hospice, and the focus is on the patient and the hospice staff in the period leading up to the patient’s death. However, these clauses do not constitute an instance of Labov’s ‘complicating action’, for two reasons. First, they do not refer to singular events that occurred sequentially at particular points in time, but rather to attitudes and happenings that occurred repeatedly during the patient’s stay at the hospice. Second, all four clauses involve negation, and hence indicate what did not apply or happen in relation to the particular patient. The first two clauses (beginning with nothing) implicitly attribute to the patient a totally dismissive assessment of and attitude towards the hospice setting and services. The remaining two clauses describe what the hospice staff were unable to achieve, respectively for the speaker’s physical and emotional well-being (cf. ‘negated narratives’ in Baynham, 2011). A ‘location’ metaphor is also used to express the impossibility of providing psychological help for the patient (couldn’t kind of get in a good place to meet her psychologically). Nonetheless, these four clauses constitute the ‘core’ of the narrative for our purposes because they indicate that the hospice staff attempted to intervene as they normally would, but saw their efforts frustrated by the patient’s attitude. This resulted in a ‘bad’ death.

While in ‘good’ death narratives the first-person plural pronoun ‘we’ tends to be used to refer to the agents of a successful intervention, here the second-person pronoun ‘you’ is used to refer to the hospice team (including the interviewee), whose intervention was ultimately unsuccessful. This use of the second-person pronoun is partly generic, insofar as it has a collective referent, and partly specific, in that it does not refer to people in general but to the particular hospice team at the time (Demjén, 2011). The interviewee’s use of ‘you’, as well as the general reflections before the specific narrative, therefore potentially suggest that anyone would have had the same difficulties if dealing with the particular patient (see Labov, 2013: 41, on the self as ‘generalized other’). This helps to present a reduced responsibility on the part of the professional and her team (because the pronoun’s referent is not restricted to them), while maintaining the sense of the ‘collective’ that in ‘good’ death narratives is typically signalled by ‘we’. In short, the switch in pronoun helps to maintain a positive professional identity for the speaker and her colleagues, in the face of something that could be considered a case of professional failure.

As with some of the previous examples, the story’s resolution involves the telling of the patient’s death. The difficulty of this particular death is conveyed explicitly by describing the death itself (traumatic), by providing distressing details (vomiting, bowel obstruction, wouldn’t accept any pain relief), and by attributing negatively valenced emotional states to the patient (very fearful and very anxious). The rest of the extract consists of reflections on the death as ‘bad’ that can be seen as external evaluation: the hospice manager switches back to her current first-person perspective (I think) and, in the final stretch of the extract, repeats it/that was difficult three times to spell out the negative effects of this death for the staff team and for the patient’s relatives, including in the period following the death. In Labov’s terms, evaluative devices are, however, spread throughout the whole extract, including particularly ‘irrealis clauses’ in the form of negation: the speaker mentions a number of things that did not happen with the patient, and implicitly suggests that these things would have been needed for her to have a ‘good’ death (e.g. couldn’t physically get her comfortable, wouldn’t accept any pain relief).

Overall, the narratives that our interviewees told to exemplify a ‘bad’ death are consistent with those they chose to tell in response to our questions about ‘good’ deaths. The core of the ‘action’ is the ways in which the hospice staff’s attempts to intervene for the patient’s benefit were frustrated by the patient’s attitude or characteristics, or by circumstances. The team are presented as acting as a coherent collective agent, referred to either by a semi-generic use of the second-person pronoun ‘you’, as in Example 3, or by the first-person plural pronoun ‘we’ elsewhere. Evaluation devices tend to be present throughout each narrative, but there is usually a fairly substantial evaluation section at the end of the story. Crucially, however, in ‘bad’ death narratives the difficulties faced by the staff are such that their intervention fails through no fault of their own. As we have shown, it is primarily the patient’s attitude that is foregrounded as the cause of the failure of the intervention.

There is also a tendency to point out that a ‘bad’ death affects the staff psychologically. This can be seen to some extent in Example 3, while in other cases the interviewees talk about their colleagues being paralysed or overwhelmed by a ‘bad’ death, or needing support themselves afterwards. This adds another facet to the professional identity at stake, namely vulnerability. The image that emerges is of staff who act in beneficial ways for the patient and family unless circumstances or the latter’s attitude prevent(s) them from doing so, and who face emotional issues of their own as a result of the difficulties and demands of their profession.

Taken together, the narratives of ‘good’ and ‘bad’ deaths thus realize the discourse goal of self-enhancement. This works through what in social cognition research is called ‘attributional asymmetry’ (Augoustinos et al., 2006: 164), a phenomenon that involves people accepting credit for success, that is, attributing it to their own actions or personality traits, but denying responsibility for failure, that is, attributing it to external factors. The subtle discursive strategies with which this difference in attribution is achieved (shift in pronouns, valenced metaphors, focus on circumstances in ‘bad’ death narratives) are ways of ‘juggling [. . .] two story-lines’ (Bamberg and Georgakopoulou, 2008: 392): not wanting to criticize vulnerable individuals, but also not wanting to be blamed for unsuccessful care. In our case, the agents involved are not the hospice managers as individuals, but the staff teams that the interviewees identify with and feel responsible for.

Concluding remarks

Our interview questions about ‘good’ and ‘bad’ deaths required the 13 hospice managers to address issues that are central to their profession, sensitive and personal for the people they serve, and increasingly visible and controversial in media and public debates. As previous studies have also found (e.g. Low and Payne, 1996; Payne et al., 1996), there was considerable consensus in how our interviewees characterized both ‘good’ and ‘bad’ deaths. A ‘good’ death is described primarily in terms of peacefulness, symptom control, frank conversations, acceptance and openness to physical and emotional support. Conversely, a ‘bad’ death is described primarily in terms of conflict within families, lack of acceptance, rejection of physical and emotional support, and physical and emotional distress.

As we have shown, most interviewees use narratives to exemplify their points and to support their claims that hospices can play a central role in making it possible for people to experience a ‘good’ death. The examples we have discussed show how the interviewees’ narratives of death revolve around professional interventions in hospices and tend to include a description of the patient’s background; an account of the (successful or frustrated) intervention on the part of hospice staff; a reference to the patient’s death; and some reflections on what makes the particular patient’s death ‘good’ or ‘bad’.

These narratives are generally similar to Labov’s ‘prototypical’ examples in that they are ‘past events stories of personal experiences not known to the audience’ (Georgakopoulou, 2007: 32), and were elicited in an interview context. As we have shown, some of the interviewees’ stories would also qualify as ‘narratives’ according to Labov’s stringent definition, while others are ‘atypical’ in one or more respects. The latter cases are in fact ‘atypical’ in different ways from the informal, interactional stories discussed, for example, in Georgakopoulou (2007). In particular, the actions and events that are at the centre of some of our narratives are not always conveyed explicitly and in sequence, whether chronological or otherwise. Rather, they may sometimes be assumed on the basis of shared knowledge established contextually and co-textually (e.g. the patient’s admission to a hospice) or be implied via hypothetical and negative structures that also function as evaluative devices. While, structurally, these narratives lack Labov’s defining element of ‘complicating action’, they are nonetheless told in such a way that a series of recognizable and reportable events can be reconstructed by the listener.

Nonetheless, throughout our analyses, Labov’s framework has proved useful both in accounting for the structural similarities between our narratives and those discussed in previous studies, and in pointing out the distinctive characteristics of the narratives in our data. As we have shown, the parts of our narratives that function as ‘orientation’ are highly valenced descriptions of patients who find themselves in circumstances that are likely to lead to a ‘bad’ death. In ‘good’ death narratives, hospice staff intervene successfully to address the patient’s challenges, and are able to avert a ‘bad’ death. In ‘bad’ death narratives, the hospice staff’s attempts to intervene are frustrated by circumstances outside their control, and particularly by lack of acceptance and cooperation on the part of patients. In some cases, as Example 3 shows, the interviewees feel able to describe some patients’ difficult approach to death as a consequence of their approach to life in general, even though the professionals’ actual knowledge of the patients is limited to the time when they were referred to hospice care.

In almost all cases, the singular first-person pronoun ‘I’ is used to refer to the interviewee as story-teller, while the agent of the professional intervention is the hospice team as a whole. When pronouns are used for the agent(s) of the intervention, the interviewees opt for collective reference via the first-person plural pronoun ‘we’ or for partly generic reference via the second-person pronoun ‘you’ (see de Fina and Georgakopoulou, 2012: 75–80, for collective narratives in immigrants’ discourse). The latter is used particularly in ‘bad’ death narratives to downplay the hospice staff’s responsibility for unsuccessful interventions, by using a referent that can be understood as inclusive beyond the immediate hospice team.

In both ‘good’ and ‘bad’ death narratives, evaluation plays a central role. The evaluative devices used by our interviewees include particularly negative and hypothetical structures. More specifically, hypothetical structures are particularly frequent in ‘good’ death stories, where they emphasize the importance of an intervention by contrasting it with alternative scenarios that were averted. On the other hand, negative structures are more frequent in ‘bad’ death narratives: ‘bad’ deaths are described in terms of what was not, but should have been, thereby creating a focus on expectations that were not met.

The formal characteristics of the ‘good’ and ‘bad’ death narratives, we have argued, suggest a strong sense of professional loyalty and identity which is at least in part due to attributional asymmetry (Augoustinos et al., 2006: 164): a ‘good’ death results from effective teamwork by hospice staff, while a ‘bad’ death occurs when the hospice team’s best efforts are frustrated through no fault of their own. This is not surprising, given the interviewees’ role as hospice managers and the fact that most of the interviews took place in their institutional setting. This type of attitude is also likely to be helpful in doing the type of job that hospice managers do. Dealing with death and bereavement on a daily basis can take its toll (as evidenced by frequent references to staff distress), and collectivizing/distributing responsibility when things go wrong can be seen as a coping mechanism. Moreover, end-of-life care has been critically reported on in the UK media recently, including accusations of cruelty, neglect and even institutional corruption (Watts, 2013). The narratives’ function of constructing and enhancing a sense of professional identity and worth, as well as conveying loyalty and team spirit, have to be seen against this wider social background.

At the same time, the interviewees’ stories show how much one’s (professional) perspective can affect the representation and evaluation of different ways of dying. In Labov’s terms, our interviewees use ‘narrative techniques to shape and transform [. . .] events, assigning praise and blame to the actors in a particular normative framework’ (Labov, 2013: 11). As we have seen, ‘bad’ deaths in particular are often blamed on patients’ negative attitudes towards the prospect of dying and towards the services provided by hospices. More generally, while our interviewees clearly show commitment and dedication to an extremely challenging job, the focus on intervention does suggest some degree of persistence of the ‘medical model’ in hospice care (see Payne et al., 1996). In addition, the interviewees’ loyalty to their role and colleagues leaves little space for any critique of the professional team or for alternative views. For example, one could question the view that dying patients should necessarily be willing to acknowledge and accept their imminent demise, and to discuss their intimate emotions openly with hospice staff. Indeed, other data show that, in some cases, patients and their families may feel invaded in their privacy or overwhelmed by the hospice staff’s intervention. In an online forum post included in our larger data set, a recently widowed woman comments that the last two days of her husband’s life were stolen from them by the well-meaning but unwelcome intervention of strangers who decided they knew what would help him.

As Van Brussel and Carpentier (2012) suggest, the notion of a ‘good’ death is contested and contingent, and largely constructed through discourse. The data we have discussed in this article show a particular construction that is the result of a strong professional identity, and are justified and enhanced in the face of criticism from the families of terminally ill patients, as well as increased media scrutiny. While the existence of such a unified ‘front’ is understandable and arguably even necessary in the circumstances, it is important to be aware that it may limit the range of ‘acceptable’ choices and attitudes that are open to patients and families.