Abstract
This article contributes new insights into the interplay between textual and reader factors in experiences of narrative empathy, or empathy with characters in narrative. It adds to the rather scarce empirical evidence on the relationships between textual devices and readers’ (non-)empathetic responses to characters. This empirical study involved stylistic-narratological analysis of short stories by Eduardo Galeano and thematic analysis of focus group discussions. The study considers empathy in relation to victims and perpetrators in narratives of persecution and torture. Methodologically, the article emphasises the value of a qualitative approach to collecting and analysing readers’ responses that is half way between naturalistic and experimental orientations. The main findings, which revolve around the interaction between certain narrative techniques and readers’ moral evaluation of characters, challenge some theoretical claims from the scholarly literature about textual effects on readers’ empathy. In so doing, the article considers empathy as a highly flexible and context-dependent phenomenon, and suggests the need for a nuanced approach that accommodates the complex interaction between textual and reader factors in the reading context. The discussion spells out the broader implications of the study for stylistic research on the role of language in bringing about effects in readers and also for narrative empathy research. These implications will be of interest to scholars conducting reception studies or reader response research in the neighbouring fields of empirical stylistics, empirical narratology, and empirical literary studies.
Keywords
1. Introduction
The study presented in this article contributes new insights into our understanding of readers’ experiences of empathy with characters from a stylistic perspective. Narrative empathy is here defined as the psychological process whereby recipients of narrative texts grasp and vicariously experience what they perceive are characters’ mental states. Narrative empathy has recently been given a great deal of attention as one of the psychological processes involved in reading. Several scholars support the view that empathy is often involved in the engagement between readers of narratives and characters (Coplan, 2004; Keen, 2006). More specifically, empathetic perspective taking has been regarded as the main psychological mechanism that underlies our experience of relating to fictional characters (Van Lissa et al., 2016: 43; see also Gaut, 1999; Keen, 2013).
This study examines the role of, and the interplay between, certain textual and reader factors in readers’ empathetic (or otherwise) engagement with characters in narrative. The study is a qualitative empirical stylistic contribution to the growing body of research on narrative empathy with regard to the potential links between narrative techniques and effects on readers. The research aims were to examine (1) the potential for empathy of the stories under analysis, (2) readers’ engagement with characters in relation to empathy, and (3) the role of and interplay between textual and reader factors in participants’ responses. The study considers empathy in relation to victims and perpetrators in narratives of persecution and torture written by Eduardo Galeano. The stories under analysis are non-fictional biographical stories, and thus are not prototypical fiction. Even though the study specifically addresses narrative empathy, the findings ultimately add to traditional stylistic concerns about reader experiences, textual effects on readers, and readers’ emotional engagement with characters.
I firstly review the scholarly literature on narrative empathy, including earlier theoretical and empirical research as well as potential factors that may influence empathetic responses. After that, I describe the methodological approach used in the study—a qualitative linguistic approach that combined stylistic-narratological textual analysis and empirical reader response research. I then present analysis and findings regarding how certain textual and reader factors seem to have influenced my participants’ responses to the story characters. Finally, I discuss the implications of these findings for both theory and practice.
2. Background
Narrative empathy is defined in my study as follows: Narrative empathy involves a character-oriented (emotional) response and perspective taking. The reader forms a mental representation of the character’s situation and mental1 state(s) while maintaining a self-other distinction. In this way, readers re-enact, simulate,2 or imaginatively experience in a first-person way what they perceive is the character’s mental state and mental activity. The resulting response is congruent with the reader’s perception and understanding of what the character’s experience must be like.
This definition integrates current knowledge on empathy in psychology, social neuroscience, philosophy of mind, and cognitive narratology, and was developed by drawing on work by Keen (2006, 2013), Coplan (2004, 2011a, 2011b), De Vignemont and Jacob (2012), Gallagher (2012), Caracciolo (2014), and Cuff et al. (2016).
This study addresses a series of claims that have been made in the literature, mainly by narrative theorists and literary scholars, about the potential of certain textual devices to influence readers’ empathetic engagement with characters (see Keen, 2006). It has been suggested in the literature that particular narrative techniques (such as certain modes of narration, point of view presentation, and characterisation techniques) have the potential to facilitate and/or block readers’ empathy with characters (see Section 2.1). Even though there have been attempts to identify linguistic features that might trigger empathetic effects (Wales, 2011: 133), ‘we know relatively little about the textual strategies that can encourage recipients to empathise with a character’ (Caracciolo, 2013: para. 15). Given this scarcity of knowledge, some scholars have called for further empirical work (see Keen, 2006; László and Smogyvári, 2008; Sklar, 2009; Van Lissa et al., 2016). My study adds to the little empirical research and evidence available so far on the connection between textual devices and empathetic responses.
2.1. Earlier work on narrative empathy
Previous research, both theoretical and empirical, has considered the links between textual devices and (non-)empathetic responses. In what follows, I present earlier approaches to narrative empathy. It is worth highlighting that this review of the literature focuses only on the work by scholars who use the term empathy explicitly. Other scholars might refer to similar effects by using terms such as sympathy, identification, or closeness.
Regarding theoretical work, one of the key contributions to the area of narrative empathy is work by narratologist Keen (2006, 2007, 2013) from the perspective of rhetorical narratology. A variety of narrative techniques (what Keen calls empathetic narrative techniques) have been associated with empathy, most of which are included under ‘characterisation techniques’ or under ‘narrative situation’, the latter including point of view, mode of narration, and access to characters’ consciousness (for a full account, see Keen (2006: 215–220)). However, Keen highlights that no narrative technique per se has yet been proven to facilitate readers’ empathy, and she argues that narrative techniques work alongside many other variables. Thus, Keen concludes that the empathy potential of these storytelling devices has yet to be confirmed.
As far as empirical work is concerned, four earlier major studies have considered the role of certain textual phenomena in eliciting empathetic effects in readers. The term empirical is used to refer to work that collects and analyses extra-textual data on readers’ responses (Whiteley and Canning, 2017).
László and Smogyvári (2008) investigated whether narrative empathy may be influenced by the relationship between readers’ and characters’ group identity. They used two versions of a Hungarian short story in which the group identity of the characters differed. Half of the participants read the ‘Hungarian’ version, whereas the other half read the ‘Slovak’ version. The results did not match the authors’ expectations, as no significant differences were found between the two versions regarding empathy with characters and readers’ national identification, but a correlation was found between narrative empathy and liking.
Van Lissa et al. (2016) investigated whether first-person narration has more potential to elicit empathy and trust than third-person narration. Their participants read the first chapter of Hunger, by Knut Hamsun, which deals with a first-person narrator who might invite ambivalent ethical evaluations and thus might challenge empathetic responses. Two versions of the story were used: first-person (possibly unreliable) narration and third-person narration with internal focalisation. Contrary to their predictions, they found that narrative perspective did not affect empathy for the protagonist. However, the authors found that the third-person narration increased trust for the protagonist, which they link to narrative (un)reliability.
Kuzmičová et al. (2017) explored the supposed link between empathy and literariness. Previous findings had suggested that literary fiction fosters empathy to a greater extent than both non-fiction and popular fiction. They manipulated the degree of foregrounding (as a distinctive characteristic of literary texts) and used two versions of Mansfield’s short story The Fly: one version was the original story, rich in foregrounding, whereas in the other version foregrounding had been reduced. Contrary to their hypothesis, they found that the non-literary version elicited more empathetic responses than the literary version, which the authors link to an aesthetically distanced reading.
A final relevant study is Laffer’s (2016) research on empathy dynamics in discussions of migrants, especially asylum seekers, with a focus on the novel The Other Hand by Chris Cleave. The study analyses the use of metaphor by the author during the interview, in the text itself, and by reading group participants, where metaphor is examined as enabling empathy across social boundaries. Although the focus of the study is on the potential of empathy to influence attitudes towards other social groups, part of the analysis and findings are relevant to research on the connection between textual factors and readers’ empathy. For instance, metaphors in the novel counter conventional representations of asylum seekers and encourage empathetic understanding of the Other. In Laffer’s words, ‘a particularly powerful affordance of metaphor’ is that ‘it can instigate an imaginative leap to understand the subjective reality of the Other, and so enables both the connection between the Self and Other, but also the recognition of the separation in experience and understanding’ (Laffer, 2016: 302).
2.2. Potential factors affecting empathetic responses
The two elements involved in the phenomenon of narrative empathy are readers as recipients of stories (i.e. potential empathisers) and characters as the result of textual linguistic choices (i.e. potential targets of empathy). Both of them are associated in the literature with a number of factors that could modulate the experience of narrative empathy.
Regarding the textual dimension of narrative empathy, several textual cues are considered in the literature as having the potential to influence experiences of narrative empathy. Section 2.1 mentioned some of the narrative techniques that are commonly associated with empathy effects (see Keen (2006: 215–220) for a full account).
Regarding the reader dimension, empathy researchers have pinpointed some factors that might modulate empathy, such as dispositional empathy and some control mechanisms. Empathy results from the interaction between trait/dispositional and state/contextual factors (Cuff et al., 2016: 6). The trait view holds that individuals differ in their empathic skills (Cuff et al., 2016: 6). State/contextual factors are to do with the context or the situation in which the target is situated. Thus, empathy may result from automatic (bottom-up) or controlled (top-down) processes (Cuff et al., 2016: 149–150). Recent neuroscientific findings suggest that a few top-down processes are responsible for regulating empathetic responses, such as contextual appraisal or the perspective adopted (Singer and Lamm, 2009: 89).
Contextual appraisal, or the knowledge about and assessment of the context in which the target’s experiences occur, is a form of top-down cognitive appraisal that can regulate empathy (Lamm et al., 2007: 56). Empathy depends on our understanding of the other person’s situation, of how their actions and mental states are situated in that context3 (Gallagher, 2012: 377; Sanford and Emmott, 2012: 209). In short, empathy requires the observer to ‘think and feel how it is to be someone else (...) in the rich complexity of that other person’s experience, from their perspective, in the situation’ (Cameron, 2013: 6).
Moral evaluation, as part of contextual appraisal, is emphasised in the empathy literature as an important control mechanism. There is ample agreement that empathy seems to be morally sensitive (Zillmann, 2006) and can be inhibited or suppressed in certain moral situations. Target perception and evaluation, including (un)fairness perception and blame allocation, are relevant aspects that fall within the contextual appraisal of a situation (Cuff et al., 2016; Singer and Lamm, 2009; Singer et al., 2006). In short, moral positions adopted towards others can render them as ‘undeserving of attention or perceived as morally repugnant’, and so the possibility of automatic empathy is closed down (Bandura, 2002, as cited in Cameron (2013: 15)). Crucially, moral evaluation can work as a control mechanism through which readers make ‘conscious efforts to selectively understand or distance [themselves] emotionally’ from characters (Breithaupt, 2012: 86).
3. Methodological approach
The study adopted an empirical stylistic approach to narrative empathy that combined stylistic-narratological analysis of some stories and thematic analysis of readers’ responses. Both datasets were collected and analysed in Spanish.
3.1. Stylistic-narratological textual analysis of Galeano’s stories
The stories under analysis were written by Uruguayan author Eduardo Galeano, and belong to The Book of Embraces (1991, 2002). These stories have been called ‘narraticules’ (Olsen, 2004), ‘fragments’ (González, 1998), and ‘vignettes’ (Lovell, 2004). The two biographical stories discussed in this article, which are whole texts rather than extracts, have a non-fictional status and are based on actual historical events. The study considers empathy in relation to perpetrators and victims in narratives of persecution and torture. For the purposes of analysis, characters are positioned into the character slots of ‘perpetrators’ and ‘victims’ due to their dramatic role in the stories (in line with the traditional character typology developed by Propp (1968)). Dictators and colonisers carry out particular actions (i.e. persecuting and torturing dissidents) to pursue their political goals, while the victim characters are at the receiving end of these actions. The theme of the stories is representative not only of Galeano’s writings but also of the wider literary discourse of resistance of Latin American writers against political oppression (Arango-Ramos, 2010). Importantly, the ideological slant presented in the stories is likely to shape the type of responses being elicited: the resulting (non-)empathetic responses to characters might be closely tied to readers’ moral and sociopolitical evaluations. This is why the analysis considers ideological point of view as a relevant textual factor, and contextual appraisal and moral evaluation as relevant reader factors.
The focus was on verbally mediated empathy (Hoffman, 2000: 49), where language becomes the only cue about characters’ experience. Specifically, I analysed the empathy potential of several narrative techniques that are commonly associated with empathy effects in the literature. These were examined through the following linguistically based analytical frameworks: Point of view presentation, including narration and focalisation: I analysed the perspective adopted in the telling of the stories, especially psychological viewpoint and ideological viewpoint, through Fowler’s (1996) and Short’s (1996) analytical frameworks. Characters’ discourse presentation: I analysed the techniques for presenting characters’ speech and thought, as well as their potential effects on readers’ interpretation, through Leech and Short’s (2007) analytical framework. Characters’ emotion presentation: I analysed the ways in which characters’ emotional states are conveyed, with different degrees of explicitness, through Langlotz’s (2017) framework. Characterisation techniques: I analysed textual cues that can shape readers’ perception of characters, or characterisation triggers, through Culpeper’s (2001) framework.
Using these analytical frameworks enabled me to add linguistic descriptive detail, precision, and systematicity in order to conduct an investigation of potential textual effects on readers which is stylistically informed (Kuzmičová et al., 2017) or stylistically aware (Whiteley and Canning, 2017).
The aim of this linguistic-stylistic analysis was to describe the empathy potential of the linguistic choices and patterns in the stories. In what follows, I summarise the stylistic analysis of the two stories with a focus on the stylistic features that are most relevant to the points that are discussed below in relation to the focus group responses. Some common features characterise the two stories. Firstly, the victim characters go through extremely distressing events and experiences (i.e. loss of family and torture). This is likely to bias the reader towards reactions such as empathy or sympathy. Secondly, the mode of narration is first-person, extradiegetic narration. The two stories involve the same first-person narratorial voice that seems to be closely aligned with the author.
3.1.1. Story 1: Gelman
Gelman is about the Argentinian poet and journalist Juan Gelman, whose relatives were kidnapped by the military during the dictatorial regime in Argentina (1976–1983). The story invites readers to interpret a causal connection: as a result of his political activism, he was exiled and members of his family were kidnapped, tortured, and/or murdered. The story captures these events together with the narrator’s speculation of what it must have been like for Gelman to undergo the events.
Point of view presentation
The narrator is the holder of point of view and occupies the role of the focaliser, whereas Gelman is positioned in the role of the focalised (i.e. the object of observation; Bal, 1997). The story presents external focalisation because the narrator-focaliser is outside the story being narrated, and so is not an experiencing character. In terms of Fowler's (1996) categories of psychological point of view, the story can be categorised as external narration type D. The person of the narration is foregrounded through the use of first-person pronouns. The person of the narration is also foregrounded through evaluative statements (e.g. ‘the Argentine military, whose atrocities would have given Hitler an incurable inferiority complex, hit him where it hurt the most’). Finally, the narrator seems to have limited knowledge about Gelman’s inner states (e.g. ‘I’ve often imagined that horrible feeling of having one’s life usurped’). Regarding ideological point of view, the story mediates anti-dictatorial attitudes that are embodied in the negative evaluation of the Argentine military, with which readers can establish different positions based on their own sociopolitical values.
Emotion presentation
The story presents what the narrator believes Gelman’s emotions are likely to be, and so the narrator attributes internal states to the character. Despite not being the character’s first-hand experiences, these narratorial speculations could potentially elicit readers’ empathy with Gelman because of the very concrete and vivid emotional picture provided.
The story conveys emotion mostly by means of metaphor, so inferences need to be drawn to interpret Gelman’s internal states. These inferred internal states include being emotionally devastated but resilient, worried, anguished, and possibly guilty. Emotions are also presented explicitly, for example through the mental disposition term (Langlotz, 2017) ‘where it hurt the most’, which indicates extreme, unbearable pain as a result of having his family harmed and taken away. In contrast, a total absence of emotional information about the Argentinian military may hinder empathy with this group of characters.
Characterisation techniques
The actions performed by the different characters work towards characterising them. Firstly, characters’ actions may automatically position them into the character roles of victims and perpetrators. That in itself might achieve, respectively, closeness and distancing effects. Moreover, readers are likely to attribute different goals and ideological values to characters based on their situation and actions. Readers’ contextual appraisal of the story-world situation may trigger moral and sociopolitical evaluations, all of which might influence readers’ responses towards the different characters.
In terms of naming, the characters behind ‘the Argentine military’ remain as a group throughout the story since they are only mentioned once. Cameron’s (2013) notion of lumping is very suitable here since this name category lumps all members of this group into a homogenous group. Apart from the choice of naming, the fact that very little information is given about them might also contribute to this lumping effect, which according to Cameron has distancing effects and may thus potentially hinder empathy.
3.1.2. Story 2: Professional Life
Professional Life tells the story of an unnamed torturer, who works for the French colonisers of Algeria, and Ahmadou Gherab, who fought in the National Liberation Front against France during the Algerian War of Independence (1954–1962).
Point of view presentation
The story can be labelled as internal narration type B (Fowler, 1996), where the narrator is not a story character but knows about characters’ internal states. By adopting an internal psychological viewpoint, the story grants access to the feelings of the two characters; however, the story seems to favour the Torturer’s viewpoint to a greater extent. Most of the story deals with the Torturer, and his speech and mental processes are portrayed (rather than Ahmadou’s). The access to the inner life of the Torturer may, together with the cumulative effect of other textual devices, enable readers momentarily to see what the world is like from the perspective of a torturer. The narrator is an omniscient-like narrator. Omniscient narration is regarded as facilitating readers’ empathy only in so far as it provides access to the inner world of the characters (Keen, 2006: 219).
The ideological point of view in this story is much more implicit than in Story 1 as it does not contain any explicit narratorial commentary. The denunciation of abuses in (post-)colonial contexts is one of the themes that Galeano has written widely about, and this can be said to be the implicit ideological viewpoint adopted in the story. Thus, the sociopolitical values that can be activated in recipients’ interpretations revolve around colonisation and the struggle for independence.
Discourse presentation
The story conveys mostly the Torturer’s speech, which is presented (following Leech and Short’s (2007) categories) in Direct Speech, Narrative Report of Speech Acts, and Free Indirect Speech (FIS). Stylisticians have discussed the potential effects of the different speech presentation categories in relation to degrees of narratorial interference and impressions of closeness or distancing from the characters’ original words. Stylisticians’ claims about the role of speech presentation categories should be incorporated into debates concerning potential empathy triggers. Scholars such as Adamson (2001) and Keen (2006) suggest that FIS can be taken to be an empathy trigger and is regarded as ‘the most likely to produce empathy’ (220). In contrast to the talkative torturer, the tortured character Ahmadou ‘would say nothing’. This absence of voice becomes meaningful since it might convey emotion implicitly (see below).
Regarding thought presentation, only the torturer’s thoughts are presented in the story in the form of Narration of Internal States (Semino and Short, 2004) and Free Indirect Thought (FIT). Regarding the latter, FIT is seen as a closeness device because it places us ‘inside the character’s mind’ (Leech and Short, 2007: 276).
Emotion presentation
As with point of view and discourse presentation, emotion presentation in the story is mostly to do with the Torturer. Mental states are presented implicitly and have to be inferred from the character’s speech (likely interpretations being emotional conflict and frustration). The Torturer’s emotional states are also presented more explicitly through the use of mental disposition terms (Langlotz, 2017), such as ‘astonishment’ and ‘feeling a certain sense of injustice’.
Ahmadou’s emotional states are presented implicitly and explicitly in the last sentence of the story: ‘Ahmadou, bathed in blood, trembling with pain, burning with fever, would say nothing’. Readers are likely to infer extreme physical discomfort out of the pain, wounds, and fever together with feelings of powerlessness. ‘Trembling with pain’ is an explicit behavioural surge term (Bednarek, 2008) since this physical behaviour directly expresses the sensation of pain. Regarding the empathy potential of emotion presentation, ‘direct description of a character’s emotional state or circumstances (…) may produce empathy in readers just as effectively as indirect implication of emotional states through actions and context’ (Keen, 2006: 218, my emphasis). In this case, readers might infer Ahmadou’s likely mental states from the physical description and from the context of torture.
Characterisation techniques
Concerning characters’ actions, the fact that the story presents a character’s torture and pain could be an empathy-eliciting device from the very outset. The fact that Ahmadou is being subjected to torture is likely to trigger moral reactions and offer room for readers’ empathy with Ahmadou and a lack of empathy with the Torturer. The Torturer might be seen as morally unscrupulous and blameworthy, potentially resulting in distancing effects.
Regarding naming, there are contrasting choices of naming for the two characters. The victim is given a proper name, including both first name and surname (‘Ahmadou Gherab’), whereas the Torturer is only referred to via noun phrases related to his professional role: ‘executioner’, ‘torturer’, and ‘French official’. Arguably, these naming strategies may have different effects: closeness towards Ahmadou and distancing towards the Torturer.
3.2. Thematic analysis of readers’ responses gathered through focus group discussions
Reader response research in stylistics involves the collection and analysis of extra-textual data that ‘capture aspects of readers’ behaviours, interpretations or evaluations in response to particular literary works (and in specific contexts)’ (Whiteley and Canning, 2017: 72). Due to the interdisciplinary nature of the subject, there are many contributions4 to empirical reader response research (for an overview, see Hakemulder and Van Peer (2016: 192)). On the one hand, empirical reader response studies have focused on the influence of formal features and devices on readers’ responses (Hakemulder and Van Peer, 2016: 190; Miall, 2006: 293). In this respect, research topics range from iconicity, non-literal language, and narrative perspective, to deviation (Hakemulder and Van Peer, 2016: 193–202). On the other hand, a different strand of research has looked into ways in which readers’ sense of self is affected by reading (Miall, 2006: 293), such as the impact of reading experiences on readers’ memory and self-concept (Miall, 2006: 292), the effects of experiencing empathy while reading on the development of empathic skills and prosocial behaviour (see Burke et al. (2016) for a review of different studies), or the potential of empathy to shift attitudes towards other social groups (Laffer, 2016). The present article contributes to the first set of studies; that is, those that aim to investigate the influence of textual devices on readers in the process of reading.
Methodologically, I argue for the value of a fully qualitative research approach in the collection and analysis of reader response data. Firstly, qualitative methods allow researchers to investigate their object of study from the perspective of individual participants (Hennink et al., 2011: 10), thus enabling insights into what is meaningful to readers. Participants’ subjective empathetic responses are often absent in previous quantitative studies (Kuzmičová et al., 2017: 141). Secondly, there is great value in the in-depth qualitative analysis of responses, as it allowed me to gain insights from the detail of participants’ responses, to map out the complex interplay between textual and reader factors, and to understand how responses relate to claims from the literature.
Within the available qualitative methods, the focus group method was used to collect readers’ responses. Other qualitative studies that address readers’ responses and attitudes to characters have examined responses within reading groups (Laffer, 2016; Swann and Allington, 2009; see also Myers, 2009), discussion groups (Whiteley, 2011), and online book reviews (Nuttall, 2015a, 2015b). In this study, two focus group discussions were conducted with nine Spanish readers: one discussion had five participants, whereas the other discussion had four participants. Participants were recruited through social networks and notices in public places (i.e. libraries and cafés). The discussions took place outside an academic context, which contrasts with the common practice of recruiting students enrolled at the researcher’s institution (Swann and Allington, 2009: 248). Regarding demographic information, participants were between 20 and 60 years old, mostly female with one male in each group, and most born in Spain with one born in Venezuela. In terms of education, there was a mixture in terms of highest qualification5 obtained.
The stories were read in this sequence in both groups: (1) Gelman and (2) Professional Life. Participants read one story, discussed the questions, and then read and discussed the next story. Questions focused on their general response to the text, their perception of and feelings towards characters, their moral reaction, etc. Importantly, questions (and the information given to participants in advance) did not contain any references to empathy so that participants would not be prompted to considering empathy as a response when they might not have considered it spontaneously.
The focus group method offers several advantages. Small groups maximise opportunities to talk and are suitable to explore complex topics when detailed accounts are sought (Litosseliti, 2003: 3). Additionally, the method is well suited to allow for open-ended discussions that enable more nuanced and spontaneous self-expression with, arguably, less researcher control than other methods such as written questionnaires and closed-question ratings (Whiteley and Canning, 2017: 74). As for potential disadvantages, readers’ responses can be subject to issues of social negotiation and group dynamics such as politeness (Nuttall, 2015b: 19), which might result in social desirability bias and self-censorship (Lang, 2009: 325). Finally, this method captures participants’ discussion of their experiences after reading rather than their experience while reading, which may result in forgetfulness and retrospective rationalisation (Short and Van Peer, 1989; Stockwell, 2016).
My methodology is innovative when compared with earlier empirical studies of narrative empathy, as it sits between naturalistic and experimental paradigms: it is a half-way house between the more naturalistic reading group experience and the more experimental laboratory-like conditions, in relation to the characteristics spelled out by Swann and Allington (2009: 248). The study was experimental-like, in that it explored empathy as a pre-specified aspect of participants’ interpretation, and the groups were prompted rather than naturally occurring. The study was naturalistic-like, in that Galeano’s stories were presented whole and unmodified, and participants interacted face-to-face among themselves mostly (with a low degree of moderator control over the discussion).
Finally, thematic analysis was conducted with the aid of qualitative analysis software (Atlas.ti) to examine readers’ responses. Atlas.ti enables thematic analysis, with themes being ‘recurrent and distinctive features of participants’ accounts, characterising particular perceptions and/or experiences, which the researcher sees as relevant to the research questions’ (King and Horrocks, 2010: 150). Thematic analysis is carried out through the development of a coding scheme and the subsequent tagging of relevant data extracts. Codes are interpretive categories that are relevant to research questions, and so coding refers to the process of assigning labels or codes to segments of information that are relevant to the analyst’s questions (Friese, 2014: 7, 24). Regarding the coding process, Atlas.ti enabled the analysis to be consistent across the entire dataset, for example through the ease of renaming and modifying codes and coded segments in the process of refining the code list (Friese, 2014: 19). Moreover, the software also enabled the possibility of easily retrieving information according to different criteria.
I related insights from my analysis of the stories (and associated claims from the literature about potential textual effects) to readers’ self-reported engagement with characters, and I examined the textual and reader factors that seem to be involved in such responses. The coding scheme below contains both deductively and inductively developed codes. For instance, the umbrella category to do with attribution processes was derived deductively as these ideas originated in the academic literature. However, the subcodes for that category were developed inductively since they emerged from the data. The codes that were applied to the data were the following (N.B. only those that are relevant to the findings presented below are listed).
Readers attributed different aspects to characters:
Readers make comparisons between the real world and the story-world, and frame those situations and experiences as being different (
Readers refer to or articulate their interpretation regarding the following:
4. Analysis and findings
In what follows, I report on the patterns in participants’ engagement with characters (see Section 4.1). After that, I present the main findings of the study; namely, empathy tends to happen when information about characters’ situation and internal states is available (see Section 4.2), with the caveat that a negative moral evaluation can override the textual potential for empathy (see Section 4.3). (For extracts of the focus group discussions in both Spanish and English, see Fernandez-Quintanilla (2018)).
4.1. Patterns of (non-)empathetic engagement with characters
Certain empathy patterns emerged from participants’ responses to characters in the group discussions.
4.1.1. Empathetic responses to characters
Regarding victim characters, all participants positioned themselves in favour of the victims and displayed responses such as empathy, sympathy, and solidarity. Evidence of empathy was found for example in the following extract (the code (1) I think that in today’s society, the word ‘dictatorship’ readily provokes an automatic moral rejection, and it automatically makes you
Regarding perpetrator characters, rather unexpectedly many participants (six of nine) displayed empathy with the Torturer (Story 2) despite his perpetrator status. Empathy with this character was found in the fact that participants adopted the Torturer’s perspective; attributed to him specific aspects of situation, goals, and values; and perceived a similarity of experience, as I discuss in turn below. Firstly, participants adopted the character’s perspective and viewed the story-world situation from his viewpoint (codes (2) Many times you say ‘it’s not justifiable, I wouldn’t be able to do it’. But the thing is you also have to put yourself in the other situation because at the end of the day, it’s what it says, ‘he and he are the same person but not the same person’.
Secondly, participants attributed to him a certain situation, goals, and values that facilitated an understanding of the character. Participants’ comments are paraphrased for brevity [codes Situational constraints and having no choice: he is forced to do that job to survive; he disagrees with it and is unwilling to do it. Goals and needs: earning a salary and feeding his family (wife and newborn child) are his priority; he is in such a dilemma that whatever he decides, he is in trouble–either moral questioning or not feeding his family; he is waiting for a promotion. Values and beliefs: background of having been raised in a colonial mindset during the French colonial period.
An example of attribution of goals to the character is the extract below [code (3) It can also be seen that he’s there temporarily because
Finally, participants perceived a strong similarity between the Torturer’s situation and situations in the real world, and this similarity of experience seems to have facilitated empathy. Empathy with the Torturer (i.e. participants’ ability to imagine what his situation must be like for him, from his perspective) may have been facilitated by inferences that resulted from participants’ real-world knowledge and experiences: social situation and everyday issues in Spain in 2015, such as the situation of people who have to put up with exploitative working conditions to get a salary and feed their family, and people who do have a sense of ethics and morality but have no other choice and end up doing things they do not want to do. Participants displayed a sense of shared experience (through the discussion of specific examples of their own environment that were perceived to be similar to the character’s situation) and considered that his context forces him to perform this torturer job. The conversation turns below illustrate this [codes (4) C: Yes, well, I also see it as this situation nowadays of jobs that pay pittance, that you know you’re exploited and you tell yourself and your employer ‘if you don’t take it there are 15,000 others behind you who would take it for that money’ (…) your family depends on whether you do it or not (…) B: But it is true, there are certain things that if I don’t do them another person will do it and it’s that sad (…) B: Another thing is that we put up with it because ‘otherwise I will be fired’, that is a different story (…) C: Also because they put you in such a dilemma that whatever you decide you’re in trouble. If this man decided to do it maybe it means a personal questioning, and if he decides not to do it he cannot feed his family (…) A: It’s the same as judicial staff who evict people from their houses, kick people out of their houses, often in television and in other many places they have even said that they ended up vomiting because of the helplessness of what they have to do, because they are forced to do it and if they don’t do it someone else will.
4.1.2. Non-empathetic responses to characters
As expected, no evidence of empathy with the perpetrators in Story 1 (i.e. the Argentinian military) was found in participants’ discussions. Regarding these characters, participants only articulated distancing, strongly negative evaluations of a moral and sociopolitical nature. Participants’ comments are paraphrased for brevity [codes Military establishment related to values of rigidity, religiosity, and not allowing people to think for themselves Against any kind of dictatorship, whatever the political leanings Against what goes against freedom or harms any person Anything that limits freedom of speech is wrong
Regarding the Torturer, two participants resisted and distanced themselves from the character’s behaviour. They strongly disassociated themselves from the character on moral grounds because, according to them, there is no justification for what he does, as illustrated below [code
(5) I think he shows the meanness of human beings. There are human beings, (6) It looks to me more like
4.2. Empathy tends to happen when information about characters’ situation and internal states is available, with a caveat
In my data, readers’ self-reported empathy with characters tends to happen when the stories allow access to the characters’ situation and mental states (i.e. emotions, thoughts, beliefs, intentions, goals, and values), independently of whether actual or hypothetical, whether explicit or implicit, and whether given through internal or external focalisation (i.e. whether it is the characters’ internal perspective or the authorial narrator’s perspective). Thus, the key was that this information is made available.
For example in extract 7, the participant shows that she was acutely aware of the role of textual information in enabling her to grasp Gelman’s (Story 1) experience [code (7) Here
In extract 8, participants agree that the Torturer is humanised because his thoughts are conveyed and also because we get to know that he is being forced to do his job [codes (8) B: I go back to the same thing, words are very important (…) not giving a name to the torturer who at the end of the day is a person, it marks him as ‘the torturer’, the one who tortures (…) when in fact he is performing an action that [originates somewhere else], it doesn’t come from him. C: It depersonalises him by not giving him an identity but B: It seems to me that it depersonalises him, I mean it tries to depersonalise it but C: A: It humanises him, that’s right.
Participants’ empathy with the Torturer can be accounted for in terms of the particular combination of narrative techniques that shape the Torturer; namely, a strong focus on his perspective, availability of information about his internal states, and presentation of his emotions, speech, and thought (see findings from textual analysis in Section 3.1.2). It can be argued that the combination and cumulative effect of these narrative techniques facilitated empathy with the Torturer. While there are clear victims in the two stories and almost absent perpetrators in Story 1, the Torturer has a more complex status in between perpetrator and victim. To some extent, he has victim characteristics since the text implicitly suggests frustration, emotional conflict, and moral struggle in carrying out this job, and explicitly mentions his need to feed his family. The cumulative effect of the narrative devices associated to the Torturer seems to have facilitated an empathetic understanding of what it is like to be in the position of a professional torturer, and indeed, he was seen as both a victim and product of his environment.
4.3. A negative moral evaluation can override the textual potential for empathy
An important caveat to consider is that access to characters’ internal states did not always bring about empathy. An important finding was that moral evaluation and positioning played a determining role in narrative empathy. Thus, the availability of information about characters’ mental states tends to facilitate empathy, but a negative moral evaluation can override the textual potential for empathy.
The conflict in the stories (i.e. the clash between the goals and actions of victim characters and perpetrator characters) resulted in different positionings in relation to the characters. These positionings were strongly linked to readers’ moral evaluation regarding the actions of perpetrators and the distressing experiences inflicted upon the victim characters, with the exception of the Torturer.
Regarding the victims, participants displayed a favourable moral position towards these characters, and morally oriented reasons were given for their positioning on the side of the victims. For instance, participants spelled out their moral values and reported being in favour of the weak and the oppressed; being against oppression and anything that takes away people’s rights and freedom; and that killing and harming others are never justifiable [codes
Regarding the perpetrators, participants reported a straightforward rejection of these characters based on their moral values, which participants articulated as follows: being against evil and cruelty, any type of oppression, anything that harms people and forbids freedom, and any kind of dictatorship, whatever the political leanings. Their comments are paraphrased for brevity [codes There is no limit to the atrocities committed by humanity against humanity. Dictatorships provoke an automatic moral rejection and one automatically feels empathy for the people suffering under the dictatorship. There are no limits to human beings’ evilness. It is so cruel to take Gelman’s children away. Taking Gelman’s children away is cruel and reaches such a level of refined torture.
Crucially, the six participants who reported empathy with the Torturer did not display any blaming or negative moral evaluation of the character. Even though none of them agreed with the practice of torture, none of them showed any moral condemnation of the character. This indicates that it is possible for textual devices to facilitate empathy with morally controversial characters in so far as readers’ moral evaluation is withheld.
In contrast, in some cases, the presence of readers’ negative moral evaluation seems to override the empathy potential of narrative devices, as in the case of two readers in relation to the Torturer (N.B. these two readers contributed the comments in extracts 5 and 6 above). Evidence was found of a distancing, non-empathetic engagement between these two participants and the Torturer, as seen in their morally condemning judgements of the character. They judged his behaviour as unfair and morally wrong, they often blamed the character for not refusing to perform the job, and throughout the discussion they referred to him as a cruel, sadistic, and immoral figure. These two extracts illustrate their perceptions of the character [code (9) I think he’s not trying to justify himself at any point, and that’s really what is most scary, because if at least he showed some (10) So the
I argue that moral evaluation was the most determining factor in these two participants’ rejection and lack of empathy with the character. Their negative moral evaluation of the character seems to have outweighed the potential for empathy of the story.
In conclusion, readers’ moral evaluation of and positioning towards characters played a determining role in whether or not empathy occurred. In fact, negative moral evaluation can override the potential for empathy of narrative devices (i.e. availability of information about characters’ situation and internal states). Thus, readers’ moral evaluation (as part of contextual appraisal) was an important top-down modulator of empathy in my participants’ discussions. This finding is consistent with the scholarly views discussed above (see Section 2.2) that empathy is morally sensitive.
5. Discussion
Although this small-scale study does not claim generalisability, my findings support evidence from previous observations in the literature in two respects: textual factors do not work in isolation and the role of reader factors should be accounted for and further researched.
My findings indicate that the influence of textual cues on narrative empathy is not as straightforward as scholarly claims seem to suggest. This points to the need for a more nuanced approach to the question of textual factors in narrative empathy. Van Lissa et al. (2016) conclude that the evidence from their study ‘does question assumptions about the direct effects of textual strategies on narrative empathy’ and that ‘literary scholars tend to overestimate the effects of textual cues on readers’ responses’ (p. 59). My study challenges claims about direct textual effects on readers and warns against generalisations. Likewise, Keen (2013) highlights that ‘caution should be taken not to oversimplify predictions about the effects of particular narrative techniques’ (para. 8). As she convincingly argues, ‘the commentary on narrative form often asserts (or assumes) that a specific technique inevitably results in particular effects (…) in readers. These views (…) should be subjected to careful empirical testing before any aspect of narrative technique earns the label of “empathetic”’ (Keen, 2006: 225). I fully agree with Keen’s and Van Lissa et al.’s appeal for caution when it comes to making claims about the effects of narrative techniques on reader experiences, since extra-textual factors are an essential part of the picture and should not be disregarded.
As to the question of the interplay between textual and reader factors, even though some claims in the literature are only concerned with the role of textual factors in determining certain reader responses, it seems crucial to broaden the scope to include non-textual reader factors. The theoretical position underpinning this study establishes that reader experiences are the result of the dynamic interaction between incoming textual information and the reader’s prior knowledge and experiences (see Caracciolo, 2014; West, 2016).
The interplay between textual and reader factors needs further research. My findings indicate that neither the textual nor the reader dimension works in isolation. I argue that both are equally relevant to character engagement in general and experiences of empathy in particular. Some of my findings might even suggest that in some contexts, reader factors can override the empathy potential of textual devices, as in the case of a negative moral evaluation (see Section 4.3). Thus, extra-textual reader factors alone might account for differences in reader responses: ‘as empirical research in discourse processing reveals, individual readers respond variously to narrative texts depending on their identities, situations, experiences, and temperaments’ (Keen, 2011, as cited in Keen (2013: para. 7)).
In this respect, an important implication of my study is the variability of context and, in turn, the variability of reader responses. The complexity of the reading context needs to be taken into account, and so more tentativeness is needed when making claims about one-to-one relationships between textual devices and effects on readers. Swann and Allington (2009: 250) point out that ‘readers’ interpretational activity is contingent upon aspects of the contexts in which they read’, and so the context of elicitation should be acknowledged and seen as relevant to the resulting interpretations (p. 262). The theory of narrative empathy would benefit from a more nuanced, tentative approach to the question of textual effects on readers. In accordance with the views presented above and the background literature consulted, my study further supports the idea that narrative empathy is a highly flexible and context-dependent phenomenon. Accordingly, more context sensitivity is needed in narrative empathy research, given the complex interaction between textual and reader factors.
It is also worth highlighting that, regarding the role of textual factors alone, my findings open up the possibility that textual factors may work in a more organic way than has been previously acknowledged. My analysis indicates that the different narrative devices are likely to have worked cumulatively to produce those responses in participants (see Section 4.2). An important implication is that it might be limiting for analysts to conceive of potential empathy triggers and barriers as comprising lists of separate factors that somehow work in isolation.
Finally, regarding methodology, my project reflects the increasingly widespread ‘impulse to collect extra-textual data about literary reading in order to inform, develop and reflect upon stylistic analysis’ (Whiteley and Canning, 2017: 72). I have argued that qualitative methods have plenty to offer in reader response research and that much can be gained from combining naturalistic and experimental orientations. I have shown the usefulness of the focus group method to collect responses in the form of open-ended discussion. Moreover, thematic analysis of the data enables plenty of detail in the responses that can add vital nuances regarding the complex interaction between factors.
6. Conclusion
The study presented in this article contributes to the growing body of knowledge on the phenomenon of narrative empathy from a particular angle; namely, the factors that can influence experiences of empathy with characters in narrative. My findings suggest that textual factors do not work in isolation, and so they challenge scholarly claims about potential direct effects of narrative techniques. It has argued that both textual and reader factors play a relevant role in bringing about different perceptions of and responses to characters, and that claims about empathy effects should accommodate the role of context. It has discussed the importance of readers’ moral evaluation and positioning as a key reader factor in our empathetic engagement with characters.
To conclude, plenty remains unsolved regarding the role of, and interplay between, textual and reader factors in (non-)empathetic responses to characters in narrative. Empathy being highly flexible and context-dependent, accounts of narrative empathy need to allow for the variability of texts and characters as well as readers and reading contexts. More evidence in the form of what real readers report is needed to (dis)confirm the plausibility of hypotheses about the empathy potential of narrative devices, and so refine and develop our understanding of how narrative empathy works.
Footnotes
Acknowledgements
I would like to thank Professor Elena Semino for her helpful comments on an earlier version of this manuscript. I would also like to thank the two anonymous reviewers for their thorough and valuable feedback.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was partially enabled by Funds for Women Graduates (British Federation of Women Graduates Charitable Foundation) (grant number 167017).
