Reflections on internet research ethics from language-focused research on web-based mourning: revisiting the private/public distinction as a language ideology of differentiation

2.1 Approaches to ethics in death online studies and applied linguistics

In studies of death-related practices online there are notable differences in the degree of reflexivity and systematic attention paid to matters of ethics, depending on the disciplinary grounding of particular research and the particular focus of the study. For instance, in studies which are grounded in the discipline of psychology and social care researchers view the data they are dealing with as sensitive, in that they feature highly personal expressions of loss; in addition, they recognize participants who at the moment of research are in bereavement as vulnerable persons coping with major changes in their life following the loss of a loved one. To address the particular nature of sensitivity and vulnerability raised by this type of research, additional measures are taken that often include the following:

1. allowing interviewees to self-identify for participation to a survey (Caroll and Landry 2010);

2. contacting the interviewees-bereaved after at least six months from their loss in order to minimize intrusion in the early grief period’ ^[1] (Dyregrov 2004; Dyregrov et al. 2014);

3. selecting interviewers based on their experience and knowledge in the field of bereavement and the conduct of qualitative interviewing (or ensuring they receive training before the conduct of research) (Dyregrov 2004; Dyregrov et al. 2014).

The majority of these measures is relevant to research based on interviews or elicited data also known as obtrusive data. In contrast, research on ‘what people have already said and done rather than what they say they have said and done’ (Hine 2011) often involves the collection of internet derived or ‘found’ data, also known as unobtrusive data. So far, researchers in the field of death online have not reached consensus on specific measures to deal with sensitive groups in unobtrusive data-based research.

In research articles, authors tend to avoid explicit references to ethical concerns altogether or they may refer to ethical principles in passim (e. g. Brubaker and Hayes 2011; Lee and Goh 2013), especially if the research reports on findings from the collection and analysis of large-scale data that have been anonymized before the start of analysis. In some cases, the choice of publicly visible profiles on platforms that do not feature privacy features, as for example in the case of the early days of MySpace, forms part of the initial design of the study, based on the assumption that platforms’ settings and features also reflect participants’ consent to research. In their review article of studies in the field of death online, Walter et al. (2011) note that the majority of research is conducted ‘most easily by going online and observing the sites in which dying people, their carers, and mourners participate’ (italics mine). Easy access does not mean ethical access, however. Walter et al. (ibid) are careful to note that ethical concerns remain, not only in the case of researchers entering password-protected sites but also in the case of researchers accessing open group sites whose postings are public, as it is not always clear how participants may feel knowing that a researcher is observing them.

Given the lack of consensus on ethics issues relating to death-related topics of research and the focus of my ongoing web-based mourning research on language, ethics guidance was drawn from the field of applied linguistics. The BAAL (2006) recommendations for applied linguists, more specifically, which were available at the time of research, offered a useful checklist of important issues for the design and conduct of ethical research that outlines the researcher’s responsibilities to informants, colleagues, students, applied linguistics, sponsors, and the public. The 2006 version also included a section on internet research (Section 2.9) informed by the Association of Internet Researchers (AoIR) ethics recommendations, among others (BAAL 2006: 7); the recently revised version of the BAAL recommendations (BAAL 2016) includes a more extended and nuanced discussion of some of these issues and also provides sets of key questions for researchers to consider in the design and conduct of their studies. Even though the 2006 version does not provide an extended discussion of internet research ethics issues, it acknowledges the challenges raised by research on online discourse, especially with respect to informed consent and confidentiality. The 2006 recommendations foreground, in particular, the importance of considering ‘the venue being researched, any site policy on research and informants’ expectations’ recognizing that ‘in the case of an open-access site, individual consent may not be required’ (BAAL 2006: 7).

Based on the above, one of the first steps in ascertaining ethical risks and concerns relevant to a particular study involves looking at the data policy of the online site of research, in the case of my study discussed below: the social network site of Facebook. Facebook’s data policy (2015) outlines individual users’ responsibilities vis-à-vis the content they share on the site as follows:

When you share and communicate using our Services, you choose the audience who can see what you share. For example, when you post on Facebook, you select the audience for the post, such as a customized group of individuals, all of your Friends, or members of a Group. […]

Public information is any information you share with a public audience, as well as information in your Public Profile, or content you share on a Facebook Page or another public forum.

Public information is available to anyone on or off our Services and can be seen or accessed through online search engines, APIs, and offline media, such as on TV. In some cases, people you share and communicate with may download or re-share this content with others on and off our Services. […]

Based on the above, data can be collected from a public profile page or group on Facebook without seeking informed consent from users, on account of reasonable expectation that users would be aware of the fact that their shared content is in the public domain. In addition, Facebook affords users with privacy settings that can be customized to limit or increase visibility of shared posts, depending on the individual users’ or group administrators’ preference. However, as mentioned above, users and researchers may hold a different view of the nature of the site they frequent that could clash with such assumptions about sharing online, raising ethical conundrums in the decision-making process.

Sections 2.2 and 2.3 illustrate specific ethical tensions arising in language-focused research on web-based mourning and outline the process of decision-making in a case study of public Facebook memorial sites.

2.2 The study

The case study draws on language-focused research on web-based mourning, i. e. the use of social media affordances for individual or collective grieving and remembering loved ones, public figures, or celebrities, which attests to the increased mediatisation of death around the world (Haverinen 2014; Sumiala 2013). The study of web-based mourning has attracted cross-disciplinary scholarly interest in recent years and is part of the rapidly growing field of death online studies (Christensen and Gotved 2015).Web-based mourning can be traced back to online memorials since the 1990s and includes memorial blogs, videos on YouTube, virtual world headstones, flickering virtual candles, memorialised profile pages or memorial groups on social network sites. Across the different types of online memorials, users display their commitment to sustaining the memory of the dead in and through their posting activity, which varies from multimodal types of sharing, such as uploading pictures, poems, and songs to text-based forms, such as sharing memories, stories, and significant everyday moments of life.

The present study involved the qualitative analysis of sharing practices relating to grieving and mourning on a memorial Facebook group site. The main data source was one Facebook memorial group set up by its creators as a public group, as a tribute to a young adult who tragically lost his life in a car accident. The site, which currently counts 1,228 members, was created by six close friends of the deceased, and it was selected after browsing, collecting, and considering twenty-five memorial Facebook groups in total. The study, developed in different stages, furnishes an applied linguistics perspective on death online that includes (i) a discourse analysis approach to the site’s content as sequentially organized wall post units (Giaxoglou 2014a, 2014b), (ii) narrative and positioning analysis (Bamberg and Georgakopoulou 2006; Georgakopoulou 2007) for tapping into the posts’ narrative potential and their links to sharers’ identity construction (Giaxoglou 2015a), and (iii) the use of the empirical framework of sharing (Androutsopoulos 2014) for the analysis of practices of communicating significant moments of mourning with and for networked audiences (Giaxoglou 2014b, 2015b). Each stage and analytic framework addresses slightly different research questions relating to the content, discourse and language patterns of young adults’ practices of mourning as attested on the site and raises slightly different ethical concerns (see Section 2.3).

Taking into account descriptions of online memorials as existing ‘between the gravestone and that teenage bedroom that never gets touched’ (Debatty 2007: 130 cited in Caroll and Landry 2010), it was felt that Rest in Peace (R.I.P.) groups, increasingly popular among digitally literate young adults on Facebook, would form a rich site for exploring young adults’ ways of expressing grieving online, a topic that had been largely neglected in the study of death-related language practices.

Selecting a public social network group site for study, and in particular one that involved interaction between young adolescents around their mourning for a cherished friend, posed from its start a number of important ethical questions, challenges, and conundrums: the research involved a vulnerable social group made up of individuals in bereavement. The study also dealt with a sensitive and to some extent taboo topic, namely death, and it did so in the context of a relatively new site for death-related communication where norms of behavior are just beginning to emerge.

The next section outlines the approach to ethics undertaken in this study in more detail: the decisions made at different stages, their underlying rationale, and some of the vexing questions raised in this type of research.

2.3 The approach to ethics in the case study: an applied ethics perspective

Applying ethics principles to a specific research project is never a straightforward task. In the case of the present study of web-based mourning, initial challenges to the process included the lack of relevant or specific guidance on handling unobtrusive data and potentially clashing views about the data’s sensitive and private nature.

As in any type of research, the first step in this case was to assess the risks posed by the particular type of study. In the field of internet research, studies are situated on a continuum ranging from low to high-risk for human subjects (Page et al. 2014; Buchanan 2011), whose dimensions are summatively presented in the table below.

Looking at the different aspects of the present study in relation to the dimensions of risk presented in Table 1, the study is positioned towards the high-risk end of the continuum with respect to the type of data (‘small’ data) and method (qualitative) and towards the low-risk end of the continuum with respect to the privacy settings (group set as an open, public group) and research focus (i. e. text- and discourse-focused). In other words, the study can be described as a medium-risk study. As such, it arguably offers a productive site for illustrating the implementation of ethics procedures as well as for rendering explicit some of the tensions arising in studies caught up in-between high and low risks.

To address ethical concerns in the case of a medium-risk study, due consideration needs to be assigned to the project’s different aspects and stages throughout its development. This type of approach to ethical decision-making that is dependent on the stage, scale, and focus of the study resonates with the principles of the process approach to research ethics (Markham and Buchanan 2015; Page et al. 2014), which calls for reflection on ethical issues as an integral and dynamic part of the project.

The case study’s different, though interrelated, stages aimed at shedding light on (i) the emerging language norms of sharing grief online, (ii) the key linguistic resources for identity construction of sharers as (networked) mourners, and (iii) the narrative potential of shared posts in web-based memorials. The first stage of the study involved browsing Facebook for memorial Rest in Peace pages using the platform’s search button and ‘lurking’ ^[2] on sites (Scharf 1999 cited in Madge 2007), that is spending time online to get a sense of different types of memorialization on Facebook (e. g. profiles, closed groups, open groups, pages etc.). The second stage involved the selection and analysis of one memorial group site, created in May 2012 and totalling 525 logs at the time of research. Finally, the third stage dealt with the selection and analysis of the posts of a single user shared on the memorial site.

The first stage was preoccupied with the initial ethics design for this study, which included the decision to search for public rather than closed groups in order to minimise risks from ethical and legal concerns relating to access and privacy. During that stage, sites were browsed to help form a general impression of web-based mourning but no analysis of any specific sites was conducted.

The second stage involved a general consideration of patterns of sharing, aiming to bring to the fore key types of posting activity (e. g. images, songs, text-based posts) in relation to emerging linguistic norms, as evident, for example, in the generic structure of typical R.I.P. posts. During that stage, the selection of one group for study required a more careful consideration of matters of access. Approaching the site as a ‘techno-spiritual space’ (Brubaker et al. 2013) implied that the possibility of accessing the site as a member of the group did not seem the optimal option; interference in members’ practices could not only affect the level and mode of members’ participation but also potentially be perceived as a form of inappropriate intrusion. For that reason, the decision was taken not to request access to the group as a member and to avoid any kind of visible intervention in the group’s interaction. For the same reason, the group’s administrators were not approached to request their informed consent, as their active posting on the memorial may have been affected. Given the setting of the Facebook group as public by the administrators, however, it can be assumed–however tenuous that assumption may be – that members of the group posting on the memorial page would be aware that what they share is publicly displayed and can be accessed, retrieved, and downloaded by third parties (see Section 2.1).

Finally, the third stage involved a shift in the focus of the study from the general consideration of patterns and norms of sharing across posts to the close discourse analytic investigation of a single user’s posts in relation to her identity construction as a networked mourner. This shift in focus called for a reconsideration of ethical issues and concerns with the aim to protect and safeguard the selected user’s privacy. At that point, it seemed important to ask for the informed consent from the group’s administrators as well as the user whose posts would be analysed to protect their privacy and anonymity.

As my only point of contact with the group members was Facebook I used the platform’s messaging service to get in touch with the users. What I hadn’t realised at the time, however, was that my message would be delivered to the ‘Other’ or spam folder of the users’ accounts, since I wasn’t friends with any of the administrators or anyone from their network. ^[3] As a result, I never got a reply from any of the users I tried to contact and I didn’t have an alternative way to reach them. This proved to be the most critical point in the ethical decision making process for this study as I found myself questioning whether I should move on with the study or give it up altogether. The serious ethical conundrum I was faced with had to do with whether my decision to move on with this part of the research without the informed consent of the user and/or the administrators could be justified and if so, what additional measures were needed to safeguard the individual participant’s privacy as well as that of other sharers.

Elgesem acknowledges (2015) that ‘in some cases, research on social media can be ethically responsible without consent; the interests of those involved may be safeguarded in other ways.’ In this final stage of the study, my decision was to go ahead with the study taking additional measures to safeguard the privacy of sharers that included anonymising any material quoted by redacting references to names, places and avoiding any other potential identifiers of individuals, places, and events in my general discussion of the group and its members. However, there was still a need to ascertain what was understood as ‘privacy’ in this context.

Drawing on understandings of private/public as outlined in extant ethics recommendations proved rather limiting in that the construct of privacy in such cases seems to be taken for granted, rather than being made explicit. The BAAL recommendations for good practice (2006), for instance, include reference to the expectation that ‘applied linguists should respect the rights, interests, sensitivities, and privacy of informants’, (Section 2.1; my italics) further clarifying that ‘it is important to try to anticipate any harmful effects or disruptions to informants’ lives and environment, and to avoid any stress, undue intrusion, and real or perceived exploitation’ (BAAL 2006: 4). Furthermore, Section 2.5 referring to ‘Deception and covert research’ includes reference to the right to privacy (ibid: 5) invoking legal concerns associated with the term. In these cases, the meaning of the term ‘privacy’ seems to be based on its presumably shared meanings and there is no need to make its meaning explicit (ibid: 5). Similar understandings of privacy that feature in normative and applied ethics resonate with popular and media discourses, suggesting an understanding of privacy as a function of being let alone to lead one’s personal life without interference from third parties.

Such conceptualisations are, however, problematic in that they largely erase the ideologies that underlie the construction of the barely 150 year-old category of privacy (Ferenstein 2015). Recently, deep concerns about privacy have been raised in public discourses, often in the context of waves of moral panic accompanied by calls for taking measures to protect ourselves from unwanted monitoring and surveillance by third parties, including corporations and the state (Krotoski 2016). Again, in these discussions privacy is taken as a given rather than as a concept that is in need of definition or explanation.

The remainder of this article will explore further the categories of private and public which are at the heart of the vexing issues raised by the case study and will consider the key implications of the discussion for internet research ethics.

3.1 Private/public as a language ideology of differentiation

Linguists would class the lexical items private and public as shifters (Gal 2005), which means that their denotation (i. e. meaning) changes alongside any changes in the presuppositions that come with different contexts of use. Indeed, privacy is essentially a cultural construct, rather than a seamless or instinctual transition from medieval forms of family and communal life to ‘civilized’ forms of individualized social organization as is sometimes thought (Ferenstein 2015). Taking into account the sociocultural situatedness of the two terms, it follows that their use in everyday and scholarly discourses is implicated with positionings and social anxieties and, at the same time, contributes to the construction of cultural distinctions that are central to everyday life and social theory, as well as to capitalist formations more broadly (Gal 2005).

The linguistic anthropologist Susan Gal (2005: 24) has shown how the private/public distinction is used in everyday discourse as a language ideology that discursively divides spaces, moralities, types of people, activities, and linguistic practices into opposed categories. In her semiotic analysis of private/public based on ethnographic fieldwork in Eastern European contexts juxtaposed with US contexts, Gal shows how the U.S. individualistic ideological schema is focused on spatial deictics formulating private/public as metaphors of space articulated in terms of spheres, realms, and closeness to the body of social actors. ^[5] (Gal 2005: 33–34).

Gal’s semiotic perspective proves useful when examining the pervasive use of the categories private/public in studies of death online from a metaethics point of view. In that field, contrastive uses of these cultural characterizations feature predominantly in discussions that deal with changes in the way death is experienced and shared.

More specifically, designations of private-public are invoked in defining and delimiting the scope of key death-related practices in the central concepts of grief and mourning. For instance, in the work of sociologist of death Tony Walter (2008), private grief refers to grieving for the loss of an intimate and is described as a legitimate type of personal feeling that is severely constrained in terms of its social or public expressions in the twentieth century. In contrast, public mourning refers to grieving for state-sanctioned individuals e. g. war dead, astronauts, victims of attacks etc. and is perceived as a form of expression designed for public display (Walter 2008).

The aforementioned definitions of grief and mourning illustrate the contrastive use of the categories private/public and their mapping to the type of loss, i. e. the loss of an intimate versus the loss of public figures, as well as to the form of communication appropriate to that loss, i. e. a form of personal feeling that is not to be expressed versus grieving as a form of expression for public display. As a result, a certain dissonance between feelings produced in one’s interiority and their socially regimented display gets foregrounded in and through the contrastive set of the two terms. At the same time, contrastive uses of private/public in this context construct further distinctions between relationships of proximity versus relationships of distance between the bereaved and the deceased that also become metaphorically mapped to the definition of grief and mourning respectively. ^[6]

Furthermore, contrastive definitions of grief and mourning foreground the importance of the categories private/public as vehicles for describing social and cultural changes in the experience and sharing of loss in line with more general sociocultural changes. For instance, Tony Walter (1994: 41) notes how in the past death was experienced in public within the community, then got confined to the private spaces of the home and the hospital; in postmodern times of fluidity, the private and the public increasingly got conflated, bringing the private feelings of the dying and bereaved at centre-stage and turning their public expression into an expression of individualism. Walter’s schema points to the extension of uses of private/public from spatialized metaphors (‘community’ versus ‘home’ and ‘service provision settings’) to metaphors applied to forms of emotional communication.

Another example of this extension of metaphorical use is attested in Caroll and Landry (2010) who note the blurring or blending of interpersonal communication and mass communication via the web in terms of the merging of private-public forms of expression: ‘What once was very private communication – messages to the deceased – becomes very public’. In this discursive context, web memorials are viewed as key sites for the manifestation and investigation of postmodern shifts in the experience of death, by virtue of their metaphorical perception as spaces for emotional expression, which are situated somewhere in-between the public and the private.

Such uses of the terms, however, draw discursive boundaries between the traditional, the modern and the postmodern by virtue of their reliance on linear narratives of a cultural transition from communities to individualism and currently perhaps to post-modern forms of collectivities and hyper-individualism. To the extent that the categories private/public are implicated in discursive refractions of wider social and cultural constructs, they function as emblems of cultural shifts. Therefore, their appropriateness as categories for analysis or for ethics discussions is undermined to a significant extent.

Given the shifting and contested nature of the constructs private/public, the question that still remains to be addressed is what their place can be in research and ethics – or in ethics as research – drawing on insights from the language-focused research of web-based mourning.

3.2 Private/public and internet research ethics: from individual to relational selves

Private/public distinctions as ideologies of differentiation refract fundamental social and cultural constructs that furnish the relationship between the self and the other in specific contexts. In that respect, the very notions of private/public become rather limiting as points of reference, since their conventional meaning is seen to extend from spatial to communicational metaphors obfuscating, rather than clarifying, aspects of social and cultural practices and by extension, research ethics concerns. The extent to which privacy concerns are (or not) foregrounded in institutional or professional codes of ethics is revealing of the academic, cultural, and political context one operates but seems to tell us little about the nature of communication in question (Markham 2004).

The need for refining conceptions of privacy has become acknowledged and placed at the heart of ongoing discussions about ethics regulations in internet research ethics. As Ess (2015) remarks, the drawing up of research ethics, including internet research ethics, insists on notions of privacy that are grounded on individualism. More specifically, the emphasis tends to be placed on the requirement to respect an individual’s privacy, in some cases also taking into account ‘third party information gathered from a primary subject about his friends or close relations’ (NESH 2006 B.13: 17, UCLA OHRRP 2011 cited in Ess 2015: 60). Ess proposes a shift of focus from individual to distributed responsibility that recognizes that individuals are connected to others through their relationships with them. Such a shift foregrounds a turn from a concern with emblematic images of private/public categories grounded in ideologies of differentiation that construct the self as an autonomous interiority to its extension to relational selves (ibid: 51).

More broadly, it seems to be the case that in contemporary life the experience and understanding of privacy is shifting alongside changes in the way we come to be and understand ourselves in the context of networked economies of sharing and social media interactions. In such contexts we imagine and stage forms of public selves in semi-public contexts where known and unknown audiences merge, blurring the boundaries between what would have been conceived as private and public and rendering their sharp contrast almost untenable.

In contexts of web mourning, drawing attention to sharers in their discursive construction as relational selves rather than as individual users, leads to additional questions such as: who is entitled to share tributes in a memorial group? Who (and how) do sharers connect to through their posting activity? How do participants construct and engage with their audience(s)? Such questions in turn lead to foregrounding the limits of informed consent, which is so far primarily linked to an individual user. In this case, it seems that informed consent would need to be extended to the group members and their relation to the deceased’s family but also to the deceased whose identity is being constructed by others post-mortem. By extending the net of people involved, however, any straightforward designation of a site as sensitive, private or public is discounted, since different participants are expected to have different views of the site depending on their relationship to other participants and their role in the group.

The final section looks at how privacy issues were revisited and eventually dealt with in the present study. It foregrounds the usefulness of discourse analysis perspectives in gleaning local understandings of the nature of shared content based on participants’ orientation to their audience and contributes to the furnishing of ethics as research.

3.3 Public/private in context: participants’ orientation to audience

In social science research, it is often assumed that local understandings of what counts as public/private can be directly ascertained using interviews or surveys where participants can report their own views and feelings on the nature of the content they share online. From a discourse practice perspective, however, the usefulness of participants’ direct reports is limited to merely pointing to widely circulating discourses rather than revealing the delicacies of communicative practice in context.

In assessing the nature of the data for analysis, it is arguably more meaningful to rely on the consideration of the discourse style of user-generated content seeking to trace indexical cues of sharers’ awareness of the open access context of the social network sites they participate in or any sign of their own perceptions about their sharing as ‘private information’ in the sense of information about behavior that occurs in a context in which an individual can reasonably expect that no observation or recording is taking place (Buchanan and Zimmer 2012).

In the case of the memorial group that has been the focus of discussion (see Section 2), the discourse style of sharers was found to involve the use of a mixture of formal expressions of condolences (see example 1) and vernacular language in the shared tributes (Giaxoglou 2014b) (see examples 2 & 3).

Example 1.

You will always be a great inspiration in everyone eyes. Miss you boy.

Example 2.

R.I.P. bro. hope ur having fun showing them angels a thing or two in tennis

Example 3.

rip [name of deceased]!♥ love you so much, and miss you like crazyy! i have no doubt in my mind you are doing it big up there! (:

In and through this mixed style, which is dominated by uses of direct addresses to the dead, sharers were found to stage personal and intimate forms of displays of mourning as cues of their relationship with the deceased as well as with members of the group and to construct themselves as entitled networked mourners. In addition, in that context the cumulative sharing of seemingly personal moments makes part of a wall of posts that serves as a representation of mourning, whereby a wider public is called upon to witness the continued impact of the deceased’s life and death on others.

Facebook memorials can, therefore, be said to provide increased opportunities for young adults to share personal tributes with and for networked audiences, both known and new. In other words, memorial sites furnish spaces in-between the so-called private and public, where sharers take an agentive stance and enter a performative space for broadcasting their thoughts and feelings, making them visible to others. In some way, the discourse strategies of sharers on the web memorial can be seen as akin to strategies one draws on when photographed ^[7]: responding to a felt need to protect one’s interiority in front of the camera one tends to adapt what they present in public as ‘private’ or ‘interior’ by leaving out elements of ‘intimacy’. This kind of practice points to a form of self-presentation, closely associated with the rise of broadsheet journalism and photography, which involves the construction and staging of a public self in contexts that are experienced as cross-cutting public-private boundaries.

The discourse-based assessment of the nature of the data in this case focused on audience considerations embedded in the dominant style of posts rather than on a straightforward designation of the site or content as public/private. This type of discourse-based assessment formed an integral part of the careful weighing of the risks posed to the researched. It also enriched the safeguarding measures for the protection of sharers’ privacy, understood here as respect for their emotional displays in an open group setting. More specifically, special care was taken to ensure that descriptions of the group in any research discussions is respectful to the individuals concerned and the group as a whole, in the sense of showing alignment to the group’s orientation to memorialisation and contributing to the legitimisation of such online practices against negative media representations of online grieving as, for instance, ‘faux-grief’ (Dent 2016).