Ethics revisited: Rights, responsibilities and relationships in online research

The papers in this Special Issue were first presented at the ‘Ethics of Online Research Methods’ workshop for the British Association of Applied Linguistics Special Interest Group for Language and New Media held at Cardiff University, April 2015. The workshop was in part a response to the increasing complexity that ‘new media’ poses for ethical-decision making in applied linguistics. The Guidelines for Good Practice developed by the British Association for Applied Linguistics pre-dated the mainstream uptake of social media and other kinds of digital communication, being first published in 1994 and then revised in 2006. They have only now been updated again in a way that could take into account the changing mediascape (BAAL 2016). The scale and scope of what might be considered ‘new media’ continue to grow, to the point where social network sites boast millions of members and archives of online interactions are sizeable. On the one hand, this communicative landscape appears to open up many opportunities for empirical research in applied linguistics, ranging from large scale corpus-based projects to smaller scale ethnographic explorations of digital literacy practices. On the other hand, the rights to access or reproduce the language used in ‘new media’ cannot be considered in exactly the same way as they would in offline contexts for linguistic research, for many online sites for interaction blur a clear boundary between private and public contexts, text and context, and allow their members flexible ways to represent an online identity.

The need to revisit ethical decision-making in relation to applied linguistic research focusing on online data is thus motivated by practical concerns about how to negotiate these mediated challenges. However, there are other, broader reasons why ethical decision making needs to be discussed. As the interactions on social media sites and platforms have evolved over the last decade, so too has the critical evaluation of the power hierarchies embedded within and maintained through their structures (Van Dijck 2013). Increasingly, while the uses to which online interaction are put have resulted in positive outcomes for some groups and individuals, social media has provided publically available, tractable contexts for anti-social behaviour such as trolling, cyber-bullying, online grooming and other forms of abuse such as hate speech. Whilst sensitive, difficult and sometimes downright unpleasant kinds of interactions are by no means limited to online contexts, the mediated forms in which these interactions are now played out open up questions about the legal, moral and ethical responses that researchers should take towards them.

The practical concerns of ethical research are discussed in many of the papers in this Special Issue, such as Mackenzie’s discussion of how to approach research participants, Rüdiger & Dayter’s discussion of how to gain informed consent (and from whom) and Spilioti, Pihlaja and Georgakopoulou’s debates about how and when to reproduce linguistic data. The political need to engage critically with the difficult, sensitive and sometimes distressing uses of social media are also brought to light in this Special Issue, and in other papers that were presented at the symposium. As these papers suggest, given the heterogeneity of online contexts, there is no single policy or procedure that researchers can adopt to navigate their way through these mediated concerns. Given that much research on computer-mediated communication is interdisciplinary, and disciplines regulate ethical decision making differently (Giaxoglou this volume), there are multiple guidelines and precedents which an individual researcher might consult for best practice. Instead, what these papers suggest is that if ethical decisions are made on a localised case-by-case basis, then the need for researchers to make this decision-making transparent becomes all the more pressing.

The need for researcher’s transparency in ethical research also makes explicit their position relative to the data and participants they observe. The statement of the researcher’s position is well established in ethnographic and anthropological methods of applied linguistics, but not always in other traditions. As Georgakopoulou argues in her paper, within descriptive linguistics the desire to avoid prescriptive interpretation of linguistic data has tended to obscure rather than foreground the researcher’s involvement in the research process. However, the different papers in this Special Issue make clear that researchers have various kinds of emotional, political and personal engagement with the participants and the interactions that they observe and analyse. This can influence various aspects of the research process, including selecting data, approaching participants and choosing what data to reproduce. The clear positioning of the researcher in relation to their research demonstrates healthy self-reflexivity that is evident in many papers in this Special issue, and can evolve over time. For example, Pihlaja’s discussion of choosing how to represent quotations has changed over the course of his research and in relation to the sensitivity of the site from which the quotations were taken. Rüdiger & Dayter explain how they addressed their increasingly negative stance towards their participants by making their stance explicit and flagging it up as one or many perspectives on the data; while Tagg et al. discuss the need to lay bare the researchers’ own evolving media ideologies. Ethical decision making is clearly not a one-off process, and may need to respond to unforeseen developments as the research process unfolds. The changes that Georgakopoulou describes in her choices of when to show excerpts of data in the interests of empirical accountability against a changing political context are a case in point.

Other researchers in other projects document additional reasons for changing ethical practices through the course of a particular project. For example Whiteman (2012) describes how she revised her relationship to the online community she had been observing as the site itself underwent change. As the forum she was observing lost posts which she was able to supply from her own research archive, she decided to make her presence as a researcher known. This kind of reflexive revision to ethical decisions is perhaps easiest to implement within the life-span of a particular project. It is less easy to revise a decision about a project where the publications from the research have long been in press. In my own work, there are decisions that I would make differently now. For example, when I analysed the stories told by bloggers who narrated their experiences of being diagnosed and treated for cancer, where possible I contacted the blogger and asked their permission to quote from their posts. Some of those blogs, over time, have been deleted. Some blogs tell the unfolding story where the blogger lost their struggle with cancer. I feel profoundly uncomfortable about quotations from those blogs remaining in the print publications where I discussed their language. Although the bloggers are no longer findable from their online presence, excerpts from their writing might one day be read by a friend of relative in the academic discussion of the blogs in my publications. Would that cause harm? Perhaps.

Transparency is also a pre-requisite for trust and accountability. Mackenzie’s paper in this issue points out beautifully that this must also operate on an individual basis, rather than assuming that all participants will understand contextual norms in the same way. However, as Giaxoglou’s discussion of online mourning practices suggests, sometimes approaching the group as a whole can remain important. Transparency is perhaps all the more important when social media interactions have become associated with practices of surveillance (Jones 2016 ). The analysis of online interactions can involve more than observing the words that people publish in their posts and can, in some cases, include information that might be less obvious. For example, meta-data can indicate where a post was published or who a person has interacted with. As Spilioti points out in her paper, who has ownership and control of these aspects of the social media interaction is sometimes unclear. Is the metadata publically available and if so, to whom? Access to this meta-data, and indeed to other publically available content via third party scraping tools has proved valuable as these details can be modelled in visualisations such as social networks and used to contextualise other kinds of linguistic features (such as dialect choices). Access to this kind of data is becoming increasingly less public, as commercially driven sites like Facebook and Instagram change the terms and conditions and so prohibit the extraction of data from their sites by tools such as NetVizz (Rieder 2013). As Rieder (2016) points out, the changes to the terms and conditions suggest a decreasing transparency on the part of social network sites and limits the public understanding of how these sites might function. Indeed, the use of meta-data is perhaps more strongly associated with the commercial profit where social network sites sell their analytics to advertisers. Even if researchers are not deriving any profit from analysing meta-data, there are questions here about the access to and use of these kinds of materials, even where they are publically available. In other cases, where a researcher might want to conceal parts of the meta-data surrounding a linguistic artefact in order to preserve the anonymity of particular participants, this may conflict with the terms and conditions of particular sites. For example, Twitter requests that tweets be shown in context, with the username and Twitter logo present. However, a researcher may not wish to include the usernames, or present the tweet in that way, especially if the tweet was gathered as part of a large scale corpus.

The ethics of openness and accountability also bear on how researchers publish our research and what use other people might make of our data and findings. Within the United Kingdom at least, the importance of the ethical responsibilities of researchers was brought into sharper focus in the last decade not just by the changes in technology, but by changes in the Higher Education sector. Increasingly, researchers are held accountable for the outcomes of their work, which are to be beneficial to public well-being and the wider economy, as seen in documents like the Warry Report (2006). Although dissemination alone cannot result in these wider outcomes of research, it remains critical that we make our work accessible. Social media platforms like blogs are one means of achieving this, especially for immediate reporting on work in progress as the research is taking place. This can close the gap between the researcher and the participants, but needs to be handled sensitively, as Tagg & Lyons et al.’s paper in this collection points out. As researchers, we need to be prepared to respond to the feedback on our published research, as social media allows us to be contacted directly, and indeed to become the objects of research ourselves. Whilst there can be positive outcomes from this public engagement, it is not without risks. Researchers, just like research participants, can find that their words published online are taken out of context or are misunderstood, as the responses on Twitter to Beth Driscoll about her research on middlebrow illustrate (Jaffe 2015). There are also hidden ethical tensions, even in researchers’ attempts to open up ‘free’ access to their work. For example, the site academia.edu might encourage researchers to share their pre-print publications of research available outside commercial paywalls, but, at the same time, it is owned by a company that generates revenue by selling its meta-data to third parties (Fitzpatrick 2015). The concerns about the use of social media interactions and the hidden ends to which they might be put apply not just to the ethics of our research practices, but to how we publish too.

The discussion of ethical decision making in relation to mediated forms of interaction perhaps opens up more questions than it can answer. However, raising those questions and debating the responses are important and need to be made explicit as part of our research practices and in our publications. The papers in this Special Issue illustrate the kinds of responses that researchers have made in these cases, and inspire us to keep wrestling with these difficult decisions as we plan our future research.