Researching Silence: A Methodological Inquiry

Abstract

The role of silence has received increasing attention in International Relations, ranging from silence as exclusion to secrecy and performance. Yet, there has been little effort to draw together a more practical, methodological inquiry into silence and how to engage with it in the research processes. This article builds on existing studies on silence and our own research experiences in conducting interviews and text analysis to interrogate the role silence plays in the research process. It aims to develop methodological tools for engaging with silence and offers a practical guide to analysing it from the data generation stage to the interpretation of silence. In doing so, it also contributes to attempts to redefine the meaning of silence in International Relations by including silence as more than an absence.

Une enquête méthodologique sur le silence

Introduction

The social and political function of silence has received increasing attention in International Relations and adjacent disciplines. Silence has been investigated in its theatrical power, for example, in silent vigils or minutes ¹ and as resistance to dominant discourses. ² Silence by elites has been studied as a strategy of concealment and secrecy. ³ Silence has also been studied as a form of exclusion from speaking; particularly feminist and postcolonial scholars have shown silencing dynamics that marginalise certain actors and prohibit them from gaining a voice in political discourses. ⁴ Silence has thus often been conceptualised as a ‘problem for the practice and understanding of international politics’. ⁵

While the role of silence has received increasing attention in IR, including work that analyses the productive role of silences, ⁶ there has been little work on the more practical, methodological challenges that an analysis of silence raises. Silence cannot be referenced in the way utterances usually are, and it always remains ‘between the offering and withholding of meaning’. ⁷ Given these fundamental challenges, silence is still often ignored or overlooked, understood as an obstacle to research, a failure or mistake in the research process.

In this article, we interrogate the role silence plays in the research process and discuss methodological tools for engaging with it. This also adds weight to the argument to understand silence not in general as a mere absence in international politics. It can, instead, play a powerful role. The article synthesises existing studies on methodological approaches to silence from fields such as sociology and linguistics and builds on our own research experience in IR. We draw on our work on the effects of Western-led media in global South countries, where interviews were conducted in a politically sensitive context, and on covert warfare, investigated using document analysis.

The article thus provides an inroad to methodological questions regarding the research of silence in IR. We discuss different methodological strategies for researching silence, moving from the level of data generation to the level of data interpretation; in this way mirroring the research process. We recognise that these levels are closely interlinked and always situated within the wider silencing dynamics of knowledge production, i.e. which questions can be asked in a particular context. This is a point that we elaborate on further below. Our article provides a practical framework for drawing together strategies and questions regarding the research of silence.

We start the article with a brief review of the existing literature that clarifies some conceptual aspects about approaching silence. The second part discusses strategies of engaging with silence at the level of data generation. After a brief discussion of the role of scholarship in silencing and self-silencing dynamics, the second section engages with the silence of sources during data generation, such as interview refusals or lack of textual sources. Following this, the third part draws together three strategies for evidencing silences by (i) tracing discourses about silence, (ii) recording non-reactions, and (iii) establishing frames of reference. In the fourth and final part, we turn to a discussion of interpreting silence.

Approaching Silence

A range of scholarship in different fields of social sciences has raised the importance of studying silence and has engaged with the questions silence poses. ⁸ Rather than seeing silence as the opposite of speech and communication, it has been argued that silence can be eloquent or conversational. ⁹ To categorise silence, scholars have developed typologies; Ephratt, ¹⁰ for example, distinguishes between stillness (e.g. when asleep), pause (e.g. planning what to say next), silencing (e.g. censorship) and eloquent silence (a deliberate, communicative act).

Shifting away from the idea that silence has an inherent substance that can be ontologically grasped as opposed to speech, pragmatic approaches have suggested that we need to analyse more practically the communicative functions that silence can gain. ¹¹ We build on this idea and move away from the ontological preoccupation with what silence is, instead understanding ‘silence as doing’. ¹²

The significance of different functions of silence has been studied in a wide range of fields in social sciences. Researchers have looked at the role of silence as acquiescence in international law, ¹³ its performative role in social situations, ¹⁴ silence as resistance in political discourses ¹⁵ or silence as concealment by politicians. ¹⁶ In IR, silence has often been studied as an outcome of exclusion and silencing. Feminist and critical race scholarship have done much work to investigate how certain voices and issues are marginalised in discourses. ¹⁷ These marginalisations and exclusions happen through highly institutionalised processes ¹⁸ as well as in daily encounters. ¹⁹ They can encompass topics within a society, like taboos, ²⁰ or can be linked to the exclusion of particular bodies and voices who are not recognised as speaking subjects. ²¹

As we will discuss in the final section, the way in which silence is often interpreted, as either exclusion of marginalised actors or an intentional act by powerful actors, problematically depends on our own assumptions regarding who can be expected to speak. It is important to recognise that silence is ambiguous and always linked to the assumptions of the listeners and interpreters, which can be based on a variety of personality, experience, educational background, or culture. Silence, as de Behar puts it, always reflects the ‘horizon of expectations’ of the interpreter; ²² with this, it also provides an opportunity to reflect on our assumptions as researchers. Scholars in varying disciplines have discussed the epistemological opportunity that silence, absence, and nothing presents. ²³ Scholars have engaged with the question of what counts as significant in the first place ²⁴ and the need for reflexivity for the interpretation of silence. ²⁵

Some scholars have directly addressed the relationship between silence as absence and exclusion on the one hand and silence as a communicative act on the other hand. ²⁶ Different contextual parameters can help to determine whether silence gains communicative functions within a discourse or remains a ‘mere absence’. The intentionality of silence has thus often been noted as a crucial factor in determining the eloquence of silence. ²⁷ The unsaid also needs to be perceived as relevant within a particular discursive situation. ²⁸ This is linked to the intersubjective dynamics of communicative silence, which hinges on the expectation of speech by the listeners. ²⁹ While building on these frameworks, our article is less focused on the categorisation or definition of silence.

We use the concept silence quite loosely as an absence of utterances which includes both communicative silences and absences stemming from exclusion, in order to accommodate a range of strategies that can help tracing such absences. The goal of our article is thus not to contribute to the ontological discussion of what silence is but instead to focus on a range of methodological strategies for tracing silences. While some ontological and epistemological aspects are necessarily linked to this, we want to offer a discussion of more practical, methodological questions.

There has been some scholarship in disciplines adjacent to IR that has addressed more practical questions around silence. Poland and Pederson, ³⁰ for example, discuss the sociological implications of interpreting silences in face-to-face interviews. Schröter and Taylor ³¹ have developed an edited volume that collects different methodological approaches to studying silence in the political context and other scholars have worked with silences across a range of disciplines in a less comprehensive way. Many of these studies are rooted in their disciplines and focused on particular issue areas, such as silences in Chinese smog discourses, ³² ethnographic tracing of silences by the Cherokee Nation leading up to the signing of the Treaty of New Echota, ³³ or the exclusion of the voices of youth in news coverage of teenage sexuality in New Zealand. ³⁴

Our article builds on such work to synthesise different ideas and strategies for IR, encompassing methods ranging from interviews to documents analysis. ‘Interviews’ as a method is here broadly understood. While the case we are mainly drawing on used individual face-to-face interviewing, our conclusions explicitly include, e.g., focus groups and potentially even survey interviewing. The distinguishing feature to documents analysis is here that with such conversational methods the research is dependent on the voluntary participation of the (potential) interviewees. The aim of our article is to provide a practical discussion for researching silences in IR that are produced by a wider range of methods and draw together strategies from a range of different subdisciplines. We structure this along the research process from data generation to data interpretation. After a brief discussion of silences at the level of data generation, the following part of the article will present three practical strategies for tracing silences.

Silence and Data Generation

As an absence claim, the invocation of silence depends fundamentally on the limits of what is listened to, the range of voices from which speech is expected, the issues which are regarded relevant. Research itself is often deeply implicated in processes of disempowerment and even the attempt to eradicate silence risks to reinforce asymmetric relations of power between researcher and researched. ³⁵ This section will analyse silence at the level of data generation, first by examining the importance of silences used by scholars and second by investigating silence of sources, such as the absence of documents or refusals to participate in interviews.

Silencing and Self-Silencing

Increasingly in IR, methods are understood as not only descriptive devices but to actively shape the research process in the interplay between the researcher and the researched. ³⁶ This understanding of method goes further than the idea that research happens in a social environment and is consequently shaped by the relationship between the people involved. The choice of research method is not only a consequence of our ontological and epistemological stance but has a direct impact on the answers we get to our research questions, and on the questions we ask in the first place.

The disciplinary subfields in IR have limits of what can be said and what is considered to be irrelevant. The boundaries of these fields often reproduce dominant assumptions, which lead to the systematic exclusion of voices. ³⁷ This is relevant for silencing by researchers regarding what topics are (not) talked about in the first place because they appear to be unimportant for IR. Despite significant changes in recent decades, the discipline of IR, as well as its various subdisciplines such as security studies or terrorism studies, has been critiqued for remaining centred on a white, elitist, western and male-centric worldview that silences other understandings. Terrorism studies, for instance, has been critiqued for a restricted focus on cases that have been covered by Western media and information given by the state, leading to effects that ‘can be seen in both methodology and in the types of questions that remain unasked’. ³⁸ The silence of the researcher hence includes the topics that are researched but also stretches into more implicit methodological considerations, such as which platforms are (not) accessed in the research process, which languages are (not) spoken and listened to, which actors are (not) considered valid knowledge producers.

Disciplinary boundaries shape what researchers can(not) or do not say. Researchers, for example, often exercise self-silencing of emotions to conform with a particular scientific understanding of research. Neutrality, detachment, and rationality are seen as the hallmark of scientific and hence proper research; ³⁹ consequently, academics engage in self-silencing, even if disclosure of emotions and experiences might help to build rapport. Seear and Mclean ⁴⁰ describe how they silenced their personal experiences and emotions while investigating the experiences of sexual minorities and the chronically ill in order to fulfil the academic standards of detachment and neutrality. That disclosing emotions to participants can be useful in research has been described and established in the literature. ⁴¹ However, researchers fear that they might damage their chances to publish in high-ranking journals if their research and writing style is not devoid of emotions. And while this situation has improved over the last decade with diverse voices being heard and the rational, positivist view of sciences being supplemented by an acknowledgement of the usefulness of subjectivity, it is far from resolved. Today, economic pressure, the loss of jobs and the consequent need to publish, drive researchers to aim for ‘save’, meaning publishable, topics and methods, and successful research.

It should, however, also be noted that non-disclosure or self-silencing during data gathering can be important to protect researchers and research participants. Ryan-Flood and Gill argue that disclosing researcher emotions and becoming close(r) to participants ‘can present dilemmas, in research settings of unequal power where empathy can become a means of exploitation’. ⁴²

Silence can also be a methodological choice for researchers to employ. As already the classic methodological study on interviews by Mayo suggests, ‘Listen – don’t talk. Never argue; never give advice’. ⁴³ Mayo’s advice is still valid today, with the interviewer’s silence seen as opening a space for responses or as the proper reaction to topics coming up in the interview. Weiss, for example, talks about the proper way to react to respondents’ grief, advises for silence, saying that ‘[t]he interviewer is a work partner, not a therapist, not a friend, not an appraising audience’. ⁴⁴ Research on interviewing thus takes careful consideration of both speaking and being silent, arguing that ‘well-informed judgements on the part of the researcher are important as to when and when not to interrupt the participant as he or she responds to a question’ (emphasis in original). ⁴⁵

A particular dilemma for the researcher arises when a research participant raises opinions counter to her values. This is described by Scharff ⁴⁶ drawing on her experience of interviewing six women one of which made a xenophobic remark. Scharff did not challenge this, but the situation led to later reflection on whether or not there was a necessity or even need to speak up. A closely connected point is made by Morison and McLeod ⁴⁷ regarding the non-disclosure of personal decisions while interviewing men about parenthood. The researcher’s silence here is a strategic choice and an attempt to managing the – complex – relationship between interviewer and interviewee. Nevertheless, there are also instances when challenging research participants might be asked for. Khalil names researchers’ tendency to take their respondents’ answers as face-value instead of probing them and trying to establish facts as problematic in research with violent extremists. ⁴⁸ The methodological problem here is apparent, with the questions of if, when and how to react. Whether or not a researcher stays silent – on a difference of opinion, a personal question, or to establish ‘truth’ and challenge what her respondent says, is a choice with important methodological repercussions.

In documentary analysis as well as in conversational research methods, such as interviewing, the silence of the researcher is important at the level of data generation. Disciplinary boundaries and cultural conventions play a big role in what is (not) spoken about by the researcher. Common understandings of scientific research can lead to self-silencing, for example, of emotions, and while silence can be used by the researcher as a strategy in interview research, it also raises difficult questions. In all cases, the researcher’s silence has both a reason and an impact on the process of data generation.

Silence of Sources

While the previous part of this section has investigated the silence of the researcher, this part looks at the silences of sources. We use the term ‘sources’ instead of ‘research participants’ to include the silence of documents, pictures, and datasets. There are various reasons why these sources cannot speak to us and keep silent. Even in times of Google Translate and Skype, lack of language skills and distant locations pose barriers, or information might be locked behind paywalls. These silences are deeply linked to the dynamics discussed above and demand a close examination of one’s own positionality and how they link to wider political power structures. Rather than dismissing the silence of sources, it can be crucial to explicitly engage with them. They can be significant in their own right and give important information on how to adapt the process of data generation.

Sources might be silent through omission of particular topics. For example, one of our students tried to investigate the representation of the motives for Islamic terrorism in Western media and found that reasons for the acts were not discussed in the newspaper reports. The student understood this absence as a failure and thought she had to change her research question. In fact, the silence she had traced in the Western press was a significant finding in itself that points to certain structural biases and forces in reporting on political violence. As Mazzei has argued: ‘We, as researchers, need to be carefully attentive to what is not spoken, not discussed, not answered, for in those absences is where the very fat and rich information is yet to be known and understood’. ⁴⁹

Because of hegemonic power structures, it is vital to record what issues and stories are not uttered and told in certain discursive settings rather than only to reproduce the agenda that dominant voices have already set. This is perhaps most obviously the case for omissions by powerful actors, such as the state. Researchers have provided important insights by recovering information that had been destroyed or classified by governments. Elkins, ⁵⁰ for example, unearths evidence for British concentration camps in Kenya during the Mau Mau uprising. Rather than just taking the absence of official government documents at face value and taking it as a failure of her research question, she stayed with this absence and used it to adapt her methods of data generation (i.e. collecting oral records by survivors). At the same time, the government’s efforts to destroy the documents is also a significant finding in and of itself. In the next section, after investigating the silence of other sources than documents, we will discuss in more detail how such silences can be recorded.

Silences at the level of data generation are not just important in documentary research. In visual research, the message of a picture is determined by its frame, and hence also by what it is not showing and thus keeping silent. As Knightley states, ‘although in most cases the camera does not lie directly, it can lie brilliantly by omission’. ⁵¹ While visuals do play a role when it comes to silence, as we will show below, a deeper engagement with the issue of changes of meaning and messages through post-production of photographs is beyond the scope of this article.

In research with participants, the silence by respondents in interviews or questionnaires can also signal a significant absence. It is relatively rare that research papers based on interviews include information about the process of interview recruitment, namely of how many people were asked and have rejected requests to be interviewed and why. ⁵² This might be seen as unimportant information; it might also be that researchers redact the silences they encountered during data generation in order to provide a picture of a smooth research process and thus authoritative conclusions. But an exploration of who keeps silent and why can provide valuable insights and – as in the case of textual analysis – should not be assumed to be a failure in the research process. Instead, it can be helpful both as a source in its own right and to adapt the process of data generation.

This is all the more important when interviews are seen as a way to ‘give voice to the many; the marginalised who do not ordinarily participate in public debates may in interview studies have their views and situation communicated to a larger audience’. ⁵³ If this is so, disclosure that people did not want to participate in interviews would be worthwhile. The idea that ‘the voiceless’ can have their views stated in and through interviews is grounded in and reinforces unequal power relations. In this view, it is assumed (i) that people want their voices to be heard in particular discourses and (ii) that they cannot make their voices heard by themselves but need to be empowered by the researcher to do so. This is a questionable view that can lead to a conflation of silence with powerlessness, a point we will discuss in more detail in the final section.

Potential research participants might decide not to speak and make their voices heard because they have no interest in either the topic of the research or in participating in research at all. They might be disillusioned by what they can achieve by participating in research, a phenomenon known as research fatigue. ⁵⁴ Answering a researcher’s questions is time consuming; it might, depending on the research topic, require remembering emotionally challenging memories or phrasing thoughts on issues the interviewee is not feeling familiar with. All this might prevent potential interviewees from opening up in the research situation; it does, however, by no means translate into ‘being voiceless’. ⁵⁵

The relationship between researcher and researched – or potential researched – is a complex one, and it is well-established in the literature that what is said in an interview is shaped by the relationship between the two. ⁵⁶ It seems reasonable that by the same token, the refusal to be interviewed might also be a consequence of this relationship. The importance of gatekeepers ⁵⁷ and of gaining trust ⁵⁸ for interview recruitment have been described. Lately, the issue of research fatigue has been received increasing attention, with people rejecting being interviewed on the grounds of being over-researched, with a lack of change, indifference towards engagement, and practical barriers often named as reasons. ⁵⁹ In line with this, when conducting interview research in South Sudan’s capital Juba on the question of media and media development, one of us realised that people from a lower or disadvantaged socio-economic background were regularly refusing to be interviewed. This project used, among other methods, interviews to investigate the effects of Western-led media development interventions in a global South country, such as South Sudan. ⁶⁰ To understand the local perspective on the media and its role in society, members of government, journalists and media activists, and civil society members were interviewed. While members of these groups were open to interviews, people from lower socio-economic standing that were not included in media production but were basically media audiences, were reluctant. Of a total of 25 interview requests made towards such people, only four interviews took place, which gave rise to the question of why people of a disadvantaged socio-economic background were unwilling to share their opinions.

An examination of field notes and informal interviews with the people who had refused to be interviewed on the question of why they were rejecting being interviewed revealed a lack of interest to engage with the research and the researcher. People said that they did not have anything interesting to say and raised the concern that they would not understand what was asked of them. Their unwillingness to participate was an important silence in its own right (and the next sections will discuss this in more depth), but it was also important in order to be able to adapt the process of data generation. In this case, the refusal to participate could be overcome by a change of setting; and in this way also showed the importance of space when it comes to interview recruitment. Space, as well as material objects, such as the presence or absence of voice recorders in the case discussed here, influenced people’s willingness to speak. People were willing to answer questions and engage with the topic of media when asked in an environment they were used to. They were made aware of this still being research and confirmed their willingness to have their opinions appear in a research paper. They justified the willingness to talk in bars and corner shops by saying that, ‘this is different, this is bar-talk’, ⁶¹ ‘it’s just chatting’, ⁶² and that, ‘I talk to a lot of people here, so why not with you?’ ⁶³ People made a difference between a formal interview, for which they were invited, which was recorded and which would take place in a ‘special’ environment, usually an office room, and a somewhat unofficial chat which took place on their’ home turf’: a situation where they felt at ease and home. The research design was accordingly adapted, and interviews took then, unrecorded, mostly place in local bars, during chess games in a corner shop, and at interviewees’ homes. ⁶⁴ The power relations changed with the location; being in their environment and participating in their everyday activity made them willing to break their silence. They were, thus, not suffering from an inability to make their voice heard – they were, instead, making a decision when, to whom, and where to talk.

The experience in South Sudan questions the conventional wisdom by researchers that marginalised people are directly silenced by stakeholders, researchers, or decision-makers because of these agents’ preference for engagements with elites or groups of rather high socio-economic standing. Instead, this case shows how people from disadvantaged groups showed an unwillingness to share their experiences and have their voices heard because of the distance between them and the researcher. The academic standards for interviews - written informed consent forms, undisturbed locations, voice recorders – can exclude research participants whose lifeworld is removed from such symbols of the academic world. To gain access, researchers here must instead enter their interviewees’ lifeworlds. Silence is then not a sign of weakness; instead, it is a reaction to distance and different social worlds that cannot be overcome by merely stating a wish to empower. To study these social worlds, the researcher has to enter them. Trying to break the silence or overcome it might not only be unhelpful but also overlooks the importance of silence and can in itself be a coercive act. Here, an engagement with the questions of who is refusing interview requests and why in consequence led to an adaptation of the research design. As in the above explained examples of silence in documents, this engagement led to further insights, here, insights about the effects of the research design on the research process. Formal interviews were not a good fit for gaining an understanding of common people’s understanding of media.

Staying or keeping silent thus can be a decision. Such a decision should not be questioned by people alien to the researched persons and external to their decision-making processes, nor should it be disposed of as powerlessness. In her research on gender-based violence, Parpart, for example, argues that keeping silent can be a survival strategy for victims of sexual violence. ⁶⁵ With public as well as domestic violence against women, increasing, keeping silent might be the only option for victims of such abuse. Keeping silent in or about such a situation does not mean that the victim is passive; it means that she is assessing her situation and how best to survive it. The probing to speak up is in itself an act of coercion, and the conflation of silence with powerlessness shows and reinforces power relations. It also shows how we think about (potential) participants in the research process. We do not have a right to people’s stories. Whether people share them is a decision they made; further, it is a privilege. While we can and should investigate why people (as well as documents) are silent, we can, and should not, equalise a decision to keep silent with being powerless.

For the researcher, this means that during data generation, silences must be taken into account; they deserve attention and examination in their own right as well as to reflect on and adapt the process of data generation. This does not mean that we necessarily should aim to break or overcome silences. It does, however, mean that they should be taken as important aspects of the research process. Following this exploration of researchers’ and researchees’ silences, the next section will explore how to evidence and trace silences in their own right.

Evidencing Silences

The previous two sections interrogated the importance of silencing dynamics in the research process, silence by researchers themselves and silence at the level of data generation, such as rejections of interview requests or absence of documents or issues in documents. We argued that it is necessary to engage with these silences both in their own right and to adapt methods of data generation (e.g. changing the setting of interviews, examining different platforms or outlets). This section will focus on the methodological problem silence poses in that it cannot be referenced in the way utterances normally are. Synthesising approaches from disciplines such as sociology and linguistics, we suggest three strategies for dealing with that problem in IR: Tracing meta-discourses about silence, recording non-reactions and establishing frames of reference.

Tracing Discourses About Silence

Discourses about silence are utterances that explicitly reference silence. An example is the United States’ (US) explicit refusal to apologise for a US spy plane in Chinese waters in 2001. Following the incident, ‘the crisis was prolonged over weeks by demands of the Chinese government that the United States apologise for its acts, while the United States refused to do so’. ⁶⁶ The US government’s silence was not only explicitly referred to by the US itself (who refused to apologise) but was also noted and discussed widely in the media and research documents.

Discourses in which silence has been noted and talked about have been called meta-discourses of silence. ⁶⁷ Circumventing the more fundamental methodological challenge of referencing silence, researchers can trace the recordings of silence in different outlets and analyse the meaning that silence is given. Discourses about silence are crucial for analysing silence because even if silence did not ‘actually’, or empirically, exist, it might play an important political role. Building on the sociolinguistic work of Schröter, ⁶⁸ we can thus draw an analytical distinction between logical and phenomenological silence. Logically silence exists when someone does not speak, whether they are alone in their room, among people at a dinner party, or in a diplomatic meeting. Phenomenologically, silence exists when it is perceived by someone else, whether the person was silent or not. ⁶⁹

Phenomenologically, silence can play a powerful political role, as one of us found in research on covert warfare. Silence by governments on the practice of ‘targeted killing’ has been frequently invoked in recent years as acquiescence to the practice in international law discourses. In fact, there has been quite a lot of protest against the practice. Yet, independent of whether this silence by governments logically exists, it does exist within international legal discourses where it has played a powerful function. ⁷⁰

Whether silence is noted at all in particular discourses is closely interlinked to power relations that structure knowledge production more generally. Some silences are heard more loudly than others. These are then recorded more often and thus become traceable as meta-discourses. This means that focusing on meta-discourses about silence runs the risk of reproducing the analytical gaze directed at the utterances and silences by powerful actors at easily accessible platforms. ⁷¹

There are exceptions to the focus on the silence by powerful actors. Sometimes silences by more marginalised actors are noted and gain significance in meta-discourses. Explicit invocations of the silence of victims of sexual and gender-based violence have, for example, played an important role in media and policy discourses as well as scholarship where ‘it is frequently assumed that women who do not publicly reveal experiences of conflict-related sexual and gender-based violence are silenced’. ⁷² As discussed above and in more detail further below, silence can become a symbolic shorthand for powerlessness and, as such, is often invoked in meta-discourses, which are recorded and thus become methodologically accessible for academic inquiry.

Explicit references to silence can also be accessed through interviews or reports of actors who speak about their experiences of not speaking. In a study of Bedouin students, for example, researchers collected stories by over 150 students and identified meta-discourses on silence as a recurrent theme among the female students who write about experiences of being sanctioned for speaking up and feeling pressured to remain silent. ⁷³ Similarly, in interviews on the Bougainville peacebuilding process, researchers had found interviewees talking about the ‘untold story’ of the war. ⁷⁴

It is important to engage with meta-discourses of silence. Tracing how silence was invoked and noticed within particular discourses offers an immediate way for researchers to record silences in interviews, reports, or other sources. Meta-discourses of silence can help the researcher to understand what assumptions are made about this silence and what meaning is given to it by the audience. They are also important because even if silences that are invoked do not logically exist, they can still gain important functions within discourses. Yet, the strategy of tracing meta-discourses has the disadvantage thatresearchers can only engage with those silences that have already been noticed by others. A more immediate strategy for evidencing silence is to record non-reactions.

Recording Non-reactions

For silence to be noted, it can be conceptualised as a reaction, a second move in an adjacent pair. The prompt is a question that receives no answer, a claim or request which gains no reaction ⁷⁵ and can be seen as a marker or indicator for the ensuing silence. This is most obviously the case in dialogues or small discursive settings, in which questions are asked directly, such as interviews. When transcribing interviews, some researchers record the seconds during which interviewees are silent in response to particular questions, differentiating between short pauses (3 seconds or less) and long pauses. ⁷⁶

In more complex discursive settings, the equivalent of this comes in the form of requests that are not reacted to. The United Nations (UN) Special Rapporteur Ben Emmerson, for example, requested information from states in the UN General Assembly ⁷⁷ and four years later stated that ‘no formal answers have been received to date’. ⁷⁸ Similarly, the International Centre for Counterterrorism published a report in which they recorded the lack of replies by certain governments. ⁷⁹

Silences might ensue as a reaction to requests to disclose documents or other evidence or direct questions about a particular issue. They can be recorded by keeping a copy of the original request, such as an email or letter, or by stating the question in a public realm (e.g. an assembly or on social media). Silences can be recorded as linked to this request in the form of no reply at all or an explicit refusal to disclose certain documents. But it can also appear through avoidance of some questions or aspects of questions.

The social norm for speaking is so strong that, especially in direct interactions, people will rarely not answer at all. Instead, answers might avoid particular topics and circumvent questions. Analysing interviews with politicians, Clayman ⁸⁰ traces different strategies, from roundabout responses to minimal or partial answers to ambiguous responses and the re-phrasing of the question. While speaking, these politicians were not actually answering. We thus need to be aware that there can be situations in which there is a response, but ‘the response was in fact no response, that is to say, it did not address in any substantive manner the question posed’. ⁸¹ These are what Kurzon calls ‘thematic silences’ in which ‘a person when speaking does not relate to a particular topic’ ⁸² and what Huckin defines as ‘textual silence’ or ‘the omission of some piece of information that is pertinent to the topic at hand’. ⁸³

Recording non-reactions is an important way for researchers to trace and evidence silence and can avoid some of the problems which the analysis of meta-discourses poses regarding the reproduction of dominant voices (as discussed in the previous section). Rather than relying on others having already spoken or written about the topic or about silence on the topic, the researcher can raise a question, can request information or the disclosure of documents. While a lack of reply to such requests might be considered a failure of the research process at first sight, tracing such non-reactions can be revealing, particularly within the asymmetrical power relations of international politics. The next section will discuss strategies for developing a frame of reference for silences by comparing foregrounded and backgrounded issues/keywords/voices.

Establishing Frames of Reference

A third strategy scholars can adopt to trace silences and absences is by establishing a particular frame of reference. Even more than with other types of research, the recording of silence requires specific explication of the selection of sources. Analysing a particular number of reports on a topic, at a certain platform, the researcher can trace silences – for example, an absence of a group of keywords or the lack of participation by some actors within this range of sources. To be able to do this, it can help to develop a comparative frame of reference, demonstrating how certain topics or voices are backgrounded in what Schröter and Taylor call ‘comparison as a means to identify absence’. ⁸⁴ We suggest five frames of reference in this section: a comparison between (i) different speakers in the same discursive settings, (ii) different sources leaving certain aspects out when covering the same topic, (iii) different aspects being foregrounded or backgrounded within the same sources, (iv) omissions of topics at different time periods and (v) silences reverberating with wider silencing structures.

The first frame of reference is provided by comparing different speakers within the same discursive setting. Feminist researchers have compared the amount of speaking time that men took in relation to women in particular settings to trace how institutional structures can work to silence female speakers. ⁸⁵ While the amount of speech is quantifiable (e.g. seconds of speaking, number of words used, etc.), silence of some speakers here becomes audible by providing a frame of reference in comparison with the volume of other speakers.

A similar methodological strategy for evidencing silence can be employed by comparing how the same issue is covered in different sources. In a study on media reporting on homelessness, Huckin established a list of topics and subtopics on homelessness as an inventory of what has been written on the topic in general, to then trace silences and omissions in some newspaper publications, such as the silence on the problem of affordable housing in more conservative newspapers. ⁸⁶ Comparison with other sources can also take place through an examination of official texts in relation to leaked documents or in contrast to other accounts, such as in comparison to oral recordings in Elkin’s study on British concentration camps in Kenia. ⁸⁷

A third strategy involves the investigation of some aspects of a practice or some concepts which are backgrounded in comparison to others. Studying UN Security Council debates on targeted killings, one of us found that the term ‘targeted killing’ had been used much less frequently for the same counterterrorism practices than terms such as ‘assassination’ or ‘extrajudicial execution’. ⁸⁸ This is an important omission because the term ‘assassination’ has much clearer connotations of legal and moral impermissibility.

The fourth strategy involves a comparison of different time periods, as Wang and Kádár did when investigating media reporting on air pollution in China. They found that smog is covered in newspapers after political announcements but not during the times of greatest pollution. ⁸⁹ The comparison of different time periods can also include an investigation of the process of drafting political statements, documents, treaties and trace what is there during initial discussions but is left out in the final document.

The main advantage of the backgrounding/foregrounding approach is that using a frame of reference for what can be expected to be said provides an explicit methodological inventory against which silences can be compared. The disadvantage is that this inventory tends to rely on what has already been said and/or is expected. But what about topics or voices which are so thoroughly silenced that they do not even appear at alternative forums or in different time periods? For a more critical research approach, it might be necessary to invoke a radically different contextual framework to understand what is more systematically excluded (and might thus not be expected at all).

Linking back to the wider, epistemological questions of silence, it can sometimes be helpful to situate text within wider silencing dynamics as the frame of reference. In her research on racial identity among white school teachers, Mazzei traces the absence of Whiteness in the teachers’ answers and shows how silence stems from a situation ‘when we speak from a position of privilege that affords us the luxury of not even considering the question of Whiteness and its attendant advantages unless we choose to do so’. ⁹⁰ As discussed in the first section, a challenge for us as researchers thus also is to investigate what privileges we might enjoy that might lead us to not even consider a topic and to investigate what issues and voices we might be silencing. This is not only relevant at the level of data generation but also at the level of interpreting silences, to which we turn now.

Interpreting Silences

As argued above, silences are not just encountered by the researcher; they are not only external but are also produced by her, in relation to her and through the research process. The researcher also plays a part at the level of data interpretation. There are complex questions of interpreting silence, and this section offers a discussion of three aspects that are important for this: (i) the question of agency and intentionality, in particular regarding the common conflation of silence with powerlessness; (ii) the importance of reflecting on the context broadly conceived; and (iii) the inherent ambiguity of silence which is linked to the projection of assumptions of the researcher.

Agency and Intentionality

The meaning we read into silences is closely linked to the presumed agency and intentionality of those who are silent. It has been argued that silence can only be understood as eloquent or communicative if it is the outcome of choice. Ephratt, ⁹¹ for example, distinguishes eloquent silence from stillness and pause through the choice of those who are silent or impose silence. Schröter ⁹² similarly argues that deliberateness is a crucial factor in determining when silence can be understood to be communicative. The interpretation of the meaning of silence and the interpretation of whether or not it was a deliberate choice hinges quite fundamentally on underlying notions of agency – which is particularly important in asymmetric relations of power, such as are at play in international politics.

Silence has been interpreted as a strategic choice that can be correlated to a range of functions in international politics – from creating ambiguity ⁹³ and diplomatically maintaining relationships ⁹⁴ to tacit consent in international law. ⁹⁵ Perhaps most notably in IR, silence has been researched through its function as secrecy, a strategic choice by state actors to conceal, withhold, destroy or hide information. ⁹⁶ Silence of powerful actors, in particular states, thus tends to be interpreted as a deliberate choice and a significant, communicative act. Silence by marginalised actors, in contrast, is often conflated with powerlessness.

This binary projection of agency in interpreting silence omits the complex ways in which silences can function. Postcolonial and feminist researchers have shown how silence can be a strategy by marginalised actors in precarious situations, such as post-conflict societies. ⁹⁷ Silence can be an act of resistance within a discourse to ‘discomfit those who regulate social behaviour with speech’. ⁹⁸ Silence can even be a performative act of resistance. Parpart ⁹⁹ points to the silent vigils of Argentinian and Kurdish mothers of victims of military regimes and of Israeli and Palestinian women against the occupation to show the importance of silence in resistance.

This has arguably become even more salient in times of social media. In March 2018, Emma Gonzalez, a survivor of the mass shooting at Marjory Stoneman Douglas High School in Parkland, Florida, kept silent while standing at the lectern at the March for Our Lives. She remained silent for six minutes and 20 seconds, the time the shooting took. Her silent protest has been shared millions of times on social media and named ‘the loudest silence in the history of US social protest’ in the press. ¹⁰⁰ Performative silences might be more powerful than lengthy written statements because the pictures and snippets can be shared with just a click on Facebook, Twitter, or TikTok.

Performative silences tend to point to a specific grievance or cause, symbols are normally chosen carefully, and explanations are often given in the form of statements or even press releases. With many other silences, it is more difficult to trace the intentionality and deliberateness underlying silence. This is by no means a theoretical question and can have quite powerful effects, which show the importance of listening to silences and to how silences are used by the sources, interviewees as well as documents. For example, Harel-Shalev and Daphna-Tekoah, investigating women soldiers’ experiences, identified ‘silence-in-speech’ that was used to deal with complex and conflicting experiences. ¹⁰¹

In practice, it is thus extremely difficult for researchers to differentiate eloquent silence as a choice from silence as speechlessness and exclusion. Silence as a form of exclusion has been called ‘silencing’, defined as the process through which actors, voices, perspectives, issues, and terms are excluded from certain discourses. This can take covert forms since ‘silencing is a process that works best when disguised’ ¹⁰² and entails forms of exclusion that needs not be coercive and are often not conscious at all. Silencing is linked to power relations and exercises of ‘control over the bounds of acceptable discourse’. ¹⁰³ Yet, these silencing dynamics inadvertently underlie all acts of communication and are always linked to eloquent silences as well.

The binary categorisation of communicative silence as a choice on the one hand and silence as an outcome of exclusion on the other hand might not always be helpful when researching silence and might overlook the crucial role of the expectations of the interpreter in presuming agency. The meaning of silence is often presumed as self-evident through a binary notion, in which silence by powerful actors is understood as a strategic choice, while the silence of more subaltern actors is presumed to stem from exclusion, marginalisation and passivity. This binary projection of the meaning of silence is problematic. It might not only be misguided, as discussed above, but risks to further marginalise subaltern actors. As feminist and postcolonial research has pointed out, the conflation of silence with powerlessness can work to further disempower these actors by denying agency. ¹⁰⁴

Silence can be an outcome of marginalisation, but it can also (and sometimes at the same time) be a strategic choice and can even work to resist and subvert dominant discourses. This section has discussed the importance of underlying assumptions of agency in interpreting silence and has argued that a conflation between silence of the marginalised as powerlessness is problematic. It not only overlooks powerful communicative effects silence can have – such as performative silences in resistance – but also risks reproducing assumptions about agency that might further disempower marginalised actors.

The Importance of Context

When researching silences, context plays an important role. We need to reflect on the particular environment in which silence takes place: rules and laws of a state and society, power relations but also societal pressure and unwritten rules and conventions. For performative silences, visuals are essential. An example of this was the silent protest by women dressed in the scarlet cloaks and oversized white bonnets of the TV show The Handmaid’s Tale at silent demonstrations in the US and UK who protested against patriarchy and oppression. In documentary research, previous texts or unofficial documents can provide further insights, as well as the context of the particular platform at which texts appear. In interviews, body language, sound of voice, and utterances provide an important frame of reference. Further research and inquiries with participants can also be conducted to understand silences more clearly.

For example, when one of us encountered the silence of research participants in the form of refusals to participate in interviews, a dominant expectation of speech was disappointed. The project was, as mentioned above, on effects of media development in South Sudan. To grasp a wide local perspective, the plan was to interview people from different socio-economic background and standing in society. Surprisingly, arranging for interviews with local elites was a relatively simple process. On the contrary to this it was extremely hard to find South Sudanese of a lower socio-economic background that were willing to participate. Following the logic that interview research enables marginalised actors by giving them a voice, it is difficult to explain this silence. The refusals could thus have been seen as a failure and written out of the research process. We discussed above how this silence was instead mobilised as a moment of self-reflection to adapt the process of data generation. But even in its own right, this silence was significant: It provided the analytically important insight that the topic – media development done by international and non-governmental organisations – was not of particular interest to some people from a lower socio-economic background.

This became clear through further conversations with participants who had refused the interview invitations. Practically speaking, the remark of ‘this [refusing to be interviewed/the authors] is perfectly fine, but may I ask why?’ was helpful. Some people who had refused before were in response disclosing their discomfort with the interview situation, pointing to the need for a change of scenery. Others mentioned that the topic was of no interest to them, that they did not have a particular interest in the media, that they listened rarely, and had no clear opinion. ¹⁰⁵ This insight is opposed to the arguments for media development made by the organisations engaged in the endeavour. These evaluations with regards to media development are questioned in the literature, ¹⁰⁶ but here the mainstream view that media is a prime concern and urgent need for populations in the global South was questioned through interview refusals. ¹⁰⁷

How people answer a question or refuse to answer can thus provide important insights. In conversational research, further questions can be asked about reasons for silence. Body language and other visual signs can provide more information. In documentary research, the wider context of the platform is important to reflect on, for example, by looking at the history of the text and how it links to adjacent sources. The wider societal context, in particular regarding how that links to the power relations underlying agency to speak and be listened to, is crucial to analyse. Yet while in researching silence we need to look at the particular context of silence; this cannot be understood independent of the assumptions of the researcher.

Silence’s Inherent Ambiguity

Silence is always linked to the ‘horizon of expectations’ ¹⁰⁸ of the interpreter. The researcher’s expectations and assumptions can have powerful effects on the communicative function; particular silences might come into play. One of us found in research on US drone warfare that silence by states was interpreted by some international law scholars as acquiescence to ‘targeted killing’ practices. It can be questioned whether this silence exists in the first place and whether it can be interpreted as a deliberate, legal choice, ¹⁰⁹ yet this interpretation of silence as acquiescence has worked to legitimise the practice and has by some lawyers been understood as confirmation of a new right to self-defence.

Silence is inherently ambiguous. ¹¹⁰ It can never be retrieved with certainty whether silence was meant to express X or Y. It cannot be proven whether silence was meant at all. And even if the researcher establishes that those who were silent had particular intentions in mind, this silence might have taken on a completely different meaning to what was intended. While all of this is to some extent also true for words, silence reveals this uncertainty more clearly because silence has no overt, textual form.

This ambiguity of silence can be frustrating. It can be seen as a stumbling block towards researching silence. It might motivate us to either write silence out of our research or to fix it by finding the meaning of silence. Yet, the inherent ambiguity of silence is also an opportunity for the researcher. It presents an opportunity to reflect on and become aware of the assumptions and expectations that are often taken as self-evident – including the assumptions and expectations of the researcher. Who is presumed to have agency when not speaking? What do we hear when there is silence? What assumptions do we make? The ambiguity of silence indicates a disruption of the neat, logocentric order that seems to present itself to us as if we come from the outside. ¹¹¹ Instead, silence is always messy. It resists our attempts to categorise, interpret, and analyse objectively, thus reminding us of the complexity and messiness of the world we interpret.

Conclusion

How do we methodologically deal with silences we encounter in the research process in IR? This was the question we asked ourselves at the outset of this article. We did not primarily want to contribute to current discussions regarding the ontology and meaning of silence but instead provide a more practical framework of how silences can be traced. While research has debated the ontological questions of silence, whether silence is a choice and strategy or an imposition, this article was more interested in the methodological challenges of researching silence. Our starting presumption was that silences are not just obstacles to overcome but carry meaning and can thus provide useful insights for research, both regarding adaptations to the research process and in and of themselves, regarding the role they can play in social and political discourses. Based on this assumption, we offered some strategies to trace and analyse silences in the research process.

Importantly, we believe that to engage with silence meaningfully, we need always to take the situatedness of research within wider power structures seriously. Silence is always linked to what is (not) listened to by the researcher. The topics that are researched in the first place, but also the platforms that are accessed or the actors who are engaged with, lead to silences that are directly linked to wider global power structures, for example, regarding language barriers or accessibility of sources. Disciplinary expectations contribute not only to who and what is considered important but also to which research ideals are upheld, for example, regarding the self-silencing of researchers during the interviewing process.

This is not to say that silence is something negative and should be overcome, but that silence is significant and needs to be methodologically addressed and not written out of research papers. At the level of data generation, the article particularly engaged with the silence of sources, such as the absence of reports or documents or refusals to participate in interviews. We argued that it is important to take the silence of the sources serious to open up reflective spaces to adapt methods. Refusals to participate in interviews can indicate that the setting of interviews (e.g. formal, recorded, etc.) is not suitable for research participants. Absence of documents is often linked to structural relations of power and might require a shift to alternative platforms, languages, or methods.

These silences are furthermore important in their own right and despite the challenges this poses, it can be worthwhile to trace them. We thus proposed three strategies for recording silence. The first strategy involves tracing discourses about silence, which are explicit references to silence. This strategy is important because silence can play an important phenomenological role even if it did not logically exist. Unlike this first strategy, the second strategy does not rely on what has already been recorded about silence. Instead, it more proactively records non-reactions through documenting questions or requests that remain unanswered. The third strategy we discussed relies on establishing a frame of reference. Clearly delineating sets of sources, absences are compared to a reference point, such as different aspects that are discussed within the same set of sources or absences that become visible in comparison with an alternative set of sources or a different time period.

We argued that silences always remain ambiguous, which does, however, not mean that they cannot play powerful roles in social and political discourses. The ambiguity of silence is, in fact, an opportunity for researchers to investigate the speech expectations that were already there in the discursive setting in which silence became significant. While this was not the focus of this article, we suggested some points to keep in mind when interpreting silence, such as the complicated questions of agency that arise in regard to silence and the importance of the context. As an attempt to draw together existing research and insights from our own research experience with interviews in South Sudan and document analysis of covert warfare, this article thus provided a framework for methodologically engaging with silence in IR.