Abstract
Emotions are processes with social origins and manifestations. However, the challenges of obtaining data on such volatile phenomena might restrict empirical research. This article presents methodological recommendations for the study of emotional processes during interactions, comprising an approach influenced by ethnomethodology. Key requirements include (a) detailed interactional data; (b) examination of whichever emotions emerge instead of studying predefined categories; and (c) nuanced “insider” understandings. Rather than focusing on individuals or the broad social milieu, useful insights are available via nuanced examination of emotional trajectories as they are manifested in interactions. An example of the translation of these perspectives into method is briefly demonstrated using excerpts from a study which explored interactions in online discussion boards.
This article examines emotions as processes, and demonstrates how research might cast light on the trajectory of emotions during interactions. It brings together methodological approaches relevant to the problem and presents methodological recommendations to clarify how such research might be carried out.
As a background, two aspects of emotion research are addressed. First is a contrast between examining emotions as individually felt (Johnson-Laird & Oatley, 2000; Scherer, 2004), and examining emotions as social (Goffman, 1973; Hochschild, 1979). Second is a contrast between viewing emotions as states which are triggered by events (Scherer, Wranik, Sangsue, Tran, & Scherer, 2004; Thagard & Kroon, 2006) and viewing emotions as processes (Denzin, 1984; Frijda, 2000). This article explores the methodological challenges of viewing emotions as social processes manifested during interactions, while nonetheless demonstrating the potential fruitfulness of that perspective. Though a range of methodological perspectives might guide our endeavours, the principal approach explored here draws on Garfinkel’s ethnomethodology and his use of Schütz’s concept of verstehen, whereby events and activities are understood by members of a social group. Ethnomethodology attends to unfolding processes as people interact, and is specifically useful here for its focus on interactive rather than individual sense making. Though the focus of the article is methodological, I briefly demonstrate how this approach was translated into a study of observable emotional processes in online discussion boards for people living with kidney failure. The methods and findings from that study are sketched here to demonstrate the kind of insight into emotions one might obtain from examining interactions. Finally I explore the limitations and practical value of this methodological perspective.
What Are Emotions?
There is no standard definition of what emotions are (Izard, 2010) but contrasting approaches exist in the literature. In one approach, emotions are described in terms of how they manifest. They are associated with physiological changes like breathing and heart rate, changes in facial expressions or activities, cognitive appraisals and feelings which are given labels like “fear” or “anger” (Frijda, 2000; Turner, 2009). Much discussion is devoted to categories and classifications of emotions (Averill, 2004; Goldie, 2004). Complex emotions are seen to consist of combinations of basic emotions (Johnson-Laird & Oatley, 2000). In all this, emotions are mainly conceived as states of being within individuals, triggered by an event (Scherer et al., 2004; Thagard & Kroon, 2006). Researchers explore the attributes of emotions, or the actions, events, and factors that trigger different emotions (Lutz, 2008; van den Broek, van der Sluis, & Dijkstra, 2011).
Problems with conceiving emotions in these ways have been identified by other writers. First, when one considers emotions as triggered by events, it is easy to ignore the agency of people feeling the emotions (Lupton, 1998). Second, one might begin to imagine emotions in deceptively clear, categorised ways—anger, fear, joy—neglecting nuances and complexities, regardless of how finely the categorising scheme is designed (Scherer, 2004). When researchers look for associations between emotions and triggering events or demographics, they may talk about “accounting for the complete variance in behaviour” (Denzin, 1969, p. 930), and the conditions required for generalisability (Wilhelm, Schoebi, & Perrez, 2004). Accounting for complete variance and trying to establish general truths suggests a conception of emotions as manifesting in ways which can be understood reliably, unhistorically, and across populations (Lupton, 1998). Researching emotions as stable categories draws attention away from their nature as processes and the microprocesses from which they are produced (Katz, 2004; Parkinson, 2004). Importantly, one might ignore the changing ways that emotions shift as we experience them.
Instead of emotional states, other authors address emotions as processes (Denzin, 1984; Frijda, 2000). For example, the heuristics used to describe emotional states—cognitive, physiological, behavioural, and subjective components and appraisal processes (Izard, 2010; Johnson-Laird & Oatley, 2000)—can be imagined in dynamic relationships (Haviland-Jones & Kahlbaugh, 2000; Milton, 2005; Scherer, 2001). From this perspective, emotions have reflexive, cogenerative relationships with the world, operating as part of the way we experience situations (Holmes, 2010; Parkinson, 2004). Though some of these authors refer to physiological aspects of emotions and others to discursive perspectives, the combination of their commentary produces a view of emotions as connecting to our perceptions of the world and activating our responses.
Such perspectives on emotions are comparable to discussions of affect (Massumi, 1995; Thrift, 2004). Massumi discusses affect in terms of “a matrix of variation” (Massumi & McKim, 2009, p. 3). All elements interacting in a situation are viewed as interrelating processes—emotions, cognitions, “objects” in the environment, and one’s management of identity. All are conceived as existing in time, and this acknowledgement of chronicity is a significant, but problematic theoretical shift. Describing the experiential congregations constituting affect, Massumi draws our attention to how the dynamic relationships implicit in emotions-as-processes necessarily lead to unpredictable outcomes.
Researching volatile phenomena, one needs to do more than attempt to predict how things will play out in the same way. Instead we need tools to follow processes as if they had infinite possible paths. We also need data in which the changing and responsive aspects of processes are available. This means data in which interactions unfold: recorded or live interactions. Considering emotions as processes turns attention to the nuances which signify change. It also draws attention away from individuals by bringing forward the situations in which changes take place, including the location of individuals within the social.
The term “social” has a range of meanings, each with methodological implications. When considering social aspects of emotions, one might examine how individuals modify, interpret, or hide their socially constructed, internal feelings (Goffman, 1973; Hochschild, 1979). This retains a focus on the individual. Alternatively, one may shift to a wider focus on discourses, societal structures, or other social factors associated with emotions (Barbalet, 1998; Kemper, 2000). By this too-wide focus, we might lose view of particular instances of emotions altogether, which matters when emotional processes are the phenomena under study. Arguably, a middle ground is to study interactions (Collins, 2004). It is a middle ground in that the focus is neither on individuals nor on broader society. Interactions are useful for our purposes if dynamics between actors (“the social”) might be directly observed.
How to Study Interactions
Significant methodological approaches to the study of interactions in sociology include ethnomethodology (Garfinkel, 1967/1984), symbolic interactionism (Blumer, 1969), and the work of Randall Collins (e.g., 2004). The approach presented here follows ethnomethodology and Harold Garfinkel.
Ethnomethodology
Ethnomethodology turns the focus of social research towards the sense-making processes of people in everyday situations (Heritage, 1984; Rawls, 2008). A central premise is that people develop meanings not by simply applying prepackaged understandings, but by reworking them as they interact. Ethnomethodology’s founder Harold Garfinkel was concerned with how situations unfold contextually, in time. Actors must realign themselves and design subsequent acts against changing contexts, moment by moment (Rawls, 2008). Garfinkel viewed meaning as contextual and developed by participants in ways particular to each unique situation, and in ways which would be recognised and understood by their peers. Meaning-making is thus an interaction between local understandings and events as they unfold. Therefore, situations should be studied as they unfold, with activities contextualised live (Maynard & Clayman, 1991).
To study such contextualised meaning-making, a researcher must be able to understand and interpret situations from the perspective of those present. Ethnomethodologists refer to this as needing a “member” perspective. This mandates a critical examination of the role of the researcher, which can be understood using Schütz’s conception of insider knowledge or verstehen, to which I will return presently. Ethnomethodology pays attention to longitudinally unfolding processes and calls for live data rather than post hoc interviews, both for the sake of engagement with context and for the observable presence of relevant actors. At the same time, ethnomethodology directs attention to people interacting in shared situations—to interactive rather than individual sense making.
Studying Observable Manifestations
Pointing out that it is individuals who feel emotion, Kemper remarked that “we can measure emotion nowhere but in the individual” (2000, p. 46). The main contribution of this article is to disagree, on a technicality: we can observe emotion between individuals. It appears to be a very small move from “inside” the individual to observations “of” individuals—but it is significant by making emotions researchable in everyday situations. We can see some emotions without having to ask people about them. Observable, socially manifested phenomena demonstrate and reenact emotions, eliciting responses from others in the interaction. Language makes experiences readily transmittable, and thereby, social (Berger & Luckmann, 1966). Emotions manifest in talk, via our choice and use of words (Galasiński, 2004). Moreover, facial expressions and bodily movements indicate emotions (Harré & Gillett, 1994; Lupton, 1998). I argue that we also betray-perform-reveal emotions in the way we write, via textual cues like terseness or exclamation marks and in written representations of physicalities. The following short excerpt from a post to a kidney discussion group demonstrates how emotions emerge from text. It is unedited to show how the contributor presented herself at the time of writing. She was facing the start of invasive treatments for kidney failure.
Please talk me down. I am unconsolable. I can hardly see to post this. I feel like I am being led to my execution. I just can’t willingly allow someone to do this to me. I just cannot. This is just a bridge too far for me.
Descriptions of emotions, of physicality, and of thoughts are combined here to communicate distress on a range of levels, showing how emotions can be expressed and accessed via text.
Observable emotional manifestations and responses make presentations of “feeling” accessible to research (Gherardi [2008] developed a similar perspective on researching “knowing”). The subject of such research would be neither social discourses nor individuals’ feelings, but their interactions en vivo: the manifestation of feelings as readable signs and how interlocutors interpreted them.
To summarise, socially manifested emotions-as-processes can be studied if one has:
Detailed interactional data; and
member-understandings of the expression of emotion, which can allow
recognition of changes, enabling nuanced perception of emotional trajectories.
Example: Exploring Emotional Aspects of Interactions in Online Discussions
To demonstrate how these methodological recommendations might be translated into methods for studying emotions in interactions, I briefly present examples from a recent study into online kidney failure support groups. These examples illustrate the capacity for flexible examination of emergent emotional themes enabled by an ethnomethodologically informed approach.
Online discussion groups are Internet forums where people discuss ideas by typing “posts,” usually a paragraph or two long. Posts are gathered into “threads” which maintain the conversational order in which posts appeared, and keep them on semipermanent display. For their frank, detailed exchanges and frequent emotional displays (Hine, 2005; Liu, 2002), online groups were an appropriate venue for the study of emotions in interactions. I joined three renal discussion groups which focussed on the ongoing lived experiences of chronic kidney failure, making clear my dual roles as a researcher and as the wife of a renal patient. I took part in discussions daily for 2 years. Through my ongoing participation, I developed relationships with other members online, and a member’s understanding of interactions in the groups.
In the following section, I briefly explain the methods used to analyse interactions as an example of one way to respond to the methodological challenges explicated before. A detailed description of the method is outside the scope of this article, but can be found elsewhere (e.g., Godbold, 2013).
Creating Local Thematic Charts
In online discussions, people might make use of thousands of possible ideas, so following all possible themes across threads or looking for the appearance of particular themes in every thread was of limited use. Instead I examined individual threads for whatever local themes they contained, charting the reuse of local themes or elements of speech within threads. “Themes” included repeated words as well as repeated or similar ideas (including metaphors, sayings, and cultural references) and tones.
By “tones” I mean manifestations of emotions in the text. These are manifestations which appeared to me, as a long-term, active member of the discursive community. I did not presume to know what emotions were actually felt by contributors, but created local labels to describe the local tones I perceived. For example, earlier I quoted from a post beginning “Please talk me down…” If I had been coding that thread, I might have described the tones as distraught, fearful, or agitated. The labels I might have used are not as important as my recognition of tones reappearing throughout the thread, and their interrelations with other themes and tones.
The analysis translated threads into charts displaying when different elements appeared in the discussion. 1 While space precludes a detailed description of emotions in online discussions here, an example chart is reproduced next.
“Enough”
A thread entitled “Enough” was begun by a young man beginning a common treatment for kidney failure called haemodialysis. His commencing treatment schedule was three 4-hour sessions per week. He wrote:
Hi all! I must put out my emotions, otherwise I’ll explode. Yesterday, I had my first 4 hours sessions. Afterwards, I was a mess. I cried and said I won’t do it anymore. My parents and friends tried to calm me down, but no success. I’m so frustrated to be depended to a machine.… I went to a doctor yesterday and asked, what do I need to sign to get just paliative care. Yes, I want to die. Sorry for being depressed, but you’re the only one, who understand me.
The chart at Figure 1 represents interactions of tones in the discussion that ensued. Contributions from the first contributor (Posts 1, 14, and 20) are marked with an A.
Sample local thematic chart from the analysis of “Enough.”
Post 1 is marked as having a tone which is “depressed,” by the coloured square at row 3, column 1. His depressed tone is stated outright (“Sorry for being depressed”), but also demonstrated by the contributor having cried, wanting to die, and by the title of the post (“Enough”). I could not contact the contributor of the second post for consent, so it is blanked out (column 2). The third post included the following remarks:
what can I say, you sound so down.… I am saddened to here you want to end it all … even being on dialysis you can have a good life, it is what you make of it. You are only young, you will get the chance of a transplant … there must be lots of things you want to achieve and they are achievable if you give your self a chance. Thinking of you and sending you hugs.
Post 3 is marked as “worried and concerned” (row 2) based on comments such as “I am saddened.” The post is also coded as “positive” (row 4), because it puts forward possible improvements to come: the chance of a transplant, and possible future achievements. Each post was analysed in a similar way, creating new rows for new tones as required.
Observation of patterns in the online discussions showed the importance of repetition in the development of “overall” tones of an interaction. People “picked up” and passed on the dominant tone of a thread, sometimes echoing it, sometimes emphasising it, and less often, contrasting it. Cumulative tones in posts created an emergent consensus in the thread, as can be seen here, where every subsequent post conveyed positive tones. Advice provided by contributors not only involved words (for instance, by saying directly, “I am saddened”) but was also conveyed in tone: lighthearted tones when there was nothing to worry about, serious or urgent tones to draw attention to a problem. Post 12 remarked “[dialysis is] not the end, its just a lifestyle adjustment”: the tone is cheerful, demonstrating cheerfulness without saying “be cheerful.” Thus meaning was developed in two ways: as tones added colour to words, and as people echoed each other’s tones. When he next posted, even the first speaker expressed a more cheerful tone. He wrote: “Today, I reached the bottom of my life and now the path can only lead me up … So, let’s go on.” He demonstrated muted positivity, matching the emergent consenses 2 (Figure 1; see Godbold, 2013, for a more detailed analysis of emotions in threads).
The sequential charts used for the study are a form of visual thematic analysis inspired by ethnomethodological perspectives on the development of interactional sense. They make patterns and sequences visually apparent, highlighting sequences ahead of content. They bring forward how posts interact, allowing observation of interactional processes.
This application of ethnomethodology demonstrates the explicitly interactional, longitudinal focus required to study emotional processes. Acknowledging that emotions manifest in complex interactions with situations, it follows the trajectories of whatever emotions were manifested, helping to free the analysis from preordained emotional states. It shifts the subject of study from individual feelings to the emotional trajectories of social interactions. The sensitivity required to work at this nuanced level was attained by becoming an active member of the groups under study. However, the detailed analysis I undertook came at the expense of being able to analyse only conservative amounts of data.
Discussion
Let us compare this ethnomethodological approach with other perspectives on emotion. To study how individual contributors felt (viewing emotions as individual and interior) one might ask them. To understand how people’s emotions were affected by social constructions or structural factors, one might examine which emotional displays were acceptable in this community (Goffman, 1973; Hochschild, 1979). Here, to examine how emotional processes affected interactions (and vice versa), I examined emotional trajectories within conversations: the focus was on the interaction (Collins, 2004; Garfinkel, 1967/1984).
Significantly, while I noticed manifested emotions, I did not theorise about the “real” internal feelings of individuals. The term “real” suggests that there is a fixed, single way we might be able to describe what was felt; this is counter to the understanding of emotions as intrinsically shifting. It also focuses on the individual; though individual feelings are important and interesting, sociology studies societies and how they are created via interactions (Collins, 2004); therefore, socially manifested feelings and their effects on interactions are the significant phenomenon.
I studied written interactions. Though comments could be separated by long pauses ranging from minutes to days, the resulting interactions were “live” in the sense that the data available for study was the (recorded) stuff of the interactions. One could see where pauses occurred as messages are date-stamped, but I did not analyse the effects of time on interactions in this study, nor concern myself with the various mind-changing or emotional trajectories which contributors may have experienced between messages. Of interest were the interactions as they occurred, particularly interrelations between posts.
The analysis is not objective research. Reading posts between peers required understanding of nuanced, shared meanings. Objectivity is a stance which would render me blind to such details. So my analyses represent only my interpretation of the data. Instead of generalisability and reliability, I aimed for verisimilitude and well-documented procedures (Silverman, 2006).
Hence, a limitation of the analysis hinges on whether I recognised meanings in the online interactions in the way that others would—verisimilitude. A rejoinder which Garfinkel adapted from Schütz (1962, 1964, as summarised by Heritage, 1984) contextualises the importance of member-understanding to ethnomethodological researchers. Schütz used verstehen (literal translation: understanding) to describe how people acquire a working knowledge of the social world via socially located interpretations. Understanding is built from local social categories and constructs. Verstehen operates as a gathering-together of these “common understandings.” Via verstehen, people who know the local understanding of situations are able to interpret situations as their peers would, and can interprete the actions of peers in a situation. Thus, verstehen is central to how the world is interpreted by social participants. For example, as a long-term member of the groups, I was able to recognise when people were making jokes or referring to sad situations previously discussed, and I could “read” the short-hand and slang they used. People (including researchers) who are not “insiders” are not able to reliably interpret nuanced social meanings of a social group. However, when a researcher is “member” enough, they become an instrument for the interpretation of common understandings of social situations in that group, by their own access to verstehen (Heritage, 1984). Schütz argued that verstehen is “an indispensable method of the social sciences” (Heritage, 1984, pp. 46, 50). Social scientists interpret and understand the actions of subjects to the same degree of inexactitude as any other social actor, because verstehen is social. This is not access to what others experience; rather, it “involves a ‘making out’ of ‘what’s going on here’” (Heritage, 1984, p. 49) using knowledge of what is typical or reasonable in a situation.
It is worth noting that Garfinkel himself used Schütz’s verstehen to contextualise ethnomethodology, and the discussion here is modelled on Garfinkel’s (via Heritage’s) interpretation of the term. Verstehen forms a central part of the set of theoretical perspectives described as ethnomethodological.
The next step might be to enlist a cohort of “members” to compare perceptions of emotional shifts in the same interaction. This is not for the sake of producing ever finer gradations in our categories of emotion. Emotions, being complex and situated, have a mercurial quality that always eludes categorisation. Emotional categories—anger, fear, and so on—are useful, but only to gesture towards something that has already changed.
Conclusion
Understanding emotions in interactions requires understanding them as processes affecting other processes. This article explicated methodological issues related to studying such volatile, shifting phenomena, demonstrating an approach based on ethnomethodology. The sequential charts used here serve as one example of the possible ways in which researchers might study socially manifested emotions-as-processes. The focus was on tones manifesting in interactions, not emotions felt by individuals, nor the wider social discourses or structures within which they were located. Such a perspective requires detailed data as well as researcher verstehen, or “member” understandings of meanings in interactions. These elements provide some recognition of nuances in interactions, which allow the perception of changes necessary to study emotions as enactive processes. The subtlety and shiftability of emotions is part of what makes them so important when interacting within situations, and it is the shifts and their interpretation by actors which warrant closer empirical attention.
Footnotes
Author note:
The author gratefully acknowledges the support of IHateDialysis, Australian Dialysis Buddies, and KidneyKorner, as well as the members of those forums. This research was supported by an APA Scholarship from the Australian Commonwealth Government.