Micro-ethnography: Towards An Approach for Attending to the Multimodality of Leadership

ABSTRACT
 This paper addresses the need for further developing an understanding of leadership as practice in its multimodality by means of theoretically motivated qualitative methods, allowing researchers to come close to the doing of leadership. Empirical studies of this kind are still relatively rare. By articulating a micro-ethnographic approach, we encourage short-term-focused engagements in empirical work and the writing of closed vignettes. Through this, current theoretical developments are connected to recommendations for fieldwork and for writing practices. We thereby articulate one possible coherent and consistent position from which to study the multimodality of leadership and to understand leadership as an accomplishment of direction.


Introduction
In this article, we articulate micro-ethnography as an approach for contributing to what we label Multimodal Leadership Studies (MMLS): studies attending to multiple modes and resources for accomplishing leadership, thus going beyond a focus on only language for producing meaning. By 'articulate' we mean that the contribution we set out to make is about providing a systematic and coherent approach for studies of leadership as a multimodal processual and relational phenomenon. We do this by linking theoretical conceptualization to methodological concerns, including how to write up and present research results. To this end, we gather and organize theoretical and methodological ideas that already exist, but that have not yet been coherently articulated in a systematic approach. The aim is to stimulate more empirical research based on practice perspectives by describing one possible way to study and write about leadership. We thus respond to recent calls for more consistency between theoretical conceptualization and methodological approach in leadership studies (Ashford & Sitkin, 2019;Gardner et al., 2020;Simpson et al., 2018).
MMLS attends to the multimodality of leadership by developing knowledge on how leadership is accomplished by language and by space, bodies and artefacts (Fairhurst & Grant, 2010). In this paper, we first define leadership as a socio-spatial process of accomplishing direction. We then propose that micro-ethnography is a viable way of developing knowledge by eliciting an ethnographic sensibility that uses the idea of 'going micro' in several ways. Micro involves both time and space: a short-term ethnography of processual phenomena (time) that focuses on the details of interaction and where it occurs (space). Thus, micro is also about the empirical material produced, providing such fine-grained details of actions and interactions that a close analysis is made possible. The empirical material is thus limited in amount yet comprehensive. Furthermore, micro refers to a distinct approach to writing up research and presenting empirical findings that we call 'closed vignettes'. Table 1 summarizes these different meanings of 'micro'. 1 The approach elicits ethnography since it shares with traditional ethnographies a concern for contextually embedded practices and meanings, and attention to practices of writing and representing (Van Maanen, 1988).
In the following, we will first expand on the background for this paper and recent calls for a more integrated approach to theorizing and studying leadership. From this, we proceed to outline a conceptualization of leadership, defining it as a socio-spatial process of accomplishing direction, and we carry on by elaborating on the implications of this definition in terms of situatedness, agency and power. In the section that follows, we outline how to practically go about in doing MMLS by micro-ethnography: short, intensive fieldwork attending to situations and interactions. After this, we introduce the notion of closed vignettes as a way of writing and efficiently presenting data, and discuss their potential in generalization and in representing multimodality. The paper concludes with reflections on the usefulness of the approach we suggest.

Background
In his seminal book Leadership for the Twenty-First Century, Rost (1993, p. 183) argued for 'a new school of leadership that is grounded in what is real, what actually happens when leaders and followers do engage in leadership'. Although there has certainly been progress made in this direction since the publication of the book, there is still a shortage of research engaging with leadership as a processual and relational phenomenon, studying what happens and what actors do when they participate in leadership practices where they occur. This leads to a persisting need for better understanding of leadership in order to renew contemporary organizations, lead change processes and work with societal transitions Clegg et al., 2021;Crevani et al., 2021).
Such an endeavour also demands connecting theory to research practice. There is 'growing awareness in the leadership literature that theoretical innovation is not in Table 1. Meanings of 'micro' in micro-ethnography.

Aspect of research
Meaning of 'micro' Fieldwork (time) Short periods, short-term interactions Fieldwork (space) Observations of interaction in context Fieldwork (material produced) Limited in size, but comprehensive Fieldwork (detail) Finely grained accounts of actions and interactions Level of analysis Close analysis of practices Writing and presenting Vignettes: short but comprehensive stories on local practices and situations itself enough to move the field forward. New theories must be accompanied by new methodological considerations' (Simpson et al., 2018, p. 658). Whereas leadership scholars have assisted to an increasingly rich theoretical elaboration on leadership as a processual and relational phenomenon (see, for example, the Special Issue in Human Relations, April 2020), we still have relatively few articles articulating in greater detail methodological approaches consistent with the theoretical perspectives proposed. Moreover, such considerations are seldom connected to how to present research, although this is a challenge acknowledged by scholars trying to stay true to processual and relational views (Hallin, 2015;Helin, 2015). MMLS is a label that we mobilize building on Fairhurst and Cooren's (2010, pp. 177-178) sailing guide for leadership studies, in particular, what they call 'the social construction of reality', where action and interaction through many modalities are central, rather than the cognitive and discursive sensemaking processes that are in focus for approaches informed by 'the construction of social reality'. Whereas we would agree that sensemaking processes are important, leadership studies cannot be reduced to only study such processes. The romance of leadership complicates the study of leadership from an actors' point of view (Bligh & Schyns, 2007;Jackson, 2005;Meindl et al., 1985), to the point that leadership is sometimes a mere fantasy (Sveningsson & Larsson, 2006). Thus, relying on actors' cognitive products is precarious, as 'talking about leadership and doing leadership does not always equate to the same thing' (Carroll, 2016, p. 92). This does not mean that actors' meaning-making and attributions are not worth studying: on the contrary. As the label micro-ethnography implies, they are key for understanding leadership processes. But just as researchers do not normally reduce, for instance, power to only comprise that which actors understand in terms of power, neither should leadership scholars limit their analysis of leadership to actors' potentially romanticized understandings of what leadership is and how it is accomplished. By detailed empirical analysis of mundane practices accomplishing direction in organizations (Crevani, 2018), we argue that it is possible to unpack processes that actors' may or may not label as leadership and thereby to engage in collaborative dialogue for developing practical wisdom on the phenomenon of leadership (Flyvbjerg, 2001).
Empirical studies exploring the multimodality of leadership are, however, still scarce (although we will draw attention to some examples in this paper). There may be numerous reasons for this lack of attention, but we would wager that two of those reasons are particularly significant (cf. Sutherland, 2018): (1) the prevailing understanding that 'ethnographies are too resource-heavy-requiring considerable time in the field' and requiring extensive 'financial backing and data analysis' (Sutherland, 2018, p. 5), and (2) a lack of clear conceptualizations of what to look for and how to write about it when studying leadership with a multimodal approach. Gardner et al. (2020) also note that given 'the complexity of the leadership phenomenon, the dearth of studies that employ […] context-rich approaches represents an opportunity for future research' ( Gardner et al., 2020, p. 20), and there is a 'need to broaden the methods employed' (Gardner et al., 2020, p. 24) in leadership research. However, as Ashford argues, there are 'considerable' challenges (Ashford & Sitkin, 2019, p. 457) in developing empirical approaches for studying leadership 'as it happens' in a 'context-rich' approach. This paper, therefore, takes the form of an invitation to further dialogue, made in the spirit of opening up for new empirically grounded ways of discussing and studying leadership.

Conceptualizing Leadership
The intellectual roots of our approach can be traced to constructionist leadership studies that conceptualize leadership as a collective achievement (Cunliffe & Eriksen, 2011;Ladkin, 2010;Ospina et al., 2020). Such approaches have evolved over the last half century and include studies of group processes and empowerment (cf. Fitzsimons et al., 2011), of shared managerial leadership (cf. Döös & Wilhelmson, 2021) and of approaches focusing on leadership as a relational process (Cunliffe & Eriksen, 2011;Dachler & Hosking, 1995;Hosking, 1988). These were, in turn, heralded by scholars such as Mary Parker Follet and her notion of 'power with' rather than 'power over' (Follett, 1940) and in Gibb's distinction between 'focused' and 'distributed' leadership (Gibb, 1954). In this paper, we specifically build on the leadership-as-practice stream of research (Carroll et al., 2008;Crevani et al., 2010;Raelin, 2016b) and approaches based on conversation analysis (Larsson & Lundholm, 2013) that have advanced an explicit focus on leadership as a process accomplished in interactions addressing 'how leadership emerges and unfolds through day-to-day experience' (Raelin, 2020, p. 3). Since leadership is theorized as located in the emergent and unfolding process of interaction, understanding talk has been key to understanding leadership (Crevani, 2018;Larsson & Lundholm, 2013;Simpson et al., 2018). More recently, leadership researchers have also brought attention to material dimensions of the leadership process by building on the more general turn to materiality in organization studies (see Carlile et al., 2013;Czarniawska, 2013;Gherardi, 2001;Robichaud & Cooren, 2013;Schatzki, 2006;but also Turner, 1971). This has opened up several avenues for addressing materiality in leadership (see, for instance, Fairhurst & Grant, 2010;Hawkins, 2015;Raelin, 2020;Van De Mieroop, 2020), including approaches in the leadership-as-practice stream of research (Alvehus, 2019;Carroll, 2016;Case & Śliwa, 2020;Sergi, 2016).

A Definition of Leadership for MMLS
For the purposes of the proposed approach to MMLS, and as a sensitizing concept (Blumer, 1954), we suggest that leadership is defined as a socio-spatial process of accomplishing direction. The notion of direction is central to many conceptualizations of leadership (e.g. Drath et al., 2008;Grint, 2005, p. 15;Smircich & Morgan, 1982, p. 258). By accomplishing direction, we refer to the way interactions shape unfolding developments of events and create a common space for action, thereby highlighting 'the ordinary doings of ordinary people as they co-produce directions for their work together' (Simpson et al., 2018, p. 646). Leadership is thus understood from within the 'flow of practice' (Simpson, 2016, p. 171)it is 'constantly in-the-making' (p. 172) and should be understood as fluid and emergent phenomenon. Preconditions for action continuously emerge and change as different co-related actors and their actions are woven together. In this sense, leadership is a relational process that can be characterized as spatial, not only situated but also taking place as actors and actions are ordered and reordered. For example, a meeting that ends in an agreement on 'what to do next' opens up some spaces for action and simultaneously closes others (cf. Larsson & Lundholm, 2013). This is an open-ended process that proceeds without necessarily being decisively influenced by any single actor (Crevani, 2018), nor definitively determining the course of future action. Divergence, multiplicity and even contradiction must be considered as part of what may emerge (Alvehus, 2019).

The Situated Character of Leadership in MMLS
Like any practice, leadership is situated in time and space. The social and spatial aspects are here treated 'not as a variable nor as background, but as a constitutive dimension of the relational dynamics that call forth leadership' (Ospina & Foldy, 2010). Hence how leadership is accomplished varies across contexts and to understand how direction is produced, we need to understand the context in which this takes place. Importantly, this means that the situation cannot be considered a neutral container, but it is rather a constitutive actor in the accomplishment of leadership, with potentially ambiguous character (Goffman, 1974). While never determining coming courses of action, situations suggest some while constraining others (Gherardi, 2019a;Massey, 2005). Moreover, whereas we propose a focus on situated interactions, we caution from reducing everything to local interaction, what Levinson (2005) has called 'interactional reductionism'. Situated practices and accomplishments are both local and global at the same time, since particular instances are resourced by, connected to and affect other practices in a broader texture of practices (Case & Śliwa, 2020;Gherardi & Rodeschini, 2016;Nicolini, 2009).
Drawing on Goffman (1974), we suggest one way of taking situatedness in consideration in MMLS by understanding leadership as emerging through how social situations are framed. This includes activities undertaken to shape how the situation is understood or experiencedpolitical manoeuvering (Alvehus, 2021;Ammeter et al., 2002;Empson & Alvehus, 2020) or stage setting (Alvehus, 2019), for example. As Goffman (1974) argued, staging situations and influencing the framing by which they are experienced is a complex and multi-layered process. As Alvehus (2019, p. 551) argued, 'we can never be sure that the leadership we experience-as managers [...] or even researchers-is the definite answer to what it is that is going on'. What we can do is try to understand any such framing from a particular point of view, acknowledging plurality (Fairhurst & Cooren, 2009) and possible emergent movements between different frames (Carroll & Simpson, 2012).
Multimodality is central to understanding framing. Meetings and other formal gatherings are often framed in space by material objects and their configurations, such as the walls and doors of meeting rooms, the positioning of chairs, tables and projectors, etc. (Schwartzman, 1989). Carroll (2016, pp. 93-94), for example, describes how a team in an open-space office moves from being engaged in individual work to a team meeting. They slowly start 'disentangling from their work' and 'swivel around looking around the area'. People start to move in before a bank of Post-it Notes, eventually standing in a circle. The Post-it Notes, a koosh ball, the layout of the open plan office, all provide a stage where people are 'transitioning' from individual work 'into something else without a word being said'. Body positioning, space, and material objects play a key role in social interaction (Goffman, 1963) and are a constitutive aspect of accomplishing direction as materiality is in and of itself a relational phenomenon (Gibson, 1979). Interaction, in turn, influences situations and objects in a dynamic process: 'Things can be translated into words, and words can be translated into things' (Czarniawska, 2013, p. 365).

Humans and Non-human Actors in MMLS
Different scholarly traditions have shed light on organizing processes being more than talk practices, involving both human and non-human actors. Within leadership, aesthetic leadership researchers have for example explored the role of the body and the senses (Hansen et al., 2007;Sinclair, 2005); scholars focusing on practice and/or communication have brough to the fore the role played by different objects (Carroll, 2016;Fairhurst & Cooren, 2009;Griffey & Jackson, 2010;Sergi, 2016), space (Ropo & Salovaara, 2019) and place (Hambleton et al., 2021;Robinson & Renshaw, 2022) in co-accomplishing leadership.
Building on this, the approach we suggest acknowledges how humans and objects in time-space configurations together accomplish leadership in interactions, and thus how leadership emerges 'even through non-deliberate material-discursive practices' (Raelin, 2020, p. 496). Such a theoretical positioning implies decentring of agency from humans to practices (Simpson, 2016). We thus acknowledge the reflexivity of humans, but see human agential capability as a result of taking part in socio-material practices (Nicolini, 2009;Reckwitz, 2002) in the same way as non-human actors are part of sociomaterial practices (Gherardi, 2019a). For example, Sergi (2016, p. 122) shows how a document in a project meeting provides a 'temporary link with the responsibility of assembling the otherwise dismantled team', how it 'encourages the team members to talk during the meeting, stimulating collective reflection' and how it plays a key part in organizing interaction and accomplishing direction.

Power in Our Approach to MMLS
The approach to MMLS suggested here not only addresses such aspects as time, space, text and toolsnon-humans, for shortbut it also acknowledges their political nature (Hawkins, 2015). Power is, of course, in and of itself, a vast and contested concept, spanning various ontological and epistemological positions. In the approach to MMLS suggested here, different aspects of power may come to the forefront and become relevant in different empirical studies. However, a key element is understanding power as performativesociety and in this case, direction, 'is performed through everyone's efforts to define it' (Latour, 1986, p. 273). This entails a non-entative view of power (Simpson et al., 2021) focusing on the emerging practice of humans and non-humans, and an understanding of power relationships as both deeply embedded in institutional contexts, and as fluctuating and fragile in processes of interaction. Such a view of power focuses on its 'emergence in relationships' and on its consequences as it 'come [s] to possess a lived validity' (Gergen, 1995, p. 36). An illustration from the leadershipas-practice stream of research is Simpson's (2016) example of the New Zeeland rugby team All Blacks. Simpson illustrates how stories of the past, and the potential of leaving a legacy, is embodied in, for example, the teams' signature black jerseys. The 'meaning of leadership … continues to evolve as the story of the team' (Simpson, 2016, p. 171). Another example is Van De Mieroop's (2020, p. 613) study, drawing on multimodal conversation analysis. She shows how roles and relationships are continuously negotiated and accomplished 'in interaction using both verbal and non-verbal means' and how even seemingly timid followers 'turn out to be highly agentive in the leadership process' (p. 615). In both these examples, power emerges in the relationships and in 236 J. ALVEHUS AND L. CREVANI interaction as events unfold; it is not an entity that can be held or wielded by autonomous entitative agents. Drawing on the terminology of Follett (1940), it is a question of power withnot power to or power over (Simpson, 2016). Relationships are embedded in, and by themselves constitute, power relations. Resources employed in human organizationstools and technologies such as bureaucratic arrangementssometimes stabilize power relationships (Latour, 1991). Bureaucratic organization, for example, can provide a blueprint that 'truncates' power relations (Smircich & Morgan, 1982, p. 259) and thereby influences the accomplishment of direction. Other non-humans can also assist in stabilizing social order (Latour, 2005). Some scholars argue that whereas verbal interaction is fluid and emerging, tools and spatial arrangements are often less so (Alvehus, 2019), as changing their configurations may require more effort (Case & Śliwa, 2020). We must remember, however, that stability is always temporary, in flux and potentially fragile; non-humans may be 'disordering devices' as well as 'ordering devices' (Vásquez et al., 2016, p. 652). A written document can provide stability ('We agreed on this'). It can also open new spaces of action ('What did we really mean by that?') and as noted by Holm and Fairhurst (2018, p. 716), 'finegrained struggles over meanings' destabilize formal and previously agreed-on arrangements. What stabilizes the social order, and how, we suggest should primarily be approached as an empirical question, related to notions of power.

Studying Leadership
To move close to direct data in terms of interactions and practice (Ashcraft et al., 2009), many leadership scholars have suggested ethnography, as 'we cannot really learn a lot about what "actually happens" or about "how things work" in organizations without doing the intensive type of close-observational or participative research that is central to ethnographic endeavour' (Watson, 2011, p. 204; see also Sutherland, 2018). Ethnographies traditionally involve prolonged engagement in fieldwork both by 'hanging out'observing everyday interaction in situand by 'asking questions', thereby gaining an understanding of people's lived everyday experiences (Dingwall, 1997).
Micro-ethnography characterizes a special kind of ethnographic endeavour. The approach is directly linked to the way in which we conceptualize leadership. Ethnographies have their roots in anthropology and are a way of dealing with the difficulties of understanding the meaning system of the Other (Geertz, 1973). Imported into organization and leadership studies, this is how ethnography is primarily used. In MMLS, leadership has a tangible ordering character. Therefore, the deeper societal and cultural meanings attributed to events by actors are not the only significant factor in understanding the accomplishment of direction. Actors' meaning-making is part of the process, but the process cannot be reduced to meaning-making alone. The approach involves attention to the sense making of the actors, for example, in how they interpret space for future action or how they extrapolate experienced courses of action into future action, but the main interest is how actions and interactions actually unfold, in line with the ethnomethodologically informed perspective in ethnography (Atkinson & Hammersley, 1994). What is to be explored with a micro-ethnography, therefore, is not whether human actors interpret a certain situation as being about 'leadership' or not, but if and how leadership (as defined in our MMLS approach) is accomplished in terms of direction. To this end, we articulate some recommendations for fieldwork.

Focus on Short, Intensive Fieldwork
'Micro' means shorter fieldwork informed by continuous dialogue with theory and multimodal means remembering to take into consideration more resources than discursive ones. The quality of ethnographic work is then conceived in a multifaceted way, in which 'long term' fieldwork is no longer synonymous with 'thickness' and the phenomenon of study guides how the ethnography is shaped. Recent debates on ethnography have challenged the common equation in organization and leadership research between long-term presence in the field and high quality of ethnographic fieldwork (Knoblauch, 2005). Also, the notion of the lone heroic ethnographer has given the way for an understanding of ethnographies as including all the entanglements in which ethnographic knowledge is produced: research questions are framed as access is negotiated, field notes are discussed among researchers, results are presented and discussed in parallel with fieldwork, writing starts before the analysis is completed and texts circulate (Pink & Morgan, 2013). This process may be rather intense, may quickly affect the course of ongoing fieldwork, and is heavily influenced by theory from the start. In multi-sited ethnography, for instance, the long-term presence in one place is abandoned in favour of following the object of study, which may be mobile or spatially dispersed (Falzon, 2009).
We therefore understand micro-ethnography as comprising shorter, intensive engagements in fieldwork combined with tight dialogue with theory. Micro-ethnography is not the only possible method, of course, but consistently with the theoretical approach for MMLS we have presented, it encourages direct engagements in fieldwork focusing on the situated accomplishing of direction as it unfolds.
The duration of the fieldwork is mainly an empirical question that must be determined by the purpose at hand. For example, Carroll (2016) presents a situation in which direction is accomplished by spatial arrangement, and the duration is but a few minutes. Sergi (2016) also focuses on a single meeting but chooses to present this meeting as a scene in a longer stretch of events. In line with these examples, we suggest that the time frame of MMLS should ideally be determined by empirical concerns: a question or problem that 'travels' in an organization could be followed until it is resolved or it dissolves. Since the flow of everyday life is bracketed, in both temporal and spatial terms (Boden, 1994;Goffman, 1963), social actors are able to identify beginnings and ends. These are, however, always arbitrary and should be used as guidelines rather than absolutes (Goffman, 1974). Any bracketing in MMLS will consequently be subject to analytical considerations and to such practical concerns as the access to and duration of fieldwork, as well as ambitions to zoom out on the texture of practices and effects of the practices under analysis (Nicolini, 2009).
'Where and when to start and finish?' is therefore a pragmatic rather than a dogmatic question. It depends both on the aspect of leadership that the researcher wants to explore and on the type of access possible. For example, Holm and Fairhurst (2018) note that in order to understand the power effects of 'slices' of interactions in meetings, it was necessary to follow the process over time, for five and a half months. Another example: in the design of an ongoing study involving one of the authors on digitalization and its consequences for the doing of leadership, the 'where' could be any situation in which digital technologies are presentmeetings via digital tools, for instance. The 'when' would not matterany starting point is finethe interesting things would be for 'how long' in relation to the focus of the study. In Table 2, we provide examples of the ways in which different study design decisions would affect the outcome and which aspects of leadership we would highlight.
The focus of a micro-ethnography approach is, therefore, not to trace what has led to what over timeanother research design would then be needed, for example, foregrounding different states succeeding each other on a time-line (see e.g. Empson & Alvehus, 2020, for an example in leadership). The focus we embrace in this approach is direction-in-the-making, change in practice as it happens. The aim is not to account for a process over time in the shape of progression or correlation linking the particular to the 'macro', but rather to develop a rich understanding of how the particular unfolds, thus linking it to the 'macro' by means of instantiation: it is in the unfolding of practice that macro-processes are constituted (Kouamé & Langley, 2018).

Focus on the Multimodal Through Studying Interaction
Conceptualizing leadership as a socio-spatial accomplishment means empirical work should focus on interactions between human and non-human actors in their multimodality in everyday life. Multimodality means that several resources contributing in this accomplishment are taken into consideration by the researcher, not only language, as is often the case in constructionist studies. Building on the approach here delineated, these resources are conceptualized as processual and relational, which means that objects, technologies, bodies, spaces, etc., exist and are reproduced in relationthey are not considered to be structural elements pre-existing to the interaction. Through ethnographic work, it is thus possible to account for what these resources become in the flow of practice, rather than what they supposedly are, and how they matter in the unfolding accomplishing of leadership.
Focusing on how direction is accomplished means following practice in situ in order to identify changes in direction. This empirical strategy depends on utilizing a sharp theoretical lens in order to identify changes (or lack thereof) in direction, however, as the researcher can no longer rely on the informants to determine if leadership was accomplished. Recent studies have used conversational turning points (Simpson et al., 2018) or body positioning and spatiality (Carroll, 2016), for example, to identify any such changes. Simpson et al. (2018, p. 656) show how different kinds of 'turning points in the flow of talk' affect 'movements and changes in trajectory'. Their study thus highlights the way in which talk transforms the situationit is a 'transformative practice'and how turning points shape the accomplishment of common direction. Carroll (2016) draws our attention to how a team's attention is directed to a common focus in an open workspace, accomplished by everyday and seemingly trivial material arrangements such as table layout, Post-It notes and swivelling chairs. In both these examples, careful attention to the minutiae of practice helps us understand how common direction is accomplished. To attend to this detailed level of interactions in MMLS, shadowing is an important method that enables the researcher to follow everyday activities as they evolve (Czarniawska, 2007). This method specifically allows for mobility, and, as organizing does not stop at the boundaries of formal organizations, it may require movement between sites. It also allows following non-humans, thus entering situations from different angles (Bruni, 2005). Furthermore, observations of interaction sequences is a central method that enables the researcher to develop an understanding of the minutiae of interaction. Studies drawing on conversation analysis and, more broadly, organizational discourse analysis (Fairhurst & Uhl-Bien, 2012;Schnurr & Schroeder, 2018) have highlighted the key role that language use plays in accomplishing leadership and shown how rich analyses of short sequences may be (e.g. Larsson & Lundholm, 2013;Simpson et al., 2018). As recent contributions illustrate, however, observations canand in our view, shouldinclude non-human actors and how they come to matter in the unfolding of the situation (Alvehus, 2019;Carroll, 2016;Sergi, 2016). Inspiration can also come from how conversation analysis has recently embraced multimodality in order to produce richer understandings of interactions (Oittinen, 2018;Van De Mieroop, 2020). In Oittinen's (2018, p. 49) study of video-mediated meetings, for example, dissent is enacted in local physical spaces and therefore, '[r]estrictions in access to bodily resources limit participants' ability to display and resolve disagreements in distant meetings'. The multimodal analysis draws attention to redistributions of power relationships where non-human actors play a central part.
To account for the participation of human and non-human actors, recording voices in interactions is not enough. In recent years, a number of ways of attending to multimodality in fieldwork have been proposed, for instance, using photos for capturing how space is configured (Case & Śliwa, 2020). Attention has been also brought to how to capture atmospheres and rhythms, for instance, by combining note taking foregrounding how the researcher is experiencing a situation with pictures of the situation and of specific artefacts (Crevani, 2019;Katila et al., 2020). The importance of the researcher's body and the understanding of fieldwork as the embodied practice has thus been highlighted in relation to affect and to space, among others (Gherardi, 2019b;Ropo & Salovaara, 2019). On the other hand, other approaches have focused less on the researches' embodied experience and more on the minutiae of interaction. In Van De Mieroop's (2020) study, bodily non-verbal interaction is brought to the forefront in interpretation (head shakes, gazes, eyelids closing, using a pen), enabling an extremely detailed account of practice and for how leadership is accomplished in subtle and inaudible manners, and thereby how non-verbal actions 'actually do leadership' (Van De Mieroop, 2020, p. 615). 240 J. ALVEHUS AND L. CREVANI

Focus on Situations
Since in our take on MMLS leadership is conceived as situated, researchers are invited to focus on specific interaction sequences in situ and to describe the interaction and its context in detail, bearing in mind that interaction and context co-constitute each other (Endrissat & von Arx, 2013). Alvehus' (2019) study of a core values workshop illustrates how tools, such as a board game and a slide presentation, and room layout co-create opportunities for framing. A manager invites to reproduce a 'casual, inviting, listening, and laissez-faire self, an image appreciated by his subordinates, which plays well into [the company] culture' whilst at the same time deliberately orchestrating the situation to accomplish this. This is in turn related to the singular event being embedded in a broader system of meaning and thus provides an in-depth illustration of that very system. Concrete interaction, situations and broader contexts are co-constituted although the relation between them is ambiguous and fragile. As with all fieldwork, it is important to be open to what emerges as interesting once the researcher becomes part of a context; some 'drift' is an integral part of the ethnographic effort (Van Maanen, 1979). In any account of a situation, the researcher must consider the role that the situation plays in the larger flow of organizational activities. Czarniawska (2007) has noted, however, that the researcher will always occupy a certain place in time and space, and therefore any account will always, by definition, be restricted. Thus, the researcher will likely also have to rely on the accounts of others, whether humans or non-humans. Such accounts are critical in order to develop a thorough understanding of both the organizational context and the way in which this relates to the broader social context. Moreover, it is key in order to develop an understanding for the reflexivity of actors and the meanings they ascribe to actions and events (cf. Garfinkel, 1967). Activities in practice are always embedded in a context of meaning and significanceof history and of imagined futures. That is why we retain the concept of 'ethnography' as we argue for the need to approach the phenomenon with an ethnographic sensibility, tuned to understanding with others rather than about others (Pink & Morgan, 2013). The starting point for empirical efforts is, therefore, concrete situations and their contextual embeddedness, and the interaction between these. An example of this can be found in Crevani (2018, p. 96), where different roles (e.g. 'sales person' or 'project manager') are enacted in practices accomplishing direction, shaping future activities. Negotiations take place around questions such as 'what should a salesperson do and how should a salesperson be? … [I]s the priority to give the best service possible to the customers or to spend time on administration?' In this way, the collective accomplishment of direction shapes future spaces for action, and in this, the context in terms of roles is reshaped. The interaction is thus embedded yet also reproduces and influences context.
Moreover, situations are not isolated events in a context, but are connected to other places and times (Hawkins, 2015), meaning that understanding the 'micro' allows us to move between emergent local practices and those that can be analytically characterized as having broader implications for the reproduction of organizations and society. For example, participation in a top management meeting may reproduce the taken-forgranted 'fundamentally undemocratic character of companies' in society as a whole (Alvesson, 1996, p. 167). The human and non-human actors gathering in the place where the researcher is present can be understood as stories-so-far, bringing the past and the distant into the here and now of the interaction (Massey, 2005;Vásquez & Cooren, 2013). Any such situation will have its own idiosyncrasies and contextual circumstances, but to the extent that it is plausible to the reader, it provides a fundament for understanding similar situations, as the instance illustrates the general pattern (Kreiner, 2012;Van Maanen, 2011). This epistemological stancegeneralization by theoretical implication of empirical similarityhowever, requires considering how we construct and relate to empirical material, something we will expand on in the next section.

Writing Leadership
The 'micro' in micro-ethnography also means 'short accounts' and we suggest that one way of writing such accounts is to focus on 'closed vignettes', that is short and comprehensive accounts that present all material relevant for an interpretation. Any account of an event will inevitably be a montage (Denzin & Lincoln, 2011) and textual representation accounts for only a small amount of what is happening in a situation. Utilizing technologies other than a notebook and audio recorderstechnologies such as a video camera or other tools for visual methodscan assist in micro-ethnography, both for coping with the level of detail and for developing the way we account for materiality (Pink & Morgan, 2013;Warren, 2002). Even when using photos or video, however, we should not fall into the trap of naïvely thinking that we are reproducing reality accurately (Bell & Davison, 2013). Narrative forms and tropes will shape accounts and affect their representation (Van Maanen, 1988). In this paper, we do not develop this point further, as it is well established in qualitative research; rather, we elaborate on ways of managing such issues of representation.

Present Closed Vignettes
Vignettes are short and comprehensive accounts of empirical situations, of the unfolding of practice. By 'closed vignette', we mean that the empirical story presented is that which is interpreted. This means that contextual aspects of relevance for the analysis must also be included in the vignette. Thereby the researcher is also forced to show how these contextual elements are relevant for the situation at hand. This approach echoes the position taken by Alvesson (1996, p. 45, note 5): 'the documentation is presented […] in its entirety. The reader thus has access to the same written material as I myself have had'. In other words, the reader has the same interpretive position as the researcher. The point of the closed vignette approach is to prevent the researcher from explaining empirical events by other empirical events unknown to the reader, and to focus, instead, on the utilization of theory.
This restriction can seem confusing at first. Will there not be underlying motives and tactics of influence by the actors? Will there not be events preceding and following the situation presented? Will other significant processes not be conducted in parallel? The answer is, of course: Yes, there is. And there always will be. Contextual elements that are of importance for the events described in the vignette must therefore be included in the vignette, and the way they enter the picture should be accounted for. Building explanations on events beyond that which is accounted for in the vignette is a position where readers are presented with explanations they cannot relate to. Explanations will then be depending only on where empirical disclosure begins and ends, rather than relying on robust theoretical and explanatory constructs. Therefore, the spatial and temporal bracketing of the vignette must be explicitly accounted for.
The vignette is the entire dataset in our MMLS approach, and in this sense, it is 'closed'. The point of the interpretive endeavour is to provide thought-provoking interpretations that can, for example, sensitize the reader to particular situations and encourage theoretical insight (Alvesson, 1996). Closed vignettes are created and selected in the dialogue between fieldwork and theory development, and enable detailed accounts of practice (see Mailhot et al., 2016). Besides the examples already mentioned, another study following this kind of writing is Lusiani and Langley (2019) about enabling leadership practices and the construction of strategic coherence. The authors present several vignettes, using graphical elements to separate them, that they then analyze focusing on mundane activities in interaction among people and tools.
How can we expect to see the accomplishment of direction in such vignettes? As we have conceptualized it, the direction is accomplished by opening up for certain spaces of action and not for others, and the concept enables us to see which spaces supplement others. For instance, one might make observations in a technology firm with open-plan offices, and a vignette could start with a physical description of the office space, what people do there and how it feels to be there. It could then move on to describe people and objects in the focus of the story before accounting for an interaction sequence, perhaps terminated by the researcher having to leave with her subject. Depending on the detail reported and the time covered, such a vignette can differ in length. The situation described here is, in fact, an example from Carroll (2016) that comprises about 1000 words in its original form. Comprehensive and rich, it enabled Carroll to discuss notions of practice, physical space, routines, artefacts and how they together coconstruct leadership.

Generalize Theoretically from Closed Vignettes
Closed vignettes provide a basis for theoretical generalization by authentic, plausible and critical accounts of practice. They talk to us by representing what we understand leadership to be and by challenging our understanding of what it can be. As anyone doing empirical research can confirm, what happens in the field tends not to conform easily to our theoretical presuppositions, which are abstract by nature. There will be friction between the ethnographic account in the closed vignette and the theoretical conceptualization, and this friction allows us to extend and nuance our vocabulary and our ways of understanding how direction is accomplished. Golden-Biddle and Locke (1993) argue that ethnographic texts convince the reader in three ways: (1) authenticity, or 'being genuine to the field experience as a result of having "been there"' (Golden-Biddle & Locke, 1993, p. 599); (2) plausibility, which means that the story has to make sense to readers but not necessarily conform to their expectations; and (3) criticality, or 'the ability of the text to actively probe readers to reconsider their taken-for-granted ideas and beliefs' (Golden-Biddle & Locke, 1993, p. 600). These three criteria, in turn, depend on an agreement between reader and author on the ontological and epistemological assumptions of a study, such as those provided here. Kreiner (2012, p. 402) has argued that the validity of case studies is based on their representing 'a localized and time-specific practice and aspire to see it as part ofand as an illustration ofa more general pattern of practice'. Most events in organizations are not idiosyncratic. Even extreme or rare events will potentially help illustrate more general patterns to be found in other places and times, and thus sensitize us to the way such patterns are enacted in our everyday lives. As Alvehus' (2019) study illustrates, rarity of the situation is per se something of significance for its interpretation, and the meeting itself becomes a space for the manifestation and reproduction of power relationships. Carroll's (2016) study, on the other hand, shows a routine occurrence in an organization and demonstrates why this routine-ness is in itself significant for its emergence. Thus, the kind of generalization would generally depend on the empirical case and the theoretical issues at hand.
Aim at re-producing the Multimodal and Processual Through the Vignettes The researcher writing closed vignettes must deal with the reproduction of practice in a way that resonates with the readerthat provides adequate tension among authenticity, plausibility and criticality. As with ethnography, the researcher needs to organize the text carefully in order not only to report utterances and actions but also to convey a sense of emergence, movement and rhythm. Empirical material is often presented as linear narratives, which may inadvertently imply linear causal chains of events (Helin, 2015). Here, researchers may look for inspiration in fiction and how authors work with these issues; drama, thriller and comics are genres that could be explored (Hallin, 2015). This requires other skills than traditional academic writing. Researchers could work with the style of the presentation in order to create a certain rhythm, for instance, by working with the length of sentences (Hallin, 2015), or they could present rhythm in other formats, for instance, with diagrams (Robinson & Renshaw, 2022;Simpson et al., 2018). Conversation analysis has always brought forward the importance of details in interaction and we can learn from this tradition how important sequences of illustrations of empirical situations can be. Both Oittinen's (2018) and Van De Mieroop's (2020) articles show, for example, the usefulness of illustrations when drawing the reader into the stream of events: by illustrating, for instance, hand gestures and gazes with drawings, a deeper understanding of the physical embodied situation is enabled.
The embodied emotional dimension of writing may also be important to consider to reproduce how it felt to be "there" and to vividly depict the situation by narrating details that we may otherwise omit. As, for instance, in this example (albeit not explicitly on leadership) by Katila et al.'s (2020) study of a Nordic start-up accelerator, where the emotional state is conveyed: 'My heartbeat quickens and I feel hot, sweaty, pained, and desperate' (p. 1319). Documentaries are also a genre that can be used to present research and attend to this dimension of the phenomenon studied, something that some academic journals accept nowadays (see Salovaara, 2014). Another example is the following, from a vignette about virtual meetings: Then I became aware of myself on the screen: blown up and pale face, tired eyes with dark rings and a hair-do I was not aware of having. The white T-shirt I was wearing did not contribute positively to the worn-out face I was looking at in the small picture of myself on the screen. I was really disgusted with my looks and felt very unprofessional, untidy and unprepared. I looked at the woman from Asia and felt even worse. There she was, smiling with hair neatly and fashionably arranged, beautiful white skin with elegant make-up and wearing a nice white dress. (De Paoli et al., 2014, p. 60) As closed vignettes can be relatively short, this can also create space for other forms of representation. Video analysis provides 'a "microscope" for an in-depth study of the ongoing production of situated social order' (Knoblauch & Schnettler, 2012, p. 335), for example, and can be presented as a combination of talk and images, or even as filmed reenactments (thereby preserving anonymity). Whereas conversation analysis focuses on conversations rather than the unfolding of multimodal practice, multimodal conversation analysis with anonymized pictures that show bodies, objects, spaces, and how they are positioned in relation to each other, moment-by-moment (Van De Mieroop, 2020) can offer inspiration for how to produce an account of the multimodality of accomplishing leadership.

Concluding Remarks
In this paper, we have attempted to formulate a coherent approach for multimodal leadership studies (MMLS), from theoretical conceptualization, through empirical work, to the writing and presentation of research. We understand leadership as a socio-spatial process of accomplishing direction, embedded in a spatial and temporal context. The framing of leadership situations is ambiguous and dynamic and to understand this, multimodal approaches involving time, space, materiality and embodiment are needed. In understanding the interaction process, attention needs to be paid to practices involving both human and non-human actors, thereby decentring agency and power.
The main contribution of this paper is to draw together some key trends in current research on processual and relational leadership, and in particular leadership-as-practice, thereby bridging the perceived gap between novel understandings of leadership on the one hand, and methodology and writing on the other (Ashford & Sitkin, 2019;Gardner et al., 2020;Simpson et al., 2018). We have in this combined ideas from various strands of research. Although none of the components we present is new, we have assembled them in a coherent approachmicro-ethnographythat provides a clear ontological and epistemological stance on which to build empirically grounded MMLS. Bringing multimodality into leadership studies gives rise to challenges for researchers in both field work and in writing up and presenting research findings, and the approach suggested here is a way of meeting up to such challenges. What empirical findings will emerge from such approaches remains to be seen, but clearly, leadership researchers will need to grapple with terminology that turns our attention towards 'movement, flux, flow, emergence, passage, continuity, confluence, diffluence, turbulence, and smoothness' (Simpson, 2016, p. 165), and similar ideas, thus inviting new vocabularies for understanding how direction is accomplished collectively.
As we mentioned initially, this is not an attempt at providing the approach to studying the multimodality of leadership practice; it is about providing an approachand we hope others will describe other approaches. What we have described here is an attempt to present a clear and coherent account of a particular ontological and epistemological stance for MMLS, weaving a pattern from threads already existing. By micro-ethnography, we suggest an approach that is sensitive to context and embeddedness, focused on practice unfolding in actions and interactions between human and non-human actors and that presents its empirical findings in a comprehensive and theoretically informed manner. Most importantly, the approach invites empirical studies and therefore encourage more empirically grounded insights on the way leadership is accomplished. The ultimate results of these efforts are to help us see leadershipand the absence of leadershipmore clearly. We hope to contribute to the development of the vocabulary of leadership theory, and thereby expand our space of action and reflexivity in studies of leadership and organizational changebut also to develop opportunities for informed, collaborative dialogue on leadership in organizational practice, something needed in order to handle change necessary to address the societal challenges of our time.

MAD Statement
This paper sets out to make a difference by providing a coherent approach for studying leadership as an emerging practice empirically. In order to reinvigorate leadership studies in line with recent theoretical developments towards relational and processual conceptualizations of leadership, researchers need approaches that enable and encourage engagement with leadership in situ, focusing on how direction emerges in interaction among several actors. We propose one such approach by gathering and articulating existing ideas and methods, thus creating a coherent line from the conceptualization of leadership, through empirical methods, and to the writing up of leadership research. Note 1. The term 'micro-ethnography' was used already by Basham and DeGroot (1977) to describe the character of 'urban anthropologies' that focused primarily on individuals rather than societies ('macro') or communities ('middle'). Our use of the term here is in line with the idea of a smaller scope, but we develop and refine the concept significantly in relation to Basham and DeGroot; see also Wolcott (1990).

Disclosure Statement
No potential conflict of interest was reported by the author(s).