Abstract
Process approaches are increasingly applied in qualitative studies in many fields within social sciences. Yet, few studies have seriously elaborated on the ontological premises of process theorizing. This study addresses this void by suggesting a process philosophical framework. The framework is ontologically grounded with the concepts of causality, spatiality, and temporality in process theorizing. We use these tenets for developing three process theorizing techniques – articulating, relating, and conjugating. Articulating denotes to effectively expressing the potential identifying and generative properties of the process. Relating is the technique by which one maintains continuous connections within and between reified properties of a process. Conjugating is the technique by which a process’ identifying and generative properties are pulled together from various temporal and spatial sites in order to form a novel nexus. Each of these techniques builds on process philosophy and process theory and is illustrated through examples from prior process studies.
1. Introduction
The word ‘process,’ in its most general sense, means something that proceeds, goes forward, or moves along. More concretely, one may say that ‘the process’ of something is the passage of that thing (Whitehead, 1920). Whereas process approaches were originally applied to understand the evolution of human and natural species in natural sciences and existential thinking in religion and philosophy, process thinking seems to be increasingly entrenched within the social sciences. Processes are commonly understood as a series of actions, operations, or functions continuously performed over the course of time so as to produce, develop, or treat a change directed to some end. Studies of organizational communication (Ashcraft, Kuhn and Cooren, 2009), cooperation (Okhuysen and Bechky, 2009), and the management and organization of work activities (e.g. Malone et al., 1999), to name a few areas, have gained increased attention from a processual viewpoint in the past decade. This is perhaps because ‘[p]rocess research enables researchers to address important questions that lie at the heart of management and organizational life’ (Langley et al., 2013: 10). However, the authors note that constructivist and process ontological studies still need more elaborate methods to ‘unravel processes as they happen so as to develop an understanding of their underlying logic while providing a theoretical interpretation that reaches beyond description and can speak to other situations’ (Langley et al., 2013: 11).
Fortunately, much work has been done to anchor process studies into existing process-based ‘grand theories,’ including structuration theory (Giddens, 1984), the actor–network theory (Latour, 2005), the mangle of practice (Pickering, 1995), and practice theory (Schatzki, Knorr-Cetina and von Savigny, 2001), among others. These ‘grand theories’ have been employed as lenses that shed new light on phenomena that had previously not been fully understood (e.g. Barley, 1986; Holmström and Robey, 2005; Orlikowski, 2000). However, process-based theories are often complex, with concepts and general propositions operating on a high level of abstraction that give the researcher little guidance in practice (Pozzebon and Pinsonneault, 2005). Process data is associated with particular challenges: it often involves multiple levels and units of analysis with ambiguous boundaries; it is eclectic; and its temporal embeddedness often varies in terms of precision (Langley, 1999). These challenges place demands on the researcher to capture a complex reality by simplifying it and structuring it in an iterative fashion (Pettigrew, 1990). Furthermore, the received literature, which aims to help researchers with tools and techniques, builds on the traditional ontological assumptions of a world consisting of fixed things rather than processes. Thus, they do not equip scholars with the necessary discourse needed for the process theorizing of social phenomena. It is therefore unfortunate that techniques for process theorizing are still in their infancy (Langley et al., 2013), and have not been particularly adapted to understand complex organizational phenomena in general, and individual level process analyses in particular (Klag and Langley, 2013; Langley et al., 2013).
The aim of the present study is to address this void by developing a meta-theoretical account addressing the gap between process ontological premises and process theorizing. The paper is structured as follows. In section 2, we discuss the ontological premises for process theorizing. In section 3, we develop a process-based framework of process theorizing by drawing on the ontological assumptions made about the concept, and we then provide examples of how prior studies have elaborated on the framework’s components. Furthermore, we demonstrate how our framework may contribute to better capture the processual nature of our everyday life in theory. We round up by summarizing our key points and address some potential areas for further development.
2. Ontological premises for process theorizing
The process perspective represents a major shift in one’s perception of the world. From the dominant perspective, focus is placed on the entities the world is made up of. But from a process perspective, we’re interested in the processes in which these entities have been shaped and are continuously re-shaped. The Greek philosopher Heraclitus’ saying that ‘No man ever steps in the same river twice’ captures this change in perspective: the river is not a stable entity, but a continuously changing process. The same perspective can be applied to social organizations – what then becomes interesting is not the structure or the form of the organization, but the processes of organizing (Weick, 1969/1979). Thus, in order to expound the process approach, it is appropriate to distinguish between two different perspectives. From an entitative point of view, the world consists of things that change over time. When change is studied, the focus lies in how a thing is transformed from one state to another; for example, how a small firm goes through certain states as it grows (Churchill and Lewis, 1983). From a processual point of view, however, the world is a process, and thus in continuous flux and change. What we perceive as entities is nothing but stable manifestations of processes (Rescher, 1996). Thus, the things studied from an entitative point of view are from a processual point of view nothing more than manifestations of processes at a certain point in time (Whitehead, 1929/1969). Throughout this paper, when we use the words ‘entity’ and ‘thing’ we refer to this understanding of the concepts.
However, the change of things is intrinsically bound to the process to which they belong, and the conditions that make the process what it is over time and space. Taken seriously, this shift in focus has radical consequences for process theorizing because it considers action rather than outcome, contingency rather than entity, time rather than state, and spatiality rather than locality. These shifts open up new ways of seeing not just familiar objects in everyday life, but also their continuous becoming. Thus, the process view offers an opportunity to study phenomena in a process of becoming (as opposed to being); or more stringently, organization as a process (organizing) rather than a structure (Weick, 1969/1979), knowledge as an ongoing process of knowing/learning rather than a stable entity or object (Cook & Brown, 1999; Orlikowski, 2002), and to pose questions related to the interface of stability and change (Farjoun, 2010). We therefore see a true challenge in how to make use of these opportunities without resorting to a secure and stable entitative view of the world.
Ontologically, the process view holds that ‘physical existence is at bottom processual; that processes rather than things best represent the phenomena that we encounter in the natural world about us’ (Rescher, 1996: 2). This resonates with Whitehead’s (1925/1967) critique of ‘scientific materialism’ by outlining two basic fallacies: (i) the error of treating an abstraction as if it were real; and (ii) the assumption that anything that is real has a simple spatial location. In other words, abstractions should be related to processual sites rather than located in physical events, and they should also be understood as metaphysical constructs rather than physical (objective) facts. These ontological assumptions have implications for (a) the locus and direction of causality; (b) the perception of time; and (c) the treatment of space or sites. Let us now briefly present each one of these assumptions.
2.1. Causality
Causation is perhaps one of the most elusive concepts of science. It has thwarted thinking and the development of sound theories ever since the first systematic attempts at abstracting and bundling relations between and within complex social and natural phenomena. Still, many scholars approach it in an unproblematic fashion, and see it as a unidirectional concept. For example, by assuming that when entity (or event) X is placed in relation to entity (or event) Y, it exerts certain effects so that entity (or event) Z emerges or happens (as proposed by Hume, see Weyl, 2009). A processual approach to causation offers a different conception, which can be summarized in the following three distinctive ways.
First, process philosophy places the locus of causality in relations between phenomena, such as between subjects and object (Gibson, 1979; Merleau-Ponty, 1962), or in heterogeneous networks of humans and non-humans (e.g., Latour, 2005) rather than in the entities themselves. Put differently, ‘there are no elements without relational connections or relations without elements’ (Luhmann, 1995: 20). Viewing causality as a relational matter therefore ignores substance as something with inherent causal powers. Causality is conveyed in novelty (Whitehead, 1978) or differences that may occur when subjects and objects are observed in relation to one another (Bateson, 1972). Thus, to convey novelty or differences we need to focus our attention not just on identifying the entities and the events that are so intrinsically bound to the causation of things, but more importantly, to understand how causation emerges and matters to the observed phenomenon: ‘The outside observer cannot stop where the network stops, but digs deeper into the layers and chains of causation. Outside observers do not observe first-level whats, but second-level hows’ (Fuchs, 2001: 39). According to this view, attention should be given to events rather than entities, to multiplexity (overlap and diversification) rather than unobtrusive repetition in causal relationships.
Second, although we take it as a given that process causality figures in relations, we need to exercise some degree of discrimination between entities and events of different kinds in order to unravel relations that effectively ‘bind.’ This is important because causation is confined in relations between specific types of events, not just general types of events. As proposed in the theory of logical types (Whitehead and Russell, 1925), causal relationships are confined in discrete levels of abstraction such that the word ‘cat’ belongs to the discursive class of objects, but does not have fur and cannot scratch, and thus does not qualify as that which it denotes, i.e. the animal made of flesh and blood (Bateson, 1972). To use another example of causal judgments: ‘When I put my hand in the fire I burn myself,’ concerns a typical performance described by the words ‘to put one’s hand in the fire,’ not an individual act in which the motion of the hand and that of the flames is determined in the minutest detail. (Weyl, 2009: 56)
We may say that such types of events have a direct ‘proximate relevance’ in a causal relationship (Whitehead, 1978). This proximate relevance of events is not confined in space or time; yet it is somewhere, and it denotes the potentiality rather than the actuality of causation in relations between specific types of events.
Third, and as a consequence, causality is not linear with a fixed starting point in space or time; it is circular (Sanders, 1993), and points both towards the subject and the object (Gibson, 1979), ‘induc[ing] two or more mediators into a coexisting causality’ (Latour, 2005: 108). According to this view, causal efficacy is seldom judged instantaneously, but instead measured a posteriori by the consideration of three vectors that mutually constitute the presence of causality: (i) the past actualities of events; (ii) the future potentialities of new related events or the continuation of past and present events; and (iii) contemporary events, which happen in causal independence of each other (Whitehead, 1978). Thus, for each movement or observation of an event in a time–space continuum, these three vectors of causal efficacy will follow accordingly, each with its relative adjustment and novelty such that new layers and chains of causal networks are conveyed. With this follows the notion of multiple contributory singular causes. As clarified by Whitehead, when we feel the heat from a fire, the sense of heat is not the sole cause of the burning fire, nor is it the cause of our body feeling the heat, but ‘it is an interaction within nature’ (Whitehead, 1920: 31; emphasis added). Put simply, causation is not unilateral in the sense that we can plot its entities along a timeline.
Circular causality is, however, not unique to process philosophy, but it does entail certain ontological considerations regarding the progression of events, i.e. the generative properties of a process. For example, when considering the generative properties of a process, we tend to think of action as a logical consequence of the end or purpose of certain aspirations or intentions – its teleology. However, as proposed earlier, the teleology of actions and events is never linear and unidirectional. To illustrate this point, the conception of the realization of organizational strategies has a tendency to fall short of the causal links between the intended process to realize goals – for example, to increase market shares in a certain way – due to the difficulty of calculating and being concerned with the environment surrounding the focal organization and its members. Instead, as Mintzberg (1978) observed, strategies more often than not tend to be realized through the interplay between ex post facto results of decisional behavior and a priori guidelines to decision-making. Therefore, process theorizing is inclined to avoid linear thinking that typically generates such ‘teleological fallacies’ (that the ends determine the process) by means of an ‘inversion of causality in time’ (Bateson, 1979: 60).
Having made the case for causality as something confined to relations of specific types and with circular characteristics in process theorizing, let us now turn to the conception of time.
2.2. Temporality
Needless to say, our perception of time, and our corollary treatment of it, has been subject to linear thinking whereby events and episodes occur according to a process. Process research is therefore bound to generate longitudinal data: be it ethnographies or recursive interviews to elicit critical events, turning points, or simply the biography of an organization or individual over time such that ‘each wave of data collection is used to inform the next’ (Neale, Henwood and Holland, 2012: 5). Such prospective designs enable us to understand the causes and consequences of change ‘backwards,’ but they tend to negotiate away the important balance between the identifying properties (those that distinguish certain things from others) and generative properties (those that tell what things can do and cannot do) of events, causes, and consequences of change along the process. Both are important. Identifying properties do not tell us what they are capable of, only what they consist of. Generative properties tell us of their capacities, but leave their distinguishing features black-boxed.
Processual theorizing requires an intervention both into the identifying and generative properties of events, causes, and consequences. There are two reasons for this from a process ontological perspective. First, Whitehead’s (1920) critique of empiricist accounts of reality led to the proposition that the world is made up of ‘actual occasions;’ that is, the intrinsic character of an event at a certain time and in a certain place (or site). Second, time is subject to measurement, but only as a derivative form of duration (Whitehead, 1920). This means that process analysis requires an appreciation of the ‘passage’ of the time of an event, as passage both marks the durability of space (or nature) and the procedure of knowing – the ‘sense-awareness’ of an event. In this regard, time becomes the measure of the identifying properties of an event, which helps distinguish certain happenings from others (or at least the possible interconnection of certain happenings with others).
Put differently, the duration of an event yields a certain knowledge content that separates it from the duration of another event. The length of an event, as measured by its time (actual or perceived), signifies the expected knowledge accumulated (or refined) in the activities that constitute that event. However, because duration is not necessarily signified by undisrupted time, we should also expect that, despite the absence of observable activity, the knowledge accumulated (or refined) signifies the event. Knowledge is unbound to observable activity-time, yet it constitutes it through its own duration. Accordingly, as Whitehead (1920: 59) puts it, ‘there are no maximum durations and no minimum durations.’ In this way, there is a circular causality between time and event.
As an illustration, consider the following extract from Czarniawska’s (2004: 776) reflection on the introduction of new structures in the city administration of Stockholm, Sweden: All important events happen at some other time, in some other place. In the beginning the researchers tend to panic and try to chase ‘the action,’ but in time they learn that ‘important events’ become such in accounts. Nobody is aware that an important event is happening when it takes place, although in most cases people are aware of the time of day and the day of the month. Events must be made important or unimportant.
It is tempting, as the quote above illustrates, to think that all important events have been ‘captured,’ and that what has been excluded does not play a role; or if it does, at least plays a marginal role in understanding change, the flow of time, and how things relate to one another. However, as put forward by Whitehead (1920: 34), ‘[w]hat happens in time occupies time.’ It is the event that signifies the measure of time rather than the ordinary flow of chronological time that purports the generative properties of an event. Yet, without the ordinary flow of time it would be difficult to relate to the potentiality of action, i.e. that this or that should take place once this or that is realized (at some point in time).
2.3. Spatiality
Space is a problematic phenomenon, yet social scientists tend to disregard its complexity, and the similar yet distant meaning of the related words place, location, and site. It is a common mistake in analyses of organizational events and phenomena to treat space (i) invariably as a physical place, (ii) as an instant, and (iii) that these phenomena and events are unbound to time. It is not the primary aim of this study to outline a correct meaning, as the concepts are both philosophically disputed (Malpas, 2006; Schatzki, 2002) and theoretically challenged (Massey, 2005). However, it seems reasonable to agree that time and space mutually constitute one another with respect to measures of human activity (Schatzki, 2010). Our brief elaboration of space is therefore not aimed at complicating this further, but to provide an account that resonates well with a processual theorizing of events.
Most contemporary thinkers agree that space is more abstract than place (Tuan, 1977), yet the concepts of space and place are dependent on one another (Malpas, 2006). Whereas place can be understood as whatever stable object catches our attention (Tuan, 1977: 161), space has both a literal and figurative meaning: ‘Space lies open; it suggests the future and invites action’ (Tuan, 1977: 54). Yet although it provides a sense of choice of action, it is a threat to human action. Schatzki (2005) makes yet another distinction, suggesting that ‘[a]bsolute space is the site of all spatial locations (e.g. the location of a building or battle) because any spatial location is inherently part of it. Absolute space is not, however, the site of events and entities’ (Schatzki, 2005: 468). In contrast, places are made – deliberately or by sheer emergence – to satisfy our need for extensions of our bodily dwellings. Place is therefore ‘an organized world of meaning’ (Tuan, 1977: 179).
The intricate relationship between space and place can be described further in terms of ‘situations’: [T]o talk of ‘situation’ almost invariably introduces topological, that is, place-related, considerations. To be in a situation is to be ‘placed’ in a certain way, and, typically, such ‘placing’ involves an orientation such that one’s surroundings are configured in a particular way and in a particular relation to oneself – just as one is also related in a particular way to those surroundings. (Malpas, 2006: 40)
Thus, to be in a ‘situation’ is not only to be randomly situated in a place, but also to be engaged, dwelling in something or, more precisely, being immersed in practice (e.g. teaching, consulting, operating a machine). However, ‘human activities do not have a spatial site – there is no type of space of which they are inherently part’ (Schatzki, 2005: 468). The spatial location of an activity is only one genre of sites among many others (Schatzki, 2010). Practices are therefore the site of activities, but not their spatial site. Similarly, the progress or going forward, the process by which a practice is lived, or the process to which many interlinked practices belong has neither a particular spatial site nor physical location which they are inherently part of. This notion of site echoes Whitehead’s (1920: 86) notion of space: ‘there is a separate timeless space corresponding to each separate temporal series, that is to each separate family of durations.’ Durable material sites are different yet related to activity sites.
In this regard, the site of a process is its context, both in terms of the material and the activity sites through which the process transpires and becomes recognizable at every temporal instance of observation. Therefore, the site of a process, just like the site of a practice, ‘enjoys powers of determination’ (Schatzki, 2005: 468): it makes possible the separation and assimilation of components that belong to the process’ identifying and generative properties. What makes this view of spatiality powerful for processual analyses is that it provides us with an ontological lens, which helps us distinguish variably between space, place, site, and situation, as well as how these mutually constitute one another. Given the various types and levels of abstraction of spatiality, we suggest that rather than focusing on one process site as an instant or proxy for the locus of activity, process theorizing is bound to take multiple sites into account. In this regard, process theorizing is polycontextual, and comprises the research field or object as a selective totality of process contexts, as well as the research activities intrinsically bound to it.
2.4. Summary
The process approach provides an opportunity to pose new questions about familiar phenomena by focusing on the processes in which things are shaped and re-shaped, as opposed to just the things themselves. On a fundamental level, this shift challenges our assumptions about three elementary concepts of science: causality, time, and space. We argue that the way these building blocks are conceptualized and incorporated into the research process will have consequences regarding the extent to which the processes under investigation are unraveled and the underlying logic is understood, and thus the extent to which the theoretical statements generated might be generalized. In the next section, we will outline an analytical framework for how these concepts constitute process theorizing.
3. Mangling the process: towards a process analytical framework
Having established the underlying assumptions of our framework, we will now turn to the specific parts of it by first defining them, discussing them systematically, and then clarifying them with examples. However, let us first start with what it means to ‘mangle’ the process. We borrow the term ‘mangle’ from Pickering (1995) to denote the dialectic structuring of resistance and accommodation across time and space. Resistance and accommodation are reciprocal processes, and are not merely social phenomena as we understand them from our lived everyday experience; they also have both material and temporal dimensions. Resistance at a certain point in time and in a particular place or site of activity can foster a response in another point in time and through alternative means. For example, as researchers we often encounter resistance towards our ideas. However, over time and inspired by the critics (reviewers and editors), we appropriate these ideas and make them more compelling and logically coherent: we accommodate to the requirements of the past and the spatiotemporal conditions of the present. Mangling the process of events and actions is thus ‘a performative one’ (Pickering, 1995: 21), as resistance and accommodation are performances of both human (how involved individuals act and react) and material (e.g. notes, transcripts) concepts (e.g. theories, frameworks), and temporal (how the past, present, and future relate to one another) conditions of the research process.
It is important to note here that we depart from the process part; the verbs that capture the doing of theory rather than the end result or outcome of our actions. We should also emphasize that the framework offered below is an attempt to flesh out Bateson’s (1972: 75) ideas about theorizing: [W]henever we start insisting too hard upon ‘operationalism’ or symbolic logic or any other of these very essential systems of tramlines, we lose something of the ability to think new thoughts. And equally, of course, whenever we rebel against the sterile rigidity of formal thought and exposition and let our ideas run wild, we likewise lose. As I see it, the advances in scientific thought come from a combination of loose and strict thinking, and this combination is the most precious tool of science.
Indeed, we argue that this loose and strict approach to thinking has some important implications for process theorizing. Our process approach has an ultimate function. It is not ultimate in the sense of revealing some absolute truth, but rather suggests some stages in the thinking process that make ‘unconcealment’ (Wrathall, 2011) possible. We consider the mangling of process as a technique by which researchers of qualitative material unconceal; that is, in the Heideggerian sense of ‘making available’ the identifying and generative properties of ‘actual occasions.’ Such properties, we argue, are not predisposed to the domain of loose, creative, or otherwise less conceptually driven approaches to theorizing. Striking a balance between loose and strict thinking helps to unravel properties of ‘actual occasions’ that both address strengths and limits of prior thought on social phenomena. However, the task is not completed here if it fails to reflect the theorizing process itself, i.e. if it black-boxes the procedure of process theorizing. Therefore, our framework is constituted by three distinctive parts: articulating, relating, and conjugating. These distinctive process methodological techniques are consistent with our previous discussion on the need for unconcealing the identifying and generative properties of a process. We use the term articulating to denote the dialectics between loose and strict thinking throughout the process of identifying, collecting, coding, and analyzing qualitative process data. Similarly, we propose the technique of relating to establishing grounds for maintaining undisrupted linkages within and between reified conceptual properties of a process on the one hand, and their linkages to the contingency of the actual process on the other. Finally, we suggest the technique of conjugating as a way of pulling together nexuses of identifying and generative properties of a process. It should be clarified that these three techniques are not steps that can be placed in a spatiotemporal linear fashion, but rather, as quoted above, as means that enable loose and strict thinking at once and throughout the theorizing process. We therefore suggest that articulating, relating, and conjugating are in direct relation to one another.
3.1. Articulating
Articulating is to express things, events, phenomena, etc. in such a way that they clearly unconceal or make explicit and available to oneself (and others) the identifying and generative properties of chains of actions, events, or occasions. Articulating is therefore an act of effectively expressing the potential identifying and generative properties of the process through the definition and properties of things. At this stage, however, it is important to maintain the possibility to reverse or alternate between the articulated words, codes, or signifiers with the signified, i.e. the entity, expression, or wording that appears in the raw data that is unconcealed. As Lefebvre (2004: 13) noted, articulating things unbearably means ‘chasing out’ the answers that ‘hide themselves behind words, in locutions and expressions.’ However, to unconceal the answers and dress them up – to articulate them – is also to abstract them from their ‘natural’ context. Thus, one always needs to make a judgment on their relevance and importance (Whitehead, 1938) in order to avoid dressing up things that cannot effectively connect (Bateson, 1979) with other types of properties of a process. Key to this activity is iteration. By iteration we mean the interweaving of change and permanence (Whitehead, 1938); the back and forth activity of dwelling into the data that is assumed to contain an answer to the research questions asked, labeling phenomena expressed in it, and relabeling the answers as we revisit the data. Dwelling into the data is the process/activity that embodies change, whereas the labeling of phenomena – the articulation of properties – is aimed at stabilizing or making our understanding of things permanent and temporally durable.
Having set what we mean by articulation, let us now turn to some practical examples of how articulation has come about in process research before we make our final point. Langley and Abdallah (2011) recently discussed two widespread process techniques: the post-positivistic (Eisenhardt, 1989; Yin, 1994) and the grounded theory (Glaser and Strauss, 1967) traditions of qualitative research. Making this distinction is important, but cannot be exhaustively discussed here due to space limitations. However, the post-positivistic technique for articulation is based on displaying variance and difference between and within cases on the one hand, and between theoretical concepts and empirical findings on the other. For example, in a study into the emergence of cognitive schema in the life insurance industry around the time of the introduction of personal computers, Bingham and Kahl (2013) were only partially guided by prior studies in the field of psychology and organization studies as to the identifying properties of schema, and much less about the generative properties: ‘Yet, although this research has been explicit about the definitions of schemas, as well as about their structure, use, change, and relevance, it has been far less explicit about schema emergence’ (Bingham and Kahl, 2013: 15).
Thus, to take one example of schema emergence, Bingham and Kahl based their study upon the notion of ‘analogy’ as portrayed in the psychology literature, albeit with some reservation: ‘existing psychological theory for schema emergence at the individual level may be imprecise for portraying schema emergence at the collective level’ (Bingham and Kahl, 2013: 16). This left them with only one methodological approach – to combine theory elaboration (e.g. Glaser and Strauss, 1967) and theory generation (Eisenhardt, 1989) techniques. In order to unconceal schema related to analogy, the authors ‘focused on nouns and verbs that denoted what came to be known as computers and their operations’ (Bingham and Kahl, 2013: 18). The focus on nouns helped them to pin down the identifying characteristics of the schema, whereas the search for verbs localized the generative properties of the cognitive schema. This was a critical step. As noted above, it fulfilled the importance criteria, helping them to focus on meaningful nouns and verbs that best captured varieties of the schema ‘computer’ and its uses across time and space. In addition to these measures, it is worth considering what we mean by articulating with respect to the ontological premises of our approach to process theorizing.
3.1.1. Causality and articulating
In articulating the various properties of analogy as a cognitive schema, Bingham and Kahl (2013) developed intra-level causal relations between the various identifying (nouns) and generative (verbs) characteristics, respectively, as well as their inter-level causal relations. More specifically, they grammatically ‘interpreted nouns as categories and verbs as relations, which required capturing all of the nouns in a text as well as the verbs and the nouns to which the verbs connected’ (Bingham and Kahl, 2013: 17; emphasis added). In doing so, they managed ‘to isolate categories and their relational structure’ (Bingham and Kahl, 2013: 17). Although these measures were important for articulating categories and relations, the causal relationship between time and space is not fixed, but established through the articulations made. From a process theoretical view, articulating comes about when there is a logical leap between each level of abstraction of a process’ identifying and generative properties. Moreover, since all levels of abstraction are interconnected, a change in one level always implies changes in the following levels of abstraction. Causality is therefore inherently part of articulating, as it ensures or deters the logical coherence between layers of abstraction or the degree of consistency in the representational chains.
3.1.2. Temporality and articulating
Indeed, Bingham and Kahl’s (2013) study was particularly concerned with the question of how a new collective schema emerged over time. To be able to do this, the authors adopted two techniques. First, in order to be able to identify changes to the identified concepts and new emergent concepts over time, the authors recorded nouns ‘exactly as represented in a text’ (Bingham and Kahl, 2013: 18). Second, in order to elicit schema emergence over time, they had to make frequency counts of various categories relating to the term ‘computer.’ These techniques helped the authors to visually see (through numbers and graphs) when certain categories were first initiated, when their use increased, and eventually when they diminished or markedly reduced. In this way, the authors were able to elicit the emergence of schema by showing ‘how certain categories and relations became more or less prevalent over time’ (Bingham and Kahl, 2013: 18). However, because articulating is a form of reification – a discursive means of making durable (Whitehead, 1920) the social phenomena that are otherwise under flux and change over time – we need to bear in mind that articulating an identifying or generative property of a process is to signify a certain property as distinctive for the ‘ongoingness’ or flow of the process, rather than for its duration. This is an important feature of articulating, as it necessarily serves as a measure or knowledge component separated from the actual dynamism and multiplexity of activity. We therefore suggest that articulating is maintained in a tentative state until data generation has reached some degree of saturation (Miles and Huberman, 1994), and the various parts of the process theory emergent from the study assimilate enough to make up a meaningful context (Bateson, 1979) for the phenomenon under investigation.
3.1.3. Spatiality and articulating
Returning to Bingham and Kahl’s (2013) study, articulating the identifying and generative properties that formed the cognitive schema ‘computer’ was not only temporally stretched over a period of 30 years (1945–1975), but also bound to a specific context (the authors themselves refer to this as the historical context). It is, however, possible to speak of an activity context or site (Schatzki, 2002). From a process theoretical point of view, schema emergence is polycontextual: vertically bound to the measure of time (and all of its subcategories such as moment and epoch) and horizontally bound to the measure of activity (and all of its variations such as arrangement, event, episode, incident, sequence, and situation). Whereas Bingham and Kahl’s (2013) study took the life insurance industry as the locus of schema emergence activity, they were aware that the discourse about computers was polycontextual, and had been recorded in several related activity sites: ‘the Society of Actuaries (SOA), Life Office Management Association (LOMA), and Insurance Accountant and Statistical Association (IASA)’ (Bingham and Kahl, 2013: 17). Thus, being sensitive to both vertical and horizontal sites offers the possibility to articulate both novel and general identifying and generative properties within the same historical instance and across sites of activity. This was indeed a technique that vastly helped increasing the empirical validity of Bingham and Kahl’s (2013: 29) study.
3.2. Relating
Relating is the process through which one brings together or establishes associations, connections, or relations between matters, events, circumstances, subjects, objects, etc., to locate probable causal relations. When connections are possible, they commonly convey value such that when bringing together two or more pieces of information that connect, they express new information previously unknown to the observer (Bateson, 1979). Thus, relating is instrumental for substantiating identifying and generative types of a process’ properties, and it is a criterion for all information that yields value to be of a relational kind. As Whitehead (1938: 9) noted, ‘It is of the essence of types, that they be connected … No fact is merely itself.’ According to this logic, two pieces of information serve as context for each other: one is something separate yet connectable in light of the other, and vice versa. Thus, such unity comes about if the information contained in each piece of information is different from the other, and the sum of both is not reducible to one or the other. Relating either yields information of another logical type or a metamessage (Bateson, 1979: 128). A function – or even an effect – of such metamessages is to ‘classify the messages that occur within its context’ (Bateson, 1979: 129). This way of connecting, defining, and classifying information never stops, and has in fact no objective beginning or end. But how exactly do we identify and establish connections?
Relating can be made at different levels. At the lowest level of analysis, relating concerns those types of properties that are logically distinct, and which clearly differ from other seemingly logically distinct properties. However, for each connection there is both a vertical association and a hierarchical ordering; the former being distinctions of logical types, while the latter are differentiations of levels of explanation. Logical types can be understood as various types of actions and behaviors within a certain context. For example, when we seek to classify various activities, say swimming or having a shower, it is the activity that emerges as a logical type that signifies what the various activities are or the purpose for which the activities are carried out, and not the water in which the activities are realized.
In contrast, levels are actions and behaviors that define or make context intelligible (Bateson, 1979). Context here is not the water per se, nor is the purpose of swimming or showering central for our understanding of it as something different: it is the aggregation of these various connectable pieces of information – the human body; the movements required for making the body float and move through the water; the water itself; the physical location of the activity; the moods, emotional states, aspirations, goals, etc. – that make the context intelligible. Depending on the variation of these types of information, we can then differentiate swimming from showering as two separate logical types of action. But we can also start classifying the information contained in the activity of swimming so that a new level emerges for each connection of logical types – say a leisure activity, training, competition, etc. Each level emerges from the underlying processes as a logical consequence rather than an ambiguous judgment. In both cases – logical types and discrete levels – the information that once yielded a description of a certain kind is conditional (Bateson, 1979). This means that each piece of information changes in relation to its context (in the case of logical types) and in relation to the relation between two or more pieces of information (in the case of levels).
However, exclusion by means of systematizing proof, facts, and evidence at an initial stage only constrains thinking: ‘one should never start from systematization’ (Whitehead, 1938: 2). Rather, as Whitehead suggested, one should engage in ‘assemblage,’ or the act of sticking to only a few large-scale notions so that one’s attention is focused on the variety of other explanations that ‘arise in the display of those chosen for primary emphasis’ (Whitehead, 1938: 2). In this way, we can mitigate the risks of displaying scientific knowledge as something fixed and given (Rescher, 1996). It is by way of establishing distinctions and differences within and between different properties of a process that connections are realized (Bateson, 1979). For the sake of clarification, let us now turn to how relating is constituted within the realm of our ontological premises.
3.2.1. Causality and relating
As noted above, establishing causal relations between a process’ properties is particularly vulnerable within the technique of relating. Let us therefore turn to Bingham and Kahl’s (2013) study to illustrate this point. In an attempt to defy the risks of ascribing certain entities causality without relation to some other entity, the authors first listed a number of schema elements (see Bingham and Kahl, 2013: 25) related to the identifying (or ‘category’ in their terminology) characteristics and the generative characteristics (‘relation’ in their terminology) of the word ‘computer,’ and displayed the frequency of these various characteristics in their data set. This helped them to minimize the causal fallacy of establishing connections where there were no connections between the characteristics and the mental schema of ‘computer as machine’ and ‘computer as brain.’ The tabular display of the various elements helped them to connect the answers found in the archival data they used for their analysis to each other, as well as to the basic assumptions made about analogy as a cognitive schema: ‘we read through each piece of text to capture what categories each verb connected’ (Bingham and Kahl, 2013: 18). However, establishing such causality is vulnerable unless it can yield information of a different kind, or unless the connected entity exhibits a distinctive degree of discrimination that cannot be easily reversed to its constituents. Thus, what Bingham and Kahl (2013) did was to measure the degree of concentration of certain groupings of categories and relations in order to establish connections that could no longer serve as individual stand-alone components.
3.2.2. Temporality and relating
As we see it, the temporality of relating is one that typically exposes the causal pattern described above over time. This way of relating, however, places greater demands on the consistency of connection over time, and not necessarily the consistency of results’ discreteness or connections. In other words, consistency in the connection between a process’ identifying and/or generative properties reveals the passage of time (Whitehead, 1920), which is a necessary judgment of an undisrupted flow of activity. Once the consistency of connections between the properties of a process is disrupted, it is no longer possible to accurately speak of the ongoingness or flow of activity, and so the general definition of process is no longer valid, leaving any attempt at process theorizing at stake. A simple illustration of this argument is again found in Bingham and Kahl’s (2013) study. They showed that the connectivity of properties related to the schema ‘computer as machine’ declined markedly during the same time period as that of the ‘computer as system’ schema. The latter, on the contrary, revealed increasing connectivity over time, and thus served as a more solid proof of the ongoingness of the analogy that various activity sites made about computers. What this simple illustration further indicates, as Bingham and Kahl (2013: 26) righteously suggest, is that ‘[t]he increasing connectedness indicates a unit structure emerged around the categories and relations.’
3.2.3. Spatiality and relating
If relating is ultimately about establishing connections between the properties of a process such that they yield information of a different kind – and that such connections need to be consistent over time – then spatiality has a critical role in maintaining this consistency. Let us recall that events have no absolute space or place (Schatzki, 2005: 468); as such, spatial locations only become meaningful when captured along with intelligible process activities or sites in connection with a process’ identifying and generative properties. How can we then speak of undisrupted consistency in connections between properties of a process? From a processual viewpoint, activity sites maintain consistency in their connections with identifying and generative properties only when the nexus of properties can be related to undisrupted action over time. However, we should be careful here to ascribe action some sort of absolute rationality or normativity at the expense of ‘feeling’ – that which smoothens the ongoingness of action and the transition between actions belonging to the same chain of actions (Whitehead, 1978). In fact, such feelings or affectivity (Schatzki, 2010) bind together actions with certain aspirations and goals – the teleology of activity – and thus do not easily replace one action site or preferred place with another. Consider the following extract from Bingham and Kahl’s (2013: 29) study: ‘Although anchoring new ideas in a familiar context appears useful to start change, our study suggests it may be less useful in sustaining change, where accentuating the novel appears key.’ Without these discrete variations of preferences for familiar action sites, it seems time would be a derivative concept.
3.3. Conjugating
Although this framework is supposed to be non-sequential in its application to process studies, the final technique of conjugating points back to the prior two modes of articulating and relating in that it requires some depth of understanding about the features of the identifying and generative properties of a process. Conjugating simply requires a certain degree of understanding of the flow of a process in order to be able to bring it forth or join together its related parts such that they appear as a unifying whole in a specific instance of time.
Our point is that no single identifying or generative property of a process stands alone, and it is neither self-contained nor replaceable by another property. This suggests analyzing process data in such a way that various facts and pieces of evidence are joined together in pairs of two or more entities distinctive enough to count as logical types, but which reciprocally and mutually constitute one another as such, i.e. identifying or generative properties of a process. We call this activity ‘conjugating,’ and suggest that this be understood as joining together, or ‘bundling up,’ according to the logic of the mutual constitution of properties. In Whitehead’s process philosophy, entities and events have a spatiotemporal causality that is unbound to a particular location, and which do not necessarily have causal effects in a specific coinciding time. Rather, entities and events have ‘concrescent occasions’ (Whitehead, 1978); that is, their immediate presence is a conjoined fact that is irreducible to their past states and constituents, although these past states and constituents form their immediate present.
Thus, the challenge of conjugating entities and events does not lie in identifying their actual occasion in a single spatial location or single temporal instance – a time slot readily measurable in chronological time – but rather in conjoining their mutual relatedness across time (past, present, future) and space (proximate, distant, actual, imaginative) in a ‘common duration,’ or a ‘cross-section of the world’ (Whitehead, 1978). For example, the common duration of the entities and activities of realizing an organizational strategy – a stated goal or objective that is clearly defined and implemented throughout the organizational unit(s) concerned – is found in those spatiotemporally distributed actions which are joined together so that their common past and present cannot be separated, and their various spatial locations coalesce into one coherent process site rather than distinctive locations.
The point we wish to make is that conjugating is the part of process theorizing where the spatiotemporality of events is enmeshed, so that the process now appears as the common duration of the process’ identifying and generative properties. We should clarify that the common duration of an event’s identifying and generative properties is usually a layered conjugation; a logical type of the first-order which in turn is a conjugation of second-order properties; and so on. This logic, however multidimensional it is, helps us to establish links between and within levels, to reconsider causality, and to abstract and concretize processual properties.
3.3.1. Causality and conjugating
We have previously argued that causality is a relational concept, thereby refuting the idea that isolated entities have a causal power in and by themselves. This holds true for conjugating the identifying and generative properties of an event. Through conjugating properties we have to some extent already established a common discourse (articulating) around the smallest units of an event’s identifying and generative properties, as well as established consistent connections within and between such properties. We may therefore think of causality as a multiplex phenomenon when conjoining two or more nexuses of an event’s properties. For example, Bingham and Kahl (2013) had already identified that the concept ‘computer’ was strongly assimilated to two separate identifying properties in the initial phase: the ‘computer as brain’ and the ‘computer as machine’. Each one of these was further distinguished by its own set of identifying and generative properties. The ‘computer as machine’ analogy was articulated through identifying properties such as ‘punch card,’ ‘tabulator,’ and ‘machine,’ as well as generative properties such as ‘merge,’ ‘punch,’ and ‘match,’ while connected formed the higher-order level property ‘data’ and ‘computer’ (Bingham and Kahl, 2013: 20). The ‘computer as brain’ analogy was articulated through the same set of identifying properties as the machine analogy, although with some novel identifying properties, such as ‘problem’ and ‘operation,’ and completely unique generative properties such as ‘solve,’ ‘examine,’ and ‘think,’ which together were related or connected to the higher-order level property ‘decision-making’ (Bingham and Kahl, 2013: 21).
This example clearly shows that when conjugating nexuses of an event’s properties, causality no longer remains as a single source and single direction phenomenon. Multiplex causality is therefore located in part in novel events that overlap and lie between types of discrete events. In either case, conjugating occurs through measuring the direct ‘proximate relevance’ (Whitehead, 1978) of nexuses of identifying and generative properties with a causal relationship: ‘The more connected a group of categories and relations become, the stronger they emerge as a stand-alone cognitive unit’ (Bingham and Kahl, 2013: 27).
3.3.2. Temporality and conjugating
Conjugating resembles the technique of relating in that it transpires from consistent connections between multiplex causal relations within and between a process’ identifying and generative properties. However, because conjugating emanates from the realm of multiplex causal relations, the technique is particularly bound to the temporality of process theorizing; or, in other words, conjugating makes the ‘nonsuccessive’ (Schatzki, 2006) nature of time visible in process theorizing. Because the properties of a process may have both an overlapping causal relation and a discrete and novel causal relation, they mark time differently. Both instances are marked by some motivating state of affairs, despite the fact that they may be located in different processual and activity sites. In conjugating such states of affairs, it therefore implies bringing together nexuses of properties of an event that may or may not have consistent temporalities. Similar to this understanding, Bingham and Kahl (2013) found that multiple and sometimes competing schema emerged simultaneously rather than serially, as though schema emergence were a successive process. This also brings us to the methodological challenge wherein the measure of time cannot be separated from the measure of activity. Rather, the latter makes the former possible. Thus, there is a hierarchical relation between the measure of time and the activity that mitigates the risks of locating activity in the measure of time, rather than vice versa.
3.3.3. Spatiality and conjugating
Concurrent with this method of theorizing process is the treatment of events as spatiotemporal potentialities. We conceive of events as spatiotemporal potentialities because both time and space have the potential to function as generative mechanisms: for example, that action always coincides in a ‘proper’ time (e.g. intended or unintended, planned or affectively charged); that an individual or a group acts and interacts intelligibly (i.e. that actions and interaction make up a coherent flow); that the site in which individuals act invites or inculcates materials (e.g. a proper material site, office, documents, machines, computers, etc.) into action; that action yields a difference compared with related actions, or is only contemporaneous with other actions; and that these multiple contributory singular causes function as measures for this time–space complex of an event. When these conditions are evoked, we may speak of an ‘evental-site’ (Badiou, 2005), or the taking place of an event in a proper location that cannot be the site of another event, at least not simultaneously. However, a methodological challenge still remains, as this causal efficacy is irreversible. It is unfeasible to conceive of a site as evental in the absence of an event – we cannot simply pursue a verificationist approach to unconceal a site as an event. Instead, we need to approach it pragmatically and heuristically (Badiou, 2005).
4. Concluding remarks
The underlying premises of the present study were the need for more elaborate yet distinctive ways of theorizing process data in social sciences in general. In particular, we have set out to bridge the gap between process ontology and process theorizing by outlining a meta-theoretical account. Drawing on process philosophy (Whitehead, 1920, 1938, 1978) and process theory in general (Bateson, 1972, 1979), and process studies in particular (Klag and Langley, 2013; Langley, 1999; Langley & Abdallah, 2011; Langley et al., 2013), our approach has advanced the understanding of the basic components and their internal relationship in process theorizing from qualitative data. We have advanced why process theorizing offers a richer conceptualization and development of relations between change and constancy, potentiality and actuality, action and interpretation, and event and substance. We began by outlining epistemological and ontological considerations for process theorizing. In particular, we distinguished the functions of causality, temporality, and spatiality in process theorizing, and showed how these are mutually constitutive in three constitutive parts of a tentative process theorizing framework. We named these parts ‘articulating,’ ‘relating,’ and ‘conjugating,’ and discussed their respective functions and how they operate by giving examples and providing advice for realizing these activities in practice. In doing so, we also elaborated on the possibility of process theorizing by engaging in these three activities, as well as how they help us emanate processual sites. The strength of this tentative framework for process theorizing by means of building more solid process theories of social phenomena lies in its capacity to maintain multiplexity within and between units and levels of analysis. Like most evolving approaches, the present framework itself needs to be ‘mangled’ in practice, both conceptually and empirically, such that more detail and rigor is brought about. However, our initial attempt should serve as a sound basis for such elaboration in future studies.
Footnotes
Acknowledgements
We would like to thank Janet Johansson, Steffi Siegert, Sara Öhlin, and two anonymous reviewers for helpful comments on earlier drafts. The usual disclaimers apply.
Funding
The first author was supported through a post-doctoral research scholarship by the Jan Wallander and Tom Hedelius Foundation.