A Figure is Worth a Thousand Words: The role of visualization in paradox theorizing

Introduction

Paradox is central to management and organization scholarship if only because tension and opposition are an inherent part of organizational practice (Putnam, Fairhurst, & Banghart, 2016; Schad, Lewis, Raisch, & Smith, 2016). Dynamic expressions of paradoxical balance are evident in various analyses of proliferating contradictions between multiple logics in organizations and in society (Carmine et al., 2021). Paradoxes have been described as “contradictory yet interrelated elements—elements that seem logical in isolation but absurd and irrational when appearing simultaneously” (Lewis, 2000, p. 760). Paradox theory addresses diverse challenges arising from individual or group tensions (Berti & Simpson, 2021a; Pamphile, 2022; Pradies, Aust, Bednarek, et al., 2021) to global challenges (Sharma et al., 2021). Paradox is an umbrella concept covering multiple facets, including dynamic expressions of paradoxical balance, as are typically assumed in the both-and approach, as well as lack of balance, as in stuckedness-circularity and dialectical transformation. The development of the field has been both convergent and divergent (Schad, Lewis, & Smith, 2019). Centripetal forces foster theoretical convergence around core concepts, while centrifugal forces push the field’s boundaries (Hahn & Knight, 2021b) through interdisciplinary innovation. Both dynamics are “vital to the development of paradox theory” (Schad et al., 2019, p. 108). Despite their interrelated nature, their binary conception minimizes theoretical convergence (Cunha & Putnam, 2019), revealing limits to understanding paradox, as Schad et al. (2019) note. A vibrant cross-disciplinary intellectual community of paradox scholars has emerged, employing a variety of methods to address a wide array of topics (Berti & Cunha, 2022; Putnam et al., 2016; Schad et al., 2016), tensions and their management (Keller et al., 2021; Schad et al., 2019).

Visuals are an effective channel to convey paradoxes (Halgin, Glynn, & Rockwell, 2018; Vince, 2021; Vince & Broussine, 1996), with specific visuals, exemplified by the widespread use of the yin-yang symbol (e.g., Smith & Lewis, 2011), being widely associated with the concept. Figures are constitutive of the core messages presented in many academic papers (particularly those dealing with complex issues). Rather than being mere embellishments, they complement written text and concepts, offering an illustration of the phenomena researched as well as the conceptual frameworks explaining their causes, dynamics, and consequences.

A rich tradition in visual organizational research has shown (e.g., Quattrone, Ronzani, Jancsary, & Höllerer, 2021) that visuals serve multiple purposes. First, visuals have a strong performative effect when they orient our attention to what counts and what is legitimate in the literature (Meyer, Höllerer, Jancsary, & van Leeuwen, 2013; Quattrone et al., 2021), thus shaping theorizing. Second, conceptual depth is increased when scholars and their audiences gain insight not immediately communicated by textual exegesis (Avakian, 2020). Visualization facilitates aesthetic communication of highly abstract concepts (Ravasi, 2017). Hence, as Messaris and Abraham (2001, p. 225) state, “pictorial framing is worthy of investigation not only because images are capable of conveying unverbalized meanings, but also because awareness of those meanings may be particularly elusive.” Finally, visual devices are powerful and effective communications for building academic–practitioner bridges (Trumbo, 1999). The use of well-known visuals, including symbols that have deep historical roots and wide global awareness, such as the yin-yang symbol, serve diverse traditions in contextualizing universal human experiences (Lindgreen & Maon, 2019).

In this article we asked the broad question: How do visuals in paradox academic work address the tension of convergence and divergence in paradox theorizing, focusing on (1) what do visual representations in paradox academic work reveal about paradox and its assumptions and (2) what new possibilities and areas of exploration does mapping the literature reveal? Our analysis contributes to paradox theory in three ways. First, we show that visuals constitute the lynchpin between convergent and divergent forces, allowing scholars to simultaneously reinforce and challenge current understandings of paradoxes. Second, we put forth a theorizing tool that defines the terrain on which paradox research has been built and identifies three antinomies, or dualities, as the cardinal points. Third, we reveal the performative effect of visuals by identifying a few classes that have dominated the imagery used in paradox research, and we subsequently draw attention to classes that reflect uncharted territories in the field.

We structure our paper as follows. We briefly outline the trajectory of paradox theory, presenting the opportunities offered by examining the use of visuals as a core component of theory development and communication. Building on a comprehensive review of the literature on paradox, we identify the basic visual grammar used in figures published in influential academic journals and relevant book chapters. Moving iteratively between this basic visual lexicon and that of narrative text, we inductively identify three fundamental antinomies—equilibrium/disequilibrium, linearity/recursivity, and boundedness/openness—that allow us to map alternative conceptualizations of paradox. From the eight possible combinations (or “classes”, as we name them in the Findings section) of these three antinomies, we derive an exhaustive typology associated with essential iconic representations in paradox theory. While our analysis focuses exclusively on visualization in scholarly works, it allows us to identify conceptual undercurrents shaping the evolution of theory as well as areas requiring further investigation. Finally, we discuss the implications of our findings for future research and practice.

Paradox Theory: Rich in Reflexivity, yet Blind to the Use of Visuals

Scholars and philosophers have speculated on paradox for centuries. In management studies, analysis of persistent opposite and interrelated tensions started with the contributions by Quinn and Cameron (1988), Poole and Van de Ven (1989), and Lewis and Smith (Lewis, 2000; Lewis & Smith, 2014; Smith & Lewis, 2011). Subsequent years have seen a significant increase in scholarly research on paradox (Schad et al., 2016) which, of late, has grown exponentially (Cunha & Putnam, 2019; Fairhurst, 2019). Scholars from various disciplines, working with different methodologies (Putnam et al., 2016; Schad et al., 2016, 2019), contribute to “conversations on a shared objective—understanding organizational tensions and their management better” (Schad et al., 2019, p. 108).

Scholars seek to strengthen paradox theory by refining definitional clarity (Berti & Cunha, 2022; Smith, 2014; Smith & Lewis, 2011), identifying central concepts of opposition, interrelation, persistence, and undecidability, as well as differentiating paradox from dilemmas, dialectics, or dichotomies (Putnam et al., 2016). Some have analysed key processes at the heart of paradox dynamics, such as vicious and virtuous cycles (Pradies, Tunarosa, Lewis, & Courtois, 2021; Sundaramurthy & Lewis, 2003), or provided reflection on paradox’s either inherent or socially constructed nature through insights on the salience or latency of paradox (Hahn & Knight, 2021a; Pradies, 2022; Schad & Bansal, 2018). Not only have responses to paradoxical tensions been widely discussed (Jarzabkowski & Lê, 2017; Jarzabkowski, Lê, & Van de Ven, 2013) but they have also emphasized the need to work through paradoxes (Lüscher & Lewis, 2008; Pamphile, 2022; Pradies, Tunarosa, et al., 2021) to find a balance between integration and differentiation (Cunha & Putnam, 2019; Smith 2014). Lewis and Smith (2014) recognize paradox theory and its multifaceted uses as a metatheory. Schad et al. (2019, p. 108) describe centripetal forces reinforcing core elements of the theory, while centrifugal forces have “pull[ed] away from the core to spur exploration and creativity, challenging, spanning, and extending its boundaries.” By demonstrating the role of power dynamics in generating pragmatic paradoxes (Berti & Simpson, 2021a), or by exploring questions of disequilibrium (Hargrave, 2020; Hargrave & Van de Ven, 2017), researchers have expanded paradox scholarships’ theoretical boundaries.

Paradox scholars’ dialogues in Academy of Management Review between Li (2021b) and Berti and Simpson (2021b) and between Li (2021a) and Hahn and Knight (2021a) capture the vibrancy of current debate. “[P]aradox scholars ‘agree to disagree,’” fostering “conflicting yet interdependent views on key concepts that co-exist and energize community debates” (Schad et al., 2019, p. 110). The paradox scholarship research program’s use of language, research craft, thinking style, and use of material tools underpinning theorizing (Clegg, Cunha, & Berti, 2022) displays a vitality that invites further inquiry. To ensure that the vitality of divergent efforts counterbalances too much convergence (Cunha & Putnam, 2019), both centripetal and centrifugal efforts are essential to “surface and open the remaining ‘black boxes’ of paradox theory” (Schad et al., 2019, p. 108).

Visual representation of the findings or conceptual contributions presented characterize many works published in the field; indeed, visuals have been integral to the evolution of paradox theory. A foundational work by Vince and Broussine (1996) relied on drawings to articulate the paradoxes that managers faced in the wake of organizational change, while the yin-yang symbol was used by Lewis (2000) and Smith and Lewis (2011). Such visualizations capture holistic and simultaneous forms of signification that are difficult to express but are essential for a better understanding of the notion of paradox (Halgin et al., 2018).

While visuals characterize key contributions, they are rarely the focus of attention, a surprising blind spot, as visuals enable researchers to capture the ineffable absurdity (Lewis, 2000) and logic-defying dynamics sometimes present in interrelated contradictions. Visualization assumes different roles (Berti, 2017), with Gibson (1978) noting that depiction frames description, meaning that how an object, such as a paradox, is represented visually inscribes interpretation of its meaning. For Gibson (1978, p. 233), “depiction captures an awareness without describing it.” We see visualizing as central to the theorizing process, able to generate novel theories, as suggested by the visual turn in management studies (Meyer et al., 2013; Quattrone et al., 2021; Swedberg, 2016). Editors and reviewers pay great attention to the consistency between text and figures (Fulmer, 2012). Even if one accepts that “diagrams are not theory” (Sutton & Staw, 1995, p. 376), figures and other visuals, together with textual headings and propositions, are essential in conveying the essence of academic work (Lange & Pfarrer, 2017). Not only do visuals help clarify other inscriptions but they also provide formal models that allow assessment of the “balance between parsimony and completeness” in a theoretical contribution (Whetten, 1989, p. 491).

Ravasi (2017, p. 243) suggests that visualizing qualitative data is essential to solving the “mystery of theorizing from qualitative data.” Similarly, Parmentier-Cajaiba and Cajaiba-Santana (2020) argue that visual mapping enables researchers to engage with data by selecting, removing, or adding elements, as a steppingstone in the theorizing process. Visualizations allow researchers to theorize without being “bound by convention or preconceived notions of linear cause and effect” (Langley & Ravasi, 2019, p. 188). Images have narrative power (Kassinis & Panayiotou, 2018); they embed knowledge (Ewenstein & Whyte, 2009), bringing new meaning to textual understanding.

In sum, visuals are central to knowledge production (and dissemination) processes (Meyer et al., 2013; Quattrone et al., 2021). Indeed, many scholars, noting the performative function of visuals (e.g., Meyer et al., 2013; Steyaert, Marti, & Michels, 2012), observe “how visualization practices produce organizational outcomes” (Quattrone et al., 2021, p. 1197), such as theorization. Visualization frames what is seen as worthy (Arjaliès & Bansal, 2018), not only facilitating sensemaking but also mobilizing actors’ (and spectators’) attention. In addition, it may be used as a means for contestation (Kornberger, 2017). As such, visualization not only defines appropriateness but also offers opportunities to debate what is worthy of attention by establishing “arenas for performance” (Quattrone et al., 2021, p. 1205).

To sum, we will investigate how visuals in paradox theorizing work to address tensions of convergence and divergence. In particular, we focus on two questions: (1) what do visual representations in paradox academic work reveal about paradox and its assumptions, and (2) what new possibilities and areas of exploration does mapping the literature reveal?

Methodology

Coding the visual representation of paradox: a classificatory device

Step 1. Identifying the basic lexicon of paradox visualization

Our analytical process was nonlinear and iterative between data and theory (Strauss & Corbin, 1998); however, for readability, we describe it as a series of sequential steps (illustrated in Figure 1). Once our corpus of 193 academic works and 266 figures was identified, two co-authors began inductively coding the first 150 figures to develop a consistent approach to “breaking down” complex models into basic forms in a process akin to open coding (Strauss & Corbin, 1998). The purpose was to break down figures into more accessible, interpretable objects, focusing on their essential elements, i.e., the elementary symbols used to compose them.

OPEN IN VIEWER

Figure 1.

Data structure.

While visual representations can be analysed in different ways, considering their representational, interactional, and compositional meaning (Kress & Van Leeuwen, 2020), we focused on recurring symbols. Basic signs were analysed in a manner akin to “word frequency analysis” in critical discourse analysis. The method allows identification of general assumptions or trends in a particular textual corpus; for example, it can show how positive or negative ideas are associated with a particular object (Bednarek & Caple, 2014). We identified some essential basic signs: boxes, arrows, straight and curved lines. We selected frequently found combinations of these basic signs consistently conveying specific meanings: different types of straight lines with arrows (one-directional, bi-directional, converging, diverging, zigzag), bounded annular shapes (circles, loops, knots), spirals, triangles, steps, and cartesian diagrams. The two co-authors performed the operation individually and then discussed and resolved ambiguities. In this way, they treated visual representations as any other form of signification, enabling the communication of complex ideas through the combination of a conventional set of elements (a visual repertoire, that comes together in more complex patterns). They also identified several images that need to be considered holistically, such as yin-yang, scale, bridge, cloud, waves, etc.

Step 2. Identifying and refining antinomies to make sense of the basic lexicon

Going back and forth between data and theory, and expanding our coding to the other figures, some of the basic elements of our visual lexicon emerged in an iterative process. Akin to a process of axial coding (Figure 1) as suggested by Strauss and Corbin (1998), we explored relationships among groups of basic visuals, iterating back and forth between data analysis and the literature the better to understand emerging themes and to group the basic elements. Initially, we surveyed the paradox literature and identified relevant concepts at the heart of paradox depiction. In doing this, we mirrored methods of discourse analysis that are based on the frequency of types of words (e.g., Tausczik & Pennebaker, 2010). Inspired by the fundamental assumption of paradox theory, the idea that social organization is invariably characterized by interdependent contradictions, we tried to identify opposite pairs that could be used to typify the meaning of these basic symbols. In so doing, we identified various fundamental contrasting dualities or antinomies. Antinomies, defined as pairs of concepts with opposite meanings, constitute a typical device to organize knowledge; they are a primal determinant of those competing tensions that paradox theorists explore (Keller & Chen, 2017). Using opposing pairs of concepts to codify meaning in paradox scholarship constitutes an effective way of representing convergent or divergent currents (Schad et al., 2016) (see Table 1 for our definition of antinomies). We came up with three pairs: equilibrium/disequilibrium, open/bounded, and linearity/recursivity. These ideas capture fundamental meanings conveyed by different symbols aligned with debates in the paradox literature, as well as the organizational literature more broadly, as we will highlight in our findings.

OPEN IN VIEWER

Table 1.

Definitions of antinomies.

EQUILIBRIUM Balance and symmetry between opposites, leading to stable outcomes		DISEQUILIBRIUM Unbalance and asymmetry between opposites, leading to unstable outcomes
OPEN A process that is divergent, with indeterminate outcomes and without a defined end, or a system which boundaries are not set		BOUNDED A process that is convergent, with a predictable end, or a system which is self-contained, well-bound and operating within strict boundaries
LINEAR Sequence of events that progresses in a single direction		RECURSIVE A process that iteratively loops into itself, leading to virtuous or vicious cycles

Antinomies emerged as we iterated between data and theory. Some alternative antinomies were considered (e.g., flux versus static) and then discarded because they were ambiguous or redundant when it came to making sense of our basic sign lexicon. The whole team were involved in refining and populating the emerging classification device (Table 2). Discussions on disagreements helped refine our typology. Moreover, the six polar elements composing the three selected antinomies can be unequivocally associated with specific graphic signs. For instance, circles are a clear representation of recursivity, while straight lines and arrows are eponymous of linearity. Similarly, signs representing openness (e.g., steps) suggest open systems with no defined boundaries and no predictable ends, while signs representing boundedness (e.g., cartesian diagrams) evoke bounded systems and a predictable end. Finally, some images (e.g., knots) are associated with stable balance, while others (e.g., pendulum) suggest the potential for instability and disequilibrium because of exogeneity. Some signs provoked questions; for example, we had originally positioned the scale as a sign of disequilibrium but not if the scale was “balanced.” We thus differentiated between the balanced scale (= equilibrium) and the unbalanced scale (= disequilibrium). Similarly, we first considered cartesian diagrams as closed but subsequently realized that arrows in cartesian diagrams may signify openness; despite this, in most cases we categorized cartesian diagrams as bounded (e.g., in Ashforth, Rogers, Pratt, & Pradies, 2014 or Besharov & Smith, 2014). Raisch, Hargrave, and van de Ven (2018) offered a rare example of arrows signalling openness in a cartesian diagram, reinforced by the existence of an additional arrow at the end of the graph.

OPEN IN VIEWER

Table 2.

Basic signs used in the visual grammar employed to describe paradox in the literature.

CLASSIFICATION		Basic forms	Metaphorical images	Notes
1	Bounded Linear Equilibrium	Opposition (Cartesian diagram); Overlapping circles; Triangles	Cliff; Mirror images	The visual symbol suggests a balanced, static relationship between different elements/forces, which are in symmetry/equilibrium
2	Bounded Linear Dis-equilibrium	Convergence		The visual symbol suggests a simple (measurable/predictable) relationship between two forces that are in direct opposition, creating at best a precarious balance maintained through struggle
3	BoundedRecursiveEquilibrium	Bidirectional arrows; Interconnected elements;Cycle; Knot; Interweaving elements; Loop	Ying & Yang; Janus; Weave	The visual symbol suggests a recursive/iterative relation between elements which outcomes are stable and predictable (stability through cyclical reproduction)
4	BoundedRecursiveDis-equilibrium	Counterbalancing forces	Pendulum; Scale	The visual symbol suggests a recursive relationship between two elements, characterized by trade-offs that maintain the system in disequilibrium (any balance is only temporary)
5	OpenLinearEquilibrium	Arrow; Sequence	Bridge	The visual symbol suggests a sequential, continuous transformation that has no specified end but that does not imply contrast
6	OpenLinearDis-equilibrium	Divergent lines or arrows	Scribble; Cloud	The visual symbol suggests a sequential, continuous transformation that has no specified end and that produces an instability in the system
7	OpenRecursiveEquilibrium	(Figure/Background graphic) Zig zag	Wave	The recursive relation between element leads to continuous transformation that takes place in the context of a preserved equilibrium
8	OpenRecursiveDis-equilibrium	Spiral; Helix; Escalation	Cascade	The visual suggests a recursive relation between element leads to divergent transformation that disrupts/transform the existing system

Step 3. Synthesizing themes into aggregate theoretical dimensions

Then, we sought to understand the connections between the themes that made up the antinomies and theorize their interrelations (Whetten, 1989) by engaging the paradox literature. We realized that the equilibrium/disequilibrium antinomy spoke to the consequence of tensions (and responses to them), while the bounded/open antinomy pointed to the evolution of these tensions and linearity/recursivity to their dynamics.

Step 4. Outlining and refining an emerging classification device thanks to the three antinomies

The choice of these three antinomies enabled development of an exhaustive typology to categorize all the visual signs (or basic elements) included in the works we examined, covering different phases of paradoxes. By combining these three pairs we developed a typology of eight possible classes, in which we have collocated the 30 basic figures (see Figure 2). All elements of imagery, including complex images that could not be broken down in their constituent components (e.g., the picture of the bridge, as opposed to the boxes and arrows in a model), have been associated to one of the categories that result from the combination of the opposites present in the three antinomies. Once the classificatory device was refined, disagreement between the authors coding the images independently dealt with how to “decompose” a complex picture, including multiple signs, rather than how to “pigeonhole” a single visual element. For example, we had to interpret how to code a loop incorporating arrow signs; after discussion, we agreed that the presence of the loop superseded the arrow, reinforcing the circular flow without suggesting linearity. Only when an arrow emerged out of the loop, was it coded as a separate element. All attempts at organizing reality through typologies (Cornelissen, 2017) create a vivid but necessarily artificial separation. Ours were ordered by using the well-known typological device of the two-by-two matrix (Lowy & Hood, 2004). The classic two-by-two diagram was enriched by the inclusion of a third dimension (linearity vs. recursivity), allowing ordering and placing in relation to alternative symbols that can be used to describe paradoxes without bundling in the same category visual representations that are semantically distinct.

OPEN IN VIEWER

Figure 2.

Classification device.

Step 5. Classifying all elements in the classification and making sense of it

We counted how many of these constitutive elements were present in each academic work, looking at the use of different classes of signs. Note that we conducted our analysis per figure. Thus, academic work including a single figure composed of heterogeneous signs (i.e., belonging to different classes) was coded for several classes. Similarly, because we conduced our analysis per number of figures, academic works including two figures composed of homogeneous signs would be coded twice. If multiple signs belonging to the same class (e.g., multiple arrows) were present in the same figure, they were counted only once. Table 3 illustrates the frequency of appearance of each class of basic forms. Finally, we considered how each type of sign featuring in a figure contributed to the three antinomies: for instance, a yin-yang sign was counted as representing “equilibrium,” “boundedness,” and “recursivity.” For a detailed overview of the coding per figures, please see our Online Appendix.

OPEN IN VIEWER

Table 3.

Frequencies of appearance of a class across the 266 figures analysed.

CLASSIFICATION		Basic forms	Occurrences
1	BoundedLinearEquilibrium	Opposition (Cartesian diagram); Overlapping circles; Triangles	73
2	BoundedLinearDis-equilibrium	Convergence	30
3	BoundedRecursiveEquilibrium	Bidirectional arrows; Interconnected elements;Cycle; Knot; Interweaving elements; Loop	156
4	BoundedRecursiveDis-equilibrium	Counterbalancing forces	3
5	OpenLinearEquilibrium	Arrow; Sequence	104
6	OpenLinearDis-equilibrium	Divergent lines or arrows	60
7	OpenRecursiveEquilibrium	(Figure/Background graphic) Zig zag	2
8	OpenRecursiveDis-equilibrium	Spiral; Helix; Escalation	7

Note: In a few instances, we identified metaphorical images which could not be categorized. These have not been included in the count.

We addressed variance in classification depending on whether the article was empirical or conceptual; whether the article represented paradox dynamics (i.e., focused on describing how a paradox evolves) or paradox description (i.e., focused on describing a paradox). ¹ The results presented in Table 4 suggested little variation between empirical and conceptual articles, with more variation between articles representing paradox dynamics or paradox description. Using this method, visual discourse analysis revealed evolutionary directions in paradox literature as well as less frequent areas of discussion (in relation to the three fundamental concerns identified by the aggregate theoretical dimensions).

OPEN IN VIEWER

Table 4.

Count of signs in each antinomy.

			Consequences		Evolution		Dynamics
			Equilibrium	Disequilibrium	Bounded	Open	Linear	Recursive
Overall	(Signs: n = 435)	n	335	100	262	173	267	168
		%	77%	23%	60%	40%	61%	39%
Empirical vs. conceptual papers	Empirical only (Signs n = 244)	n	191	53	141	103	145	99
		%	78%	22%	58%	42%	59%	41%
	Conceptual only (Signs: n = 191)	n	144	47	121	70	122	69
		%	75%	25%	63%	37%	64%	36%
Paradox description vs. paradox dynamics	Paradox description (Signs: n = 199)	n	149	51	130	70	136	64
		%	75%	25%	65%	35%	68%	32%
	Paradox dynamics (Signs: n = 236)	n	187	49	133	103	131	105
		%	79%	21%	56%	44%	56%	44%

Note: When a sign appeared twice in a figure, it counted as one occurrence.

Findings: The Antinomies

The first antinomy that surfaced is equilibrium versus disequilibrium (Cunha & Putnam, 2019). When facing competing demands, this pair represents opposite tendencies to (1) find balance and synergy, while preserving the status quo (Smith & Lewis, 2011), or to (2) recognize the state of constant imbalance brought about by oppositions that are interrelated, as well as how opposite tendencies are constantly in flux, disrupting the status quo (Hargrave & Van de Ven, 2017). The tension between the “orthodoxy” that privileges a symmetrical and hence power-neutral view of paradox (Van Bommel & Spicer, 2017) and studies that highlight asymmetry within opposite pairs and the intertwinement of paradox and power is also represented (Berti & Simpson, 2021b; Gaim, Clegg, & Cunha, 2021). Equilibrium/disequilibrium is a key tension in analysis of economic policy interventions as well as a foundational tension in structural functionalist theory (Parsons, 2013). Conveyed through signs such as the pendulum (i.e., disequilibrium of stasis) or overlapping circles (i.e., equilibrium), this antinomy was particularly relevant in representing paradox processes. In our analysis, equilibrium refers to a sign that points to balance and symmetry between opposites, leading to stable outcomes. By contrast, disequilibrium describes a sign that suggests imbalance and asymmetry between opposites, leading to outcomes other than stasis.

The second antinomy, open versus bounded, captures the problem of considering paradox concisely in the perspective of a bounded or open system (Schad & Bansal, 2018). Such an opposition points to the problem of studying paradox at multiple levels (Berti, Simpson, Cunha, & Clegg, 2021; Gotsi, Andriopoulos, Lewis, & Ingram, 2010), as nested and interwoven (Andriopoulos & Lewis, 2009; Cunha, Simpson, Clegg, & Rego, 2019). The degree of boundedness/openness, particularly relevant in looking at paradox processes, is the key tension in systems theory of organizations (Scott & Davis, 2015). Open system epistemology emphasizes how paradox interacts with other elements of organization; open-endedness occurs because of the ontological unfolding and indeterminate nature of the world and its events (Hadjimichael & Tsoukas, 2019), stressing paradoxes’ constant evolving. By contrast, a bounded system approach tends to focus on systems that are isolated from their environment, emphasizing paradox as a self-contained occurrence (Tsoukas & Cunha, 2017), an antinomy captured in visual circles and zigzags. More globally, we use the attribute “open” to describe a sign that suggests a divergent process in which outcomes are indeterminate, without a defined end, or a system whose boundaries are not set. By contrast, being “bounded” refers to a process that is convergent, with a predictable end or a system which is self-contained and immanent.

The third antinomy, recursive versus linear, represents the tension between the need to signify paradox’s distinctive epistemological foundations, characterized by the self-reproducing nature of social phenomena (Hernes & Bakken, 2003), while managerial legitimacy strives for ways to “break” the vicious circles that paradoxicality generates, thus highlighting their practical transformative potential (Smith & Lewis, 2011). Literature explores both the iterative ways in which interdependent tensions self-reproduce, becoming persistent paradoxes (Schad et al., 2016), as well as the progression from one step to another entailed in working through and transcending paradoxes (Abdallah et al., 2011; Bednarek, Paroutis, & Sillince, 2017). A clear relation can be identified between paradoxes and responses to paradoxes or steps to escape paradox (Pradies, Tunarosa, et al., 2021). In positivist and phenomenological accounts of causality in organizational analysis, it is linearity/recursivity (Hernes & Bakken, 2003; Meyer, 1972) that is at issue. Single arrows signify linearity while recursivity was a recurrent theme in visuals with spirals. Signs that indicate a sequence of events that progress in a single direction are linear; by contrast, signs that illustrate a process that iteratively loops into itself, leading to virtuous or vicious cycles, highlight recursivity.

Findings: The Evolution of Paradox Theory through Visuals

The above-mentioned antinomies identify eight classes categorizing basic visual signs. One rule we constructed was not to consider individual shapes per se but clusters of “similar” shapes identified by classification. The eight classes may be viewed as ways of seeing paradox as an umbrella concept, each portraying dimensions absent in the others. Below we expand on the eight classes identified, drawing on illustrative figures to exemplify our thoughts (also see the Online Appendix). As each figure may have multiple elements belonging to different classes, we tried to pinpoint figures in which the class in question is particularly striking.

Class #4. Oscillation (disequilibrium—bounded—recursive)

The visuals in this class point to a recursive relationship between two elements, characterized by trade-offs that maintain the system in disequilibrium, in constant imbalance, as any balance is only temporary and precarious. These signs point to a bounded system, in that the elements are self-contained, limited by how far from the middle point they diverge, evident in visuals of counterbalancing forces, images of scales or a pendulum. Authors use the scale or other visuals from this class in two ways in their theorizing. First, a swinging pendulum points to a constant lack of stasis within a bounded system with recurrence of disequilibrium at the center (see for example, Figure 3 in Lewis & Dehler, 2000). Second, this sign is used to theorize either/or choices or dilemmas that paradoxes create and the related vacillation of processing them, as in Smith and Lewis (2011), stressing either/or ways of thinking about paradoxes. Smith and Lewis’ (2011) example in a disequilibrium state (with one side heavier than the other) aligns with our sensemaking of the class. Despite the subtleties offered by such signs, they are extremely rare in our review, occurring in only 1% of the figures coded.

OPEN IN VIEWER

Figure 3.

(a) Evolution of classes over time and (b) Distribution of classes between visuals that represent paradox description and paradox dynamics.

Discussion

In this article we asked the broad question: How do visuals in paradox academic work address the tension of convergence and divergence in paradox theorizing, focusing on (1) what do visual representations in paradox academic work reveal about paradox and its assumptions and (2) what new possibilities and areas of exploration does mapping the literature reveal? To answer these questions, we approached visual representation as a semiotic activity that occurs alongside written texts and draws on different categories of signs to produce and share meaning (Cobley & Jansz, 2010). More specifically, we developed an original method for analysing the visual elements of the paradox academic literature, akin to discourse analysis of written texts, based on the frequency of use of the basic symbols that compose visual representations of paradoxes and their dynamics. Analysing the meaning conveyed by these recurring symbols allowed us to advance a typology of eight classes of visuals, organized around three sets of contrasting dualities or antinomies.

As shown in Table 3 (frequencies of appearance), our findings demonstrate that some assumptions and ideas about what a paradox is and how actors respond to them are more prevalent than others. For instance, class #3 (cyclical reproduction) is prevalent. This seems logical as interdependence and persistence of opposition underlie the definition of paradox and are aligned with the notion of circularity and cyclical reproduction. More broadly, the prevalence of classes that convey equilibrium (particularly #1, #3, and #5) supports the critical view that paradox researchers tend to “tame” paradox, underrating the role of conflict and over-emphasizing synergy (Cunha & Putnam, 2019; Fairhurst, 2019). In fact, as shown in Figure 3a (evolution of classes), over time, the only class that increased in frequency emphasizes equilibrium (class #5), while the only one that decreased emphasizes disequilibrium (class #6). But even within a prevalent category of visuals, such as class #3, there is ample space to explore different forms of dynamic circularity (Tsoukas & Cunha, 2017). New questions might include: How can paradox create stuckedness (Hage, 2009)? Or how can paradoxes generate irresistible forces of transformation? The study of these expressions of out-of-control circles promises to offer new angles on the dysfunctions of paradox.

Our analysis also draws attention to the variety of visuals used by figures describing paradoxes versus those unpacking paradox dynamics. As shown in Figure 3b (distribution of classes), figures that unpack paradox dynamics tend to rely almost exclusively on two classes (class #3, cyclical, and class #5, sequence). By contrast, figures describing paradoxes rely on a wider range of classes (class #1, balanced; class #3, cyclical; class #5, sequence; and class #6, chaotic change). Broadening the sets of visuals used beyond those that have become conventional may facilitate divergence in theoretical ideas, as theorizing and visualizing go hand in hand (Ravasi, 2017). This is especially important for exploring paradox dynamics (Jarzabkowski, Bednarek, Chalkias, & Cacciatori, 2022; Pradies, Tunarosa, et al., 2021).

By employing a heuristic device (the antinomy-based classification) that is normatively exhaustive (it covers all possible visual forms, regardless of their appearance in literature), we also identify unexplored territories, showing which classes of visuals have been underused. For example, class #4 (oscillation) contains very little exemplars, even though visuals such as a pendulum or an unbalanced scale could be used to capture important developments in paradox theory, like the asymmetrical relationship of opposite poles and the need to “protect” the weaker pole of a paradox (Huq, Reay, & Chreim, 2017).

Our findings thus contribute to paradox theory in three main ways. First, we show that visual representations constitute the lynchpin between convergent and divergent forces, allowing scholars to simultaneously reinforce and challenge current understandings of paradox. Hence, we show that examining the theoretical development of the field without paying specific attention to the visuals is incomplete—analogous to analysing a speech act by focusing solely on the content of the communication and ignoring the non-verbal elements (e.g., pauses, gestures, emphasis, etc.) that bring meanings to life. Indeed, visuals reinforce the role of analytical discovery by positioning the phenomenon under study, showing the reader what is figure and ground, and highlighting key dynamics. But visuals also go beyond analytical discovery. Unlike written texts, which rely on a discursive mode of communication, visuals operate in a representational mode (Kress & Van Leeuwen, 2020) that does not require lengthy description or detailed elaboration to articulate an idea. As such, visuals facilitate an intuitive grasp of the ways in which researchers push the boundaries of extant knowledge.

This is consistent with research on visual argumentation (Groarke, Palczewski, & Godden, 2016), which suggests that visuals may open up previously closed sites of contestation (e.g., Sundaramurthy & Lewis, 2003), challenge taken-for-granted assumptions (e.g., Lewis, 2000), and create “mind-bombs” (DeLuca, 1999, p. 4) that significantly transform current understandings in a field (e.g., Jarzabkowski & Lê, 2017). In sum, visuals may directly call attention to those aspects of paradox that are not explained by prior research, making novelty apparent within the context of the familiar. For instance, the modified yin-yang in Gümüsay, Smets, and Morris’ (2020) study of the first Islamic bank in Germany captures dynamism in the context of equilibrium; the knot in Jarzabkowski et al. (2022) orients attention to the multiple paradoxes that characterize grand challenges in an interorganizational system; the two scales in Pradies’ (2022) study of French veterinarians highlight that any balancing act when navigating paradox entails finding dynamic equilibrium, not only between responses but also between emotional traces.

Second, our visual analysis contributes a tool for scholars to theorize competing demands. We induced three key antinomies, or dualities: equilibrium/disequilibrium, bounded/open, and recursive/linear. We see these antinomies as defining the terrain along which tensions have been theorized; they constitute the field’s “cardinal points” (see Figure 2), which represent the fundamental principles used to categorize or understand the codes or messages in the paradox literature. The three sets of antinomies are an important theorizing tool, enabling scholars to engage in disciplined imagination (Weick, 1989). On the one hand, the antinomies themselves afford the consistent application of guidelines (i.e., the “disciplined” part of the theorizing process), but the repository of visuals depicted invites scholars to generate alternative perspectives (the “imagination” part of the process).

Third, by mapping the frequency of visuals used in the literature in a typology of eight classes, we contribute an understanding of the performative effect of visuals. Building on the premise that the visual language used in research papers has the capacity to shape social reality (Austin, 1962), our work suggests that, over time, researchers have relied on a limited number of classes, inadvertently producing recurrent conceptualizations of paradox that have shaped subsequent debate in the field and cemented certain assumptions. Paradox visuals have thus recursively influenced the social phenomena they purportedly capture and communicate (Marti & Gond, 2018).

This echoes the current literature on the performativity of theories, which focuses on the processes through which theories shape the social structures and practices they are supposed to describe (D’Adderio & Pollock, 2014; Marti & Gond, 2018). Yet we reveal one additional process that leads to path dependence and self-fulfillment of theories: the surreptitious translation of ideas from older to newer theories. If traditional management theories are equilibrium based, aimed at describing relations between stable variables, paradox theory is process- and recursivity-based, aimed at examining the mutual constitution of action and structure, of phenomena and response, of imagination and materiality (Hernes & Bakken, 2003). Nonetheless, our analysis shows that assumptions of linearity and of one-directional causal relationship are persistently conveyed in paradox theorizing through unambiguous arrows and boxes. Similarly, the idea of equilibrium and status maintenance is conveyed by the continued use of symmetrical, well-ordered graphic representations. Thus, some assumptions related to the very paradigms that paradox theory is aiming to challenge are ironically projected through the sensegiving effects of visuals.

Being mindful about performativity might encourage scholars to explore uncharted territories purposefully, broadening the repertoire of paradox visuals. In line with the principle of requisite variety, we argue that a complicated phenomenon, such as paradox, requires a commensurately varied repertoire of representing concepts because “it takes richness to grasp richness” (Weick, 2007, p. 16). We therefore open an “arena for performance” (Quattrone et al., 2021, p. 1205) in that we invite reflexivity on what is visible and invisible in current paradox visualizations. For example, the rare visibility of visuals, such as a pendulum or an unbalanced scale (class #4), invites us to explore the asymmetrical relationship of opposite poles (Cunha & Putnam, 2019). Indeed, converging towards familiar symbols (e.g., circles and loops), at the expense of introducing divergent ones (e.g., pendulum, zigzags, steps, and clouds), fails to match the complexity of what we seek to explain, with the risk of producing visuals that oversimplify the ideas argued in written texts.

Additionally, our findings show that even if the use of complex, realistic symbols, and images (e.g., bridges, clouds, cliffs, etc.) is not as widespread as the use of simpler graphic signs, it is still quite significant. Visual metaphors are indeed a powerful tool for theorizing (Cornelissen, 2006; Weick, 1989) because they “force us to make semantic leaps” (Cornelissen, 2006, p. 1584). This facilitates the emergence of new meanings and insights thanks to the juxtaposition of ideas from different domains. Visual metaphors act as analogies, making “abstract subjects comprehensible” and “giv[ing] the researcher permission to focus on the relevant part of the problem, to abstract out other parts” (Ketokivi, Mantere, & Cornelissen, 2017, p. 237). Our work raises questions about how visual metaphors count in paradox theorizing. It also invites scholars to consider other forms of visualization, such as the sailing visuals used by Lê and Pradies (2022) to unpack improvisation when navigating paradox in stormy conditions, or the African symbols and their relation to paradox theorizing explored by Koli and Lê (2022). The introduction of more complex imagery could also allow scholars to transcend the three antinomies.

Conclusion

Our paper encourages the community of paradox scholars to be as imaginative as possible in addressing paradoxes visually. Along with the abstract and the contribution, we need to pay particular attention to visual images because they may be the strongest messages we send to our audience. In sum, our paper invites discussion enriching representation of paradox, and thus opens multiple avenues for future research. First, while our paper focuses on single basic visual elements of a figure, it may be that when signs are combined, they reveal aspects of paradox that each sign alone does not. We see this as the potential paradoxical visual nature of a single figure; that is, separately, two or more signs in the same figure might appear disconnected, while together they convey novel elements of paradox. Second, our analysis has explored visuals across different types of papers (e.g., empirical vs. theoretical) and figures (e.g., figures visualizing paradox description or paradox dynamics); however, other dimensions, such as the type of paradox being discussed, the level of analysis (e.g., individual, team, organization—and interactions between them), or the number of poles at play, could be investigated. Future research could also refine analysis further by looking at these dimensions. Third, our work addresses the (in)commensurability of paradox. By using a two-by-two-by-two device, boundaries are created that may suggest that expressions of paradox are relatively stable and commensurate. Yet, tensions are dynamic and can evolve and metamorphose, exposing different meanings at different levels and time scales (Jarzabkowski et al., 2013), or result in cascading effects (Gilbert, Michaud, Bentein, Dubois, & Bédard, 2018). In essence, visualization is a way of seeing (Berger, 1972) but as scholars continue to (re)produce visuals, one must never forget that a way of seeing is also a way of not seeing.

Supplemental Material