Categories, Identities, and Cultural Classification: Moving Beyond a Model of Categorical Constraint


Address for reprints: Mary Ann Glynn, Boston College, 140 Commonwealth Ave., Chestnut Hill, MA 02467, USA (


Categorization processes have gained currency in organizational theory. Categories are endemic to organizations and markets, serving as touchstones for organizational identity claims and for audience attention, legitimation, and valuation. Durand and Paolella argue for an expansion of current perspectives on categories, particularly that of prototype theory. Although we agree in spirit, we advocate an expansion of their perspective, which seems to focus primarily on the cognitive aspects of categorization and the force of their constraint, particularly at the individual level of analysis. We suggest three revisions to Durand and Paolella's arguments in order to extend the conversation. First, we advocate that categorization processes might usefully be understood by socio-cultural perspectives that explicitly consider the role of audiences and the embeddedness of categories in wider classification systems. Second, we connect categorization processes to identity formation and maintenance at the levels of both the organization and the market. Third, we move beyond the constraining power of categories to consider their generative capabilities in processes of emergence and change. Overall, we discuss these in the context of organizational identities and cultural classification systems.


Categories are social constructions that provide a supple ‘conceptual system’ (Rosa et al., 1999, p. 64) that can be constructed and reconstructed, with regard to the realities that are encountered in the world (Mervis and Rosch, 1981) or the particular purposes that are pursued (Zuckerman, 1999). Categorization is a complex process that can span multiple levels of analysis (from individuals through organizations and markets), and that has been linked to a variety of organizational phenomena, including: stock valuations (Zuckerman, 1999), entrepreneurial proposals (Cornelissen and Clarke, 2010; Navis and Glynn, 2011); competitive rivalry (Porac et al., 1995); market emergence (e.g. Kennedy, 2008; Khaire and Wadhwani, 2010; Navis and Glynn, 2010; Rosa et al., 1999); and broader social movements (Rao et al., 2003; Weber et al., 2008).

Durand and Paolella (2013) argue against the view that ‘categorizing is a mechanical cognitive process …’ (p. 6); like them, we approach categorizing as a phenomenon that is often more complicated, less mechanical, and sometimes more problematic. We seek to elaborate the foundation outlined by Durand and Paolella in three directions, by theorizing categorizing processes in terms of: (1) socio-cultural perspectives that explicitly consider the role of audiences and the classification systems in which categories are embedded; (2) identities, at the levels of the organization and market; and (3) not only the constraining power of categories but also their generative capabilities.

Although we have individuated these three theoretical extensions on categorizing, we recognize that they are fundamentally inter-related: categorization is a cooperative venture between category members and their relevant audiences, rooted in cultural understandings and expectations, and which, because of the interpretive potency of categories, can allow for more tolerance and less discipline than Durand and Paolella (and others) convey. In the remainder of this article, we break down this explanation of categorizing to explore the constituent ideas; then, we synthesize their implications for categories, identities, and cultural classification systems.

Categorization as Socio-Cultural Classification and (Audience) Perception

Categorization furnishes an answer to the fundamental question of sensemaking, ‘What is it?’ Or, more specifically, ‘What kind of thing is it?’ as a way of locating an entity – a person, an organization, an object, an event, an experience, or most anything – within a broader system of meaning or classification. And so, one might answer the sensemaking question (above) by identifying the person as a mother, the organization as a bank, the seat as a chair, the song as country–western, and the celebration as a birthday party. The answer tends to privilege an entity's sameness or resemblance to others in the focal category, over the distinctiveness that they may evidence relative to other categorical members. Implied in the answer is that, for example, a person is a mother and not a father; an organization is a bank and not a school; the seat is a chair and not a sofa; the song is country–western and not jazz; and the celebration is a birthday party and not a wedding. Thus, the act of categorization is hinged to the broader classification system in which it is embedded, implicitly or explicitly.

More generally, classification schemes imply both membership in a particular category and not in others; thus, they encapsulate relational oppositions. This is true, even for the groupings identified by Durand and Paolella, i.e. causality (birds that fly and those that do not) and ad hoc goals (things to stand on to change a light bulb and things not to stand on for this task). Categories are constructed so as to minimize the differences among members within the category while maximizing the differences between other categories; the effect is to create sharp or clear boundaries between categories that can forestall problems of categorization (e.g. Zuckerman, 1999).

Thus, a well-bounded classification scheme clarifies the inclusion (and exclusion) of members and is used by relevant audiences to understand and sort things into categories; indeed, fitting things into such well-established and well-understood categories creates a ‘categorical imperative’ (Zuckerman, 1999) such that audiences prefer to base their valuations on established categories. Indeed, the imperative has been shown to be powerful enough that things that fall outside the classification scheme are typically overlooked or devalued (Zuckerman, 1999). In this case, categorizing is generally as Durand and Paolella have described it, as a prototype matching process that is fairly mechanized or routine.

And yet, recent research suggests that the categorical imperative may not be as forceful across different kinds of contexts, particularly those with different kinds of boundary conditions, including: when categorical classifications and boundaries are unclear or in flux (as in emerging markets or industries); when the purpose of categorization is something other than valuation (or the application of well-understood or taken-for-granted criteria), or the target of categorization is highly ambiguous (as is often the case for new entrepreneurial ventures or organizational forms). Under these conditions, the perceiver has few, if any, benchmarks against which to sort, classify and assign meaning. Thus, the lack of clarity in the target, boundaries, and categories does not offer the easy or simplified answer to the key question of categorization, ‘What kind of thing is it?’ or an understanding of the relational opposition to other categories that undergirds the process.

For instance, when category boundaries are more fluid (as in emerging markets, e.g. Navis and Glynn, 2010, or entrepreneurial ventures, e.g. Navis and Glynn, 2011) or when categorizing has a different objective (as in understanding a creative or relatively unfamiliar musical product; e.g. Phillips and Kim, 2009), the categorical imperative may be decreased or irrelevant. In this case, when categorical boundaries are unclear or ignored, prototype based processing may be even more critical in classification because it offers a benchmark for comparisons. The prototype is the most central member in the category (Rosch, 1978), as Hogg and Terry (2000, p. 123) explain:

Prototypes are typically not checklists of attributes but, rather, fuzzy sets that capture the context-dependent features of group membership, often in the form of representations of exemplary members (actual group members who best embody the group) or ideal types (an abstraction of group features).

The process of categorizing renders an entity more understandable because it furnishes a conceptual label, or a set of meanings, that are applied to the entity, thereby distilling it into a condensed form; as such, the conceptual order of the category substitutes for the perceptual order of experience (Weick, 2012). Were we to treat each experience or entity we encounter as new, unfamiliar, and uncategorizable, we would be cognitively overwhelmed, behaviourally paralysed, and devoid of meaning or action expectations. Thus, prototypes are useful because they generalize and abstract: ‘experience is named by words, gathered into generalizations’ (Weick, 2012) that constitute the meaning of categories.

Once established and shared, classification schemes encourage us to think about experience in terms of categories, rather than direct experience; as a result, categorizing can become a ‘disciplining exercise’ (e.g. Zuckerman, 1999). But categorizing need not become the ‘straightjacket’ that Durand and Paolella describe. For instance, Wry et al. (2011) show how categories can exhibit ‘acceptable variation’; King and Whetten (2008) demonstrate that ‘minimum standards’ bound categories; Lounsbury and Rao (2004) attest to how product categories in the American mutual fund industry can be reformulated; Kennedy (2008) empirically shows how new categories may coalesce around aprototypical firms that do not fit established categories; and Rao et al. (2005) demonstrate the weakening of categorical boundaries with category crossing in French gastronomy. Taken together, this body of work suggests that categories may be more dynamic than Durand and Paolella depict. More than that, it indicates that the genesis of new category emergence may have fairly diverse sources, a consideration we take up subsequently, when we discuss the generative capabilities of categories.

Beyond conferring comprehension and meaning, categorizing reduces uncertainty and confers validation by recognizing an entity's membership. Prototypes are the engine for this categorization: ‘It is the prototype that actually reduces uncertainty. Hence, uncertainty is better reduced by prototypes that are simple, clear, highly focused, and consensual, and that, thus, describe groups that have pronounced entitivity’ (Hogg and Terry, 2000, p. 124). A corollary to this uncertainty hypothesis is that, under conditions of greater uncertainty, there is more of an urge to categorize, in order to reduce that uncertainty; such an urge can easily frustrate because categorization is no longer clear or easily accomplished. This may offer an alternative explanation to Zuckerman's (1999) finding that analysts, when confronting an aprototypical firm, tend to de-value it; such discounting may be due to the greater uncertainty (and presumably greater risk) rather than simply an imperative to conform to the category prototype.

Thus, categorizing is not purely cognitive, but socio-cultural as well because it is anchored in the context in which categorizing occurs (e.g. Rindova et al., 2011). Hogg and Terry (2000, p. 125) explain that an individual's ‘cognitive system, governed by uncertainty reduction and self-enhancement motives, matches social categories to properties of the social context and brings into active use (i.e. makes salient) that category rendering the social context and one's place within it subjectively most meaningful.’ As a result, categorizing is contingent on the social context in which it takes place, embedding it culturally and institutionally. Categories that best fit the field or context are activated, by firms claiming membership to that category and audiences granting that membership. Thus, categorization can confer a social identity, the topic we take up next.

Categorization and Identity

In contradistinction to theories that define an organization's identity in terms of a bundle of central, distinctive, and enduring attributes (e.g. Albert and Whetten, 1985), a categorical perspective defines the organization's identity in terms of its membership in one or more groupings (or categories) and, in particular, its embodiment of (or assimilation to) the categorical prototype. While attribute-based approaches to identity may (or may not) presume that particular behaviours will ensue, categorical approaches to identity are more explicit in this regard, bringing the organization's perceived identity and behavioural norms into alignment via prototype matching (e.g. Fiol and Romanelli, 2012; King et al., 2011). Because prototyping invokes stereotyping, action expectations, and a categorical identity, it is explicit about how organizations in a category should behave.

In spite of their obvious differences, conceptualizations of organizational identity and organizational membership categories have an affinity and complementarity (e.g. Navis and Glynn, 2010). Drawing on Scott's (2003, p. 7) conceptualization of organizations as ‘actors in their own right, as “collective social actors” ’, Whetten and Mackey (2002, p. 395) view organizations as social actors, ‘authorized to engage in social intercourse as a collectivity and possessing rights and responsibilities as if the collectivity were a single individual.’ This model of the organization – as a social actor – implies that ‘organizational identity is appropriately conceived of as a set of categorical identity claims (who or what we claim to be, categorically) in reference to a specified set of institutionally standardized social categories’ (Whetten and Mackey, 2002, p. 397), a perspective that is consistent with that initially articulated by Albert and Whetten (1985), which emphasized the claim-making aspects of organizational identity (Glynn, 2000). Similarly, Rindova et al. (2011, p. 423) show how such identity claims – or what they refer to as ‘self-categorizations’ – are enduring and responsive to updating, as organizations incorporate new cultural resources into their cultural ‘toolkit’ (Swidler, 1986). Thus, beyond a claim to a singular attribute, organizational identity involves claims to categories of membership.

Here, we highlight two key implications of a categorical approach to organizational identity: (a) managing the balance of ‘optimal distinctiveness’ between categorical similarity and differentiation, and (b) acknowledging the role of the audience in identity dynamics.

Identity as Optimal Distinctiveness Within a Category

If category prototypes are the metaphorical straightjackets described by Durand and Paolella, supplying the stringent criteria to which claimant members must conform or risk being internally ostracized or externally devalued, then it seems we would observe starker similarities between category members than we actually do. For instance, in the category of ‘Warehouse Clubs’, SAMS Club and Costco have identities that are readily distinguishable: among other distinctions, the two clubs differ in the extent to which they appeal to business owners and individual consumers in their store designs, product and service offerings, and personnel policies (e.g. Cascio, 2006). The same can be said of The Home Depot and Lowe's in the ‘Home Improvement Retail’ category, where similar distinctions are found. Indeed, even these category exemplars exhibit defining qualities that set them apart, seemingly loosening the presumed straightjacket of a prototype. In general, category members demonstrate different degrees of resemblance to a prototype – or ‘grade of membership’ (Hsu et al., 2009) – that permits some organizational diversity within the category.

From a definitional perspective, category prototypes represent the minimal rather than maximal standards for valid category membership. Prototypes reflect an ideal type, not necessarily the standards that all organizational members need to meet (King and Whetten, 2008). For example, while the sale of food represents a minimal standard for all organizations in the restaurant category, member organizations retain considerable discretionary autonomy to differentiate their identities and offerings (e.g. on the basis of food type, preparation style, dining experience, etc.) and still remain legitimate category members. This definition is consistent with the idea that categories will exhibit considerable acceptable organizational identity variation within their boundaries (Wry et al., 2011).

From a relational perspective, categories and their associated prototypes are not unitary constructs that exist in a vacuum, but instead are embedded in a much broader and richer classification system, as we discussed in the preceding section. The minimal standards for membership in the restaurant category as a whole, for instance, are quite different from those in its various sub-categories (e.g. fine dining vs. fast food; Mexican vs. Thai restaurants). The importance of these categorical relationships is reflected in identity research, where organizations are seen to claim identities at different (vertical) levels of category specificity, from superordinate to basic level categories (Rosch, 1978), and to pull in distal (horizontal) categories that can introduce legitimate variation into already established categories (Wry and Lounsbury, 2012). Indeed, a centrepiece of research at the nexus of identities and categories questions how and with what implications identities are variably configured (e.g. Navis and Glynn, 2011; Rindova et al., 2011; Wry and Lounsbury, 2012; Wry et al., 2011). Thus, organizations manage competing tensions to ‘fit in’ and garner category legitimacy through isomorphic mechanisms of similarity, and to ‘stand out’ and garner competitive advantage through strategic mechanisms of differentiation. As we discuss next, there are clear implications of the optimal distinctiveness of an organization's identity for audience perceptions and valuations or legitimation of the organization's identity.

The Role of Audiences in Categorizing

Perhaps nowhere else are the dynamics of organizational identity and audience assessments more transparent than in the case of new ventures seeking needed resources. Navis and Glynn (2011) argue that these dynamics are contingent on the market context in which the ventures operate: in established market categories, new venture legitimacy is attained by adherence to the minimal standards of the market category, thus leaving opportunity for credible variants in the business concept; in new market categories, however, legitimacy is attained by adherence to the minimal standards of the entrepreneurship category, such that distinctiveness derives from the novelty of the business concept. Navis and Glynn (2011) highlight how the cultural and institutional meanings affixed to categories prime both audience sensemaking and judgments of venture plausibility. In addition, they point to the importance of entrepreneurial narratives (Lounsbury and Glynn, 2001) – as a metaphorical ‘glue’ – that weaves together disparate categories of new venture membership to lend coherence and resonance to the organization's identity. More generally, this research points to the embeddedness of categories in a broader classification system, including how organizational and market-level category prototypes may interact to affect meanings and their potency.

Organizational identity claims, particularly for new ventures or hybrid organizations that cross two or more categories, often necessitate explanation or justification; without this, they can risk the illegitimacy discount (Zuckerman, 1999). As rationalizing accounts or explanations, narratives seem to be a particularly potent identity mechanism (e.g. Lounsbury and Glynn, 2001), cueing audiences about what kind of organization it is (e.g. Navis and Glynn, 2011). For instance, Martens et al. (2007) have shown that organizations' identity narratives can positively affect resource acquisition for entrepreneurial firms; Wry et al. (2011) reveal how narratives enable collective identity formation of entrepreneurial categories; Vaara and Tienari (2011) analyse how narratives construct identities in multinational corporations; Carlsen (2006) relates narratives to identity change over time; and Anteby and Molnár (2012) show how narratives can resolve seemingly contradictory identity elements in an organization's history. Like framing (Navis and Glynn, 2010), narratives or storytelling can function to construct, change, or make coherent an organizational identity; the importance of narratives, as rationalizing accounts of the organization's identity is thus likely to be more important when those identities are hybridized, under threat, or novel.

As much as organizations within a category may vary in their alignment, audiences too may vary in their assessments of the fitness of an entity's membership in the category and the associated behavioural expectations (e.g. Hogg and Terry, 2000; Navis and Glynn, 2011). When organizations claim identities in new or abstract categories (such as nanotechnology), the conceptual abstraction of those categories may have particular appeal to early market investors for their cultural currency (Kennedy et al., 2010) and the lenience they offer member firms to diversify within category boundaries (Pontikes, 2012). However, the same categorical fuzziness may dissuade many potential transaction partners due to the lack of a salient category prototype, i.e. one that supplies clarity around category boundaries, defining practices, and member firms. This logic indicates that categories serve multiple purposes. Moreover, judgments of organizational worth – when tethered to firms' identity claims – are, at least in part, formed in the eye of the beholder. What this logic does not indicate, therefore, is a view of categories as constraining cognitive forces.

One theoretical perspective that has been quite explicit in acknowledging the role of the audience in determining an organization's identity is that of population ecology (Hannan et al., 2007; Hsu and Hannan, 2005; Pólos et al., 2002). Theorists in this tradition postulate that categorizing an organization, and thereby endowing it with an identity, typifies its form, or the rule-like code through which it will be externally valued. As McKendrick et al. (2003, p. 61) describe:

… a form is an external identity code, meaning that it is the perceptions and opinions of ‘outsiders’ that matter. The external identity code possesses rule-like status, so that its observable violation is negatively sanctioned – it causes outsiders to drop discontinuously their valuation of the entity to which it is applied.

Identities then, consist of codes, which audiences apply to organizations to evaluate their categorical fitness. The imperative – or the disciplining effect (Durand and Paolella, 2013) – arises from the perceived organizational alignment with the category prototype. As Weick (2012) describes it, categorical experience substitutes for direct experience (with the organization) and the nuance of organizational individuation within the category is largely lost to the audience. The drive in categorizing here is to reduce identity uncertainty and ambiguity; organizations that invoke categorical confusion invite audience sanctions while those perceived to be sharp or focused in their category membership are rewarded (e.g. Hannan and Carroll, 1992; Hsu et al., 2009; Zuckerman, 1999). Thus, organizations are seen as passive players in categorizing processes; identity classifications are ‘coded’ by audiences.

A different treatment of the audience role is seen in Navis and Glynn's (2010) study of the emergence of satellite radio. Here, the audience is viewed in an interactive role, co-creating the category in conjunction with the member organizations in the collaborative venture of category emergence. Through the mutual sensegiving and sensemaking activities of the category's member firms and their audiences, the boundaries, practices, and membership criteria of satellite radio were established as increasingly ‘real’ and legitimate (Kennedy, 2008; Kennedy et al., 2010). These shared social construction efforts – which Navis and Glynn (2010) showed to be robust across a variety of actor types – supplied the basis for subsequent intra-category identity comparisons of member organizations. These findings highlight the temporal fluidity of category prototypes and the active role of both firms and audiences in their establishment and evolution.

Weber et al.'s (2008) study of the grass-fed meat and dairy product movement provides another example of audiences' participatory role in category emergence. As the grass-fed movement gained traction, cultural codes hinged to the authenticity, sustainability, and naturalness of cattle products became the prototypical markers of the nascent space; in turn, these appealed to and mobilized potential producers and consumers. Such advancements encouraged entrepreneurial production, led to the formation of a collective identity around these codes, and helped to legitimate the grass-fed meat and dairy product category. The audiences (e.g. investors, suppliers, customers) used the emergent category prototype as a signal of resonance with their own interests and as an organizing mechanism to structure the system of market exchange. These findings highlight the role that audiences play in category emergence and establishment, as well as how prototypes can organize the classification system. By bringing in the audience more explicitly, it becomes more evident how the categorical perspective on identity invokes considerations of the category's credibility, legitimation, or resonance.

The Generative Capabilities of Categorization

To this point, we have emphasized that categories are embedded within broader systems of meaning and classification. And while the embeddedness concept is often associated with constraint, we contend that categorization can be generative because of, and not in spite of, the structurally and culturally embedded features of categories. Generative dynamics have the ‘power to originate and propagate something that would not exist otherwise’ (Rousseau and Ling, 2007, p. 3). Categorization can be a generative process for organizations when it expands collective resources and capabilities so as to create new identities, new possibilities for action, or new membership categories that are valued and legitimate.

Acknowledging the socio-cultural aspects of classification systems affords a view of categories as cultural elements or toolkits (Swidler, 1986) that are ‘public resources’ that actors can selectively ‘pull down’ (Weber and Dacin, 2011) in the service of constructing identities (e.g. Rindova et al., 2011), rendering them comprehensible and thus legitimate (e.g. Navis and Glynn, 2011). Categories can thus serve as the ‘raw material’ for bricolage, category spanning, or identity formation for organizations. Categorically familiar meanings, imported from disparate domains, can prime new understandings and evaluations, thereby blunting the unfamiliarity associated with novelty, and enabling entrepreneurial success and wealth creation (e.g. Lounsbury, 2001; Navis and Glynn, 2011). Conversely, bringing categorically unfamiliar meanings into more familiar domains similarly enables creative recombinations, which can be blended and naturalized using cultural narratives, frames, or labels. Thus, categories can be generative, when they serve as the cultural building blocks for identity bricolage or as creative recombinations for organizational innovation and change (King et al., 2011).

Furthermore, categories have generative potency because of their contextual fluidity. Although they are situated in particular contexts, categories may not be fully bound to a particular context, but rather may be transported (or diffused) to different types of situations or contexts. This kind of diffusion often plays a role in the development of hybrid identities, which consist of identity elements that are not usually found in combination (Albert and Whetten, 1985). Battilana and Dorado (2010) show how the category of ‘business’ was coupled to that of ‘social enterprise’ to create an organizational identity for two pioneering commercial microfinance organizations. Similarly, Glynn (2000) demonstrated how the categorical expectations of the business category applied to a symphony orchestra created identity challenges. Thus, the diffusion of the categorical prototype can infuse other, and sometimes unrelated, categories so as to reframe our understandings of an organization and our expectations of its behaviour.


Leveraging the ideas advanced by Durand and Paolella in their article, we sought to move beyond a view of categorization as primarily cognitive, highly constraining, and lodged at the individual level of analysis. Instead, we argued for an understanding of categorization that is cultural, as well as cognitive; enabling as well as constraining; and organizational or collective in nature. To do so, we advanced three amendments to Durand and Paolella's arguments. First, we advanced a theorization of categorization processes that was informed by socio-cultural perspectives that explicitly accounted for the broader classification system in which categories are embedded. Moreover, we recognized that such classification systems are available to both organizations and their audiences; as such, we conceive of categories as a cultural resource or toolkit (Swidler, 1986) that organizational actors and their audiences use to make sense of, attach meaning to, and evaluate, the organization. More broadly, we argued that categorization is a cooperative venture between organizations and their audiences, rooted in cultural understandings and expectations, and which, because of the interpretive potency of categories, allows for more tolerance and less constraint than Durand and Paolella suggest. Thus, we extend the conceptualization of categorization beyond the individual self to that of the others (i.e. audiences), depicting the organization as a social actor (Scott, 2000; Whetten and Mackey, 2002) that makes claims to, and is granted membership (by audiences) in, categories that can yield and legitimate the organization's identity.

Building on the concept of categorization as a process of social construction between actors and audiences, our second argument focuses on the formation and maintenance of identity at the levels of both the organization and the market. We show how categorical membership is tethered to an organization's resemblance to the prototype, but that this prototype need not envelope organizations in an inflexible straightjacket. Instead, we view prototypes as setting the minimal standards for membership and as a kind of touchstone for organizations to manage both their similarity to, and differentiation from, other members of the same category. The result is to yield identities that are optimally or legitimately distinctive (Brewer, 1991; Brewer and Gardner, 1996; Navis and Glynn, 2011) and to recognize that categories paradoxically encapsulate both homogeneity and heterogeneity in organizational membership. Moreover, it implies that prototypes can both constrain variation and enable differentiation, thus seeding generative capacity in organizations, our third revision to Durand and Paolella.

Third, we sought to move beyond the constraining power of categories and to acknowledge their generative capabilities, particularly in processes of creative recombination or bricolage. As much as conformity to a prototype conditions organizational membership in a category, categorical membership can be a font of cultural resources that can be generative of other organizational resources that may shape identity, advance growth, or enable change. Combining existing categories, like ‘satellite’ and ‘radio’, to create a new market category (‘satellite radio’) can generate new understandings, identities, and expectations for the firms in the new space (Navis and Glynn, 2010). Similarly, Battilana and Dorado (2010) show how such combinations (i.e. ‘commercial’ and ‘microfinance’) can yield new hybrids (‘commercial microfinance’).

Our work has interesting implications for theory and empirics. Importantly, our ideas call attention to the importance of clarifying the unit of analysis under study. For instance, researchers have examined whether, and with what implications, organizations have memberships that span categories, showing that cross-categorizations carry risks for organizations (e.g. Hsu, 2006; Hsu et al., 2009; Kovács and Hannan, 2010). Our work suggests that such risks may be contingent on the level at which focal categories are specified. Nanotechnology organizations, for instance, may not be seen as spanning categories when understood at the superordinate category of nanotechnology; however, as the unit of analysis shifts more narrowly to a basic-level category or even subcategories at lower levels of analysis (e.g. carbon nanotubes), those same organizations may be seen as category spanners. Thus, unit of analysis matters because it specifies the level of the categories invoked, e.g. basic level, superordinate, or subordinate categories (Mervis and Rosch, 1981), which in turn locates organizations in broader systems of meaning associated with the classification system.

In addition, we have emphasized that categories are dynamic and not simply static; this may be especially true for industry or market categories, which are in continual flux as new entrants introduce acceptable variation, significant jolts occur from the external environment, and performance dimensions change. Such dynamism can originate from political processes involving the state, media, firms, or professions, for example, which can alter the frames through which categories are understood and intra- and inter-category comparisons are made. In some cases, these fluctuations can even lead to the formation of new categories. Accordingly, categorization scholars must be careful to account for relevant differences in the composition and application of categories over time (cf., Navis and Glynn, 2010): as categories evolve, so do the bases by which members and audiences understand and use them. Broadly speaking, ‘research on categories is distinctive in its focus on meaning rather than structure’ (Negro et al., 2010, p. 9). Although categories have structural properties as members of classification systems, it is their meanings – emotional, behavioural, social, or economic – that are likely to play the most important role in explaining organizations in terms of their identities and actions.

Theoretically, our arguments suggest several useful directions for future research. Importantly, we illuminate how categorization processes are not only cognitive, but cultural. By focusing too narrowly on the cognitive moorings of categories, we can risk overlooking their embeddedness in broader systems of meaning and their embeddedness in societal, market, and industry contexts. Categories are cultural constructions, but, over time and frequent use, they take on an institutional or taken-for-granted character; recognizing this allows us to see that not all categories are culturally potent and that some may be more legitimate than others. Thus, organizations' identity claims to categories are not random, but rather are ones that lend advantage and credibility. We can suppose that categorical claims may be predicated on a certain form of cultural resonance, in that organizations may seek to appropriate categorical memberships that align them well with their social, environment, or technical environments.

Recognizing the cultural constitution of categories permits us to see how categories furnish a repertoire of meanings for categorical members that they can appropriate in constructing their own identities. Conceptualizing the organization as a social actor (Whetten and Mackey, 2002), Glynn and Watkiss (2012) show how it ‘functions as a skilled cultural operative, sufficiently agentic so as to select those cultural elements that align with its internal character, but not so insensitive as to adopt those elements that are culturally inappropriate or illegitimate.’ Drawing from work in cultural sociology (Small et al., 2012), Glynn and Watkiss (2012) identify several cultural mechanisms that can operate in identity construction. Two of these seem particularly relevant in theorizing categorization processes and especially, the link to organizational identity: culture as a ‘toolkit’ and culture as a narrative.

One particularly promising path of inquiry is to conceptualize culture as a ‘toolkit’ or reservoir of meanings (Swidler, 1986; for applications of this approach for organizations, see Weber, 2005 and Glynn and Giorgi, 2013). Some recent empirical work suggests the feasibility of this approach (e.g. Rindova et al., 2011) and, particularly, how the organizational enrichment of its cultural repertoire (by appropriating cultural tools) can define and redefine an organizational identity over time. By explicitly acknowledging the cultural underpinnings of categories, we can understand organizational claims of membership to specific categories as attempts to enhance (or increase) their cultural toolkits and resources. Thus, categories are not merely devices used by audience to understand and sort organizations, but also resources that organizations can appropriate to their advantage.

Conceptualizing culture as a narrative (Glynn and Watkiss, 2012) enables us to understand the interpretive mechanisms that enable categorization processes and categorical claims by organizations. Narratives bundle meanings into interpretive packages that offer ready-made accounts of why membership claims are valid, coherent, and appropriate. Earlier, we discussed how narratives can offer rationalizing accounts for categorical memberships, perhaps when they may not be easily evident, as when organizations or markets are new or changing. Moreover, organizational storytelling that narrates and justifies cross-categorical claims would seem to be particularly effective in legitimating pluralistic identities or hybrid forms.

Empirically, this approach suggests that inquiry into categorization might usefully focus on cultural resources and cultural narratives, and especially their link to the claimed organizational identity. For instance, researchers might examine cultural resources using methods similar to those used to examine strategic, human, or economic resources. And, narratives can be content analysed for their organizational claims to categories, the meanings of those categories for the organization, and for the relationships among multiple categories the organization might claim. To conclude, we advanced three theoretical extensions intended to build on and amend the arguments for a reorientation of category research articulated by Durand and Paolella. By shining a spotlight on the classification systems in which categories are embedded, connecting category and identity concepts at multiple levels, and pointing to the generative capabilities of categories, we hope to add to their efforts to expand our theorization of categorization processes. Overall, we aim to move beyond a model of categorical constraint and redirect future research to explore some relatively unexplored issues in organizational theory.