The exercise of informal influence through intra-organizational impression management Open Access

In recent years, a vast body of research has emerged which investigates how managers attempt to influence stakeholders’ perception of the organization and themselves through an organization’s communication with its constituencies (e.g. Cho et al., 2015; Guillamón-Saorín et al., 2017). The literature refers to such endeavors as impression management, which is “the process by which individuals attempt to control the impressions others form of them” (Leary and Kowalski, 1990, p. 34). From the perspective of institutional theory, the major impetus for impression management is to ensure legitimacy (Bansal and Clelland, 2004) by managers who seek to influence stakeholders’ approval for their organization and themselves (Jin et al., 2022).

While prior research has unveiled numerous facets of impression management directed at an external audience, accomplished primarily through the narrative parts of accounting communication (see, for instance, the synthesis by Merkl-Davies and Brennan, 2017), little attention has been paid to managers’ efforts to ensure their internal legitimacy through impression management in intra-organizational communication. Internal legitimacy can be defined here as the perception of internal units that the actions of another internal unit are desirable, proper or appropriate within the organizational system of norms, values, beliefs and definitions (see Suchman, 1995, p. 574). The securing of legitimacy is especially important for internal units in “post-bureaucratic” organizations (Courpasson and Reed, 2004), which are defined as flexible and decentralized governance structures, in which the position of organizational units and their managers is not assured by the rational-legal bureaucracy alone (Maravelias, 2003). Such governance structures can be found in various alternative organizational forms such as network organizations (Huybrechts and Haugh, 2018), joint ventures (Cäker and Siverbo, 2011) and cooperatives (Oczkowski et al., 2013). Post-bureaucratic organizations combine traditional bureaucratic governance with principles of informal networks and democratic decision-making, and their “managers are [typically] confronted with a situation where imperative authority and its fiat is no longer a solution” (Josserand et al., 2006, p. 56). In this setting, organizational members are in positions to suspend their support for internal units and to challenge the informal influence that is the basis of their efficacy (Styhre, 2008). Consequently, these internal units need to explicitly legitimize their existence and position vis-à-vis other units.

One way to do so is to produce discourses, referred to elsewhere as “softer forms of domination and control” (Courpasson and Reed, 2004, p. 7), which can promote the internal legitimacy of units and the formal instruments they operate by aligning them with favorable values and ideals espoused by particular managerial narratives. The generation of internal legitimacy is therefore largely a discursive effort (Kostova and Roth, 2002; Landau et al., 2014) and reflects that “language is a medium of social control and power” (Leclercq-Vandelannoitte, 2011, p. 1251). However, the accounting literature has largely overlooked impression management in post-bureaucracies; a gap that matters in light of the increasing proliferation of organizations exhibiting post-bureaucratic traits (Bolin and Härenstam, 2008; Maravelias, 2003), and the consequential prominence of internal legitimacy conflicts in organizational control dynamics.

In this paper, we explore how intra-organizational impression management (IOIM) is leveraged in a post-bureaucratic organization. The study is anchored in a rhetorical perspective, given that rhetoric, as “the art of persuasion” (Harmon et al., 2015, p. 77), is “a type of instrumental discourse used to persuade audiences” (Green, 2004, p. 654), which is thus aimed at ensuring legitimacy (Hoefer and Green, 2016; Suddaby and Greenwood, 2005). Following Aerts and Yan (2017) and in line with the idea that language is a medium of power and control, our analysis focuses on linguistic style characteristics. We mobilize the idea that linguistic devices enable rhetorical appeals, which are intended to guide recipients in interpreting and evaluating the information provided in ways preferred by the sender and build relationships between the sender and its audience (Crismore et al., 1993; Hyland and Tse, 2004).

Our historical study examines rhetorical IOIM in two Dutch cooperative banks over a timespan of several decades. A reliance on historical sources is required here because legitimation practices tend to disappear from contemporary awareness as they become taken-for-granted over time (Suddaby et al., 2017, p. 470). Despite its labelling suggesting a relatively new phenomenon, cooperatives have embodied various post-bureaucratic characteristics for many years. A cooperative bank is a particularly illustrative instance of a post-bureaucratic organization because it is not governed by a traditional rational-legal bureaucracy (which we will refer to as formal bureaucracy henceforth) but is a members-based democracy instead. Therefore, its managers are “caught in the ambiguous dialectics of democracy and bureaucracy” (Josserand et al., 2006, p. 56) that are the basis of post-bureaucracies. Formally autonomous local member banks own their supra-local organizations, which are referred to as their central banks (Sluyterman et al., 1998). Despite being central decision-making units, the central banks have limited formal power over the autonomous member banks. This complicates many tasks that require cooperation from local member banks, including group strategy implementation, risk management and IT management. Additionally, as regulatory bodies do not wish to deal with member banks individually, the latter have delegated control over regulatory compliance to their central banks, but without providing them with the formal means to fully enforce this compliance. Due to the inverse hierarchical relation of the cooperative, member banks are positioned to suspend their support for their central bank and its initiatives in the cooperative group. In the past, they have repeatedly done so, blocking strategic and reorganization initiatives by the central banks (Sluyterman et al., 1998). However, the central banks are not without (informal) means; they control much of the information circulating in the cooperative, providing them with opportunities for IOIM.

To ensure the internal legitimacy of themselves, and their informal influence over member banks, the cooperative central banks regularly issued so-called “relational reports” to their member banks. We analyze more than half a century of monthly and weekly issues of these reports (1911–1971), to study if and how rhetorical IOIM was mobilized. In this analysis, we recognize that, in addition to IOIM, internal units in post-bureaucratic organizations may also draw on alternative sources of internal legitimacy. In the cooperative democracy, the relative bargaining power of an internal unit and its subsequent effectiveness in lobbying – taking advantage of informational and resource asymmetries in a group of highly diverse autonomous member banks – is an especially salient resource for generating intra-organizational legitimacy (Bontems and Fulton, 2009; Maroua, 2015). Moreover, an internal unit’s rhetorical alignment with broadly shared ideological concerns may also contribute to its legitimacy. For cooperatives, this especially concerns alignment with (1) cooperative ideology aimed at the advancement of underprivileged societal groups (Heras-Saizarbitoria and Basterretxea, 2016) and (2) religious ideology on which the cooperative ideas of some organizations are founded (Sluyterman et al., 1998).

By examining IOIM and its interaction with these alternative resources for internal influence, we address two research questions which guide our paper:

1. When its position is not assured in a formal bureaucracy, does an internal unit, such as a cooperative central bank, engage in (different types of) rhetorical IOIM?

2. How is the internal unit’s propensity to engage in different types of rhetorical IOIM related to the other sources of informal influence at its disposal?

We mobilize Aerts and Yan’s (2017) content analysis method to examine linguistic style characteristics contributing to rhetorical appeals in intra-organizational communication. Building on the assumption that rhetorical appeals do not manifest in single linguistic style dimensions but rather in combinations of several dimensions (Yan et al., 2019), we examine rhetorical profiles indicative of co-occurrence patterns among the linguistic style characteristics. Our findings suggest that the cooperative central banks mobilized three separate rhetorical IOIM profiles, each of them complementing or substituting alternative sources of informal influence: bargaining power or rhetorical alignment with religious or cooperative ideologies.

Our paper contributes to prior research on impression management by providing evidence of rhetorical impression management through linguistic style characteristics in an intra-organizational context. In this regard, we highlight how IOIM may be attuned to the characteristics of the audience, something that is less feasible with extra-organizational audiences. In addition, we contribute to the management control literature by highlighting that IOIM may be used in “packages,” complementing or substituting other informal controls, including bargaining power and rhetorical alignment with religious and cooperative ideology.

The remainder of the paper is structured as follows. In Section 2, we review the relevant literature and theory to develop our research questions. In Section 3, we explain the methodological choices underpinning this study. Our findings are presented in Section 4. In Section 5, we discuss these findings, and in Section 6, we conclude with the implications and limitations of our study.

From the perspective of institutional theory, managers are motivated to ensure legitimacy for their organization and themselves to secure continued stakeholder support (Bansal and Clelland, 2004; Jin et al., 2022). Here, an organization is legitimate to the extent that it conforms with social norms, values and expectations. However, the achievement of organizational legitimacy is inherently problematic, because social values and expectations are often contradictory and ambiguous (Ashforth and Gibbs, 1990). Therefore, organizations seek to influence the normative evaluations of their audiences in various ways. One way is through an organization’s communication with its stakeholders (Brennan and Merkl-Davies, 2013), which “enables [them] to establish and maintain relationships and to create [their] social worlds” (Merkl-Davies and Brennan, 2017, p. 433).

A significant body of literature builds on the premise that managers utilize the narrative parts of accounting communication to manage the impressions external audiences have of their organization and themselves (Merkl-Davies and Brennan, 2017). In its attempts to ensure its legitimacy, an organization develops communication strategies that shape the audiences’ judgment by presenting itself in a favorable manner and by convincing them that the organization complies with their objectives and expectations (Hoefer and Green, 2016; Jin et al., 2022). Building on literature that considers impression management a central part of legitimation (Ogden and Clarke, 2005), prior research has investigated various channels of communication, including the narrative parts of annual reports, such as letters to shareholders (e.g. Aerts and Yan, 2017; Moreno et al., 2019), SEC filings (e.g. Hooghiemstra et al., 2017), press releases (e.g. Guillamón-Saorín et al., 2017) or conference call transcripts (e.g. Frankel et al., 2016).

This stream of research employs a range of methodological approaches. Smith and Taffler (2000) suggest that there are two generic approaches to analyzing accounting narratives – form-oriented and meaning-oriented approaches. Form-oriented studies focus on linguistic style characteristics by analyzing word occurrences as indicators of specific stylistic patterns. For instance, they capture the frequencies of positive and negative words, which are indicative of the tone of narratives (Brennan et al., 2009). By contrast, meaning-oriented studies highlight content as they analyze the underlying meaning of a text. For example, this type of study suggests that managers claim responsibility for successes and attribute failures to extraneous influences. Likewise, it shows that managers refer more frequently to the future in periods of low performance to divert readers’ attention from unfavorable outcomes at the present time (e.g. Clatworthy and Jones, 2006; Keusch et al., 2012). The disparate foci of form-oriented and meaning-oriented analyses result in differing levels of objectivity and subjectivity, respectively (Smith and Taffler, 2000; Steenkamp and Northcott, 2007). Overall, the form-oriented approach provides a structured, measurable way to capture linguistic style characteristics, distinguishing it from meaning-oriented analyses that rely on interpretative content assessments.

To date, several papers have been published that use historical data to explore impression management in extra-organizational communications of single firms (e.g. Corazza et al., 2020; Moreno et al., 2019). More sizable is prior research that relies on larger-scale empirical methods to identify institutional factors, firm characteristics and conditions to engage in impression management, such as companies’ listings (Aerts, 2005), financial performance (Yang and Liu, 2017) or equity offerings (Davis and Tama-Sweet, 2012). By contrast, as the next subsection will show, a much smaller literature addresses intra-organizational communication and tends to draw mostly on meaning-oriented studies of specific controversial decisions in individual firms.

Many modern-day organizations not only rely on hierarchical structures for their governance but are typified by more flexible and decentralized governance structures. Such post-bureaucratic organizations (Courpasson and Reed, 2004) are governed by combinations of formal bureaucracy and more “entrepreneurial” governance forms, involving informal micropolitics of gaming and decentralized negotiations over power (Courpasson, 2000). This type of organization includes network (Huybrechts and Haugh, 2018), multinational (Brenner and Ambos, 2013) and cooperative organizations (Oczkowski et al., 2013). Their units have in common that their internal legitimacy is not only derived from their position in the formal hierarchy but also from their ability to deploy informal strategies to influence their position in the organization’s pecking order. For example, Lu and Xu (2006) note how internal units in multinational joint ventures need to compete for resources from external and internal stakeholders and thus need to obtain external and internal legitimacy. The same holds for fully owned units in multinational organizations (Kostova and Roth, 2002). In such organizations, internal units deploy strategies to position themselves favorably in relation to other units (Balogun et al., 2019).

Many of these strategies are discursive in nature (Erkama and Vaara, 2010). In this regard, Landau et al. (2014) argue that organizational units mobilize multiple “legitimation narratives” as part of internal struggles to establish the legitimacy of competing actions. In line with this idea, some papers analyze internal documents and explore different legitimation strategies unfolding in the dynamics surrounding controversial organizational decisions. Although these studies focus on the contents of communication and therefore provide especially meaning-oriented analyses (Smith and Taffler, 2000), they do occasionally refer to the rhetorical appeals that managers make. For instance, Erkama and Vaara (2010) identify several rhetorical strategies aimed at legitimizing a single controversial decision in a manufacturing plant. Balogun et al. (2019) find that senior managers sought to ensure the internal legitimacy of a relocation decision by highlighting strategic benefits and emphasizing the successes of prior decisions by headquarters. Sapir (2020) explores so-called rhetorical legitimation work in a higher education institution which needed to generate internal legitimacy for the decision to branch out to other sectors. Referring to the emergence of a new organizational form, in terms of multidisciplinary partnerships between accountants and lawyers, Suddaby and Greenwood (2005) find that the proponents and opponents of such partnerships mobilized specific rhetorical strategies combining different institutional vocabularies and theorizations of change.

These papers provide the first empirical evidence of discursive attempts to ensure internal legitimacy by making rhetorical appeals. However, the literature has yet to move beyond the legitimation of single controversial decisions and consider the legitimation strategies of internal units for themselves through rhetorical IOIM, particularly regarding its linguistic style characteristics. It is important to consider these aspects of internal communication, as effective internal accounting communication is considered a significant, albeit under-researched organizational asset (Jack, 2013), which is not yet fully understood by the accounting community (Daff and Parker, 2021). In this paper, we seek to address this research void by exploring if and how IOIM is consistently leveraged in a post-bureaucratic organization.

Rhetoric refers to persuasive language with the objective to establish legitimacy (Green, 2004; Harmon et al., 2015). According to Sillince and Suddaby, “rhetoric is the core form of communication that coordinates social action” (2008, p. 6). Organizations mobilize rhetoric to influence others’ perceptions of themselves or an organizational unit (Edgar et al., 2022). Therefore, rhetoric represents “communicative resources” which may be mobilized to achieve specific outcomes, such as convincing stakeholders about the adequacy of an organization’s actions or achieving acceptance of specific corporate decisions (Brennan and Merkl-Davies, 2014).

Our analysis is anchored in Classical Aristotelian rhetoric, which comprises three strategies – logos (emphasizing rational reasoning), pathos (appealing to emotions) and ethos (reflecting the communicator’s credibility and authority) (e.g. Aerts and Yan, 2017; Edgar et al., 2022). According to this framework, organizations or, as in our setting, organizational units may attempt to persuade an audience based on rational arguments, by evoking affective responses or by emphasizing their credibility and authority. These appeals may materialize in different ways. Research in applied linguistics suggests that communication consists of the content itself as well as accompanying linguistic style characteristics intended to guide the recipients in evaluating the content in ways preferred by the sender (Hyland and Tse, 2004). Accordingly, rhetorical appeals can be made directly through specific content in communications. The papers by Edgar et al. (2022) and Paugam et al. (2021) are recent examples of meaning-oriented approaches to investigate profit warnings and activist short sellers’ reports, respectively. However, rhetorical appeals can also draw on linguistic style characteristics guiding the readers’ interpretation and perception of the content provided (Hyland, 1998). In line with this idea, Aerts and Yan (2017) and Yan et al. (2019) provide form-oriented analyses of letters to shareholders and identify holistic rhetorical profiles consisting of co-occurrences of linguistic style characteristics.

In this paper, we follow Aerts and Yan (2017) to examine rhetorical IOIM through linguistic style characteristics. Our starting point is a set of linguistic style characteristics anchored in the rhetorical strategies of logos, pathos and ethos. Next, we explore rhetorical profiles consisting of distinctive co-occurrence patterns of the individual linguistic style characteristics. The form-oriented approach to rhetorical IOIM is suitable here for at least three reasons. First, we propose that informational asymmetries within an intra-organizational context are comparatively less significant than those observed in communications with external stakeholders. This distinction arises because internal recipients are generally more familiar with the subject matter being conveyed. In line with Bergh et al. (2019), who conceptualize impression management as leveraging informational advantages, reduced informational asymmetries may consequently limit opportunities for impression management based on content alone. Therefore, a more nuanced strategy – namely, the use of rhetorical techniques that emphasize linguistic style – emerges as a more viable approach, underscoring the importance of this perspective. Second, we focus on form-based manifestations of rhetorical IOIM that are largely detached from specific themes and, hence, are broadly applicable (see also Aerts and Yan, 2017). In so doing, we consider the broad scope of content that forms the subject of the intra-organizational communication under investigation. Third, research in applied linguistics indicates that the linguistic style characteristics included in our study are “interactional resources” (Hyland and Tse, 2004) revealing the author’s stance toward the information provided, but also implying an engagement with the recipients, which contributes to building relationships between the author and the readers (Hyland, 1998). The building of such relationships appears paramount in an intra-organizational setting as they may be more intimate than in an extra-organizational setting and may reinforce the generation of internal legitimacy.

We also consider that internal units in post-bureaucratic organizations can rely on alternative sources of informal influence to legitimize their organizational position. Therefore, we also explore the interdependencies between IOIM and these alternative sources of influence.

The first alternative source of influence is bargaining power – the relative capacity of different parties in a negotiation or dispute to influence or secure agreement on their own terms. This power comes from several formal, but also informal sources, including a party’s expertise about local affairs and levels of attachment between parties (Inkpen and Beamish, 1997), its control over resources and others’ dependence on these resources (Crook and Combs, 2007), and its own prominence in alliance networks (Ozmel et al., 2017). Bargaining power is relational – it is generated by differences in parties’ negotiating abilities (Yan and Gray, 1994). This relational aspect makes bargaining power an especially salient resource for units in post-bureaucratic organizations because their relative power depends at least in part on their ability to bargain for a favorable organizational position (Courpasson and Reed, 2004). This holds especially for cooperatives, where individual member organizations can lobby for positional advantages – taking advantage of any informational and resource asymmetries that may exist between these members and the central units in the group (Bontems and Fulton, 2009; Maroua, 2015).

In addition, cooperative organizations are especially open to control through ideological power, which is the deployment of ideology as an instrument of domination (Bourdieu, 1979, p. 78). It is either based on an ability to shape norms of conduct and ritual or aesthetic practices in communities (Jones and Jones, 2004, p. 412), or to use favorable aspects of existing ideologies to exert influence on individuals or groups. The management control literature refers to this approach as cultural control and suggests that the values, beliefs and social norms shared by a group may be used to regulate the behavior of its members (Malmi and Brown, 2008). In this respect, Alvesson and Kärreman (2004, p. 426) refer to socio-ideological controls to denote managerial attempts to manipulate and use values, ideals and understandings about specific parts of social reality to justify certain commitments and actions. Cultural- or socio-ideological controls draw on selected ideological stories and narratives in attempts to manipulate employees’ privately held norms, values and sense of which behaviors are considered appropriate.

Attempts to draw on or manipulate ideology to legitimize an internal unit are particularly effective in post-bureaucratic organizations on which specific ideologies are historically imprinted. Here, managers may attempt to align the purpose they claim for their internal units with particularly favorable aspects of corporate ideology (Maravelias, 2003). In cooperatives specifically, ideology serves both as a mobilizing normative framework to engage members, and as a “cultural-cognitive” framework to render the post-bureaucratic organizational form, and the position of internal units therein, self-evident (Soetens and Huybrechts, 2023). With their allegiance to the uplifting of vulnerable populations, the ideologies in Dutch cooperative banks have always encompassed two mutually consistent components: (1) Raiffeisen’s cooperative belief system and (2) the Christian belief system (Sluyterman et al., 1998), both of which sought to emancipate the farming population. These belief systems were initially espoused by the founder of Dutch cooperative banking who defined these banks as instruments to “prevent usury, to help the farmer in his need, but also to promote thrift, charity, industriousness and temperance” (cited by Poli, 2019, p. 353). Subsequently, their values expanded to encompass neighborly care, frugality and solidarity (Colvin et al., 2020). Although religious movements are among the most commonly cited groups to draw on ideological power, such power can also be wielded by any group which is able to monopolize a claim to specific meanings for a particular group (Mann, 2012). In our case, cooperatives’ social remit is founded on a strong ideology, the rhetorical alignment to which may be a potent source of informal influence. Consequently, we consider ideological power, resting on Raiffeisen’s cooperative belief system and the Christian belief system, an important alternative source of influence.

Overall, we argue that organizational units in Dutch banking cooperatives can draw on two alternative sources of influence to occupy specific informal power positions vis-à-vis other organizational units: bargaining and ideological power. We expect that IOIM and these alternative sources of influence are substitutive: central banks which have greater access to the latter to secure their internal legitimacy have a lower need to engage in IOIM for the same purpose. Therefore, our exploration of IOIM in post-bureaucratic organizations focuses on the interplay between impression management and these informal sources of influence.

We draw on historical data from two Dutch cooperative banks which were founded in 1898 and ceased to exist independently in 1972: Alpha and Beta, both anonymized. Farmers had limited access to capital markets, being charged extremely high interest rates and, thus, faced severe constraints on their further development. Consequently, these farmers and local dignitaries came together to create two local credit cooperatives. Both banks were established to emancipate the agricultural population of the day and drew on both Raiffeisen’s cooperative and the Christian religious belief systems to serve as sources of ideological power.

Alpha was a cooperative of member banks located in the middle and north of the Netherlands. At its peak in 1949, 733 local banks were members of Alpha. Like many organizations of the day, it drew on Christian values of togetherness and thrift, but it also identified strongly with the cooperative values of Friedrich Wilhelm Raiffeisen, a German mayor who created the idea of cooperative self-help, revolving around three concepts: self-help, self-governance and self-responsibility. These concepts translated to several principles of local member bank governance. These principles included local solidarity – more affluent village dignitaries and local members were personally liable for losses; low-cost management – the member bank was initially staffed and managed by volunteers, who frequently held office hours at their own homes; and local intimacy – an intimate personal knowledge of clients drove down control costs and increased the effectiveness of the member bank.

In the southern part of the Netherlands, Beta was a credit cooperative which had 601 member banks at its peak in 1961. Beta’s belief system incorporated similar cooperative values, but, as it was situated in a predominantly Catholic part of the country, its values were also founded on the Catholic religion. Beta espoused many of the same principles as Alpha, which were informed by a combination of cooperative and Catholic values, such as thrift, solidarity and a call to community participation. Beta drew on the socioreligious pillarization of the day by making use of the Catholic infrastructure, keeping office hours in church buildings and community centers. Although the central banks of Alpha and Beta drew on the cooperative and religious belief systems in different proportions as bases of their ideological power, both cooperatives were comparable in many other respects, including financial performance and trends in numbers of member banks.

As mentioned, each cooperative founded a “central bank,” a supra-local organization to support member banks with services which were not tenable at the local level. As the banking industry became subject to more complex and stricter regulation, these central banks were charged with control over their cooperative groups and were given a role as group representatives toward national and international regulatory bodies. Despite these roles of monitoring and control, the central banks were not provided with a position of power in a formal bureaucracy. Both Alpha and Beta can therefore be considered post-bureaucratic organizations, in which the central banks needed to actively ensure internal legitimacy for themselves and their supervision over the cooperative of member banks.

In support of this pursuit, both central banks issued “relational reports” to their member banks. These reports – edited and managed by the central banks – communicated group policies, the vision of the cooperative central bank on wider developments in the field, and changes in laws and regulations for member banks. The reports included articles about the activities of the central banks, opinion pieces about developments in the cooperative movement, stories about the ways member banks had made a local impact, pieces about the different ways central banks offered support to local member banks and articles about national rules and regulations and the ways the central banks promoted compliance by their member banks. Frequently, the reports connected cooperative ideology, Christian religion, the interests of local member banks and the central banks. By prominently positioning themselves in these stories and narratives, the central banks highlighted their significance in the group. In this way, they sought to legitimize their existence and authority over the collective of autonomous member banks.

The first author obtained the relational reports from the years 1911–1971; a timeframe which was selected, because starting in 1911, both Alpha and Beta regularly produced relational reports. 1971 was the final year in which both cooperatives produced these reports separately. All reports were digitized and converted into computer-searchable text using text recognition software. In total, the 61 years of monthly and weekly issues constituted 30,311 pages. We removed all graphical elements, including photographs, colors and other layout elements, to enable a comparison uniquely based on textual characteristics. In 1942 and 1943, Beta did not publish reports due to paper shortages caused by World War II. For completeness, we extrapolated these years linearly, but this extrapolation had no effect on our results.

For the reasons outlined before, we conduct a form-oriented analysis, in which we explore rhetorical IOIM based on linguistic style characteristics and take into account that such characteristics are likely to be used interdependently. Our empirical approach consists of three steps. First, we identify single linguistic style characteristics by utilizing automated semantic analysis software: Linguistic Inquiry and Word Count (LIWC) by Pennebaker et al. (2015). Second, these linguistic style characteristics are entered into an exploratory factor analysis which we conduct to detect more comprehensive profiles consisting of multiple dimensions and representing co-occurrence patterns among the linguistic style characteristics. We thus mobilize the idea that it is the co-occurrence of linguistic style characteristics that contributes to rhetorical appeals rather than single linguistic style characteristics (Yan et al., 2019). Finally, we perform exploratory hierarchical regression analyses to study rhetorical IOIM and its interplay with other sources of informal influence. The rhetorical profiles identified in the second step serve as the dependent variables of these regression analyses.

LIWC is frequently used in the accounting literature, particularly to capture impression management in communication with external stakeholders (Aerts and Yan, 2017; Merkl-Davies et al., 2011; Yan et al., 2019). It draws on a dictionary file that consists of different word lists, each list reflecting a specific word category. The processing feature scans the text available for analysis word by word. During this process, each word occurring in the text is compared with the words included in the dictionary file. At the end of the analysis, the proportion of words belonging to each word category out of the total number of words in the text is provided (Pennebaker et al., 2015).

While prior studies used the English-language dictionary file of LIWC, we refer to the 2007 Dutch version, given that the relational reports were published in the Dutch language. The composition of the dictionary file is equivalent in both language versions (van Wissen and Boot, 2017). Therefore, the word categories utilized are the Dutch equivalents of the word categories included in the English-language version used in prior impression management research.

Our selection of LIWC word categories is based on Aerts and Yan (2017), which is anchored in Aristotelian rhetoric and its three types of strategies – logos, pathos and ethos. These authors distinguish causal language, attitude markers, self-references and cognitive mechanisms as devices of rhetorical impression management (see also Yan et al., 2019 for a similar approach). In what follows, we relate these linguistic style characteristics to the three overarching rhetorical strategies distinguished by Aristotelian rhetoric. Notably, a single style characteristic can serve more than one rhetorical strategy (Aerts and Yan, 2017) and distinguishing between the strategies is inherently difficult, as they are “distinct, but not separable” (Higgins and Walker, 2012, p. 197). This fuzziness underscores the value of considering broader rhetorical profiles, as we do here.

Causal language appears essential in establishing logical appeals (logos), which target an audience’s “rational commitment” (Nørreklit, 2003, p. 595) by providing logical arguments and focusing on means-ends relationships (Green, 2004; Hossain et al., 2019). It thus involves an “appeal to reason” (Higgins and Walker, 2012, p. 198). Regarding internal legitimation, relying on logos implies that an organizational unit seeks normative validation by logically presenting its circumstances and justifying its courses of action rationally. As logos entails “persuasion through reasoning” (Edgar et al., 2022, p. 34), our analysis includes causal language (CAUSAL), the measurement of which is based on the word list “Causation” in LIWC.

Attitude markers, measured as the difference between “positive emotion” and “negative emotion” words in LIWC (EMOTIONAL), capture emotional language and comprise affective appeals (pathos) (Aerts and Yan, 2017). This difference captures the balance between positive and negative emotions expressed in communication and thus the direction of emotion (García Osma and Guillamón-Saorín, 2011). Pathos targets audience attitudes and emotions to evoke emotional responses (Hoefer and Green, 2016; Paugam et al., 2021). Pathos thus mobilizes “emotional justifications” (Hossain et al., 2019, p. 565). Hyland (1998) argues that the signaling of the communicator’s attitudes may induce readers to adopt similar attitudes, which contributes to the building of relationships. In terms of internal legitimation, the mobilization of pathos can thus help to create an emotional bond between an organizational unit and its audience, thereby strengthening the former’s legitimacy.

Personalization and references to cognitive processes presenting the communicator as a “credible source of ideas and views” (Edgar et al., 2022, p. 34) are associated with credible appeals (ethos) (Aerts and Yan, 2017). While logos and pathos aim to trigger rational or emotional reactions to the message provided, the core of ethos relies on trust in the communicator (Edgar et al., 2022; Nørreklit, 2003) by emphasizing her credibility and authority (Hossain et al., 2019; Paugam et al., 2021). Regarding internal legitimation, this involves presenting an organizational unit as trustworthy and assertive. Self-referencing through first-person pronouns contributes to this effect by connecting the communicator to the message. Personal pronouns may also foster relational bonds with the reader and thus contribute to more affective messages (Aerts and Yan, 2017). Accordingly, we draw on the LIWC word list “Total first person” (PRONOUNS) which comprises first-person pronouns, which “reflect the degree of author presence” in the communication (Hyland and Tse, 2004, p. 170).

We also consider cognitive mechanisms that enhance the communicator’s credibility and perceived authority. We include the word list “Insight” (INSIGHT), encompassing words related to processes of learning and understanding, which can foster an audience’s trust in the communicator and thus contribute to ethos. These words may also complement causal language in establishing logical appeals and address the affective stance of readers, in line with pathos (Aerts and Yan, 2017). Moreover, we account for cautious language through the LIWC word list “Tentative/cautious/hesitant” (TENTATIVE), which reflects openness and honesty of the communicator (Aerts and Yan, 2017). Hyland (1998) argues that such hedges may also reflect tact and deference to the reader and may contribute to the building of a relationship with readers which allows for affective appeals. In addition, the word list “Certainty” (CERTAINTY) captures expressions of conviction that may further strengthen the communicator’s credibility as they underline her conviction in the contents communicated.

While the linguistic style characteristics above relate to the communicators’ credibility, we also consider language that signals authority. This includes modal verbs (e.g. should, could), which guide the audience’s cognition and actions, serving as proxies for directive language. Hyland (1998) argues that modal verbs may also substantiate affective appeals by aligning the goals and understandings of the communicator with those of the audience when the communicator expresses beliefs and views the audience presumably agrees with. Hence, our variable DISCREPANCY is based on the LIWC word list “Discrepancy.” The variable INHIBITION relies on the corresponding LIWC word list including words which indicate restrictions and prevent actions, while the variable EXCLUSION is based on the LIWC world list “exclusive” capturing words that signify distinctions. These categories are indicative of critical thinking and, thus, may be involved in appeals anchored in ethos as they emphasize the communicator’s authority and credibility (Aerts and Yan, 2017).

We perform exploratory hierarchical regression analyses to estimate the effects of bargaining power and ideological power on central banks’ propensity to engage in IOIM. Strictly speaking, a hierarchical regression analysis is not so much a statistical technique but rather a framework for model comparison. Multiple regression models are built by the stepwise addition of variables to assess improvements in model fit. This approach is particularly useful for exploring the effects of adding variables and possible moderating effects in the data. Consequently, hierarchical regression analyses help to clarify complex, poorly understood relationships among variables. In our analysis, we include variables reflecting bargaining power and ideological power and we add various moderator models to explore the interaction between these sources of informal influence and their influence on IOIM.

The first independent variable, BPOWER, represents the number of member banks in each cooperative and reflects the bargaining power of each central bank relative to the member banks. All else equal, a larger number of member banks is associated with greater bargaining power of the central bank. Spreading resources and decision rights among more autonomous member banks dilutes the level of control by each member bank over the group. As a larger number of member banks makes their coordination and organization more difficult to achieve (Sluyterman et al., 1998), the central bank’s ability to provide such coordination and organization puts it in a better bargaining position vis-à-vis the member banks. This comes at the expense of the wielding of bargaining power by individual member banks.

Second, we measure the two components of ideological power introduced in section 2.4 for Alpha and Beta. RELIG includes central banks’ rhetorical appeals to religious concerns as measured by LIWC, counting Dutch equivalents of words such as “altar” and “God.” The variable COOP is the percentage of words that reflect cooperative ideological rhetoric. This dictionary has been constructed specifically to capture this rhetoric and includes Dutch equivalents of words including conviction, principles, morale, ideology and ideal.

Third, we measure the regional religiousness (the proportion of the population that identifies as religious) in each cooperative’s working area. We do so because the relationship between the use of cooperative and religious rhetoric (COOP and RELIG), and the deployment of IOIM may be different for member banks who identify as religious to different degrees. Member banks located in highly religious regions are likely more responsive to central banks’ rhetorical alignment with religious values than member banks in regions that do not identify with religion to the same extent. Therefore, the former may experience a lower need to engage in IOIM. We measure regional religiousness in two stages: first, Statistics Netherlands (CBS) provided time-series data on the religiousness of Dutch provinces over the period of study. These data are the proportion of the population who identify as Christian or Catholic. Second, these data were matched with the primary areas of operation of Alpha and Beta. Based on this data, we constructed the variable AREA_RELIG, which is the percentage of people who identify as religious in each of the central banks’ working areas.

We subsequently mean-centered the BPOWER, RELIG, COOP and AREA_RELIG variables to avoid issues with multicollinearity, especially in the moderator analyses that we perform. Given that impression management practices may be related to performance (Clatworthy and Jones, 2006), we included the control variable Return On Assets (ROA). ROA is one of the most meaningful indicators of financial performance in the banking industry (Reger et al., 1992). Following Deephouse and Carter (2005), ROA is computed as the net income of the collective of member banks in period t divided by their average assets in the periods t and t-1. Using an average yearly measure of assets as a denominator is consistent with regulatory practice and prevents distortions due to possible year-end window dressing. All performance data was collected manually from the annual reports of Alpha and Beta.

Table 1 reports the descriptive statistics of the linguistic style dimensions and the independent variables, and Table 2 shows the bivariate correlations for the independent variables.

We begin by presenting the findings of the exploratory factor analysis to answer RQ1. Next, we interpret and report the findings of the exploratory hierarchical moderator analyses related to RQ2. These analyses are conducted separately for each rhetorical profile identified by the factor analysis.

We apply an exploratory factor analysis to detect co-occurrence patterns of the previously introduced linguistic style characteristics (Aerts and Yan, 2017; Yan et al., 2019). A principal component analysis with varimax rotation suggests that the linguistic style characteristics represent three underlying factors shown in Table 3 with eigenvalues of 3.56 (Factor 1), 2.55 (Factor 2) and 1.33 (Factor 3), cumulatively explaining 82.7% of total variance ^[1]. These factors group the linguistic style characteristics that tend to occur together (Yan et al., 2019). We use a cut-off value of 0.55 for interpreting the values, following guidelines from the statistical literature that factor loadings of this level or higher are considered good (Hair et al., 2019, p. 151; Tabachnick and Fidell, 2013, p. 654).

For each factor, at least one rhetorical strategy can be identified as the primary point of reference, but the factors also include elements of the others. The first factor represents a configuration of linguistic style characteristics anchored in logos and ethos. The heavy use of causal language signals the communicator’s efforts to justify actions through logical narratives. The reliance on insight language complements these efforts and reveals the communicator’s willingness to show her considerations. Additionally, tentative and discrepant words introduce a sense of honesty and openness, softening the tone of certainty language and potentially building a relationship of trust with the audience. This factor thus emphasizes the communicator’s credibility through logical reasoning and openness. We label the factor “Participative reasoning profile” (PRP) to emphasize the communicator’s willingness to share her cognitive processes (“participative”), and her strong reliance on causal language (“reasoning”).

The second factor reflects ethos and refers to a communication pattern characterized by a strong emphasis on self-references. Along with certainty expressions, this positions the communicator as a confident authority. However, the mobilization of inhibition words signals her awareness of challenges or boundaries. Exclusion words help define the scope of the argument, indicating what she considers relevant and what not. These word categories are indicative of a structured and nuanced approach culminating in critical thinking that reflects the communicator’s credibility, but they also demonstrate the communicator’s sense of direction, in line with the emphasis on her authority. We thus label the factor “Reflective authority profile” (RAP) because this label mirrors tendencies of critical thinking (“reflective”) as well as the strong author presence (“authority”). In sum, while PRP blends logos and ethos to build credibility through logical reasoning and reflective openness, RAP centers more on the communicator’s authority.

The third factor is characterized by a strong reliance on emotional language with a positive “net affective tone” (Aerts and Yan, 2017). We detect a low utilization of causal language and cognitive mechanisms. Therefore, this factor appears mostly in line with pathos, appealing to the readers’ emotions. Accounting for this focus, we label the factor “Affective appeal profile” (AAP).

In summary, the cooperative central banks mobilized three distinct rhetorical profiles. RAP and AAP primarily reflect ethos and pathos, respectively, while PRP blends logos and ethos. Overall, the profiles explored can be interpreted through combinations of Aristotelian rhetoric. While these findings are insightful regarding RQ1, we next turn to the interplay between the rhetorical profiles and other sources of informal influence to answer RQ2.

The exploratory hierarchical moderator analyses for each of the previously identified rhetorical profiles are reported in Tables 4–6. Each table includes six models: Model 1 is a regression of each of the rhetorical profiles against BPOWER, RELIG, COOP, AREA_RELIG, controlling for ROA. Models 2–6 each include single interaction terms of these independent variables. We focus on the interaction model with the highest explanatory power, as measured by R², and for each rhetorical profile, its interaction term is illustrated graphically in Figure 1.

Table 4 shows the findings of the hierarchical moderator analysis of PRP. Model 1 highlights the main predictors of the use of PRP in the cooperatives. As expected, PRP is negatively associated with BPOWER (b = −0.004, p < 0.01), reflecting how the central banks mobilized PRP when they lacked sufficient bargaining power. This finding suggests that organizational units put more effort into convincing their audiences through logical reasoning and openness when they had fewer alternative means to exert intra-organizational influence. Similarly, PRP is negatively related to RELIG (b = −2.161, p < 0.1), indicating a lower use of PRP when religious rhetoric was mobilized. By contrast, COOP shows a positive association with PRP (b = 19.65, p < 0.1), as the reliance on cooperative rhetoric was complementary to PRP.

This finding points to the importance of cooperative ideology for member banks in both cooperatives. The effectiveness of the logos- and ethos-based rhetoric of PRP could be enhanced by linking it to cooperative values, such as local solidarity and social justice. Such a combination appears more effective for securing acceptance of central banks’ authority than relying on logical and reflective reasoning alone. Additionally, Model 1 suggests that the use of PRP was more pronounced in more secular areas, where appeals to religious identity were likely less effective. Hence, PRP was likely attuned to specific audience characteristics, in this case, their religious self-identification.

Table 4, Model 3 shows the single moderator model explaining the highest proportion of PRP variation (R² = 0.275, F(6, 115) = 7.2696, p = 0.000). Figure 1 (Panel I) illustrates the significant interaction term RELIG × BPOWER (b = −0.049, p < 0.01). Religious rhetoric strongly moderates the relationship between bargaining power and the use of PRP. For central banks relying heavily on religious rhetoric, greater bargaining power was associated with lower PRP use. Thus, bargaining power substituted PRP, suggesting that either was sufficient for strengthening central banks’ informal position. By contrast, increasing bargaining power for central banks with comparatively limited religious rhetoric resulted in no significant changes in PRP use. Hence, bargaining power substituted for PRP mainly when religious rhetoric was prominent. Overall, these findings highlight that appeals to reason and credibility were complemented by cooperative ideology but substituted by religious rhetoric. However, this was not the case for the RAP.

Model 1 in Table 5 shows no significant relation between RAP and BPOWER. However, the highly significant and positive relationship between AREA_RELIG (b = 0.053, p < 0.01) and RAP suggests that especially central banks operating in highly religious areas tended to mobilize RAP. Moreover, both RELIG (b = 2.568, p < 0.01) and COOP (b = 25.054, p < 0.01) are associated with a higher use of RAP. Hence, the use of religious and cooperative rhetoric was complementary to the use of RAP, suggesting that the central banks accompanied self-referential justifications for their authority claims with cooperative and religious rhetoric.

In Table 5, Model 5 is the single moderator model that explains the highest proportion of RAP variation (R² = 0.435, F(6, 115) = 14.7545, p = 0.000). In contrast to the single moderator model of PRP, where religious rhetoric moderated the relationship with bargaining power, this model highlights how cooperative rhetoric moderates the relation between RAP and bargaining power. Figure 1 (Panel II) illustrates this moderation graphically. RAP appealed to internal audiences by highlighting the credibility and especially the authority of the central banks. The central banks motivated these qualities by aligning themselves with the ideological roots of the cooperative – the use of cooperative rhetoric was complementary to the use of RAP.

For cooperative central banks, the relationship between bargaining power and RAP varied with their reliance on cooperative rhetoric. For central banks with low reliance on such rhetoric, bargaining power and RAP were substitutive, with higher BPOWER linked to lower RAP use. Central banks with high bargaining power thus may feel less pressure to emphasize their authority through impression management. By contrast, for central banks with a high reliance on cooperative rhetoric, BPOWER was positively associated with RAP. We can explain this unexpected result by pointing to the suspect position of bargaining power in the cooperatives. The central banks may have possessed high levels of bargaining power, but this power could not be exercised directly in the cooperative democracy. Member banks are known to resist central banks’ direct claims to power, as those claims challenge local autonomy (Sluyterman et al., 1998). Hence, in central banks’ use of RAP, cooperative rhetoric served as “conduit” through which bargaining power was manifested and justified. Extensive reliance on bargaining power alone as a source of central banks’ authority carried the risk of being perceived as overly dominant. The deliberate incorporation of cooperative rhetoric may thus have served as a mitigating mechanism to address this concern.

Model 1 in Table 6 shows a positive relation between AAP and BPOWER (b = 0.002, p < 0.1). The use of AAP accompanied BPOWER, presumably because bargaining power itself could not be exercised directly in the cooperative democracy (see section 4.2.2) but instead needed to be accompanied with rhetoric aimed at generating an emotional attachment to the central banks’ use of such power. AAP was couched in such appeals for emotional attachment. Additionally, RELIG is positively associated with AAP (b = 6.829, p < 0.01), indicating that an increased use of religious rhetoric implied an increased use of AAP.

Model 2 is the single moderator model that has the highest explanatory power of AAP variation (R² = 0.437, F(6, 115) = 14.8473, p = 0.000). The model shows how AREA_RELIG moderates the relation between the use of religious rhetoric and AAP (b = 1.013, p < 0.01). Figure 1 (Panel III) depicts this moderation. In highly religious areas, religious rhetoric supported the emotional connections sought by the central banks via AAP. In areas with low religiousness, this supporting role of religious rhetoric is not detected. Instead, religious rhetoric was used as a substitute for appeals for emotional attachment by the central banks. This moderating effect of the religiousness in the operating areas points again to the sensitivity of the message sender to local concerns. AAP was based on appeals to emotional responses by the audience and our analysis suggests that this emotional response was amplified by adapting the message to local circumstances – in our case, the receptiveness of the audience in the central banks’ operating areas to religious rhetoric.

Overall, Figure 1 shows one of the most striking findings: for each individual rhetorical profile, the model with the highest explanatory power interacts with a different moderator. This suggests that central banks’ use of each different rhetorical profile was attuned to its ability to mobilize different alternative sources of informal influence. Taken together, the different panels in Figure 1 also reflect the unpredictability of these interactions for different rhetorical profiles. It is likely that other types of post-bureaucratic organizations will display different patterns because their internal dynamics are different, but here we demonstrate that such dynamics are likely to exist – internal units draw upon different rhetorical profiles in relation to other forms of informal influence. These and other findings are discussed in the next section.

Grounded in the view that impression management can be employed to secure internal legitimacy, we adopted a rhetorical perspective, as rhetoric constitutes persuasive discourse oriented toward legitimacy construction (Green, 2004; Suddaby and Greenwood, 2005). Our focus on linguistic style characteristics takes into account that such rhetoric may guide the recipients’ interpretation and evaluation of information and build relationships between the sender and its recipients (Crismore et al., 1993; Hyland and Tse, 2004). Our analysis reveals three different IOIM profiles indicative of rhetorical legitimation work (Aerts and Yan, 2017; Erkama and Vaara, 2010). Our findings further indicate that IOIM interacts with other sources of informal influence. While we expected that IOIM may be substituted by alternative sources of influence, we find evidence that IOIM may be either substituted or complemented by these other sources. These insights contribute to prior research on impression management as well as to the management control literature. In what follows, we flesh out these contributions.

The accounting literature has accumulated a rich body of knowledge concerning impression management in extra-organizational communication (e.g. Frankel et al., 2016; Guillamón-Saorín et al., 2017; Keusch et al., 2012) but has yet to assess whether and how organizations engage in internal impression management based on intra-organizational communication. This study provides initial findings that rhetorical IOIM can complement or substitute other forms of informal influence. These findings are particularly relevant, given that many modern governance structures rely on internal units’ informal influence over each other, in lieu of a formal bureaucracy (Stafsudd, 2009). In addition, in more traditional governance forms, units ranked higher in the organizational hierarchy may control the dissemination of information. However, in the decentralized members-based democracy of cooperatives, formal power and communication may flow in opposite directions. Our study shows how central units can make use of their control over these information flows to ensure their internal legitimacy. To date, there are some meaning-oriented analyses showing that organizations mobilize content-based IOIM to legitimize controversial corporate decisions (Balogun et al., 2019; Erkama and Vaara, 2010). Our paper adds to these insights that organizations can also mobilize IOIM anchored in linguistic style characteristics. Here, we suggest that the reliance on such linguistic style characteristics may be especially significant in intra-organizational settings, as information asymmetry is likely lower between internal units than between the organization and external audiences. In this context, impression management strategies based on content alone may ostensibly be less impactful, given that lower information asymmetries may enable internal audiences to verify contents biased by impression management endeavors, with adverse effects on internal legitimacy.

Our paper also contributes to the impression management literature, which has started to investigate co-occurrence patterns of different linguistic style dimensions (e.g. Aerts and Yan, 2017; Yan et al., 2019), by illuminating rhetorical profiles and their interdependencies with other types of informal influence. We argue that impression management is also anchored in appeals to potentially different characteristics of its intended audiences. In our study, we incorporated the bargaining power of central banks in relation to member banks (their relative power positions which characterize both central and member banks) and religious convictions (measured as the regional religiousness in member banks’ working areas) as characteristics of the audiences of central banks ^[2]. We reveal that the sender attuned the tone of its communications to these characteristics in different ways. This adjustment of rhetorical appeals to audience characteristics has not been extensively documented because external audiences are mostly considered homogeneous (Schoon, 2022). Accordingly, prior research has mostly focused on the effects of impression management (Guillamón-Saorín et al., 2017; Hooghiemstra et al., 2017) or their determinants, including environmental conditions (e.g. Keusch et al., 2012) and characteristics of the communicator (e.g. García Osma and Guillamón-Saorín, 2011). By contrast, less attention has been paid to the characteristics of audiences and their impact on impression management. Our findings suggest that post-bureaucratic organizations may use combinations of IOIM profiles and additional sources of informal influence, which are specifically attuned to the characteristics of the targets of the rhetorical appeals. For instance, the exercise of bargaining power by central banks was made legitimate to the autonomous member banks – lacking such power but known to resist central banks’ claimed authority on this basis (Sluyterman et al., 1998) – through cooperative or religious rhetoric. Also, combinations of different sources of informal influence addressing cooperative sensibilities of local autonomy and religiousness were substitutes to the rhetorical profiles we distinguish. In contrast to extra-organizational impression management in which the characteristics of the audience are largely unknown, intra-organizational audiences are more delineated and can be appreciated better. Consequently, we suggest that the sender can address the concerns of its audience more effectively than in the case of extra-organizational impression management.

Our study also extends prior research on management controls by recognizing that the different rhetorical profiles are a form of soft control that complements but also substitutes elements in the organization’s control repertoire. Current literature points to the effectiveness of organizational controls when they are combined in “management control packages” (Malmi and Brown, 2008). However, Grabner and Moers (2013) and Bedford et al. (2016) problematize this conceptualization of packages, noting that an exploration of packages explicitly needs to consider the nature of the interdependence between their constituent parts. Our moderator analysis highlights the situated nature of this interdependence, by illuminating how impression management can function both as a substitute and a complement to other sources of informal influence. Thus far, the literature lists organizational strategy (Bedford et al., 2016), organizational culture and intensity of experienced competition (Einhorn et al., 2021) as contingent factors in management control packages, and we add the local religious context which we show to affect the degree and nature of interdependence between IOIM, bargaining power and the use of religious and ideological discourse.

Relatedly, we provide a novel explanation of the workings of soft controls as they are incorporated into control packages. The different sources of informal influence in our study never appeared in isolation but were mobilized in different but coherent combinations. In post-bureaucratic organizations, “structures of legitimacy” are in place to legitimize “structures of domination” (Courpasson, 2000), and our findings empirically show how different sources of informal influence were used to legitimize rhetorical authority claims. For example, RAP drew on cooperative ideology as a “conduit” that moderated its relationship with bargaining power. RAP’s rhetorical construction of central banks as authoritative and critical and their use of bargaining power were highly suspect in the member-based democracy of the cooperatives. Therefore, this relation was accompanied by cooperative discourse to legitimize these sources of dominance by couching them in the ideological roots of the cooperatives. A similar observation could be made for the role of religious rhetoric in relation to the rational rhetoric of PRP and bargaining power. We add to the literature that packages of soft controls may constitute multiple, mutually enforcing components: rhetorical appeals to combinations of logos, ethos and pathos, as well as supportive rhetoric that legitimizes these appeals by addressing the local concerns of their audiences. These findings are also significant for the management control literature because they situate control beyond the boundaries of the firm. Local units bestow legitimacy upon other internal units, but these legitimacy evaluations are, in part, based on the ideological beliefs in their communities. Our findings show how the central banks attuned intra-organizational impression management to these local beliefs. Arguably, this effect is most pronounced in the reverse hierarchy of cooperatives, but our findings also plead for a broader understanding of the ways combinations of management controls provide legitimacy to internal units and their actions.

This paper’s key takeaway message is that although they are “hidden” in communication flows, multiple rhetorical IOIM profiles can play a crucial role in efforts to legitimize internal units. The paper suggests that these rhetorical profiles do not manifest in isolation but interact with other sources of informal influence. We find that, in combination, these constitute tailored control packages which may be attuned to contextual and audience characteristics in quite specific ways to impact the internal perceptions of these internal units.

Our findings have several implications. First, organizations with an elaborate ideological signature, such as religious-, relief- and sustainability organizations may have external stakeholders promoting the incorporation of internal units that are not obviously associated with their ideological remit. For example, regulators may ask cooperative banks to centralize their compliance- and risk control. Our findings suggest that the units involved may need to communicate extensively to justify their organizational position. Simultaneously, however, we have shown that they may draw on the discursive resources that are associated with this ideological signature. Hence, compared to their rational-legal bureaucratic counterparts, internal units in post-bureaucracies have a greater need for impression management, but they also may have access to more elaborate discursive resources to shape the resulting impressions.

Second, we propose that it is likely that the efficacy of internal units in post-bureaucracies depends on their communicative abilities, as their existence and the demands they place on other units need to be accompanied by legitimizing discourses. In our case, the central banks were positioned to distribute relational reports, but such a position may not be assured for all internal units. Therefore, we suggest that post-bureaucratic organizations must facilitate intra-organizational communication flows especially well because these appear critical to their governance and internal control.

Third, our findings tentatively suggest that post-bureaucracies may be relatively agile because their internal units are particularly responsive to the needs of other internal units. Although we only studied discursive attempts to build internal legitimacy, these attempts may be supplemented by substantive acts to serve the interests of other internal units. In this way, in their efforts to secure their own legitimacy, internal units’ inherent orientation towards the interests of other units may be conducive to higher organizational agility. Especially in unstable and dynamic contexts, this may potentially be of great importance.

Our study is subject to limitations which may provide fruitful avenues for future research. First, this paper refers to the rather unique setting of a cooperative organization, in which senior managers need to actively work on their legitimacy with their members. Organizations with more traditional forms of governance may rely less on impression management techniques, simply because they are in a better position to formally enforce the presence of internal units. Therefore, we suggest that more studies are needed to understand how IOIM is used in firms with different governance structures. Second, while we have explored the interdependencies between IOIM and other informal sources of influence, there may be other ways for units to influence their internal legitimacy. As the efficacy of informal controls may depend on the legitimacy of the managers operating them (cf. Courpasson, 2000; Courpasson and Reed, 2004), we call for studies that examine alternative controls in relation to the legitimacy of the managers deploying them. Third, our data do not allow us to qualify the response of the internal audience to the rhetorical appeals. Whether central banks accomplish the establishment of internal legitimacy is beyond the scope of our study but presents a potentially impactful opportunity for future work. Fourth, while we follow the approach by Aerts and Yan (2017), which has been shown to generate insightful findings on extra-organizational impression management (see also Yan et al. (2019)), we acknowledge that rhetorical appeals may also manifest in LIWC word categories other than those included in our study. Finally, we acknowledge that the relatively low adjusted R² for some of our models indicates that IOIM may be influenced by additional factors not captured in our analyses. Given the inherent complexity of IOIM, these results underscore the need for further investigation to explore additional determinants.

Despite these limitations, we believe that the findings in this paper contribute to ongoing theorizing on the subject, which relies on theoretical generalization rather than statistical generalization alone (cf. Eisenhardt, 1989). In this vein, this paper has made progress in theorizing the role of IOIM in conditions where governance by a formal bureaucracy alone cannot be assumed.

We are grateful for the constructive comments made during the EIASM conference on Performance Measurement and Management Control, the MCA conference in Birmingham, the EIASM New Directions in Management Accounting conference and seminars at Aarhus University, Denmark and the University of Groningen, the Netherlands. We also gratefully acknowledge the support of the organizations studied in this paper. Finally, we wish to thank the anonymous reviewers and the editor for their helpful comments and suggestions.