The Diffusion Process of Strategic Motivating Language: An Examination of the Internal Organizational Environment and Emergent Properties

Motivating Language Theory: A Review of Selected Literature

MLT was first conceptualized by Sullivan (1988) as an oral communication alternative to the prevailing management focus on uncertainty reduction in organizational leadership. Thus Sullivan asserted that at organizational, team, and dyadic levels of analysis, leaders could improve employee motivation by strategically using all three major forms of linguistic speech acts—in their oral messages (Austin, 1975; Searle, 1969). He wrote the following comparison about extant managerial motivation theory and the added value of MLT:

All of these theories require supervisors to use language in a restrictive manner. Motivational language theory calls for the opposite; all functions of language must be combined in a coherent discourse in order to have the greatest impact on employee motivation. (Sullivan, 1988, p. 113)

The three major speech act categories (Austin, 1975) were described by Sullivan as “locutionary acts, focused on the meaning of words; illocutionary acts, focused on what the speaker is doing while talking; and perlocutionary acts, what the speaker hopes to accomplish” (Sullivan, 1988, p. 108). These speech acts are firmly rooted in management theory, and can be translated into leader-subordinate oral communication as follows:

These three facets of MLT are based on a few central assumptions. First, the leader must behave consistently with what he or she says. More succinctly, the leader has to walk-the-talk. Second, ML must include the use of all three types of speech acts in appropriate ways. Third, while MLT emphasizes leader-to-follower oral communication, the follower must correctly interpret the leader’s intended messages (J. Mayfield & Mayfield, 2012; Sullivan, 1988).

MLT has so far been investigated through the development and implementation of a highly reliable and valid scale (Luca & Gray, 2004; J. Mayfield, Mayfield, & Kopf, 1995). The same instrument or modified versions have been operationalized in both quantitative and qualitative studies to support positive, significant relationships between ML, and desirable employee behaviors and attitudes including more effective decision making, higher job satisfaction, higher communication satisfaction toward a leader, higher perceived leader competence, more innovation, higher job performance, higher team creativity quality, higher self-efficacy, enhanced organizational commitment in Mexico, lower absenteeism, and lower intent to turnover (Holmes, 2012; Madlock & Sexton, 2015; J. Mayfield et al., 1998; J. Mayfield & Mayfield, 2012; M. Mayfield & Mayfield, 2015; McMeans, 2001; Sharbrough et al., 2006; Sun et al., 2008; Wang et al., 2009; Zorn & Ruccio, 1998). Furthermore, Wang et al. (2009) adopted an experimental design methodology that supported MLT’s positive and significant influence on creative idea genesis by virtual teams, hence extending ML to written communication and suggesting causality. However, since MLT investigations into multiple media channels are quite limited, this research study is confined to the domain of spoken words.

While these insights are promising, many important aspects of MLT remain untested. The extant studies are cross-sectional, and with a few exceptions are limited to the dyadic level of analysis. In addition, well-respected scholars (Fairhurst, 2001; Fairhurst & Connaughton, 2013; Macy & Willer, 2002; Zorn & Ruccio, 1998) have recommended alternative research methodologies to the frequent linear models that have been used to explore ML in the past. Also important, only one investigator (Holmes, 2012) has evaluated ML both quantitatively and qualitatively at a cross level of analysis, that of the organization (principal speech) and its interactions with lower level outcomes (teacher and student performance ratings). In brief, his study found that school principals’ (at the organizational level) use of ML was significantly related to both school (student achievement) and teacher performance ratings. And finally, there is a lack of knowledge about how ML diffuses within an organization, and how central variables interact with this diffusion process.

Increased understanding of the aforementioned processes and use of new research methodology, specifically a simulation model, are needed to both advance the theory and to successfully reap its benefits in practice. Simulation models allow researchers to develop more complex theories and especially theories that involve speech feedback loops in the relationships. When such feedback loops exist—most particularly in dynamic relationships that unfold over time—it is difficult to fully grasp their influence as evidenced by how a leader talks. Simply put, tools such as software simulations, which examine predicted variable relationships, help us better comprehend how leader speech is adopted over time. Once a simulation has been created based on individual interactions, it is subsequently possible to develop hypotheses that can be tested empirically.

For this simulation, the rate of ML diffusion within the internal organization is the dependent variable. Relatedly, previous research has identified salient factors that should interact with this outcome. These key independent influences can be grouped into three categories: the communication culture (CEO ML strength, and the extent to which an organization’s culture nurtures oral communication competence as evidenced in selection criteria, rewards, and germane processes—including formal training), institutional factors—including formal hierarchical layers and the turnover rate, and influence variables (peers, immediate supervisor, and the top-leader).

A Diffusion Model of Motivating Language Theory at the Organizational Level of Analysis

This section will draw upon management, communication, and social network theories to create a model and predict patterns of ML diffusion. A social network can be described as “a network as a social phenomenon composed of entities connected by specific ties reflecting interaction and interdependence” (Carpenter, Li, & Jiang, 2012, p. 1329). As previously noted, this step is requisite for MLT extension and responds to calls from research and practice experts for multiplex social network models (Moliterno & Mahony, 2011; Monge & Contractor, 2001; Shumate & Contractor, 2013). Furthermore, since ML has been previously viewed as a subset of leader oral-communication competence (Madlock, 2008; Sharbrough et al., 2006), this model also responds to the recommendation (Jablin & Sias, 2001) for more research on communication competence content at the organizational level of analysis. To achieve this goal, we adopt Sharbrough et al.’s (2006) generic definition of leader communication competence:

An individual’s communication competence can include general items such as having a clarity of expression, using language appropriate to a situation, providing a timely response, and being attentive. . . . An immediate supervisor’s management or leadership efforts clearly provide a broader range of communication competence than a nonsupervisory position. (Sharbrough et al., 2006, p. 326)

To begin our model specification, a working definition of diffusion is also needed. Strang and Soule (1998) described diffusion as

the spread of something within a social system. The key term here is “spread” and it should be taken viscerally (as far as one’s constructionism permits) to denote the flow or movement from one source to an adopter, paradigmatically via communication and influence. (p. 266)

The same article renders diffusion as contagion, a theoretical construct that envisions an organization’s communication network as a conduit which exposes participants to certain oral language behaviors. In turn, this exposure increases the probability that the particular speech patterns will be adopted by other network members (Monge & Contractor, 2001; Strang & Soule, 1998).

While many organizational structures exist—both formal and informal—not every form can be effectively modeled in a given simulation. There is a need for parsimony with any given simulation in order to adeptly focus the investigation, and to grasp the model’s results. Increasing complexity does not necessarily refine our understanding of oral communication practices diffusion, but does increase the chances of research opacity (Railsback & Grimm, 2011). When selecting model parameters, it is important to prioritize focal system and model characteristics as a way to generate pathways toward more complex systems. With these guidelines in mind, we based our simulation of the ML diffusion process on a nonnetwork structured, internal organizational framework. Likewise, we did not model oral communications between organizations and to external stakeholders such as customers and suppliers. The trade-off is that while these choices impose limitations on our final model results, they also offer a sharper view of spoken leader communication adoption.

This taxonomy and related axioms will be discussed in the following paragraphs. Also, the diffusion model is shown in Figure 1. Axiom definitions for Figure 1 are succinctly noted by bold type throughout this discussion.

OPEN IN VIEWER

Figure 1.

Model of factors influencing motivating language (ML) diffusion factors within the internal environment of an organization.

Before further presenting model specifics, it will be useful to discuss the meaning and role of axioms in this article. Axioms are used in developing logical propositions, and are statements that are either so self-evident as to be uncontroversial (e.g., cats like to sleep), or have been shown to be true through repeated observations (e.g., bodies experiencing the same gravitational pull will fall through a vacuum at the same rate regardless of weight). These axioms provide the building blocks for developing more complex models of a given phenomenon. For our analysis, we use axioms rooted in management, communication, and social network literature to develop our computer simulation, with the various axioms acting as behavioral rules for—and environmental influences on—diverse actors. When these simpler rules (axioms) interact over time, we can create our more complex model of ML diffusion.

While communication culture is generally viewed as a subset of its organizational counterpart (Schein, 2009), some communication theorists view communication as creating culture. This article’s model adopts a grounded in action orientation of communication culture—where communication and culture interact to cocreate perceptions of organizational reality (Keyton, 2013). On an even broader scale, Smith (1993) observed that many authors view organization and communication as interchangeable. Consonantly, this study assumes a communication perspective of organizational culture which refers to “patterns of human action and its recursive behaviors (including talk and its symbolic residues) and meaning” (Eisenberg & Riley, 2001, p. 294). Among these patterns of human action is the strength of an organization’s value for oral communication competence. Although little research has occurred on this subject at team and organizational levels (Jablin & Sias, 2001), one can logically surmise that the importance of oral communication competence in an organization can be observed through the selection of a top boss who is or is not speech communication competent with high-ML potential, lower level managerial hiring strategies, formal oral communication training availability, and rewards (both tangible and intangible) that are distributed for oral communication competence.

Additionally, organizational culture seems to be a relatively stable organizational property over extended periods of time (Ott, 1989; Schein, 2009). And organizational culture shapes who is admitted into an organization, how new entrants are socialized, and what structures and processes are supported and reinforced. Within this frame, organizations tend to have long-term, sustained approaches to actions and activities. Even though organizational cultures can transform over time, stability appears to be more the norm (R. E. Miles & Snow, 1978; Ott, 1989; Schein, 2010). This assumption of probable cultural stability creates a model that can generate more readily understandable interactions and propositions with only a minor loss of generalizability. As a result, this perspective of organizational culture (including its corollary expression in leader oral-communication) creates an underlying axiom for all model types, and this axiom is as follows:

Respected scholars assert that a top-leader’s spoken communication style has a robust impact on the behaviors of employees and overall competitive advantage (Phillips, Lawrence, & Hardy, 2004; Robbins & Judge, 2012; Westley & Mintzberg, 1989). Thus, an organizational culture that prioritizes high-ML potential in communication competence during CEO selection and where the incumbent fulfills these expectancies with subsequent, consistent high-ML use after hire, will be more likely to experience a high-ML diffusion rate. In brief, high-ML speech from the CEO, top boss ML ability, fosters the diffusion of high-ML among lower level leaders. Conversely, in cases where the organizational culture does not value such CEO oral communication competence and where the top-leader does not consistently use high-ML speech after hire, diffusion of high-ML among lower level leaders is much less likely.

Among other cultural characteristics that are incorporated into this ML simulation, formal, effective training is well-supported in its power to enhance skill learning, including in the realm of oral communication competency (Aguinis & Kraiger, 2009; Cascio, 2012; Robbins & Hunsaker, 2012). Thus, ML training should assist its diffusion. Similarly, social exchange theory (Blau, 1986; J. A. Miles, 2012) suggests that extrinsic and intrinsic rewards which are given to non-top-leaders who are proficient in high levels of ML rewards should encourage its diffusion within an organization. On the other hand, when such rewards discourage or give no positive reinforcement for high-ML use, diffusion rates of high-ML speech among non-top-leaders will be lower.

Along with encouraging of diffusion of shared behaviors, numerous organizations use cultural values in their selection decisions of lower level managers, and many job seekers are attracted by perceived organizational fit between their own beliefs, knowledge, skills, and abilities and those of the recruiting organization (Cascio, 2012). As a result, an organization whose culture fosters ML may attract and select new nontop managers who are speech communication competent (with ML hiring) instead of candidates who lack such capabilities, termed as precompetent by Jablin and Sias (2001). It should however be noted that while an organizational culture may espouse great respect for oral communication competence, ML use itself is based on sufficient knowledge of within-culture norms and values.

Consequently, new communication-competent managerial hires (who lack such inside cultural knowledge) are expected to take approximately 6 months to become generally competent in ML. The estimate of 6 months is taken from the time span that most communication researchers’ use for a rule of thumb when estimating assimilation into an organization’s culture (Jablin & Sias, 2001). An opposite effect can be expected where oral communication competence is not embedded in organizational cultural values, recruitment, and selection processes. In this scenario, diffusion of high-ML speech is significantly less likely.

Thus, the following four axioms are put forth under the rubric of an organizational culture’s oral communication orientation.

The next step in the ML diffusion model building incorporates institutional influences (factors). These variables can be defined as external and internal forces that moderate the rate of ML diffusion. Although many such factors exist, agent-based model parsimony demands simplicity (Railsback & Grimm, 2011). Therefore, this proposed model concentrates on two of the most pertinent institutional influences: the rate of leader turnover and the number of formal hierarchical layers within the organization.

Even in organizations with cultures that value quality oral communication, a high turnover rate (organizational departures which are voluntary or involuntary) among top-leaders is expected to retard the diffusion of high-ML use. Top-leaders are expected to strongly influence the entire organization by serving as role models for immediate reports, and these reports will, in turn, serve as models for their direct reports. This chain should continue throughout the organizational hierarchy, acting as a contagion factor. Each leader has a precompetency period in which he or she will not be able to employ high levels of ML. While this period is relatively brief, high turnover levels will usher in more extended time frames in which poor leader oral-communication can be emulated.

Turnover often is accompanied by newcomer replacement and subsequent organizational entry and assimilation. When either external newcomers are hired or promotions of nonmangerial employees within the organization to supervisory positions are the responses to turnover, an assimilation process, for example, learning new role behaviors (Jablin, 2001), must take place before the leader is capable of being wholly productive (Cascio, 2012). As previously noted, most researchers estimate that the average assimilation period can take up to 6 months. Since even communication competent hires and some internal hires (incumbents who are inexperienced with sending supervisory messages) are neophytes to using ML speech, one can predict that turnover rates will moderate ML diffusion.

Beyond causing an initial time delay with organizational assimilation, and the inference that turnover will significantly effect ML diffusion, turnover’s type of impact on the rate of ML adoption is unknown. For example, in the case of simple attrition, the departing manager may or may not be a high-ML user. Furthermore, replacement leaders could also become or not become proficient in ML, depending on many variables including the available talent pool, communication competence, and organizational oral communication culture practices.

Returning to the criteria for organizational characteristics selection in this study, increased formal hierarchical layers, structural organizational levels, will make it more difficult for leaders to model competent oral communication for managers at lower levels and foster a positive contagion effect. In a complimentary fashion, organizations with fewer formal hierarchical layers should be more amenable to the adoption of competent top-leader oral-communication behaviors because a greater proportion of leaders are directly influenced by the top boss.

More generally stated, the second predicted institutional moderating variable of the ML diffusion rate is the formal hierarchical structure of the organization. In brief, increased spatial and proximity distances in the communication network (a subset of the social network) of an organization may delay ML message transmission and hence reduce the likelihood of its adoption. Shumate and Contractor (2013) describe communication networks as “relations among various types of actors that illustrate the ways in which messages are transmitted, exchanged, or interpreted” (p. 449). These same authors observe that colocation may effect actors’ construction of communication networks. Relatedly, in an earlier literature summary, Monge and Contractor (2001) noted the work of Krackhardt and Brass (1994) which presents the principle of interaction, namely that more frequent contact leads to a contagion—more shared attitudes. In the same vein, Zahn (1991) discovered that close physical working proximity as well as shorter distances in the hierarchical chain of command and between offices make message exchange more likely. Drawing upon this literature, we expect that formal organizational hierarchical structures which facilitate colocation of social network actors, and fewer reporting relationship distances between them will augment high-ML diffusion. Therefore, a flatter organizational hierarchy will foster ML adoption. Based on this discussion, the following axioms concerned with institutional factors are stated.

The third category of this proposed model for ML diffusion is influence factors (the ability to motivate others to achieve/adopt desired goals, attitudes, and behaviors) including the influence of peers, immediate supervisor, and the top boss (Yukl, 2013). Again, for requisite model simplicity, peer influence is restricted to non-CEO leaders who are on the same hierarchical level. Contagion theory calls the nature of these relationships structural equivalence (Burt, 1987; Susskind, Behfar, & Borchgrevink, 2006). To be precise, structural equivalence is denoted when social network actors possess exactly the same communication ties. However, scholars allow for more flexibility by accepting similar ties or regular equivalence as structural equivalence (Carpenter et al., 2012; Monge & Contractor, 2001). This connection identifies the set of reporting leaders as a subunit whose members have kindred roles and are more likely to interact, thus generating shared communication influences (Burt, 1987; Strang & Soule, 1998). Susskind et al. (2006) note that “[i]t has been shown that frequent and repeated exchange between members increases group understanding of expectations for normative behavior” (p. 121). On the other hand, peer influence introduces a more informal message exchange and social network to this diffusion model that may also weaken regular equivalence. These informal connections can be very influential on attitudinal and behavioral change (Krackhardt & Brass, 1994; Krackhardt & Hanson, 1993; Monge & Contractor, 2001).

In comparison, the strength of immediate supervisor oral communication influence on direct reports has been strongly supported in the literature, and may well transfer from subordinate leaders’ message interpretation from their bosses into subsequent messages to direct followers (Jablin & Sias, 2001; Robbins & Hunsaker, 2012). This linking-pin effect is also related to social learning theory where followers may model an immediate superior’s behaviors, including speech patterns (Bandura, 2001; Graen, Cashman, Ginsburg, & Schiemann, 1977; Jablin & Sias, 2001). Last, the extremely important impact of a CEO’s oral communication skills on all organizational stakeholders—including lower level leaders—has already been documented (Phillips et al., 2004; Robbins & Judge, 2012; Westley & Mintzberg, 1989). Therefore, three axioms can be stated which complete development of this ML diffusion model.

These axioms can be used to logically develop an ML diffusion process model, and generate hypotheses about this model. While this process will not result in quantitative statements about organizational relationships, it will indicate what differences can be expected from different organizational characteristics. Specifics on this process will be presented in the Methodology section of this article.

Methodology

Results

Analytic Results

The regression tree results indicate varying diffusion paths contingent on whether the top-leader has high- or low-ML ability. These findings indicate that there is a substantial difference in the diffusion of high-ML use in the two scenarios. The regression tree results are displayed in Figure 2. Across all situations, there was a high-ML diffusion rate of 54%. However, when a top boss employs high oral ML communications, the maximum ML diffusion rate is 85% and the minimum diffusion of high-ML is 57%. (This minimum diffusion rate only occurs when 4.5 or fewer time periods have elapsed.) In comparison, when a top boss employs low-ML speech, the greatest diffusion of high-ML use is 52% (only when there are rewards for high-ML use, and there are many formal hierarchical layers separating top and bottom organizational levels). The lowest diffusion rate with a low-ML top boss is 25%.

OPEN IN VIEWER

Figure 2.

Regression tree analysis results for motivating language (ML) diffusion process.

In addition to differences in the amount of high-ML use diffusion, there is also substantial variance in the way that diffusion of high-ML occurs. When the top boss uses high-ML communication, the greatest influential factor in ML diffusion is time. With short time periods (less than 4.5 time periods or 2.25 years) ML diffusion reaches only 57%. In comparison, after 4.5 time periods (or 2.25 years) high-ML is diffused across approximately 73% of leaders. So if a top-leader employs high-ML, the passage of time (and continued use of high-ML by the top-leader) becomes the most important factor in diffusing high-ML use.

The next most important factor is rewarding for high-ML use. If rewards are set to discourage high-ML use, diffusion is reduced to 61%. In the cases of neutral or positive rewards for high-ML use, employee turnover emerges as a diffusion factor. In this case, higher turnover (above 5.5%) hinders high-ML diffusion (to 73%) as compared with the scenario with lower turnover rates (where there is a diffusion rate of 85%).

The results from situations where top-leaders employee low-ML speech significantly diverge from the preceding results. In the top-leader low-ML communication situation, rewards are the most important factor for increasing the diffusion of high-ML use. When there are positive rewards for high-ML use in this context, such talk will be used by 52% of the non-top-leaders, but only when there are more than 3 or more hierarchical levels in an organization. When there are fewer organizational levels, the top-leader low-ML expression seems to overwhelm the reward effect, and high-ML use only reaches a 37% diffusion rate.

When rewards are neutral or negative toward the use of high-ML in the low-ML top-leader scenario, training can partially offset the adverse reward system. By including training for high-ML use, organizations were able to increase high-ML diffusion rates to 41% when a top-leader practiced low-ML. When training was not available, high-ML diffusion dropped to 25% after 5.5 time periods when the top-leaders used low-ML.

As discussed earlier, there is an important distinction that must be made about these results: They are intended to provide a rigorous development of axiomatic statements about ML diffusion. Yet they are not statements about specific expected results. The ABM method used in this article provides an avenue for logically exploring and expanding our model. Consequently, the results should be viewed more as an expected ranking or ordering of ML diffusion outcomes rather than focusing on absolute percentages. A graphical depiction of these regression tree findings is presented in Figure 2.

From these results, a set of 9 hypotheses can be derived. Of note, the hypotheses signify whether ML is high or low due to the dominant influence of the top-leader.

Hypothesis 1: Top-leader ML behavior will have the greatest influence on diffusion of high ML use throughout an organization.

Hypothesis 2: The ML diffusion process differs between organizations with high- and low-ML top-leaders.

Hypothesis 3: When there is a high-ML top-leader, ML diffusion requires a period of time for other diffusion influences to become salient.

Hypothesis 4: Where there is a high-ML top-leader, rewards that counter high-ML use decrease high-ML diffusion (after a sufficient period of time).

Hypothesis 5: Where there is a high-ML top-leader, higher turnover rates can impede high-ML diffusion (after a significant period of time and accounting for rewards).

Hypothesis 6: Where there is a low-ML top-leader, rewards for high-ML use can increase lower-level leader adoption of high-ML diffusion.

Hypothesis 7: Where there is a low-ML top-leader, rewards for high-ML use are most effective in high-ML diffusion where there are sufficient organizational levels to insulate non-top-leaders from top-leader influence.

Hypothesis 8: Where there is a low-ML top-leader and neutral or negative rewards for high-ML use, formal ML training can increase high-ML use diffusion.

Hypothesis 9: Where there is a low-ML top-leader and neutral or negative rewards for high-ML use, the passage of time decreases high-ML diffusion in the absence of leader ML training.

Table 1 provides an overview of how relevant organizational attributes are expected to influence the level of organizational ML diffusion. Each set of characteristics has an attendant ranking (from highest to lowest) on how widespread ML language use is expected to diffuse throughout an organization.

OPEN IN VIEWER

Table 1.

Organizational Characteristics and Motivating Language (ML) Diffusion Ranking.

Rank	Top Boss	Time	Rewards	Turnover	Levels	Training
1	High	More than 4.5 periods	Positive or neutral	Less than 5.5%	NA	NA
2	High	More than 4.5 periods	Positive or neutral	More than 5.5%	NA	NA
3	High	More than 4.5 periods	Negative	NA	NA	NA
4	High	Less than 4.5 periods	NA	NA	NA	NA
5	Low	NA	Positive	NA	3 or more	NA
6	Low	NA	Neutral or negative	NA	NA	ML training
6	Low	5.5 or more time periods	Neutral or negative	NA	NA	No ML training
8	Low	NA	Positive	NA	Less than 3	NA
9	Low	NA	Neutral or negative	NA	NA	No ML training
10	Low	Less than 5.5 time periods	Neutral or negative	NA	NA	No ML training

Note. NA indicates that the organizational characteristic had no significant effect on the ML diffusion ranking.

Discussion

This study has contributed new insights into the emergent processes of MLT. As a result, this investigation has uncovered key factors expected to influence ML diffusion at an organizational level of analysis, and which also may be necessary for reaping the benefits gained by this genre of leader oral-communication. While the implications of these findings are exploratory, groundwork has been laid for future hypotheses which can be tested empirically. In addition, new nonlinear methodology has been adopted to examine ML, a longitudinal ML model was employed, and the interaction between oral communication content and network structure was analyzed. Similarly, the contagion potential for leader speech communication styles were examined. All of these research steps have been suggested by experts in management, communication, and social network research.

Surprisingly based on the extant literature, some of the model’s axioms were either supported in unexpected ways or not supported at all. Formal organizational structure had a weak significant effect on ML diffusion which was contrary to expectations. With a high-ML top-leader, formal hierarchical layers did not affect the high-ML diffusion rate. (Perhaps this observation can be attributed to the fact that CEOs speak both directly and indirectly—through their direct reports—to lower level managers.) Yet more layers (over 2.5) slightly boosted the effects of high-ML diffusion when the CEO was a low-ML oral communicator (from 37% to 52%). Aligned with this trend, axioms which predicted significant moderating effects in ML diffusion from recruitment and selection, peer influence, direct supervisor influence, and top boss influence were not supported. Reflecting on previous research, these findings are incongruent with well-respected studies.

We can only speculate about the underlying reasons for these phenomena. Such discoveries make more sense when the multifaceted nature of the ML diffusion model is taken into account. The top-leader’s ML use is usually compatible with the values of an organization’s culture. Furthermore, these attributes, expressed through practices and norms, may be comparatively more powerful than diffusion influences from the other axiomatic variables. These latter factors could well have significant influences which are subsumed by the impact of a culturally congruent, top-leader’s ML use.

Of note, the asymmetric diffusion pattern shown in Figure 2 merits discussion here. Our model indicates that with a high-ML top-leader, time, rewards, and turnover (in descending order of impact) moderated the diffusion process. In contrast, with a low-ML top-leader, rewards, greater number of hierarchical levels, training, and time (in descending order of impact) come into play for organizational adoption of high-ML. Interestingly, non-top-leader turnover did not have any significant impact on high-ML diffusion when the top-leader was a low-ML communicator, whereas longer tenure (5.5 time periods of 6 months or more) did significantly increase the high-ML diffusion rate when the CEO was a high-ML user.

There are some valuable implications for research and practice that can be gleaned from this study. For research, some testable hypotheses for diffusion of high levels of ML have been identified. First, consistent top-leader ML practice, that is congruent with an organization’s cultural norms, is closely related to the ML speech adoption of lower level leaders. Second, ML diffusion will differ in process and in moderator effects, contingent on whether the CEO’s ML use is high or low. Third, time and rewards are both moderating factors in the ML diffusion process for all assimilation scenarios, but their impacts will vary, dependent on top-leader ML use.

For practice, this model strongly supports the dominant influence of top leadership on organizational oral communication practices. Thus, if the benefits of high ML are desired, careful selection of a CEO who is oral communication competent and compatible with the organization’s culture is paramount. Taking a further step, cultural change itself may be needed if more effective organizational communication is desired. Furthermore, among other leaders, rewards for high-ML use encourage its diffusion. Training also has a positive influence on high-ML diffusion, but only when the CEO is a low-ML communicator. In addition, a respectable period of time (over 2.25 years) is needed for the greatest levels of high-ML diffusion. Finally, higher levels of leader turnover can hinder the adoption of high-ML among an organization’s leadership ranks.

Perspective is called for at this time. An agent-based model configuration is preliminary. Empirical studies are needed for more in-depth examination of these findings. Herein is the inherent limitation found in this study. These working hypotheses are intended to explore literature embedded axioms and their interactions with the objective of finding hypotheses for future empirical testing. In brief, we cannot draw definitive conclusions from these emergent properties until such rigorous testing occurs. Hence, the model percentages are only benchmarks, and are not intended to be accurate assessments. A second constraint in this research is that the current literature reviewed has been largely drawn from U.S-based scholarship. Consequently, any generalizability to global settings is restricted. Also, our model captured oral language diffusion within a traditional organizational structure. Networked and other innovative forms of organizational design were excluded for the sake of model parsimony. Another limitation occurs with the restricted domain of oral communication. Finally, model development was premised on a relatively stable organizational culture. Therefore, this model is not expected to apply to organizations that experience a dramatic cultural shift which affects target organizational characteristics (such as a change in training or rewards policies).

Such limitations can be addressed in future research through empirical analysis of our recommended hypotheses. For testing, a first step is to examine the role of top-leader ML use on organizational diffusion. Since the model predicts CEO behavior as having the dominant effect on diffusion both quantitatively and qualitatively, examining this variable will allow researchers to explore the most robustly predicted aspect of the model. If empirical results show little or no difference in diffusion rates, then the entire model will need to be redeveloped. Conversely, empirical support for this proposition will encourage further tests on propositions that may require more subtle analyses.

Another aspect for future testing is to examine how just fractal these phenomenon are. While the model examined an entire organization, conceptually, there is no reason why the model is not applicable to subsets of an organization—assuming ML use congruence between the top-leader and a leader that is in charge of an organizational division. In this scope, the model could be tested in one organizational area (such as production or marketing) to see if diffusion occurs as predicted. Also, investigations should be extended to both global and networked organizational research settings. It would also be potentially beneficial to extend the scope of future, related research to incorporate multiple media ML use, especially with written communication. Finally, since ML diffusion appears to be temporally moderated, continued emphasis on longitudinal research is in order.

In conclusion, high leader ML can potentially unleash valuable contributions to an organization through improved employee performance and quality of work life. Yet in order for these benefits to be realized, successful implementation and practice of high levels of ML must first occur. This article is an important step in that direction.