Despite their importance for organizations making decisions about promotions and pay, as well as providing employees with feedback and promoting their professional development (Cleveland, Murphy, & Williams, 1989; DeNisi & Murphy, 2017; DeNisi & Pritchard, 2006; Ferris, Munyon, Basik, & Buckley, 2008), performance appraisal interviews remain a challenging task for supervisors (Gordon & Stewart, 2009; Westerman & Smith, 2015) as well as a disliked chore for employees (Brown, Hyatt, & Benson, 2010; Lawler, Benson, & McDermott, 2012; Pulakos & O’Leary, 2011; for popular press accounts see Kenny, 2016; Rock, Davis, & Jones, 2014; Sytch & DeRue, 2010). One way to master appraisal interviews more effectively might be through increased levels of leader empathy. In fact, previous theorizing and research has repeatedly highlighted the important role of leader empathy as a main driver of successful leader-follower interactions (Bass & Riggio, 2006; Mahsud, Yukl, & Prussia, 2010; Yukl, 2010). For example, leader empathy has been described as a key ingredient of considerate leadership (Fleishman & Salter, 1963). In the context of appraisal interviews and change management, more broadly, empathic supervisors who are skilled at taking their employees’ perspective, recognizing their employees’ needs, and consequently addressing those needs are probably more likely to create a constructive discussion about change.

Regardless of a growing conceptual background on the role of empathy in the leadership literature (e.g., Burch, Bennett, Humphrey, Batchelor, & Cairo, 2016; Humphrey, 2002; Pescosolido, 2000), empirical studies that explore how leader empathy functions in the workplace remain rather sparse (see Burch et al., 2016), and studies that focus on the fine-grained behaviors that are at the core of leader empathy are especially scant (for a rare exception, see Meyer et al., 2016). In fact, most previous studies and measures of leadership empathy primarily focus on employees’ perceptions of leader empathy (e.g., Mahsud et al., 2010; Sadri, Weber, & Gentry, 2011) but neglect to capture more objective accounts of leader empathy that drive such perceptions in the first place.

We believe this issue should be addressed for at least two reasons. First, building new knowledge about the behavioral markers of leader empathy can help to derive clear practical implications in terms of increasing leaders’ empathic skills during appraisal interviews and beyond. Second, exploring how leader empathy is enacted in real-time sheds light on how leadership influence unfolds at the event level of leader-follower interactions (DeRue, 2011; Hoffman & Lord, 2013). Focusing on objective accounts of leader empathy could help to enrich our understanding of how this important leadership skill functions during leader-follower encounters and could also help to supplement rather broad definitions of leader empathy with fine-grained and more tangible behaviors.

The present study addresses this research gap. Specifically, the overall aim of this study is twofold. First, we delve into the behavioral manifestation of leader empathy and relate supervisors’ empathic communication style as a higher-order construct to fine-grained verbal mimicry processes between supervisors and their employees. In particular, we build on recent work from the fields of computational linguistics (Pennebaker, Boyd, Jordan, & Blackburn, 2015) and use computerized dictionary-based text analysis to analyze the degree of language style matching between supervisors and their employees. Language style matching describes a process by which conversational partners adapt their word use to one another (Ireland & Pennebaker, 2010; Niederhoffer & Pennebaker, 2002). As such, our study also aims to make a methodological contribution by using automatic word processing to measure verbal mimicry in leader-follower interactions. Previous research based on automatic text analysis has emphasized the special role of language style matching for establishing and maintaining successful dyadic relationships (e.g., Ireland et al., 2011). However, we are not aware of any empirical examinations aimed at exploring language style matching during leader-follower encounters in organizations.

Second, our study contributes to research on the functionality of leaders’ expressed empathy. We draw from the performance appraisal and leadership literature and develop an argument for the importance of empathic communication during appraisal interviews. Based on the idea that supervisors’ display of empathy is vital for engaging employees’ hearts and minds, we relate supervisors’ empathic communication style to employees’ intentions to change and to their perceptions of supervisor likeability. In synthesizing our theoretical arguments, we subsequently test an indirect effects model wherein language style matching encourages change and fosters the liking of leaders via leaders’ empathic communication style. We explore our hypotheses in a field sample of audiotaped performance appraisal interviews.

The Role of Empathy in the Leadership Process

The concept of empathy has received considerable scholarly attention over the last 100 years, specifically from the fields of psychotherapy and counseling psychology which resulted in a multiplicity of definitions (for overviews see Duan & Hill, 1996; Kerem, Fishman, & Josselson, 2001). In their review of the history of empathy research, Duan and Hill (1996) summarize that definitions of empathy have ranged from empathy being a personal trait or stable ability (e.g., Book, 1988; Buie, 1981; Hogan, 1969) to empathy being a situation specific state (e.g., Greenson, 1960; Rogers, 1957) and to empathy being a multiphased experiential process (e.g., Barrett-Lennard, 1981; Rogers, 1975). Adding to the complexity, there has been considerable debate in the literature on whether empathy is mainly an affective or cognitive phenomenon (Duan & Hill, 1996). The affective view on empathy emphasizes the notion of empathy as an emotional response such that empathy encompasses the ability to share or experience another person’s feelings or emotions (e.g., Feshbach & Roe, 1968; Mehrabian & Epstein, 1972). Emotional contagion (e.g., Gladstein, 1983; Hatfield, Cacioppo, & Rapson, 1994; Hatfield, Rapson, & Le, 2009) which describes that individuals rather automatically “catch” the emotions of others during social interactions is closely tied to the affective view of empathy. Scholars following the cognitive view on empathy, on the other hand, explained that empathy is a person’s intellectual understanding of another person’s internal state (e.g., Deutsch & Madle, 1975; Hogan, 1969). Cognitive efforts to recognize and understand someone else’s perspective or point of view have been labeled as perspective taking (Bernstein & Davis, 1982; Gladstein, 1983). Today, empathy is generally seen as a multidimensional concept encompassing both affective and cognitive components and leading to adequate behavioral actions (Cuff, Brown, Taylor, & Howat, 2016; Decety & Jackson, 2004; Duan & Hill, 1996; Eisenberg, 2000; Smith, 2006). The behavioral aspect of empathy includes that empathic individuals have the behavioral ability to respond compassionately to another person’s needs, motivations, or opinions by communicating their understanding (e.g., Decety & Jackson, 2004; Eisenberg & Miller, 1987). Depending on the specific research context at hand, different components and views prevail (Zhou, Valiente, & Eisenberg, 2003).

Our focus in the current study is on empathy in the workplace, and specifically on expressed leader empathy during appraisal interviews. In line with previous research on the topic (Kellet, Humphrey, & Sleeth, 2002, 2006; Mahsud et al., 2010), we understand leader empathy as an ability that combines thinking and feeling. More specifically, we define leader empathy as a leaders’ ability to accurately recognize and understand the emotional reactions and feelings of their followers (see Mahsud et al., 2010). This understanding, in turn, helps leaders to respond appropriately to the needs of their followers and to craft an appropriate (emotional) response.

Empathy has long been identified as an important keystone in the leadership process (Bell & Hall, 1954; Fleishman & Salter, 1963). However, subsequent research on the role of leader empathy in organizations has been very scarce as recently outlined by Burch et al. (2016). It was not until the early 2000s that scholarly interest in leader empathy and emotional intelligence—as a related construct—gathered momentum (for initial influential articles see George, 2000; Pescosolido, 2000). Generally, leader empathy is perceived as being central for managing social relations because empathic leaders are said to be more effective at managing the emotions of their followers (Bass & Riggio, 2006; Yukl, 2010). Consequently, it is a common theme in the leadership literature that leaders who are skilled at identifying and responding to follower emotions are also more effective leaders (e.g., George, 2000; Pescosolido, 2000). For example, it has been argued that leaders who are more attuned to their followers’ emotions are better able to spark enthusiasm in their followers, to develop collective goals and objectives, and to promote flexibility in decision-making and change (George, 2000). In addition, leader empathy has been related to higher ratings of transformational leadership (Pillai, Williams, Lowe, & Jung, 2003; Skinner & Spurgeon, 2005), employees’ well-being (Scott, Colquitt, Paddock, & Judge, 2010), and ratings of leadership performance (Sadri et al., 2011). Results from a recent large-scale survey (Businessolver, 2017) further showed that employees themselves value leader empathy and see it as an integral component of the workplace. More than half of the participants stated that they would even accept a pay cut to work for an empathic employer (Businessolver, 2017). In sum, these findings provide support for overall positive effects of leader empathy in the workplace. Little is known, however, about the underlying behavioral “ingredients” that are characteristic of leader empathy. In other words, our understanding of how leader empathy comes to live during face-to-face interactions among leaders and followers in organizations is very limited, and leader empathy tends to remain a rather fuzzy construct. In the following, we therefore ask what drives perceptions of leader empathy and consequently explore which behavioral markers might be objective indicators of leader empathy.

Communication Dynamics at the Core of Leader Empathy

Leaders frequently find themselves in more or less challenging conversations with their followers (e.g., De Vries, Bakker-Pieper, & Oostenveld, 2010). As described above, a core component of empathy involves how empathy is expressed in these difficult face-to-face interactions, such as during appraisal interviews (Asmuß, 2008, 2013). In order to achieve successful conversations, we argue that leaders must be able to not just feel or experience empathy but to craft empathic responses. Therefore, the primary focus of our study is not on a leader’s internal empathy (i.e., empathy as an intraindividual phenomenon) but rather on how empathy is communicated in situ (i.e., empathy at the interpersonal level; Burch et al., 2016). This also means that we do not explore trait aspects of empathy but follow a state view on empathy, thereby focusing on situational influences (see also Cuff et al., 2016). The follower plays a central role in this view as we regard leaders’ empathic communication to be more of a collaborative act.

An approach focused on how leader empathy is displayed and functions at the communication level also aligns with recent calls in the leadership literature. Leadership scholars increasingly value a communication-centered view of leadership as exemplified by several theoretical reviews (e.g., Fairhurst, 2008; Fairhurst & Connaughton, 2014; Fairhurst & Uhl-Bien, 2012; Ruben & Gigliotti, 2016; Uhl-Bien, 2006; Uhl-Bien, Marion, & McKelvey, 2007). A communication orientation places communication at the center of leadership research and highlights communication as a core ingredient inherent in the leadership process.

To accommodate our view on leader empathy as being contextually situated in leader-follower interactions, we propose a measurement approach that is able to tap the interactive dimension of leader empathy (Kellet et al., 2006) and capture how leader empathy is embedded in the communication process between supervisors and their employees during appraisal interviews. Specifically, we suggest to make use of the fine-grained linguistic markers that characterize leaders’ and followers’ language style which we describe in more detail below. In a second step, we relate these objective behavioral indicators to global assessments of leaders’ empathic communication style. We see this communication style as a higher-order construct that emerges from the interaction process between leaders and followers (see also Lehmann-Willenbrock & Allen, 2017).

Language Style Matching as a Measure of Verbal Mimicry

From a behavioral perspective, empathy has been related to expression of behavioral mimicry in the past (for an overview, see Chartrand & Lakin, 2012). Behavioral mimicry involves two or more individuals showing the same behavior at the same time including mimicry of mannerisms, gestures, postures, and other motor movements (Chartrand & Lakin, 2012). For example, in a series of experiments on the so-called chameleon effect in social interactions, Chartrand and Bargh (1999) found that university students high in perspective taking (i.e., the cognitive component of empathy) mimicked the behavioral mannerisms of their interaction partners more than students low in perspective taking, resulting in a sense of connection and liking among interaction partners (Chartrand & Bargh, 1999; see also Lakin & Chartrand, 2003; Stel & Vonk, 2010). Beyond overt motor mimicry, individuals can also verbally mimic one another. Verbal mimicry can include, for example, that individuals mimic their interaction partner in syntax (Levelt & Kelter, 1982) and speech rate (Manson, Bryant, Gervais, & Kline, 2013; Webb, 1969).

From an evolutionary perspective, verbal and nonverbal mimicry is an evolved psychological mechanism for social coordination and its main function is to smooth coordination between interaction partners (Van Vugt & Kameda, 2014). As such, mimicry helps to align thoughts and feelings among interaction partners, facilitating more harmonious interactions (Chartrand & Bargh, 1999; Maurer & Tindall, 1983). Thus, and in line with previous scholars (e.g., Chartrand & Bargh 1999; Meyer et al., 2016), we posit that mimicry is indicative of empathy.

Since the early 2000s, a new stream of research emerged from the realm of computational linguistics that focuses on how conversational partners match their language use to one another (Ireland & Henderson, 2014; Ireland & Pennebaker, 2010; Niederhoffer & Pennebaker, 2002). These studies extend previous research on verbal mimicry, for example, on speech rate, because they explore and classify every single word that is uttered during a conversation, providing a detailed measure of similar word use (see also Chartrand & Lakin, 2012). Of special interest is the similar use of the so-called function words. Function words are “the syntactic backbone of language” (Gonzales, Hancock, & Pennebaker, 2010, p. 5) and include pronouns, prepositions, articles, and other content-free parts of speech (Chung & Pennebaker, 2007; Niederhoffer & Pennebaker, 2002). The use of function words reflects how individuals say things but not what is said. Because function words are relatively short and carry little meaning outside of context, their use is rather automated and nonconscious in comparison to the use of content words such as nouns and verbs (Segalowitz & Lane, 2004). The way each individual uses function words in a conversation is called individual language style (Pennebaker & King, 1999; Pennebaker, Mehl, & Niederhoffer, 2003). Previous research showed that conversational partners tend to match their language styles to one another called language style matching (Chung & Pennebaker, 2007; Ireland & Pennebaker, 2010; Niederhoffer & Pennebaker, 2002). The extant literature on language style matching generally paints a positive picture showing that language style matching is, for example, positively related to relationship initiation and stability (Ireland & Pennebaker, 2010; Ireland et al., 2011), to reduced emotional distress in supportive communication (Cannava & Bodie, 2016), and to mutual liking among team members (Gonzales et al., 2010). Motivated by these previous findings, we explore the role of language style matching during appraisal interviews. Specifically, as language style matching represents dyadic coordination and a form of verbal mimicry among interaction partners (Gonzales et al., 2010; Niederhoffer & Pennebaker, 2002), we propose that language style matching is associated with expressions of empathy. In fact, one previous study on the role of empathy in motivational interviewing, a counseling approach focused on eliciting behavior change by helping clients to explore and resolve ambivalence (Miller & Rollnick, 2013; Miller & Rose, 2009), successfully related language style matching between clients and therapists to therapists’ ratings of empathy (Lord, Sheng, Imel, Baer, & Atkins, 2015). Hence, expecting to replicate prior work in a new context we conclude that language style matching seems to be a valid micro-level behavioral measure of leader empathy. We put forward the following hypothesis:

  • H1: Language style matching between supervisors and employees positively relates to supervisors’ empathic communication style.

In pursuing a dynamic perspective of leader-follower interactions, we further aim to explore if language style matching among supervisors and employees is stable across longer interaction episodes such as appraisal interviews. There is still a dearth of research on the role of language style matching in natural conversations and most studies focused on an overall measure of language style matching at a conversational level (e.g., Cannava & Bodie, 2016; Gonzales et al., 2010; Ireland et al., 2011). An exception is a study on hostage negotiations which explored temporal changes in language style matching among hostage takers and police negotiators (Taylor & Thomas, 2008). Findings showed that successful negotiations in contrast to unsuccessful negotiations were characterized by a lower variability in language style matching over time. Thus, successful negotiators were able to maintain a constant level of coordination by steering clear of dramatic fluctuations in language style matching (Taylor & Thomas, 2008). Although this particular study was rather explorative in nature and focused on a specific interaction context that is quite different from interactions among supervisors and their employees in organizations, it gives rise to the question whether more stable levels of language style matching might be more indicative of an empathic communication style. In particular, appraisal interviews cover a multitude of topics and are described as a complex communication task (Gordon & Stewart, 2009). Some topics and performance criteria might be easier to discuss than others and especially negative feedback can be challenging (Beer, 1981). Therefore, recommendations suggest that supervisors should maintain constant attention to their employees’ thoughts and feelings and use an empathic tone throughout the appraisal discussion (Gordon & Stewart, 2009). Increased attention to one’s conversational partner, in turn, has been linked to higher levels of language style matching (Ireland & Henderson, 2014; Tausczik, 2012). Taken together, our second hypothesis thus posits:

  • H2: Throughout an interaction episode, more stable levels of language style matching between supervisors and employees relate positively to the empathic communication style of supervisors.

Effects of Empathic Communication During Appraisal Interviews

After laying the groundwork and disclosing the micro-level behavioral manifestation of leader empathy, we want to shed additional light on the overall effectiveness of leaders’ empathic communication during appraisal interviews. Appraisal interviews require supervisors to discuss strengths and weaknesses in employee performance and to provide directions for future development. As such, appraisal interviews provide a context not only for initiating and managing change processes (DeNisi & Pritchard, 2006; DeNisi & Smith, 2014) but also for establishing and managing leader-follower relationships (Elicker, Levy, & Hall, 2006). The current study attempts to explore how expressed leader empathy relates to both of these functions. In particular, we propose that leaders’ expressed empathy during appraisal interviews positively relates to (1) followers’ intention to change (in terms of engaging the minds of followers) and to (2) followers’ perceptions of leader likeability (in terms of engaging the hearts of followers).

The nature of appraisal interviews entails a need for employees to develop and grow, for example, through taking on new tasks and committing to new performance goals. At the most fundamental level, the overall purpose of performance appraisal and the ensuing appraisal interview is to improve employee performance (DeNisi & Pritchard, 2006; DeNisi & Smith, 2014). Thus, change is a constant during appraisal interviews. We believe that supervisors who communicate empathy during appraisal interviews can facilitate chance processes on part of the employee for the following reasons. First, expressed leader empathy supports the process of building rapport with followers which creates the necessary foundation for initiating change processes. Although this view is not strongly discussed in the leadership literature, we can extrapolate from earlier findings outside of the organizational domain. Specifically, the role of empathy as a prerequisite for change has received considerable attention in the literature on counseling and psychotherapy (for overviews, see Bohart, Elliott, Greenberg, & Watson, 2002; Wampold, 2015). For example, expressing empathy is an important component in motivational interviewing (Miller & Rollnick, 2013; Miller & Rose, 2009). Similarly, in his work on client-centered counseling, Rogers (1975, p. 3) summarized that “a high degree of empathy in a relationship is possibly the most potent and certainly one of the most potent factors in bringing about change and learning.” We argue that this rationale can also be transferred to leader-follower interactions during appraisal interviews in organizations. In particular, empathic communication during appraisal interviews may help leaders to attain a more complete and accurate understanding of their followers’ needs. This understanding could elevate the ability to anticipate how their followers are going to react which can help leaders to adapt their own behaviors to the needs of their followers (see also George, 2000). Similar argument can be found in a study by Mahsud et al. (2010, p. 564) who maintained that “leaders with high empathy are more able to recognize when different relations behaviors are relevant.” Thus, an empathic communication style likely helps leaders to inquire more deeply into the views and needs of their followers and to achieve a better understanding of the specific topics that need more explanation during appraisal interviews. Thereby, creating a shared understanding of the performance appraisal task at hand and the need for change.

Second, an empathic communication style might also be beneficial in terms of how employees perceive the actual task of performance evaluation. Instead of receiving performance evaluation as a threat (e.g., Beer, 1981; Clifton, 2012; Kay, Meyer, & French, 1965), employees who have a performance discussion with a supervisor who shows an empathic communication style might more easily perceive the appraisal interview as an opportunity to grow and to address personal goals in development and performance. In a similar vein, leaders’ expressions of empathy have been associated with a higher ability to manage follower emotions (George, 2000; Pescosolido, 2000). Again, we argue that this is especially important during appraisal interviews as the task of evaluating follower performance might easily elicit feelings of anxiety and resistance (Beer, 1981). To illustrate a negative scenario, an employee might fear negative feedback during the appraisal discussion and feel that performance appraisal is merely a mechanism by which supervisors can exert power. Likewise, that particular employee might disagree with his/her supervisor’s performance rating received in the appraisal interview and might have strong emotional reactions during the appraisal discussion. Leaders that can communicate empathy might be more likely to address and manage such strong negative emotions successfully (Humphrey, 2002; Pescosolido, 2000) and steer the appraisal interviews towards a more participative, problem-solving discussion (Beer, 1981). Hence, we argue that supervisors’ display of empathy can help employees to translate ideas for improvement into actions and to commit to developmental measures. We thus hypothesize:

  • H3: Supervisors’ empathic communication style is positively related to employees’ intentions to change.

On a conceptual level, leader empathy is frequently linked to strong social skills that help leaders to create and cultivate cooperative interpersonal relationships with their followers (Yukl, 2010). For instance, Burch et al. (2016, p. 180) recently proposed in their review on the role of empathy in organizations that empathy is an “important social construct that can create bonds between individuals or build barriers when empathic behavior is not presented at the appropriate times.” Transferred to the appraisal interview context, we argue that a supervisor’s empathic communication style can help to strengthen the interpersonal relationship with his/her employee. In other words, having an appraisal interview with a supervisor who communicates more empathically will likely prompt the respective employee to evaluate the supervisor as more pleasant and likeable. Indeed, previous empirical research showed that leader empathy is positively related to employees’ perceptions of strong leader-member exchange relationships (Mahsud et al., 2010). Based on this line of reasoning, we hypothesize:

  • H4: Supervisors’ empathic communication style is positively related to employees’ ratings of supervisor likeability.

In summarizing our proposed effects, we subsequently test the cognitive (in terms of fostering employees’ intentions to change) and emotional (in terms of increasing liking of the supervisor) impact of supervisors’ empathic communication style while considering the lower-level fine-grained verbal mimicry process that build the foundation of empathic communication in the first place. Based on the theoretical notions leading to Hypotheses 1–4, we therefore derive the following indirect effects hypotheses:

  • H5: Language style matching has an indirect effect on (a) employees’ intentions to change and (b) employees’ ratings of supervisor likability via supervisors’ empathic communication style, respectively.

  • H6: The stability of language style matching has an indirect effect on (a) employees’ intentions to change and (b) employees’ ratings of supervisor likability via supervisors’ empathic communication style, respectively.

Method

Research Context

Data were gathered in a large German production company as part of a larger data collection effort. The participating company implemented annual appraisal interviews as part of their HR strategy prior to our data gathering. Participation was voluntary, and both supervisors and their employees were asked for their consent. All appraisal interviews were audiotapend, and each participating supervisor contributed just one interview (i.e., 48 unique dyads). 

The appraisal interviews were based on semi-structured guidelines and covered two overarching topics, namely, performance evaluation and development planning. These two topics were tied to two different interview phases. During the first part of the interview, employees’ past performance was evaluated along a set of predefined categories (e.g., work quality). Employees had access to the rating form through the organizations’ intranet and could thus prepare for the interview session. The second part of the appraisal interview was focused on the employees’ future development within the organization and was based on a 3-year time frame. Supervisors and employees discussed developmental goals, promotional opportunities, and training needs.

Participants

Our sample included a total of 48 audiotaped dyadic appraisal interviews (N = 48 supervisors and N = 48 employees). All participants worked within the fields of engineering. Four out of the 48 supervisors were female (91.67% male), which corresponds to the involved industry’s average. Supervisors’ mean age was 41.38 years, ranging from 27 to 56 years (SD = 7.70) and their organizational tenures ranged from 3 to 36 years, with an average of 19.47 years (SD = 7.27). On average, supervisors were responsible for 27.37 employees (SD = 17.03) and had 7.42 years of experience in their current leadership position (SD = 5.49). Mirroring the gender distribution of the supervisors, the vast majority of the employees in our sample were male (91.67%). Employees’ average age was 40.40 years, ranging from 20 to 58 years (SD = 9.52). On average, employees’ organizational tenure was 19.56 years, ranging from 1 to 36 years (SD = 9.43). There were no female-female dyads in our sample.

Language Style Matching

We measured supervisors’ and employees’ language style matching using the Linguistic Inquiry and Word Count (LIWC), which is a computerized dictionary-based text analysis program (Pennebaker et al., 2015; Pennebaker, Francis, & Booth, 2001). Prior to analyses, all interaction data had to be transcribed. To make sure that the interviews were transcribed in a similar fashion, we developed a set of transcription rules in line with recommendations by Pennebaker et al. (2015). For example, abbreviations were written out, minimal verbal expressions such as “uh-huh” were interpreted as “yes” or “no” depending on the context, and transcripts were anonymized by replacing names with a filler word (i.e., XXXX). Next, supervisor and employee transcripts were converted into separate text files to allow separate analyses for each conversational partner.

The LIWC differentiates a variety of categories including linguistic, psychological, and topical categories by analyzing a certain text on a word-by-word basis, comparing each word in a given text file to an internal dictionary (Pennebaker et al., 2015). The program then calculates the percentages of total words accounted for by each category. For the purpose of the present study, only variables relevant to our hypotheses were included. In particular, we focused on supervisors’ and employees’ use of function words based on the German dictionary developed by Wolf et al. (2008). The German dictionary calculates five basic-level function word categories: personal pronouns (which can be further differentiated into first-person, second-person, and third-person pronouns), negations, assent, articles, and prepositions. An overview including examples for each category is shown in Table 1. The percentages of function words used by an individual describe his/her language style (Niederhoffer & Pennebaker, 2002; Pennebaker & King, 1999). To attain a measure of language style matching among both conversational partners, we used the formula proposed by Ireland and Pennebaker (2010) which is frequently used to assess language style matching in natural conversations (e.g., Ireland et al., 2011). We initially calculated language style matching scores for each function category separately before deriving an average measure of language style matching across all categories to yield a composite score for each dyad. As a result, language style matching is essentially a dyadic index. The formula to assess language style matching is as follows (articles are used as an example):

$$ {\mathrm{LSM}}_{\mathrm{article}}=1-\left(\left(|{\mathrm{article}}_1-{\mathrm{article}}_2|\right)/\left({\mathrm{article}}_1+{\mathrm{article}}_2+0.0001\right)\right) $$
Table 1 Function word categories analyzed in language style matching
Full size table

In the denominator, 0.0001 is added in order to prevent empty sets that can occur if the value for the particular category is zero in both transcripts (i.e., neither supervisor nor employee used that category).

To capture fluctuations in language style matching between supervisors and their employees across each appraisal interview, we sought a time-specific measure of language style matching. Thus, instead of measuring language style matching based on the entire transcript of each conversation, we divided each transcript into smaller interaction episodes. In particular, we divided the conversational flow into ten consecutive supervisor and employee utterances each (i.e., nine talk turns). Previous studies (Lord et al., 2015; Taylor & Thomas, 2008) have chosen an even more fine-grained segmentation by analyzing each adjacent talk turn among two conversational partners. However, we found in our data that a lot of talk turns were characterized by just a few words or even just a “yes.” While this represents the nature of natural conversations, it might probe problems for statistical analyses as word count-based text analysis methods yield less reliable results at lower word counts (Ireland & Henderson, 2014). As a rule of thumb, analyses on texts with less than 50 words should be interpreted with a grain of salt as stated on the LIWC website. Our segmentation allowed us to analyze on average 97 employee words and 234 supervisor words per interaction episode. As a coefficient of variability, we calculated the relative standard deviation of language style matching across each appraisal interview (see also Taylor & Thomas, 2008).

Supervisors’ Empathic Communication Style

Supervisors’ empathic communication style was assessed by two external raters using an adapted version of the Rating Scale for the Assessment of Empathic Communication in Medial Interviews (REM; Nicolai, Demmel, & Hagen, 2007). This rating scale demonstrated good psychometric properties in previous studies and proved to be a reliable and valid instrument for assessing empathy during conversations (Nicolai et al., 2007). The REM comprises nine items of which six items measure empathic communication. The additional three items measure confrontational behavior. The REM was originally designed to assess empathic communication in dyadic interactions among physicians and patients. Hence, we adapted the wording of the items by replacing “physician” with “supervisor” and “patient” with “employee,” respectively. Sample items include “Did the supervisor show understanding of the employee’s point of view?”, “Did the supervisor try to put him/herself in the position of the employee?”, and “Did the supervisor show interest in the employee’s opinion?”. All items were answered on a 7-point Likert-type scale. The two endpoints of each item were described as behavioral terms such as indicating that the supervisor showed (1) no interest or (7) a lot of interest. We provided the raters in our study with additional behavioral anchors to facilitate the rating process (e.g., “frequently cut the employee off” vs. “frequently asked the employee to express his/her opinions”).

Raters were two trained psychology students (one graduate student and one undergraduate student) who independently rated the supervisors’ empathic communication style in all 48 appraisal interviews by listening to the audio recordings of the appraisal discussions and using a paper-pencil template. In order to assess whether the two raters provided consistency in their ratings of empathic communication across the different supervisors (i.e., inter-rater reliability), we calculated two-way random intraclass correlations (ICCs) with absolute agreement (average measures) between both raters (McGraw & Wong, 1996). As described earlier, each appraisal interview comprised two phases (i.e., performance evaluation and development planning). Thus, the two raters assessed supervisors’ empathic communication at two times during each appraisal interview. The ICCs for the first interview phase yielded a value of .94 (p < .001) and the ICC for the second phase yielded a value of .85 (p < .001), indicating excellent agreement among both raters (Cicchetti, 1994). Average internal consistency yielded a value of .81 (Cronbach’s α). In order to explore possible differences in the extent of supervisors’ empathic communication style across both interview phases, we calculated t tests for dependent samples. Results showed no significant differences between both interview phases (t(47) = .21, p = .83; t(47) = .78, p = .44, for the first and second rater, respectively), indicating that the extent of expressed supervisor empathy was rather stable across the appraisal interview sessions. Thus, we used the mean empathy score across both interview phases. Following recommendations from the behavioral observation literature (Yoder & Symons, 2010), we also used the mean empathy rating of both raters in all following analyses.

Self-Report Measures

After the appraisal interviews, employees were asked to complete a self-report questionnaire. Five employees did not fill out this questionnaire (10.4% missing data) which was mostly due to time constraints (e.g., they had to rush to the next meeting). One more employee did not provide the measure of supervisor likeability (12.5% missing data for supervisor likeability).

Intentions to Change

Intentions to change were operationalized as the number of developmental measures that employees identified after the appraisal discussion. Employees were asked to provide a list of developmental measures based on the following two questions: What measures have you agreed upon during the appraisal interview? What do you want to do differently after the appraisal interview? We chose to add this second question in order to allow employees to also record more personal developmental goals. Examples included specific training needs (e.g., participating in a robotic workshop) but also less tangible goals (e.g., passing on information to the team more quickly). We counted the total number of developmental measures for each employee.

Supervisor Likeability

We measured employees’ ratings of supervisor likeability using a nine-point semantic differential scale by Pelz and Scholl (1990). The scale comprised three bipolar items (i.e., pleasant—unpleasant; likeable—dislikeable; attracting—repelling). Employees were asked to refer directly to the appraisal interview session when evaluating their supervisor’s likeability (in contrast with overall supervisor likeability). Cronbach’s α yielded a value of .78.

Analytic Strategy and Control Variables

We specified a path model in MPlus version 7 (Muthén & Muthén, 2012) using maximum likelihood estimation with robust standard errors (MLR). Employing a path model allowed us to test all hypotheses, including indirect effects, simultaneously while also estimating the correlation between the two outcomes and controlling for possible confounding variables.

Due to our small sample size at the dyadic level (N = 48 interviews), we chose to only control for those variables that were significantly associated with our focal study variables in preliminary analyses (see Table 2). In particular, we controlled for the overall length of the appraisal interview discussion which was related to ratings of supervisors’ empathic communication and also showed correlations with our two linguistic measures. Moreover, we tested for a possible confounding effect of the employees’ overall performance rating received in the appraisal interview which has been linked to interview outcomes in the past (Pichler, 2012). These overall performance ratings ranged from (1) exceeds the expectations profoundly to (5) in need of improvement. However, no employee received a rating of (5) in our sample. As the performance measure used in the participating company was not interval scaled, we clustered the employees into two similar sized and conceptually different groups (i.e., high and average performer). The performance rating was related to ratings of supervisor likeability in preliminary analyses and thus controlled for in subsequent hypothesis testing. Control variables were modeled as paths on leaders’ empathic communication style and both outcomes in the model (Kline, 2005).

Table 2 Means, standard deviations, and intercorrelations of study variables
Full size table

Results

Means, standard deviations, and correlations between all study variables are shown in Table 2. Overall, appraisal interviews amounted to more than 37 hours of leader-follower interactions, with an average interview length of about 47 min. To reveal relationships between language style matching among supervisors and their employees and supervisors’ empathic communication style, we analyzed a total of 358,586 words. Precisely 73.73% of the words were recognized and categorized by the LIWC software which exceeds the average 66.15% recognition rate reported in previous studies using the German dictionary (Wolf et al., 2008). Language style matching yielded an average of .60, ranging from .29 to .78 (SD = .10). On average, language style matching was not stable but varied in each appraisal interviews as expressed by a mean relative standard deviation of .33, ranging from .11 to .87 (SD = .14). To illustrate changes in language style matching over time, Fig. 1 shows the temporal dynamics of two sample appraisal interviews.

Fig. 1
figure 1

Language style matching among supervisors and employees over the course of two sample appraisal interviews. Interview 1 lasted for approximately 35 min in total; variability in language style matching was .13. Interview 2 lasted for approximately 37 min in total; variability in language style matching was .87

Full size image

As stated above, we specified a path model in MPlus and tested all hypotheses simultaneously while controlling for interview length and the employees’ overall performance rating received in the interviews. This model was fully identified and thus showed perfect fit to the data. We report standardized path coefficients. In support of H1, we found that the amount of language style matching between supervisors and their employees was strongly related to supervisors’ empathic communication style (β = .72, p < .001, two-tailed). Contrary to our expectations, changes in languages style matching over time (i.e., language style matching variability) were not indicative of supervisors’ empathic communication style (β = .31, p = .15, two-tailed). Thus, H2 was rejected.

Our second set of hypotheses stated that supervisors’ empathic communication style was positively related to employees’ intention to change (H3) and to employees’ perceptions of supervisor likeability (H4). Lending support to H3, we found that supervisors’ empathic communication was marginally significantly related to employees’ intentions to change (β = .23, p = .09, two-tailed). Results further revealed that supervisors’ empathic communication style was positively related to employees’ perceptions of supervisor likeability (β = .48, p < .01, two-tailed), supporting H4. Concerning our control variables, we found that employees’ overall performance rating was marginally significantly related to the employees’ perceptions of supervisor likeability (β = − .26, p = .06, two-tailed). Omitting the control variables from our model did not meaningfully change the hypothesized effects.

We chose to employ one-tailed tests of significance for testing indirect effects. We saw this as justified because of the low statistical power that indirect effects generally have (Hayes, 2013) and because of our small sample size. As recommended (Hayes, 2013), we report unstandardized coefficients for indirect effects. Our first indirect effects hypothesis focused on the role of language style matching whereas our second indirect effects hypothesis zoomed in on the role of language style variability. The indirect effect of language style matching on employees’ intentions to change via supervisors’ empathic communication style (H5a) was 2.56. The respective 90% confidence interval included zero (− 0.33, 5.46). We thus rejected H5a. The 90% confidence interval of the indirect effect of language style matching on supervisor likeability via supervisors’ empathic communication style (H5b) did not include zero (1.25, 7.54; indirect effect = 4.39). Thus, H5b was supported. The indirect effect of language style matching variability on employees’ intentions to change via supervisors’ empathic communication style (H6a) was 0.80. The respective 90% confidence interval included zero (− 0.43, 2.03). Hence, H6a was rejected. Finally, the indirect effect of language style matching variability on supervisor likeability via supervisors’ empathic communication style (H6b) was 1.37. Again, the corresponding 90% confidence interval included zero (− 0.23, 2.97) and H6b was rejected. In the presence of all specified relationships, employees’ intentions to change and ratings of supervisor likeability were moderately correlated (r = .26, p < .05, two-tailed). Overall, the model explained 24% of the variance of intentions to change and 31% of the variance of supervisor likeability.

Due to a high correlation between our two linguistic measures (r = −.84, p < .001, two-tailed) and the pattern of results from hypothesis testing, we decided to run an additional path model with only language style matching as a focal predictor. A summary of the model is shown Fig. 2. In this simpler model, language style matching was again positively related to supervisors’ empathic communication style (β = .45, p < .01, two-tailed). Empathic communication, in turn, was significantly related to both employees’ intentions to change (β = .29, p < .05, two-tailed) and supervisor likeability (β = .48, p < .01, two-tailed). The 90% confidence intervals of the two indirect effects did not include zero (0.18, 3.89) and (0.72, 4.80), respectively. We see this as additional support for an influential effect of language style matching on our two outcome measures via supervisors’ empathic communication. To further aid interpretation, we also computed a post hoc power analysis for the two indirect effects using the MedPower application (Kenny, 2017). For the indirect effect of language style matching on employees’ intentions to stay via supervisors’ empathic communication style, statistical power reached .55 (α set to .10). The power of the indirect effect of language style matching on supervisor likeability via supervisors’ empathic communication style was considerably stronger at .89 (α set to .10), which exceeds the recommended .80 level (Cohen, 1988).

Fig. 2
figure 2

Effects of language style matching on employees’ intention to change and supervisor likeability via supervisors’ empathic communication style. Standardized coefficients are shown. Control variables (interview length, employees’ final performance rating) are omitted for the sake of clarity. *p < .05; **p < .01. (two-tailed)

Full size image

Additional Analyses

To gain further insights into the micro-level behavioral manifestation of supervisor empathy, we also explored how supervisors’ empathic communication style relates to supervisors’ specific use of functions words. Results revealed that ratings of supervisors’ empathic communication style were positively correlated with supervisors’ percentage use of the function word category “assent” (r = .48, p < .001, two-tailed). Thus, supervisors who used more affirmative words such as “yes” and “ok” were perceived as more empathic. In addition, ratings of supervisors’ empathic communication style were negatively related to supervisors’ percentage use of second-person pronouns (“you”; r = −.29, p < .05, two-tailed). This shows that supervisors were considered less empathic when they frequently and directly addressed their conversational partner.

Finally, and inspired by a recent study showing that pronoun use reflects the speaker’s position in the social hierarchy at hand (Kacewicz, Pennebaker, Davis, Jeon, & Graesser, 2014), we inquired further into supervisors’ and employees’ different and/or similar use of personal pronouns. Specifically, previous findings suggest that people with higher status, in our case the supervisors, refer less to themselves and therefore use fewer first-person singular pronouns (“I”). Instead, people of higher status are said to use more second-person (“you”) and first-person plural (“we”) pronouns, thereby shifting their attentional focus toward other people. We calculated t tests for dependent samples for every pronoun category. Results are presented in Table 3. Findings showed that supervisors did indeed use significantly more second-person (“you”; t(47) = 15.63, p < .001) and first-person plural (“we”; t(47) = 2.62, p = .01) pronouns in comparison to their employees. Employees, on the other hand, used significantly more first-person singular pronouns (“I”; t(47) = − 9.75, p < .001) than their supervisors. In sum, these additional findings point at nuanced differences in supervisors’ and employees’ communication styles.

Table 3 Differences in the use of personal pronouns among supervisors and employees
Full size table

Discussion

Although the leadership literature suggests that empathy is an important leadership skill (e.g., Fleishman & Salter, 1963; Humphrey, 2002), empirical research on the role of empathy for establishing successful leader-follower interactions is scarce. This study addressed this shortcoming in the literature by (1) by shedding light on the micro-level behavioral manifestation of leader empathy and by (2) exploring the effectiveness of expressed leader empathy during appraisal interviews.

Theoretical Implications

Our findings build and extend on previous research on appraisal interviews and leadership theory in several ways. First, our findings align with and extend previous research on the fine-grained behavioral manifestation of empathy (e.g., Chartrand & Lakin, 2012; Lord et al., 2015; Meyer et al., 2016). Specifically, the results provide empirical evidence for a link between ratings of empathic communication and language style matching among supervisors and their employees: appraisal discussions characterized by higher levels of verbal mimicry, such that both conversational partners adapted their word use to one another, prompted external observers to perceive the respective supervisors as being more empathic. To the best of our knowledge, this is only the second study (cf. Lord et al., 2015) that relates language style matching to ratings of empathy and it is the first study that explores this relationship in the context of leadership and followership. By showing that language style matching is in fact a behavioral marker of expressed empathy, our study builds important knowledge about the underlying mechanisms contributing to leader empathy and its positive effects. Language style matching is frequently seen as a nonconscious or automated form of interpersonal congruence and an expression of verbal and cognitive alignment (Ireland et al., 2011; Ireland & Pennebaker, 2010; Niederhoffer & Pennebaker, 2002). This alignment helps interlocutors to build a common understanding of the situation and tasks at hand by blurring self/other distinctions (Hove & Risen, 2009). Taken together, language style matching likely helped supervisors to take their employees’ points of view into account and to respond appropriately to their needs, as characterized by a strong empathic communication style. As a result, supervisors were more likely to create common ground during the appraisal discussion.

Second, our study builds knowledge about the temporal nature of language style matching. Our fine-grained behavior-based analysis allowed us to trace the temporal progression of language style matching across the entire appraisal discussion. In contrast to our predictions, we did not find that more stable levels of language style matching were indicative of ratings of supervisors’ empathic communication style. Consequently, we also did not find support for indirect effects of language style matching variability on our outcome variables via supervisors’ empathic communication style. Overall, the findings suggest that a certain amount of fluctuation in language style matching might be normal or characteristic of natural conversations. Further research is needed that explores the ebbs and flows of language style matching across interaction episodes and that can provide guidelines according to which changes in language style matching are rather normal versus dramatic (cf. Taylor & Thomas, 2008). On a descriptive level, changes in language style matching were more pronounced in some appraisal discussion than in others, and language style matching showed different growth trajectories across time as exemplified in Fig. 1. These observations suggest that the appraisal discussions in the current study were characterized by different process dynamics, which is in line with previous research on communication dynamics in appraisal interviews (Meinecke, Klonek, & Kauffeld, 2016). Future research should thus explore further if there might be linear, rhythmic, or discontinuous shifts in language style matching over time (see also Kozlowski, 2015) and if such shifts relate to supervisors’ empathic communication or other communication measures such as conversational engagement (Ireland & Henderson, 2014). In a similar vein, future research might explore if supervisors’ empathic communication exhibits different growth trajectories throughout a conversation. For example, high levels of empathic communication might be especially important at the beginning of an appraisal discussion as a positive conversation starter (i.e., “setting the tone”). It remains to be explored if a behavioral rating approach as applied in this study is sensitive enough to capture such subtleties in communication style.

Third, our findings highlight the functional role of supervisors’ expressed empathy during appraisal interview in terms of increasing employees’ intentions to change and in terms of building affiliation between supervisors and employees. In line with arguments found in the counseling literature (Miller & Rollnick, 2013; Rogers, 1975), our findings suggest that empathy helps in establishing strong interpersonal relationships and in facilitating change. Importantly, we focused on supervisors’ expressed empathic communication style instead of more internal accounts of leader empathy (Neumann, Chan, Boyle, Wang, & Westbury, 2015). Hence, our findings show that leaders who behave empathically can initiate change processes on part of the employee and heighten perceptions of likeability. Against the backdrop of the appraisal interview context with its main purpose of increasing employees’ future performance (DeNisi & Murphy, 2017; DeNisi & Pritchard, 2006), these findings build a case for leaders’ empathic communication as a powerful tool for change management. In this sense, our study relates to the growing literature base of leadership empathy (e.g., Kellet et al., 2002, 2006; Mahsud et al., 2010; Sadri et al., 2011) but expands on previous research by showing how change is evoked through expressed empathy.

Fourth, our supplementary analyses add to research on the role of pronoun use in social hierarchies (Cassell, Huffaker, Tversky, & Ferriman, 2006; Kacewicz et al., 2014; Sakai & Carpenter, 2011). In particular, we found subtle differences in the mean percentage use of personal pronouns between supervisors who used significantly more second-person (“you”) and first-person plural (“we”) pronouns and their employees who used significantly more first-person singular pronouns (“I”). One explanation for this contrasting pattern can be found in the literature on attentional states (Chung & Pennebaker, 2007; Tausczik & Pennebaker, 2010). The use of first-person pronouns has been shown to covary with increased levels of self-attention (Davis & Brock, 1975; Rude, Gortner, & Pennebaker, 2004). A heightened use of second-person pronouns, on the other hand, signals that individuals focus their attention on someone else (i.e., outward attention; Kacewicz et al., 2014). Transferred to the specific research context of our study, these findings suggest that employees were mainly focused on their own performance evaluation and their own plans for future development. Supervisors on the hand, frequently and directly addressed their employees during the appraisal discussion. A closer look at the mean percentage use illustrates that employees used about seven times as many first-person singular pronouns as second-person pronouns. Supervisors used both types of pronouns at about the same rate. Interestingly, supervisors’ percentage use of the “you” category was negatively correlated to their ratings of empathic communication. This might indicate that focusing too much on their conversational partner comes at the cost of self-reflection such that supervisors come off as too confrontational or dominant (e.g., “You must change this”). As a limitation of this interpretation, however, it should be mentioned that we did not examine the syntactic context in which each individual pronoun was embedded. That is, second-person pronouns can also be used in an affirmative way (e.g., “You are great”).

Finally, our study also makes an important methodological contribution. Our findings show that techniques from the realm of observational research methods that focus on supervisors’ and employees’ actual communication—rather than post-hoc perceptions of leader empathy—are a promising extension to more traditional approaches used in leadership research in general and research on leader empathy in particular (see also Lehmann-Willenbrock & Allen, 2017). The computational linguistic approach used in this study takes advantage of specific characteristics of language and discovers verbal mimicry based on transcripts of natural conversations (Chung & Pennebaker, 2007; Tausczik & Pennebaker, 2010). As such, it constitutes an unobtrusive approach (Hill, White, & Wallace, 2014) to measuring leadership and followership influence and can help to overcome problems of social desirability and reactivity in organizational and leadership research. Taken together, linguistic analysis as applied in the current study provides a new angle to the study of leader-follower interactions in organizations.

Practical Implications

The findings of the current study show that supervisors’ expression of empathy during appraisal interviews is an important leadership behavior. Supervisors’ consideration for the employees’ point of view increased employees’ readiness to commit to new tasks and goals. In addition, supervisors’ expressed empathy made them more likeable. From an intervention perspective, this finding underscores the importance of training leaders to engage in more empathic behaviors. While expressing empathy may be easier for some leaders than for others (see also Sadri et al., 2011), our findings showed that most leaders are capable of expressing at least medium amounts of empathy. From this follows that leaders could be coached in deepening their empathic communication skills. In fact, previous research from the clinical context suggests that empathic communication skills can be increased through training (Miller & Mount, 2001). Thus, trainings for leaders to steer more effective performance appraisal interviews could incorporate a component focused on increasing their empathic communication skills. Likewise, organizations might focus on selecting and promoting leaders who already have a certain degree of empathy.

Focusing more deeply on the behavioral markers of empathy, our findings further showed that supervisors’ and employees’ similar use of language style was linked to supervisors’ ratings of empathic communication. Hence, one way to build mutual alignment and rapport among leaders and their followers would be through language style matching. Because language style matching is typically conceptualized as an automatic process (e.g., Ireland & Henderson, 2014), we would not recommend trying to teach leaders to consciously mimic the verbal behavior of their conversational partner in order to come across as more empathic. However, there is empirical evidence that language style matching increases when individuals pay close attention to one another (Tausczik, 2012). We noticed in our sample that some supervisors tended to get lost in the interview guidelines, reading out lengthy rating definitions and paying more attention to the documents than to their conversational partner. Thus, we suggest for interview templates to be simplified or include specific prompts that guide supervisors to focus on the employees’ utterances and views. In terms of language use, our findings further suggest that leaders should attend to how they address their conversational partner during the appraisal discussion. A high proportionate use of second-person pronouns (“you”) was associated with lower ratings of empathic communication. Thus, leaders should be encouraged to underline their performance ratings with I-messages so they do not come off as too directive. In addition, leaders could be sensitized to using a more inclusive language style by using more we-references (see also Weiss, Kolbe, Grote, Spahn, & Grande, 2017). Trainings to steer more effective appraisal interviews could specifically incorporate such communication exercises.

Limitations and Future Research Directions

The following limitations deserve mentioning. First, our field study was based on data from a single organization. Even though the participating organization is a large international corporation, the specific organizational environment (e.g., the particular performance criteria that were applied, the structured interview guideline used by the organization, or the merit pay component) restricts the generalizability of our findings. Future research should explore the functionality of supervisor empathy in appraisal interviews from different industries. In this sense, it might also be worthwhile to explore if supervisor empathy is more important for some developmental measures than for others (e.g., especially for more personal developmental goals). In addition, our findings should be replicated beyond the appraisal context. Many organizations are shifting away from formal performance appraisal toward more regular and informal check-ins (Kenny, 2016; Pulakos, Mueller Hanson, Arad, & Moye, 2015). Traditional appraisal interviews, as in the present study, are characterized by a rather steep hierarchy and pronounced power differential among supervisors and their employees in comparison to other organizational interaction contexts such as regular team meetings or informal chit chat. Future research should explore if more informal conversations between leaders and followers might lead to higher levels of language style matching and empathic communication.

Another limitation concerns our small sample size. Future research should try to replicate our findings using larger and more diverse samples. In this regard, another limitation that may warrant future research is whether there are gender differences in leaders’ expressed empathy. Most of our study participants were male, which is typical for the industry involved in our study. However, previous research showed that women tend to have better empathy than men (Christov-Moore et al., 2014; Schulte-Rüther, Markowitsch, Shah, Fink, & Piefke, 2008). It remains to be explored how these gender differences in empathy translate to the organizational context (see also Burch et al., 2016), and especially to leader-follower interactions.

A final limitation concerns our choice of measures. We assessed supervisors’ empathic communication using ratings by external observers because we were especially interested in more objective accounts of supervisor empathy and real-time expressions of empathy. Future research might supplement this approach by including additional self- and other-report measures of empathy by both supervisors and their employees. This way, researchers could explore the extent to which supervisors thought they did understand their employees’ point of view and the extent to which employees’ felt understood (i.e., inward empathic experiences; Duan & Hill, 1996). Moreover, additional self-reported measures of empathy might help to capture both state and trait components of supervisor empathy.

Following up on this thought, future research should explore if supervisors’ empathic communication style is rather stable across different conversations with different employees or if supervisors’ empathic communication is also contingent upon employee characteristics. How familiar the employee is with his or her supervisor (i.e., tenure with one’s supervisor) could be an important variable in this regard and should be considered in future studies. The link to language style matching suggests that it is easier for some dyads than for others to achieve mutual interpersonal alignment. However, future research is needed to more fully understand the role of supervisor empathy and its link to language style matching and possibly other forms of behavioral mimicry (e.g., mimicry based on motion sensors; Meyer et al., 2016). Overall, research on language style matching—especially in natural conversations—is still in its infancy and requires further substantiation.

Conclusion

The findings of this field study allow at least three conclusions carrying theoretical, practical, and methodological implications. Theoretically, our findings extend past research on the behavioral manifestation and functionality of leader empathy by showing that empathic communication allows supervisors to steer more positive conversations with their employees in terms of initiating change processes and building affiliation. Practically, our findings provide a starting point for developing leadership interventions focusing specifically on supervisors’ verbal conduct. Methodologically, linguistic analyses based on recordings of actual appraisal interviews allowed us to explore fine-grained verbal mimicry processes among supervisors and their employees. Taken together, we suggest linguistic analyses—albeit time-consuming—to be a viable addition to the leadership researcher’s toolkit. The current study sets an example of how this field can contribute to our understanding of leader-follower interaction processes at a linguistic level.