carrying too heavy a load? the communication and miscommunication...

CARRYING TOO HEAVY A LOAD? THECOMMUNICATION AND MISCOMMUNICATION

OF EMOTION BY EMAIL

KRISTIN BYRONSyracuse University

Despite advice to avoid doing so, email senders intentionally and unintentionallycommunicate emotion. Email characteristics make miscommunication likely, and Iargue that receivers often misinterpret work emails as more emotionally negative orneutral than intended. Drawing on the computer-mediated and nonverbal communi-cation, emotion, and perception literature, I introduce a theoretical framework de-scribing what factors make miscommunication most likely, how emotional miscom-munication affects organizations, and how employees can improve the accuracy ofemotional communication in emails.

Employees are increasingly likely to use andprefer electronic mail (email) to communicatewith coworkers, customers, and other col-leagues. The proliferation of email for businesscommunication is likely due to some advan-tages, such as flexibility and asynchrony, it hasover other communication media. Conse-quently, email has increased information shar-ing in organizations (Rice, 1987; Sproull &Kiesler, 1986) and has improved productivityamong employees separated in time and place(Higa, Sheng, Shin, & Figueredo, 2000).

However, the proliferation of email communi-cation has also introduced some challenges notassociated with other communication media.Research and theory suggest one likely delete-rious effect of email use is harm to workplacerelationships. Friedman and Currall (2003) haveargued that the characteristics of email increasethe likelihood of conflict escalation among thosecommunicating by email. In their study of fac-ulty and staff at a university teaching and re-search institute, Sarbaugh-Thompson and Feld-man (1998) found that as email use increased,the overall volume of all forms of communica-tion decreased, mostly because of fewer “greet-ings” and other informal interactions between

coworkers. In addition, employees reported feel-ing less connected to their coworkers as theiremail use increased.

A related potential problem lies in the com-munication of emotion. For several reasons,emotions are particularly difficult to accuratelycommunicate by email. Although some have ar-gued that email should not or cannot conveyemotion, more recent theory and research sug-gest that email senders communicate emotionsto recipients, intentionally or not (Thompsen &Foulger, 1996; Walther & D’Addario, 2001).

Little theory or research exists to explain whatfactors may affect the accurate or inaccurateperception of emotion in emails. To fill this gap,I describe a theoretical framework, drawing on awide range of research and theory on such top-ics as computer-mediated communication(CMC), nonverbal communication, emotions,and perception to develop a model explainingwhy emotions are likely to be inaccurately per-ceived in email communication, which types ofinaccurate judgments are most likely, what fac-tors influence their occurrence, and what impli-cations this has for organizations.

Why should organizations care about inaccu-rate emotion perception among their employ-ees? The functions of emotion in organizationshelp to clarify the reasons. Emotions provideinformation about others and the environmentand play an important role in relationship de-velopment and group identity (Buck, 1984).Therefore, the failure to accurately communi-cate emotion, particularly positive emotions,may inhibit relationships between coworkers or

I thank Sigal Barsade, Justin Kruger, Ross Rubenstein,Batia Wiesenfeld, and the participants of the 2005 Academyof Management Organizational Behavior Junior FacultyConsortium for their helpful comments on this paper. I alsothank associate editor Loriann Roberson and three anony-mous reviewers for their constructive comments and gener-ous support.

� Academy of Management Review2008, Vol. 33, No. 2, 309–327.

309Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyrightholder’s express written permission. Users may print, download, or email articles for individual use only.

employees and their clients or customers (Bar-sade, 2002). In addition, because emotions pro-vide information to guide behavior, employeeswho inaccurately interpret others’ emotions arenot making adequately informed decisions re-garding their behavioral response. Social cogni-tion researchers argue that people need to un-derstand each other with enough accuracy todirect their actions and interactions with oneanother (Bernieri, 2001; Fiske, 1993). Althoughone could argue that employees do not rely ex-clusively on email to communicate at work, theextent to which employees rely on email is in-creasing, and its use often supplants, ratherthan supplements, other ways of communicat-ing (Hallowell, 1999; Sarbaugh-Thompson &Feldman, 1998).

The prevalence of emotional miscommunica-tion in emails at work defies exact quantifica-tion. There is some evidence to suggest thatemotions are commonly communicated—accu-rately and inaccurately—by email at work. In afocus group study all participants indicatedhaving expressed or perceived emotions inemail at work, and nearly all reported problemswith doing so (Byron & Baldridge, 2005). Evencommunicating positive emotions in emails isfar from straightforward: participants in thesame study volunteered fewer ways to expressor perceive positive, as compared to negative,emotions in email, and many of these ways werecontradictory (e.g., writing short or long mes-sages) or were identical to the cues they en-dorsed for expressing or perceiving negativeemotions (e.g., using exclamation points). Fur-ther, a small body of research examining emoti-cons—sets of typographical symbols that whenread sideways are meant to represent emotionsin emails—suggests that these symbols are notuniformly interpreted (Thompsen & Foulger,1996; Walther & D’Addario, 2001).

DEFINING THE COMMUNICATION OFEMOTION

First, it is important to clarify what I mean bythe term emotional communication and to spec-ify the overall framework that guides this paper.Although there is disagreement on the finerpoints among researchers studying emotion,emotion here refers to discrete feeling states,such as happiness, fear, and anger. Many re-searchers categorize emotions in terms of their

valence (i.e., positive, negative, or neutral); emo-tions such as happiness and joy are positive,emotions such as anger or disgust are negative,and neutral refers to the absence of either pos-itive or negative emotions. Emotions also differin intensity—the perceived strength of the emo-tion (e.g., rage and elation are higher in inten-sity than frustration and contentment). In thispaper I focus on emotions because emotions (un-like moods) are considered intentional—that is,precipitated by a specific stimulus, such as aperson or event (Frijda, 1994). Email recipientswould likely be more motivated to perceive oth-ers’ emotions (than others’ moods), because anemail sender’s emotion may be in response tothe recipient’s behavior, whereas moods are rel-atively stable within individuals and less likelyto be caused by others’ behavior.

Defining Accuracy

In this paper I am also concerned with theaccuracy of judgments regarding emotional con-tent in work emails. Several different streams ofliterature are concerned with the concept of ac-curacy in interpersonal communication—for ex-ample, research on interpersonal perceptual ac-curacy (e.g., Kenny & Winquist, 2001), nonverbalcommunication of emotion (e.g., Ekman, Friesen,& Ancoli, 1980), and communication in general(e.g., Pavitt, 1989). Some conceptions of accuracyare not relevant for the present framework. Forexample, nonverbal communication researchersoften define accuracy as agreement with an es-tablished standard based on the idea that non-verbal expressions represent a universal lan-guage (see Elfenbein & Ambady, 2002). Therelative newness of email means that no univer-sal standard of emotional communication byemail exists, rendering this criterion irrelevant.Although other definitions exist, the criterion ofaccuracy that appears most relevant to thepresent framework is to define accuracy in rela-tion to the sender’s intentions. That is, I defineaccuracy in terms of the extent to which thereceiver interprets the emotional content (interms of valence and intensity) of a messageconsistent with that intended by the sender (Ek-man & Friesen, 1969). Inaccurate judgments ofemotion then differ from the sender’s intendedemotional expression in terms of valence and/orintensity. Under this definition, then, accuracy isultimately determined by the receiver—and his

310 AprilAcademy of Management Review

or her perception of the sender’s communicationattempt (although every receiver becomes asender when responding to or sending a mes-sage). Defining accuracy is both a conceptualand methodological consideration; therefore, Ilater discuss other ways of defining and mea-suring accuracy.

Although some lament psychologists’ preoc-cupation with errors in perception (e.g., Funder,2001), the present framework considers both ac-curacy and inaccuracy. However, I focus on in-accuracy because, as I argue below, some char-acteristics of email make inaccurate judgmentslikely. This is not to say that inaccurate judg-ments are inevitable; to the contrary, I detailsteps that organizational members can take toimprove the accuracy of judgments regardingemotion communicated by email.

Berlo’s Source-Message-Channel-ReceiverModel

Berlo’s Source-Message-Channel-Receiver(SMCR) Model of communication (1960) serves asan overarching framework for the model Ipresent here. Using a psychological perspective,Berlo argued that four factors determine the ef-fectiveness of communication attempts: thesource, or sender (S), the message (M), the chan-nel (C), and the receiver (R). Senders (S) encodemessages (M) verbally or nonverbally usingtheir choice of channels (C) to receivers (R) whodecode them. Senders affect the communicationprocess because, among other things, they havedifferent communication skill levels, come fromdifferent cultures, and have different attitudestoward receivers. Messages and channels influ-ence the communication process because send-ers choose how to encode and send the mes-sage. Receivers affect the communicationprocess because they have prior informationabout or attitudes toward senders, differentcommunication skill levels, and prior beliefsbased on their sociocultural context.

Although some communication researchershave criticized the model for its simplicity andits assumption of linearity in the communicationprocess, it has proven useful in explaining com-munication in other studies (e.g., Pavitt & John-son, 2002) and remains one of the most popularmodels of communication. Communication re-searchers have praised the model for, amongother things, acknowledging that the sender and

receiver’s relationship and social context affectthe communication process— both of whichprove useful in the present model.

EFFECT OF CHANNEL ON EMOTIONPERCEPTION IN EMAILS

I adopt here the convention of some research-ers to refer to channel as the medium used (e.g.,email, phone) and to use these terms inter-changeably (e.g., Moenaert & Souder, 1996). Sev-eral theorists have proposed that the objectivecharacteristics of various media influence theaccuracy of and effort associated with deliver-ing different types of messages. Proponents ofrational choice models such as information rich-ness theory (Daft & Lengel, 1986) argue that dif-ferent media have inherent characteristics thatconstrain their use and make them more or lessappropriate for a given situation or message.According to this theory, communication mediavary in their information richness—that is, theability to allow rapid feedback, enable multiplecues, use natural language, and establish a per-sonal focus (Lengel & Daft, 1988). Whereas face-to-face communication is highest in richness,email is leaner because fewer cues are avail-able and because feedback can be delayed andis less obtainable. With leaner media, commu-nicators have greater difficulty resolving ambi-guity and facilitating understanding. Therefore,using lean media such as email for ambiguousmessages is less effective and efficient—it in-creases the likelihood of inaccurate communica-tion and more time is then spent resolving themiscommunication.

Communicating Emotion with Few Cues andLittle Feedback

One of the characteristics of email that com-plicates the communication of emotion is its rel-ative lack of cues. Email communication is textbased, which means that communicating ver-bally is easier than communicating nonver-bally. Relative to email, face-to-face communi-cation provides more nonverbal cues, such asfacial expressions, paralanguage, and socialcontext cues (Owens, Neale, & Sutton, 2000;Sproull & Kiesler, 1986). Because emotions tendto be expressed and perceived nonverballyrather than verbally (Ekman et al., 1980), therelative dearth of cues in email, compared with

2008 311Byron

some other channels, makes the miscommuni-cation of emotion in emails more likely.

This is not to imply that nonverbal cues areentirely absent from emails. On the contrary,emails may contain nonverbal cues that haveparallels in face-to-face communication, such asemoticons to symbolically convey emotion, as-terisks to provide emphasis, or all capital lettersto indicate emotional intensity. However, de-spite the availability of some nonverbal cues,research indicates that their purposeful use re-mains infrequent (Rezabek & Cochenour, 1998;Witmer & Katzman, 1997). Further, many authorsof articles on “netiquette”—etiquette related tointernet use—advise employees to use spar-ingly, or not at all, cues such as emoticons inwork-related emails because their use may ap-pear too casual and unprofessional (Calem,1995; Jett, 2005).

In addition to the relative lack of cues, the factthat communication partners are not copresentfurther complicates the communication of emo-tion by email by limiting and delaying feed-back. In contrast, communication partners com-municating face to face are copresent, allowingsenders to get immediate feedback and recipi-ents to seek clarity by asking questions or re-peating information. Because the delay in feed-back restricts the ability of communicationpartners to resolve ambiguity, the miscommuni-cation of emotion by email may be more likelythan in face-to-face communication. This is es-pecially true for the communication of emotion,because determining the emotional state of oth-ers is often ambiguous (Blanchette & Richards,2003; Nowicki & Duke, 1994). Indeed, the lack ofimmediate feedback—and other characteristicsof email—led Daft and Lengel to prescribe thatCMC, such as email communication, be used for“very simple or unequivocal” messages and notfor messages that are “ambiguous, emphatic, oremotional” (1986: 57).

Neutrality Effect

The characteristics of email may increase thelikelihood that receivers perceive emails in-tended to convey positive emotion as more emo-tionally neutral than senders intend, which Iterm here the neutrality effect. First, the reducedavailability of cues and feedback may makeemail communication in general less physiolog-ically arousing than face-to-face interaction

(McKenna & Bargh, 2000; cf. Kiesler, Zubrow,Moses, & Geller, 1985). Using an evolutionaryperspective, Kock (2005) has argued that the lackof such cues as facial expressions or vocal tonein email render it less arousing than communi-cation that permits these cues. Research usingfunctional magnetic resonance to examinebrain wave activity or other measures ofarousal, such as pulse rate or palm sweating,has shown that facial expressions, direct gaze,and other nonverbal cues are physiologicallyarousing, and their absence tends to decreasearousal during communication (Critchley et al.,2005). Although some emails are very arousing,regular email users generally report that the actof sending and receiving emails at work isduller and less stimulating than engaging inface-to-face communication (see Kock, 2005, for areview).

Second, emotional intensity is difficult to ac-curately convey in email owing to the relativelack of cues. In face-to-face communication, vo-cal tone, such as raising one’s voice, or the in-tensity of a facial expression conveys how in-tensely a person may be experiencing anemotion. However, research suggests that accu-rately conveying the intensity of positive emo-tion in email is difficult. Email recipients ratedjokes communicated by email as less humorousthan the email senders expected (Kruger, Epley,Parker, & Ng, 2005). The authors attributed thesefindings to egocentrism—”the inherent difficultyof moving beyond one’s subjective experience ofa stimulus and imagining how the stimulusmight be evaluated by someone who does notshare one’s privileged perspective” (Kruger etal., 2005: 926). Email senders “hear” their intona-tions indicating emotional intensity when writ-ing the email and fail to consider that thosereading the email cannot similarly “hear” them.For example, an employee may intend to ex-press positive enthusiasm in an email about anupcoming change, but the email recipients mayperceive it as more neutral than intended, con-veying information about the change ratherthan excitement about it. Further, with less feed-back available, such as a wrinkled brow orblank stare, senders may not find out that theirrecipients failed to understand an email as in-tended.

Third, research conducted in organizationalsettings has shown that emails tend to be taskoriented (e.g., Ku, 1996; Sarbaugh-Thompson &


Feldman, 1998), perhaps because of the relativedifficulty of communicating emotional content.In a survey of company email use, a managerdiscussed his tendency to be serious, perhapsbecause of the difficulty of conveying positiveemotion: “With email I find myself answeringw/o all the kindness necessary to keep peoplehappy with their job” (Markus, 1994: 139). Thetendency for emails between coworkers or be-tween clients and customers to be serious, taskoriented, and impersonal may mean that recip-ients come to expect less positive emotionalcontent in work emails (Lea & Spears, 1991).

Because the emails they receive may tend tobe (or at least appear to be) emotionally neutral,employees may develop a schema regardingemails and therefore may overlook emotionalcontent when it is present. Expectations regard-ing a particular context constitute a schema—arelatively stable mental framework regardingwhat is typical or usual in a given situationbased on past experience (Sternberg, 1996).Schemata can have a strong influence on per-ception in that people ignore disconfirming dataand favor data consistent with their expecta-tions (e.g., Brewer & Treyens, 1981). Similarly, inthe absence of clear cues about the emotion ofthe sender, receivers are likely to fill in the gapswith information based on a schema. Althoughthe infrequency of emotional content couldmake emotional content more salient, employ-ees tend to receive a large volume of emails(Dawley & Anthony, 2003) and, consequently,read them quickly, “often missing importantpoints intended by the writer” (Grosvenor, 1998:2).

A neutrality effect, such that emotions are per-ceived as more neutral than intended by thesender, could occur for either positive or nega-tive emotions. However, the research citedabove suggests that positive emotions would bemore susceptible to the neutrality effect. In ad-dition, the symbolic meaning of email may con-tribute to the neutrality effect when sendersmean to convey positive emotion. Whereas face-to-face communication symbolizes caring(Trevino, Lengel, & Daft, 1987), email communi-cation may suggest a less personal focus (Tim-merman & Harrison, 2005). In effect, the easewith which emails are sent, the informality ofemails as a written form of communication, andthe often ephemeral quality of emails may con-tribute to their symbolic meaning such that

emails intended to convey positive emotions areinterpreted as more neutral (less positive andintense) than the sender intends. Trevino andher colleagues offered the following example:“The manager who congratulates a subordinateon 25 years of service with an electronic mailmessage may symbolize a lack of concern, leav-ing the subordinate feeling furious rather thancared about” (1987: 558). The symbolic meaningof email may explain why employees reported apreference for receiving good news face to facerather than by electronic communication(Sproull & Kiesler, 1986).

Proposition 1: Receivers are likely toinaccurately perceive emails in-tended to convey positive emotion asmore emotionally neutral than in-tended by the sender.

Negativity Effect

The characteristics of email may also in-crease the likelihood that receivers perceiveemails as more intensely negative than sendersintend—referred to here as the negativity effect.Because emails are text based and relativelylacking in cues, their emotional tone is oftenambiguous. The ambiguity of emotional tone inemails makes the negativity effect likely by in-creasing the salience of any negative informa-tion, particularly because emotional content inemail may violate employees’ schemata ofemails as emotionally neutral. Previous re-search indicates that violations of expectanciesresult in more intense and negative evaluations(e.g., Kernahan, Bartholow, & Bettencourt, 2000).

Furthermore, research suggests that emailsand other electronic media increase the likeli-hood of negativity effects. Walther andD’Addario (2001) found strong evidence for neg-ativity effects in email such that any verbal ornonverbal negative cue tended to override othercues. They concluded that a negativity effectexisted in emails because emails often containfew cues about emotion. Similarly, other re-search has shown that evaluations based onelectronic media are more negative and lessaccurate (e.g., Weisband & Atwater, 1999). Anec-dotally, negativity effects may be the reason onecompany requires the following statement to beappended to its employees’ emails: “This e-mailmay display a telegraphic style that gives the

2008 313Byron

false impression of curtness or insensitivity”(Martin, 2004).

In addition to the lack of cues, the lack offeedback also likely contributes to the negativ-ity effect in emails. With less feedback avail-able, email senders have less information avail-able to construct effective messages. Forexample, they have less information about theemail recipient and how he or she is interpret-ing their communication attempts. This meansthat inaccurate interpretations of emails are notlikely to be righted—email senders may not findout that their emails have been interpreted morenegatively than they intended. An MBA studentsaid about emails from her boss: “I can nevertell how my manager feels. When organizing ameeting I got a sarcastic reply (’this had betterbe good’) that I took to heart” (Byron & Baldridge,2005). Only much later did she find out the man-ager intended the comment to be funny.

Therefore, based on research and anecdotalevidence, I propose a negativity effect in theperception of emotion by email such that a rel-atively common judgment in emotion perceptionby email is perceiving emails as more intenselynegative than intended by the sender. The evi-dence provided above suggests that emails in-tended as positive, neutral, or negative are sub-ject to the negativity effect, although when thenegativity effect occurs, receivers may be morelikely to perceive emails as conveying angerand its variants—annoyance, frustration, aggra-vation, hostility, and rage (Shaver, Schwartz,Kirson, & O’Connor, 1987)—than as conveyingother negative emotions—sadness or fear.

Proposition 2: Receivers are likely toinaccurately perceive emails as moreintensely negative than intended bythe sender.

Both the neutrality effect and the negativityeffect are indicated by the extent to which areceiver’s perception differs from the sender’sintent (in terms of rated valence and intensity); astronger effect is indicated by a greater differ-ence between the sender’s intent and the receiv-er’s perception, and the absence of the effect isindicated by no difference between these. Con-sidering intended emotion on a continuum rang-ing from negative to positive (with neutral at themidpoint), these two effects are similar, in thatboth predict that receivers perceive emails asmore negatively valenced than intended by the

sender. They differ in terms of emotional inten-sity; the neutrality effect predicts that receiversperceive emails as less intense than intendedby the sender, whereas the negativity effect pre-dicts a more intense perception. They also differin that the neutrality effect is specific to emailsintended to convey positive emotions, whereasthe negativity effect could occur for emails in-tended as positive, neutral, or negative. Last,these two effects may have different anteced-ents and different consequences.

Figure 1 summarizes the proposed relation-ships between sender and receiver factors (in-cluding their social context) and message fac-tors that increase the likelihood a receiver willexperience neutrality effects or negativity ef-fects. I purposefully restricted the model to onlythose variables expected to be the best, al-though not the only, predictors of neutrality ornegativity effects based on previous related re-search.

EFFECT OF SENDERS AND RECEIVERS ONEMOTION PERCEPTION IN EMAIL

Even among technologically savvy users whoare highly dependent on email at work, individ-uals and groups vary in both their perceptionsand use of email at work (Carlson & Zmud, 1999;Higa et al., 2000). These variations likely persistbecause the rapid adoption of email allowed forfew established norms and rules for its use. Ac-cordingly, these variations in terms of users (i.e.,recipients and senders of emails) and their so-cial context (i.e., their workgroup or organiza-tion) likely influence how receivers perceiveemotion in email communication.

How Sender Factors Influence EmotionPerception in Emails

In accordance with the SMCR model, emailsenders (S) influence the accuracy of emotionalcommunication in emails because, among otherreasons, senders have different attitudes towardand knowledge about the receiver. In turn, emailrecipients likely rely on knowledge about thesender, such as his or her gender, in perceivingemotion in emails from that sender. In little re-search have scholars directly considered thesefactors, perhaps because of the belief that therelative lack of cues about the sender’s identityin electronic media eliminates biases and al-


lows for more accurate perceptions of messages(e.g., Lea & Spears, 1992). However, this viewrests on the idea that users do not know eachother—an untenable notion in the workplace,where employees often send and receive emailsto and from known coworkers, clients, and cus-tomers. Even when those communicating elec-tronically do not know each other, receivers ac-tively search for information about the senderand use any available cues to form judgments(Nowak, 2003; Walther & Tidwell, 1995). There-fore, it seems likely that sender characteristicsand the sender-recipient relationship influencehow emotional content in emails is perceived.

Gender. Receivers may differentially perceiveemotions in emails sent by male coworkers thanthose sent by female coworkers based on actualand expected differences in emotional expres-sion. Compared to women, men tend to be lessemotionally expressive (Hall, 1984). And when

men do express emotion, they tend to be lessaccurate encoders of that emotion (Hall, 1984;Wagner, MacDonald, & Manstead, 1986). Thesetwo factors make it more difficult to accuratelyinterpret the emotions of men than women.

Although little research exists on the expres-sion of emotion in emails, studies have foundthat men are less likely than women to use cuesin emails representing positive emotions (Reza-bek & Cochenour, 1998; Witmer & Katzman,1997). Because men tend to be less emotionallyexpressive of positive emotion, their emails maytend to be emotionally neutral. Consequently,email receivers who perceive their emails asemotionally neutral will often be accurate. How-ever, when men intend to convey positive emo-tion, their lower accuracy at encoding emotionin general, along with their minimal use ofemoticons indicating positive emotion, may in-crease the likelihood that recipients will per-

FIGURE 1Model of Sender, Receiver, Social Context, and Message Factor Effects on Receivers’ Emotion

Misperception in Emails

2008 315Byron

ceive their emails as more neutral than in-tended. That is, receivers are more likely toexperience the neutrality effect when emails arefrom male rather than female senders (PathP3a).

The negativity effect may also be more likelyto occur with emails from male senders. Onestudy found that men were more likely thanwomen to express anger in emails (Witmer &Katzman, 1997). Other research on nonverbal be-havior has shown that although emotions ex-pressed by women are generally easier to de-code, some negative emotions such as anger aremore easily detected when expressed by menthan by women (e.g., Rotter & Rotter, 1988). Be-cause male email senders may be more likely toexpress negative emotion in email and becauseemail receivers may be more sensitive to cuesindicating negative emotion in their emails, Ipredict a negativity effect for emails receivedfrom male senders (Path P3b).

Proposition 3a: Receivers of emailsfrom male as compared to femalesenders will be more likely to inaccu-rately perceive emails intended to bepositive as more emotionally neutral.

Proposition 3b: Receivers of emailsfrom male as compared to femalesenders will be more likely to inaccu-rately perceive emails as more emo-tionally negative than intended.

Relationship length. Many studies have ex-amined how CMC between those with a longerhistory differs from that of more recently ac-quainted or unknown communication partnersand how CMC differs from face-to-face commu-nication. When communicating electronically,people less known to each other use less rela-tional communication and perceive each otherless accurately (Walther, Anderson, & Park,1994). These findings mirror more general re-search on person perception, which has tendedto find that individuals are more accurate atperceiving known than less known targets (forreviews see Fiske, 1993, and Jussim, 1991). Inaddition, individuals communicating electroni-cally, as compared to face to face, tend to eval-uate unknown communication partners morenegatively (Straus & McGrath, 1994; Weisband &Atwater, 1999).

In contrast, the longer the relationship be-tween communication partners, the more accu-rate and rich CMC becomes. Over time, commu-nication partners communicating electronicallytend to use more relational communication, tobe more sociable, and to communicate more ac-curately (Walther et al., 1994; Walther & Bur-goon, 1992). In addition, the longer communica-tion partners communicate via electronic media,the more similar that communication is to face-to-face communication: a meta-analysis com-paring CMC to face-to-face communicationfound few differences in accuracy or content be-tween the media when communication partnerswere known to each other (Walther et al., 1994).

These findings suggest that the length of therelationship between those communicating byemail will have three simultaneous effects onemotion perception. First, partners with a longeremail history may be more likely to express andperceive emotion by email. Second, they maycommunicate emotion more accurately. Third,they may be less likely to negatively evaluatethe email and its sender. Combined, these threefindings suggest that receivers of email fromless known senders are more likely to experi-ence the negativity effect (Path P4 in Figure 1):

Proposition 4: Receivers of emails fromsenders known for shorter durationswill be more likely to inaccuratelyperceive emails as more emotionallynegative than intended.

Relative status. The relative status of thesender may also affect how receivers perceiveemotion in emails. Although some have arguedthat CMC limits cues indicating status and in-hibits the salience of status, several studieshave found that cues about communicators’ sta-tus are present in CMC (e.g., Sherblom, 1988;Weisband, Schneider, & Connolly, 1995). Even inthe absence of cues, it is likely that recipientsknow the relative status of familiar senders,given the salience of status in organizations.

The negativity effect may be more commonwhen emails are received from higher-statussenders, for several reasons (Path P5 in Figure1). First, higher-status employees generally areless likely to send positive emotional content inemails. In a study of employees at a telecommu-nications company, Ku (1996: 38) found that so-cioemotional content, which he defined as using


email “to get to know someone, to keep in touchwith someone in another location, and to sendnotes that contain sociable or nonwork-relatedcontent,” was less likely in emails sent to lower-status employees and that employees with highabsolute status in the organization were theleast likely to use email for socioemotional pur-poses.

Second, although higher-status employeesmay be less likely to express positive emotions,they may be more likely than lower-status em-ployees to express negative emotions. Researchhas shown that employees are more likely toexpress negative emotions to those of lower,rather than higher, status (Flett, Blankstein,Piner, & Bator, 1988; Kipnis, Schmidt, & Wilkin-son, 1980). In reviewing research on emotionalexpression and status, Gibson and Schroeder(2002) found evidence that those with morepower express a wider variety of emotions, in-cluding anger and other negative emotions.“Lower status [organizational] members, how-ever, are not free to express negative emotionsupward” (Gibson & Schroeder, 2002: 206). Al-though the expression of negative emotiondownward in organizations may have fewersanctions than the expression of negative emo-tion upward, this by no mean implies that higher-status senders can do so without fear of reprisal.Organizational norms typically discourage theabuse of power by higher-status employees(Bies & Tripp, 1995). Higher-status senders maybe more likely than lower-status senders to ex-press negative emotion in emails, although theoverall occurrence may be infrequent.

Last, being dependent on higher-status othersfor desired outcomes may motivate lower-statusemployees to seek information about them. In-dividuals tend to seek information about higher-status others because they want to be able topredict and influence them in order to obtain theoutcomes they desire (Depret & Fiske, 1993). Intheir search for information, they may be partic-ularly sensitive to negative cues from higher-status others, because lower-status employeesare motivated to seek approval from higher-status others (Snodgrass, 1992). More generally,research on the recognition of nonverbal cuesprovides some support for the “subordinationhypothesis”—the premise that those with lowerstatus are more motivated to recognize emo-tional displays of those with higher status (Elf-enbein & Ambady, 2002).

Proposition 5: Receivers of emails fromhigher- as compared to lower-statussenders will be more likely to inaccu-rately perceive emails as more emo-tionally negative than intended.

How Receiver Factors Influence EmotionPerception in Emails

Characteristics of receivers are also likely toexplain emotion perception in email communi-cation. This idea rests on a well-supported as-sumption—that individual variations affect per-ception in general (e.g., Fiske & Taylor, 1991),and emotion perception in particular (seeO’Sullivan, 1983, for a review), in persistent andpredictable ways.

Age. The receiver’s age may be important inexplaining how he or she perceives emotionalcontent in emails. Specifically, older employeesmay be less likely to perceive emotional contentin email communication. Research suggeststhat older employees tend to use and perceiveemail differently from younger employees. Theyhave been found to use email less frequently ingeneral (Higa et al., 2000; Mitra, Hazen, LaFrance,& Rogan, 1999), to use email less to send socio-emotional content (Ku, 1996), and to have lowerrichness perceptions of email (Higa et al., 2002).These findings suggest that older employeesmay be more likely to perceive emails as emo-tionally neutral based on their own use of emailand their perceptions of the medium.

More generally, age is negatively related toemotional expression, regulation, and percep-tion. Older employees tend to be less likely toexpress and more likely to suppress emotions,and they tend to show a decline in emotionperception. Gross and his colleagues found that,in general, older people were less likely to ex-perience and express emotions and were morelikely to report controlling those emotions theydid experience (Gross et al., 1997). Similarly,other researchers found that the experience ofnegative emotions declined with age until age60 (Carstensen, Pasupathi, Mayr, & Nesselroade,2000). Last, although findings have been mixed,research results have tended to support the find-ing that older individuals are less accurate atemotion perception from verbal and nonverbalstimuli (e.g., Grunwald et al., 1999; Nowicki,2005). In summary, older employees may per-ceive others’ emails as more neutral than in-

2008 317Byron

tended based on their own use and perception ofemail as less emotional and because of theirlessened tendency to experience and accuratelyperceive emotional stimuli (Path P6).

Proposition 6: Older email receiverswill be more likely to inaccuratelyperceive emails intended to be posi-tive as more emotionally neutral.

Negative affectivity. Employees’ negative af-fectivity (NA)—their tendency to experience neg-ative emotional states—likely influences emo-tion perception in email communication. Iexpect that NA leads to persistent, predictabledifferences in emotion perception because, as atrait, NA has been found to be relatively stablewithin individuals and to influence perceptionof both people and objects (e.g., George & Brief,1992). A large body of research has consideredthe relationship between NA and perception.Two findings are particularly relevant here: NAincreases attention to congruently valencedstimuli and decreases perceptual accuracy (e.g.,Ambady & Gray, 2002).

Although much research links traits and moodto perception, findings have not always beenconsistent. One reason that has been proposedto explain these inconsistencies is the type ofprocessing used in perception. Research indi-cates that the processing and accuracy effectssuggested above tend to occur when severalconditions exist: when the perceived stimulus isaffective in nature and when automatic process-ing is used in the perception of the stimuli, as itis in nonverbal emotion perception (Ambady &Gray, 2002; Patterson & Stockbridge, 1998). Al-though the question of whether affective infor-mation is automatically processed in emailshas not been determined, to the extent thatemotion perception in email communication isprocessed automatically, as it is in face-to-facecommunication, I expect that those higher inNA will be especially likely to experience thenegativity effect in their perception of emails(Path P7).

Proposition 7: Email receivers higherin NA will be more likely to inaccu-rately perceive emails as more emo-tionally negative than intended.

How Social Context Influences EmotionPerception in Emails

In the SMCR model, the social context of thesender and receiver affects the accuracy andeffectiveness of their communication (Berlo,1960). In addition, social context factors havebeen theoretically and empirically linked to theuse and perceptions of different media (e.g., De-Sanctis & Poole, 1994; Schmitz & Fulk, 1991).Therefore, I propose that social context vari-ables will influence the perception of emotion inemails.

The existence of strong shared norms regard-ing email use may moderate the relationshipbetween user characteristics and perceptual in-accuracy of emotion in email when both commu-nication partners are in the same group. Re-search has found that strong norms regardingthe use of electronic media promote its effectiveuse (Orlikowski, Yates, Okamura, & Fujimoto,1995; Yates, Orlikowski, & Okamura, 1999) andthat norms are more strongly related to emailuse and perceptions than are individual charac-teristics (Higa et al., 2000). Consequently, dis-play rules—norms regarding emotional expres-sion—may affect how emotions are expressedand perceived in work emails. Research on emo-tional labor suggests that display rules differamong organizations and influence employees’expression and repression of emotions (e.g.,Morris & Feldman, 1997; Rafaeli & Sutton, 1987).Moreover, they influence the valence, frequency,and duration of emotions employees display atwork in different contexts, such as interactionswith customers or supervisors (e.g., Morris &Feldman, 1997).

Because display rules address the context inwhich emotions are expressed, it follows thatorganizational display rules may exist for andapply to the expression of emotion in email com-munication. The existence and knowledge ofdisplay rules for the expression of emotion inemail communication are likely to influencehow employees express and perceive emotion inemails. Past research on display rules in orga-nizations suggests that rules may seek to bothsuppress some emotions and increase the ex-pression of others. When present and adheredto, these display rules may attenuate the rela-tionship between recipient and sender charac-teristics and inaccuracy in emotion perceptionin email communication by establishing guide-


lines for the expression of emotion in emails.Receivers’ perceptions then may be more deter-mined by the display rules and less influencedby the sender and receiver factors proposedabove (and shown in Figure 1). Established dis-play rules regarding emotional expression inemails may increase employees’ accuracy ofemotion perception in emails. Conversely, whenthese rules do not exist, are not shared, or arenot followed, email recipient and sender char-acteristics will be more likely to influence re-ceivers’ emotion perception of emotion in emailcommunication, and receivers will be morelikely to misperceive emails’ emotional content(Path P8).

Proposition 8: Known and shared dis-play rules regarding emotional ex-pression in emails will moderate therelationship between recipient andsender characteristics and emotionperception in emails such that the ex-istence of display rules will attenuatethe relationship between recipientand sender characteristics and receiv-ers’ perceptual inaccuracy of emotionin emails.

EFFECT OF MESSAGE ON EMOTIONCOMMUNICATION

How senders encode their messages (M) con-tributes to the communication process (Berlo,1960). Despite the difficulties of doing so, as re-search suggests, senders may adapt email toconvey their intended emotions. Similarly, or-ganizational communication scholars have ar-gued that although somewhat constrained bythe inherent characteristics of media, peoplecan use media adaptively to achieve their aims(e.g., Barry & Fulmer, 2004; Carlson & Zmud,1999).

The most straightforward way of encoding amessage to convey the sender’s emotion is toverbalize it. For example, a sender could write,“I am so happy that you decided to sign thecontract.” Common sense suggests that inter-preting verbalizations of emotion would be as-sociated with a fairly high degree of accuracy(especially when accuracy is defined in terms ofthe receiver’s interpretation matching the send-er’s intent). Although words have differentmeanings for different senders and receivers,

research examining the affective tone of texthas found a relatively high degree of agreementamong raters (Bestgen, 1994; Mossholder, Set-toon, Harris, & Armenakis, 1995), suggesting thatpeople tend to reliably interpret verbal emo-tional content. Further, the content and avail-ability of resources, such as dictionaries of af-fect (e.g., Whissell, 1989), suggest that manywords describe emotion and that these wordshave some degree of shared meaning.

However, senders may infrequently verbalizetheir emotional state to others. Research onface-to-face communication has found that peo-ple are more likely to communicate emotionsnonverbally than verbally (Ekman et al., 1980). Infact, people have difficulty verbalizing at allwhen they are experiencing intense emotions(see Von Glinow, Shapiro, & Brett, 2004, for areview). For these reasons, considering otherways of expressing emotion apart from verbal-izing it may be fruitful. Although fewer cues areavailable in emails than in face-to-face commu-nication, senders can use cues to express emo-tion in emails, according to some research.

The most obvious cue of emotion in email isthe emoticon. Research suggests that receiversrely on emoticons, when present, to interpretsenders’ emotional state (e.g., Byron &Baldridge, in press; Thompsen & Foulger, 1996).However, Walther and D’Addario (2001) arguethat emoticons are symbolic rather than sponta-neous displays of emotion, and, as such, receiv-ers may interpret them as less authentic indica-tors of senders’ emotion (Walther & D’Addario,2001). If this argument is correct, then it may bethat receivers rely more heavily on other cues,such as response time, message length, pres-ence or lack of greeting, or degree of formality,in perceiving emotion by email. Exploratory re-search has shown that employees report relyingon these other cues to express and interpretemotion (Byron & Baldridge, 2005). However,save response time (Walther & Tidwell, 1995), Ifound no studies that empirically consideredhow receivers might emotionally interpret thesecues, either alone or combined with other cues.The issue of what cues receivers rely on to indi-cate emotion and how they interpret these cuesis a matter for future research.

Despite the lack of empirical research, I arguethat receivers will more accurately interpretemails that include more verbal and nonverbalcues of emotion. Although it is conceivable that

2008 319Byron

cues can be contradictory (e.g., representingboth positive and negative emotions), ceteris pa-ribus, receivers are more likely to accuratelyinterpret emails that contain more cues, espe-cially given that cues in email may be morecontrollable and intentional displays of emo-tion. Based on the above arguments, verbaliza-tions of emotion and the presence of nonverbalemotional cues may moderate the relationshipbetween sender and receiver characteristicsand the likelihood that receivers will inaccu-rately interpret senders’ emotion (Path P9).

Proposition 9: Emotional cues, such asverbal cues and emoticons, in emailswill moderate the relationship be-tween recipient and sender character-istics and emotion perception inemails such that these cues will atten-uate the relationship between recipi-ent and sender characteristics and re-ceivers’ perceptual inaccuracy ofemotion in emails.

CONSEQUENCES OF EMOTIONALINACCURACY IN EMAILS

Based on the previously discussed functionsof emotion, the neutrality and negativity effectslikely have both positive and negative implica-tions for workplace relationships and the trans-mission of information at work. Figure 2 summa-

rizes the proposed positive and negativerelational and informational effects of the neu-trality and negativity effects in organizations.

Positive Consequences of the Neutrality Effect

The tendency for receivers to inaccurately per-ceive emails intended to convey positive emo-tion as emotionally neutral should have positiveinformational and relational consequences inorganizations. In terms of informational conse-quences, the neutrality effect may allow receiv-ers to focus more on the informational content ofthe email than its emotional content, therebyincreasing the accuracy of nonemotional com-munication. For example, some influence tac-tics, such as inspiration and ingratiation, in-volve the display and perception of positiveemotion. Positive emotion has persuasive pow-ers by increasing liking for the influencer andreducing resistance on the part of the influencetarget (Gibson & Schroeder, 2002). The target ofthe influence attempt who “catches” the positiveemotion is presumably more likely to be per-suaded because he or she is less focused on theinformational content of the attempt. Whenemail senders use email to influence othersthrough the display of positive emotions, theneutrality effect increases the likelihood thatthe targets of the influence attempts—the emailreceivers—will focus on the informational

FIGURE 2Proposed Relational and Informational Effects of the Neutrality and Negativity Effects in Emails

in Organizations


rather than the emotional content of the emailedinfluence attempt. This may increase the accu-racy of decisions, because email receivers areless likely to be distracted by the intended pos-itive emotional content in the emails.

In terms of relational benefits, the neutralityeffect may serve to increase social distance be-tween employees. The sharing of positive emo-tion can help to strengthen workplace relation-ships, and, presumably, the failure to sharepositive emotion can inhibit relationship devel-opment. Although workplace relationships aregenerally considered integral to organizationalfunctioning, some theorists have argued thatclose relationships at work can complicaterather than simplify employees’ work lives(Eisenberg & Witten, 1987). For example, somemanagers seek to maintain a reasonable degreeof distance between themselves and their sub-ordinates. Therefore, to the extent that the neu-trality effect inhibits relationship developmentand that further relationship development is notalways preferable, a benefit of the neutralityeffect could be to maintain or increase socialdistance between employees.

Negative Consequences of the Neutrality Effect

The neutrality effect is also likely to have neg-ative informational and relational conse-quences. When subordinates inaccurately per-ceive emails intended to convey positiveemotion as more neutral, they receive inaccu-rate information about what behaviors elicitpositive emotion from their superiors. For exam-ple, when managers use email to praise employ-ees, the diminished positive emotion conveyedin the email may be less informative to subordi-nates about what behaviors are desirable andshould be modeled or repeated. A colleague re-cently recounted the following example. Her de-partment chair sent an email to college facultymembers about an award she had won from afoundation. Although he likely intended to con-vey positive emotion (e.g., joy), she perceived theemail as more neutral, leaving her disappointedby his “lack of enthusiasm” and unclear aboutwhether this award was institutionally valued.In sum, the neutrality effect might lower subor-dinates’ performance-to-outcome expectancybecause it increases uncertainty regarding or, atthe least, does not help to clarify desired perfor-mance.

The neutrality effect may also have harmfulconsequences for workplace relationships. Al-though social distance is sometimes desirable,relationships between coworkers and othersoutside the organization are often fundamentalto an organization’s success. Research on emo-tional contagion in groups suggests that thecommunication and sharing of positive emo-tions among group members benefit group func-tioning by increasing cooperation, curtailingconflict, and improving task performance (Bar-sade, 2002). It seems likely that the neutralityeffect in email communication is one of the rea-sons that employees have reported feeling lessconnected to their coworkers as email use hasincreased and replaced other forms of commu-nication (Sarbaugh-Thompson & Feldman, 1998).If employees fail to communicate positive emo-tions by email, their relationship developmentmay be stunted.

Positive Consequences of the Negativity Effect

The negativity effect may have both benefi-cial and harmful consequences in organiza-tions. Although I expect no potential positiverelational consequences of the negativity effect,it may offer positive informational conse-quences. It may motivate the receiver to seekmore information about the sender’s emotionalstate, which could result in the eventual greateraccuracy of his or her judgments. For example, ifa recently hired subordinate incorrectly per-ceived an email from his boss as conveying an-ger (although she intended it as neutral), thesubordinate might be motivated to ask cowork-ers or the supervisor about the email. When thereceiver responds to negative emotion in emailsby seeking more information—for example, byinitiating face-to-face communication— he orshe may acquire more accurate informationabout the intent of the email sender.

Negative Consequences of the NegativityEffect

On the whole, the negativity effect likely hasgreater negative than positive consequences fororganizations. According to Markus (1994), em-ployees have a stated preference for email whendelivering negative information—for example,when they don’t like the receiver or when thetopic angers them. Although sending an email

2008 321Byron

in these situations may be less anxiety produc-ing for the sender, the tendency for receivers toperceive the email as more negative than in-tended could have long-term consequences forthe sender’s relationship with the receiver. Intheir model of conflict escalation in email com-munication, Friedman and Currall (2003) theo-rized that communicating by email increasesthe likelihood of conflict escalation betweenthose using the medium to communicate. Al-though they did not address a negativity effect,it seems likely that the negativity effect is anadditional factor in exacerbating conflict be-tween those communicating by email. Employ-ees who mistakenly perceive an email sender asexpressing anger may feel provoked and retali-ate with an angry response, thus escalating con-flict and harming workplace relationships.

In addition to harming workplace relation-ships, the negativity effect may have negativeinformational consequences. When email re-ceivers perceive an email sender as more angrythan intended, they may receive distorted infor-mation about their past performance and de-sired future performance. An email intended asa reminder may be perceived as corrective feed-back. For example, one MBA student told of anemail from his manager to all employees in hisdepartment regarding sales goals. The studentinterpreted that the manager was angry with hisperformance, which confused him because hehad exceeded his quota. He was left wonderingwhat he had done wrong and what he should dodifferently.

The negativity effect may be increased by thetendency for email senders to use fewer nicetiesor to not “sugarcoat” their message. When askedto provide negative feedback to others, studyparticipants were less likely to positively distortnegative information in electronic media (Suss-man & Sproull, 1999). However, it should benoted that the more accurate expression of amessage does not ensure accurate perception ofit. Rather, email receivers may negatively dis-tort the information, which could lead to employ-ees’ increased anxiety and insecurity abouttheir performance.

DISCUSSION

My purpose has been to present a model ofemotional perception in email communicationthat accounts for channel, sender, receiver, and

message characteristics. Drawing from a largebody of literature on communication, emotions,and perception, the present model explains whyemotions are likely to be inaccurately communi-cated in emails, which inaccurate perceptionsare most likely, what factors affect their occur-rence, and what consequences may ensue.

Methodological Issues

Testing the present model requires carefulconsideration of several issues, especially re-garding how researchers measure accuracy. Ihave defined accuracy in terms of the extent towhich the receiver’s perceptions of an emailagrees with the sender’s intentions; therefore,measuring accuracy depends, in part, on mea-suring senders’ intended emotion. Fortunately,research on nonverbal emotional expressionsoffers guidance (for reviews see Ickes, 2001;Kenny & Winquist, 2001; Noller, 2001). In a fieldstudy researchers can ask senders to retrospec-tively report their intended emotion. Another op-tion more appropriate to a laboratory study is tohave senders encode a particular emotion in anemail, which is then randomly assigned and“sent” to different receivers (Noller, 2001), simi-lar to the method used by Kruger and his col-leagues (2005) in their recent studies on emailmiscommunication.

However, researchers may alternately defineaccuracy in terms of the extent to which thereceiver interprets a message consistent withthe internal state of the sender. This can differfrom the sender’s intended communication ofemotion, because sometimes people (1) lackawareness of their internal state but uncon-sciously “leak” the emotion to others or (2) inten-tionally express an emotion different from theirinternal state. Both criteria of accuracy havebeen endorsed by researchers of nonverbal com-munication; accuracy has been defined in termsof the agreement between the sender’s inten-tions and the receiver’s perceptions and in termsof the agreement between the sender’s internalstate as indicated by self-report and the consen-sus of other judges (Noller, 2001), despite thepotential conflict between these criteria and theproblems inherent in each. Researchers testingthe present model may consider examining mul-tiple definitions of accuracy to gain a fuller un-derstanding of emotional communication inwork emails.


More generally, I recommend that researcherstesting all or parts of this model consider ad-vances in research on interpersonal sensitivity.Noller (2001) has argued that standard contentmethodology allows researchers to determinethe extent to which misunderstandings are dueto the sender and the extent to which misunder-standings are due to the receiver. Alternatively,Kenny and Winquist (2001) have outlined numer-ous research designs using componential anal-ysis that allow researchers to partition accuracyinto different sources, such as accuracy due to areceiver’s tendency to respond consistently overtime across multiple senders or to a receiver’stendency to perceive a particular sender consis-tently.

Scholarly and Practical Implications

The propositions I have set forth have impli-cations for management scholars and practition-ers. First, they imply that although emotionalcontent is difficult to transmit accurately inemails, it can be transmitted in this medium.This suggests that ignoring the possibility thatemails can carry emotional content is likely toincrease the likelihood of miscommunication.Kruger and his colleagues (2005) similarly haveargued that being overconfident in our ability toeither perceive or convey ambiguous content inemails prevents us from taking steps to increaseaccurate communication. People believe theycan accurately convey intended emotion whiledoubting others’ abilities: in a recent survey,employees were likely to report that their col-leagues needed more training in email yet theythemselves did not (Dawley & Anthony, 2003).

Certainly, a first step toward improving accu-racy in emails is to recognize the possibilitythat, as the model suggests, we are fallible asboth email senders and receivers. Miscommuni-cation in emails can be caused by senders’ in-ability to accurately convey their intendedmeaning and by receivers’ inability to accu-rately perceive senders’ intended meaning. Still,more empirical research on emotions in emailcommunication is needed to gain an under-standing of how emotions are expressed andperceived and what factors affect the perceptualaccuracy of emotions in emails.

A second implication is that, despite the diffi-culty of accurately expressing and perceivingemotions in emails, inaccurate emotion percep-

tion is not inevitable. Although I have outlinedwhen particular errors may be more likely, thepropositions also suggest when accuracy ismore likely. This suggests that accuracy can beimproved; inaccuracy is not inevitable. How-ever, because few studies have examined whatfactors influence accuracy in email communica-tion, research testing the propositions of thepresent model is needed. Additionally, althoughthe relative lack of cues and slower and reducedfeedback make inaccurate emotion perceptionmore likely in email communication, increasingthe use of cues and the rate and amount offeedback in emails may reduce the likelihood ofinaccurate emotion perception in emails. Send-ers who use established, shared cues to commu-nicate emotion or who seek clarity by repeatingimportant information may more accurately ex-press emotions by email. Likewise, email recip-ients who seek clarity by asking questions orstating their interpretation of the message aremore likely to accurately perceive emotionsfrom email communication.

Finally, the propositions imply that the per-ception of emails is influenced by both individ-ual-level and group-level factors. Consideringthat employees often communicate with col-leagues outside their workgroup or organizationand with clients and customers outside theiremploying organization, further research ongroup and organizational differences is war-ranted. Specifically, research should addresshow these differences influence the perceptionof emails received from those inside and outsidetheir workgroup or organization, as well as therelative contribution of individual-, group-, andorganizational-level effects.

In addition to directions for future research,practical advice can be gleaned from the modelas well. First, employees should be aware thatemails communicate emotion and, if the numberof relative paths is any indication, that they maybe especially likely to misinterpret emails asmore negative than senders intend. In addition,employees should be cognizant of the fact thatothers may be perceiving emotional contentfrom the emails they send—whether they intendto communicate emotions or not. Increasedawareness of emotion in emails will help tomake email communication more effective byincreasing its accuracy. A second related prac-tical implication is that employers may want toconsider offering training on the use of email at

2008 323Byron

work (including, or perhaps even especially, tothose who feel confident in their use of email).Research on other electronic media has shownthat training helps to establish organizationalnorms for its use (Orlikowski et al., 1995). Asspecified in the present model, established andaccepted norms may help to limit the influenceof recipient and sender factors that increase thelikelihood of inaccurate emotion perception inemails.

Limitations

Although the model has both practical andscholarly implications, it is not without limita-tions. I focused on the likelihood of emails beingmisperceived as negative or neutral; I did notconsider the possibility that emails could bemisperceived as expressing positive emotionalcontent. Although inaccurately perceiving anemail as positive is possible, I intentionally ig-nored this possibility in the model because Iconsidered it less likely based on my review ofrelated research and theory. Future researchshould confirm the frequency of different typesof inaccurate judgments in emotion perceptionin email communication.

Second, the model focuses on the perception,rather than the expression, of emotion in emailcommunication. Unfortunately, relatively littleis known about emotional expression in emailsin organizational settings. Organizational re-searchers and theorists should consider ex-panding knowledge about how emotions arecommunicated by email at work, how frequentlywork-related emails contain emotional content,and how authentic and unauthentic displays ofemotion in emails at work differ. In particular,the prevalence of email communication in theworkplace may have important implications foremotional labor and regulation (e.g., Morris &Feldman, 1997; Rafaeli & Sutton, 1987). It seemslikely that suppressing undesired emotions maybe less effortful and stressful for employees andmore convincing to others in email communica-tion as compared to face-to-face communication.

Furthermore, I based my arguments partly onresearch concerning the perception and expres-sion of emotions from nonverbal behaviorpresent in face-to-face communication. Al-though I focused on research findings thatwould likely be replicated with regard to emails,the extent to which these processes are parallel

remains a subject for future research. For exam-ple, although women tend to be better facialencoders of emotion, I found no research ad-dressing whether women are able to encodeemotion more accurately in email communica-tion.

Last, I should note that the present paper fo-cuses on email and neglects discussion of otherforms of electronic communication. Email, how-ever, is presently the dominant form of elec-tronic communication in the workplace, andmany of the theoretical arguments presentedmay be applied to other electronic media usedat work, such as instant messaging or collabo-rative electronic technologies that similarly re-duce available cues and slow and decreasefeedback. Because emotions serve several im-portant functions in organizations, it seems im-portant to examine how emotional informationin the workplace is exchanged within a technol-ogy that is rapidly replacing other communica-tion channels.

The increased prevalence of email communi-cation in the workplace underscores the need todevelop an understanding of how emotions areperceived and expressed in emails at work. Thesmall body of research on this topic suggeststhat emotions are present in email communica-tion, but this research has focused on nonorga-nizational settings and has not considered whatfactors might influence emotion perception inemail communication. In contrast, I have fo-cused here on how emotional content in emailsis likely to be misperceived and what factors arelikely to increase or decrease the likelihood thatthese perceptual inaccuracies will occur. In do-ing so, I argue that narrow views of communi-cation media are insufficient to explain new me-dia and that considering both objective andperceptual and social processes has the great-est potential to advance understanding on thistopic (e.g., Barry & Fulmer, 2004). Gaining a bet-ter understanding of how emotions are commu-nicated (or miscommunicated) in email at workis especially important given the increasinggrowth in email communication and the in-creasing recognition of the importance of emo-tions in organizations.

REFERENCES

Ambady, N., & Gray, H. 2002. On being sad and mistaken:Mood effects on the accuracy of thin slice judgments.


Journal of Personality and Social Psychology, 83: 947–961.

Barry, B., & Fulmer, I. S. 2004. The medium and the message:The adaptive use of communication media in dyadicinfluence. Academy of Management Review, 29: 272–292.

Barsade, S. 2002. The ripple effect: Emotional contagion andits effect on group behavior. Administrative ScienceQuarterly, 47: 644–675.

Berlo, D. K. 1960. The process of communication. New York:Holt, Rinehart & Winston.

Bernieri, F. J. 2001. Toward a taxonomy of interpersonal sen-sitivity. In J. A. Hall & F. J. Bernieri (Eds.), Interpersonalsensitivity: Theory and measurement: 3–20. Mahwah, NJ:Lawrence Erlbaum Associates.

Bestgen, Y. 1994. Can emotional valence in stories be deter-mined by words? Cognition & Emotion, 7: 21–36.

Bies, R. J., & Tripp, T. M. 1995. The use and abuse of power:Justice as social control. In R. Cropanzano & M. Kacmar(eds.), Politics, justice, and support: Managing socialclimate at work: 131–146. Westport, CT: Quorum.

Blanchette, I., & Richards, A. 2003. Anxiety and the interpre-tation of ambiguous stimuli: Beyond the emotion-congruent effect. Journal of Experimental Psychology:General, 132: 294–309.

Brewer, W. F., & Treyens, J. C. 1981. Role of schemata inmemory for places. Cognitive Psychology, 13: 207–230.

Buck, R. 1984. The communication of emotion. New York:Guilford Press.

Byron, K., & Baldridge, D. 2005. Toward a model of nonverbalcues and emotion in email. In K. M. Weaver (Ed.), Acad-emy of Management Best Papers Proceedings: OCIS B1–B6.

Byron, K., & Baldridge, D. In press. E-mail recipients’ impres-sions of senders’ likeability: The interactive effect ofnonverbal cues and recipients’ personality. Journal ofBusiness Communication.

Calem, R. E. 1995. Email users advised: Put best foot forward.Crain’s New York Business, July 17: 22.

Carlson, J. R., & Zmud, R. W. 1999. Channel expansion theoryand the experiential nature of media richness percep-tions. Academy of Management Journal, 42: 153–169.

Carstensen, L. L., Pasupathi, M., Mayr, U., & Nesselroade, J. R.2000. Emotional experience in everyday life across theadult life span. Journal of Personality and Social Psy-chology, 79: 644–655.

Critchley, H. D., Rotshtein, P., Nagai, Y., O’Doherty, J.,Mathias, C. J., & Dolan, R. J. 2005. Activity in the humanbrain predicting differential heart rate responses toemotional facial expressions. NeuroImage, 24: 751–762.

Daft, R. L., & Lengel, R. H. 1986. Organizational informationrequirements, media richness and structural design.Management Science, 32: 554–571.

Dawley, D. D., & Anthony, W. P. 2003. User perceptions ofe-mail at work. Journal of Business and Technical Com-munication, 17: 170–200.

Depret, E. F., & Fiske, S. T. 1993. Social cognition and power:

Some cognitive consequences of social structure as asource of control deprivation. In G. Weary, F. Gleicher, &F. K. Marsh (Eds.), Control motivation and social cogni-tion: 176–202. New York: Springer-Verlag.

DeSanctis, G., & Poole, M. S. 1994. Capturing the complexityin advanced technology use: Adaptive structuration the-ory. Organization Science, 5: 121–147.

Eisenberg, E. M., & Witten, M. G. 1987. Reconsidering open-ness in organizational communication. Academy ofManagement Review, 12: 418–426.

Ekman, P., & Friesen, W. V. 1969. The repertoire of nonverbalbehavior: Categories, origins, usage, and coding. Semi-otica, 1: 49–98.

Ekman, P., Friesen, W. V., & Ancoli, S. 1980. Facial signs ofemotional experience. Journal of Personality and SocialPsychology, 39: 1125–1134.

Elfenbein, H. A., & Ambady, N. 2002. On the universality andcultural specificity of emotion perception: A meta-analysis. Psychological Bulletin, 128: 203–235.

Fiske, S. T. 1993. Social cognition and social perception.Annual Review of Psychology, 44: 155–194.

Fiske, S. T., & Taylor, S. 1991. Social cognition (2nd ed.). NewYork: McGraw-Hill.

Flett, G. L., Blankstein, K. R., Pliner, P., & Bator, C. 1988.Impression management and self-deception compo-nents of appraised emotional experience. British Journalof Social Psychology, 27: 67–77.

Friedman, R. A., & Currall, S. C. 2003. Conflict escalation:Dispute exacerbating elements of e-mail communica-tion. Human Relations, 56: 1325–1347.

Frijda, N. H. 1994. Varieties of affect: Emotions and episodes,moods, and sentiments. In P. Ekman & R. J. Davidson(Eds.), The nature of emotion: Fundamental questions:59–67. New York: Oxford University Press.

Funder, D. C. 2001. Three trends in current research on per-son perception: Positivity, realism, and sophistication.In J. Hall & F. Bernieri (Eds.), Interpersonal sensitivity:Theory, measurement and applications: 319–331. Mah-wah, NJ: Lawrence Erlbaum Associates.

George, J. M., & Brief, A. P. 1992. Feeling good–doing good: Aconceptual analysis of the mood at work–organizationalspontaneity relationship. Psychological Bulletin, 112:310–329.

Gibson, D. E., & Schroeder, S. 2002. Grinning, frowning, andemotionless: Agent perceptions of power and their effecton felt and displayed emotions in influence attempts. InN. Ashkanasy, C. Hartel, & W. Zerbe (Eds.), Managingemotions in the workplace: 184–211. Armonk, NY: M. E.Sharpe.

Gross, J. J., Carstensen, L. L., Pasupathi, M., Tsai, J., Skorpen,C. G., & Hsu, A. Y. C. 1997. Emotion and aging: Experi-ence, expression, and control. Psychology and Aging, 12:590–599.

Grosvenor, L. 1998. Hybrid language: A study of e-mail andmiscommunication. Paper presented at the 45th annualconference of the Society for Technical Communication,Arlington, VA.

2008 325Byron

Grunwald, I. S., Borod, J. C., Obler, L. K., Erhan, H. M., Pick, L. H.,Welkowitz, J., Madigan, N. K., Siwinski, M., & Whalen, J.1999. The effects of age and gender on the perception oflexical emotion. Applied Neuropsychology, 6: 226–238.

Hall, J. A. 1984. Nonverbal sex differences: Communicationaccuracy and expressive style. Baltimore: Johns HopkinsUniversity Press.

Hallowell, E. M. 1999. The human moment at work. HarvardBusiness Review, 77(1): 58–66.

Higa, K., Sheng, O. R. L., Shin, B., & Figueredo, A. J. 2000.Understanding relationships among teleworkers’ e-mailusage, e-mail richness perceptions, and e-mail produc-tivity perceptions under a software engineering envi-ronment. IEEE Transactions on Engineering Manage-ment, 47: 163–173.

Ickes, W. 2001. Measuring empathic accuracy. In J. A. Hall &F. J. Bernieri (Eds.), Interpersonal sensitivity: Theory andmeasurement: 219–241. Mahwah, NJ: Lawrence ErlbaumAssociates.

Jett, C. 2005. E-mail etiquette. Free-Lance Star, July 30.

Jussim, L. 1991. Social perception and social reality: A reflec-tion-construction model. Psychological Review, 98: 54–73.

Kenny, D. A., & Winquist, L. 2001. The measurement of non-verbal sensitivity: Consideration of design, components,and unit of analysis. In J. A. Hall & F. J. Bernieri (Eds.),Interpersonal sensitivity: Theory and measurement: 265–302. Mahwah, NJ: Lawrence Erlbaum Associates.

Kernahan, C., Bartholow, B. D., & Bettencourt, B. A. 2000.Effects of category-based expectancy violation on affect-related evaluations: Toward a comprehensive model.Basic and Applied Social Psychology, 22: 85–100.

Kiesler, S., Zubrow, D., Moses, A. M., & Geller, V. 1985. Affectin computer-mediated communication: An experimentin synchronous terminal-to-terminal discussion. HumanComputer Interaction, 1: 77–104.

Kipnis, D., Schmidt, J., & Wilkinson, I. 1980. Intra-organiza-tional influence tactics: Explorations in getting one’sway. Journal of Applied Psychology, 65: 440–452.

Kock, N. 2005. Media richness or media naturalness? Theevolution of our biological communication apparatusand its influence on our behavior toward e-communica-tion tools. IEEE Transactions on Professional Communi-cation, 48: 117–130.

Kruger, J., Epley, N., Parker, J., & Ng, Z. 2005. Egocentrism overe-mail: Can we communicate as well as we think? Jour-nal of Personality and Social Psychology, 89: 925–936.

Ku, L. 1996. Social and nonsocial uses of electronic messag-ing systems in organizations. Journal of Business Com-munication, 33(3): 297–325.

Lea, M., & Spears, R. 1991. Computer-mediated communica-tion, de-individuation and group decision making. Inter-national Journal of Man-Machine Studies, 39: 283–310.

Lea, M., & Spears, R. 1992. Paralanguage and social percep-tion in computer-mediated communication. Journal ofOrganizational Computing, 2: 321–341.

Lengel, R. H., & Daft, R. L. 1988. The selection of communica-

tion media as an executive skill. Academy of Manage-ment Executive, 2(3): 225–232.

Markus, M. L. 1994. Finding a happy medium: Explaining thenegative effects of electronic communication on sociallife at work. ACM Transactions on Information Systems,12: 119–149.

Martin, J. 2004. Miss Manners: Apologies beforehand signal aweak effort. Milwaukee Journal Sentinel, January 21:http://www.jsonline.com/lifestyle/advice/jan04/201304.asp.

McKenna, K. Y. A., & Bargh, J. A. 2000. Plan 9 from cyberspace:The implications of the internet for personality and so-cial psychology. Personality and Social Psychology Re-view, 4: 57–75.

Mitra, A., Hazen, M. D., LaFrance, B., & Rogan, R. G. 1999.Faculty use and non-use of electronic mail: Attitudes,expectations, and profiles. Journal of Computer-Mediated Communication, 4: http://jcmc.indiana.edu/vol4/issue3/mitra.html.

Moenaert, R. K., & Souder, W. E. 1996. Context and anteced-ents of information utility at the R&D/marketing inter-face. Management Science, 42: 1592–1610.

Morris, J. A., & Feldman, D. C. 1997. Managing emotions inthe workplace. Journal of Managerial Issues, 9: 257–274.

Mossholder, K. W., Settoon, R. P., Harris, S. G., & Armenakis,A. A. 1995. Measuring emotion in open-ended surveyresponses: An application of textual data analysis. Jour-nal of Management, 21: 335–355.

Noller, P. 2001. Using standard content methodology to as-sess nonverbal sensitivity in dyads. In J. A. Hall & F. J.Bernieri (Eds.), Interpersonal sensitivity: Theory andmeasurement: 243–264. Mahwah, NJ: Lawrence ErlbaumAssociates.

Nowak, K. L. 2003. Sex categorization in computer mediatedcommunication (CMC): Exploring the utopian promise.Media Psychology, 5: 83–103.

Nowicki, S., Jr. 2005. Manual for the receptive tests of theDiagnostic Analysis of Noverbal Accuracy 2. Unpub-lished report, Emory University Department of Psychol-ogy, Atlanta, GA.

Nowicki, S., Jr., & Duke, M. P. 1994. Individual differences inthe nonverbal communication of affect: The diagnosticanalysis of nonverbal accuracy scale. Journal of Nonver-bal Behavior, 18: 9–35.

Orlikowski, W. J., Yates, J., Okamura, K., & Fujimoto, M. 1995.Shaping electronic communication: The metastructur-ing of technology in the context of use. OrganizationScience, 6: 423–444.

O’Sullivan, M. 1983. Measuring individual differences. InJ. Wiemann & R. Harrison (Eds.), Nonverbal interaction:243–270. Santa Barbara, CA: Sage.

Owens, D. A., Neale, M. A., & Sutton, R. I. 2000. Technologiesof status management: Status dynamics in email com-munications. In M. A. Neale, E. A. Mannix, & T. L. Griffith(Eds.), Research on groups and teams. Volume 3: Tech-nology: 205–230. Greenwich, CT: JAI Press.

Patterson, M. L., & Stockbridge, E. 1998. Effects of cognitive


demand and judgment strategy on person perceptionaccuracy. Journal of Nonverbal Behavior, 22: 253–263.

Pavitt, C. 1989. Accounting for the process of communicativecompetence evaluation: A comparison of predictivemodels. Communication Research, 16: 405–433.

Pavitt, C., & Johnson, K. K. 2002. Scheidel and Crowell revis-ited: A descriptive study of group proposal sequencing.Communication Monographs, 69: 19–32.

Rafaeli, A., & Sutton, R. 1987. Expression of emotion as part ofthe work role. Academy of Management Review, 12: 23–37.

Rezabek, L. L., & Cochenour, J. J. 1998. Visual cues in com-puter-mediated communication: Supplementing textwith emoticons. Journal of Visual Literacy, 18: 210–215.

Rice, R. E. 1987. Computer-mediated communication and or-ganizational innovation. Journal of Communication, 37:65–94.

Rotter, N. G., & Rotter, G. S. 1988. Sex differences in theencoding and decoding of negative facial emotions.Journal of Nonverbal Behavior, 12: 139–148.

Sarbaugh-Thompson, M., & Feldman, M. S. 1998. Electronicmail and organizational communication: Does saying“hi” really matter? Organization Science, 9: 685–698.

Schmitz, J., & Fulk, J. 1991. Organizational colleagues, mediarichness, and electronic email: A test of the social influ-ence model of technology use. Communication Re-search, 18: 487–523.

Shaver, P., Schwartz, J., Kirson, D., & O’Connor, C. 1987.Emotion knowledge: Further exploration of a prototypeapproach. Journal of Personality and Social Psychology,52: 1061–1086.

Sherblom, J. 1988. Direction, function, and signature in elec-tronic mail. Journal of Business Communication, 25: 39–54.

Snodgrass, S. E. 1992. Further effects of role versus gender oninterpersonal sensitivity. Journal of Personality and So-cial Psychology, 62: 154–158.

Sproull, L., & Kiesler, S. 1986. Reducing social context cues:Electronic mail in organizational communication. Man-agement Science, 32: 1492–1512.

Sternberg, R. J. 1996. Cognitive psychology. Fort Worth, TX:Harcourt Brace.

Straus, S. G., & McGrath, J. E. 1994. Does the medium matter?The interaction of task type and technology on groupperformance and member reactions. Journal of AppliedPsychology, 79: 87–97.

Sussman, S. W., & Sproull, L. 1999. Straight talk: Deliveringbad news through electronic communication. Informa-tion Systems Research, 10: 150–166.

Thompsen, P. A., & Foulger, D. 1996. Effects of pictographsand quoting on flaming in electronic mail. Computers inHuman Behavior, 12: 225–243.

Timmerman, P. D., & Harrison, W. 2005. The discretionary useof electronic media: Four considerations for bad newsbearers. Journal of Business Communication, 42: 379–389.

Trevino, L. K., Lengel, R. H., & Daft, R. L. 1987. Media symbol-ism, media richness, and media choice in organizations.Communication Research, 14: 553–573.

Von Glinow, M. A., Shapiro, D. L., & Brett, J. M. 2004. Can wetalk, and should we? Managing emotional conflict inmulticultural teams. Academy of Management Review,29: 578–592.

Wagner, H. L., MacDonald, C. J., & Manstead, A. S. R. 1986.Communication of individual emotions by spontaneousfacial expressions. Journal of Personality and SocialPsychology, 50: 737–743.

Walther, J. B., Anderson, J. F., & Park, D. 1994. Interpersonaleffects in computer-mediated interaction: A meta-analysis of social and anti-social communication. Com-munication Research, 21: 460–487.

Walther, J. B., & Burgoon, J. 1992. Relational communicationin computer-mediated interaction. Human Communica-tion Research, 19: 50–88.

Walther, J. B., & D’Addario, K. P. 2001. The impacts of emoti-cons on message interpretations in computer-mediatedcommunication. Social Science Computer Review, 19:324–347.

Walther, J. B., & Tidwell, L. C. 1995. Nonverbal cues in com-puter-mediated communication, and the effect ofchronemics on relational communication. Journal of Or-ganizational Computing, 5: 355–378.

Weisband, S., & Atwater, L. 1999. Evaluating self and othersin electronic and face-to-face groups. Journal of AppliedPsychology, 84: 632–639.

Weisband, S. P., Schneider, S. K., & Connolly, T. 1995. Com-puter-mediated communication and social information:Status salience and status awareness. Academy of Man-agement Journal, 38: 1124–1151.

Whissell, C. M. 1989. The dictionary of affect and language.In R. Plutchik & H. Kellerman (Eds.), Emotion: Theory,research, and experience: 113–131. New York: AcademicPress.

Witmer, D., & Katzman, S. L. 1997. On-line smiles: Does gen-der make a difference in the use of graphic accents?Journal of Computer-Mediated Communication, 2: http://www.ascusc.org/jcmc/vol2/issue4/witmer1.html.

Yates, J., Orlikowki, W. J., & Okamura, K. 1999. Explicit andimplicit structuring of genres in electronic communica-tion: Reinforcement and change of social interaction.Organization Science, 10: 83–103.

Kristin Byron ([email protected]) is an assistant professor of management at the Whit-man School of Management, Syracuse University, New York. She received her Ph.D.from Georgia State University. Her current research interests include emotions andnonverbal communication in organizations and employees’ responses to mistreat-ment.

2008 327Byron

carrying too heavy a load? the communication and miscommunication...

Documents