Flaming: The Relationship Between Social Context Cues and Uninhibited Verbal Behavior in Computer-mediated Communication

Mauri Collins

Fall, 1992

Hypothesis: In the absence of social context cues, the level of uninhibited verbal behavior in computer-mediated communication rises.

One phenomenon occurring in computer-mediated communication is the appearance of uninhibited verbal behavior. It appears that the level of uninhibited verbal behavior indulged in by those communicating via computer-mediated communication is a function of the absence of social context cues. I will first briefly examine social context cues as they effect communication content. These includes both verbal, non-verbal and situational cues. Then I will review the literature on uninhibited verbal behavior (often called 'flaming') in computer-mediated communication, to determine if there is support for the hypothesis that it is the lack of contextual cues that allows the phenomenon to occur, and if so, how researchers have explained the connection between the absence of social context cues and the occurrence of uninhibited verbal behavior.


The term 'social context cues' refers to the various geographic, organizational, and situation variables that influence the content of conversation among persons. Persons are usually sensitive to these social context cues and they can inhibit or facilitate what is said, how, and by who to whom. When defined as a person's physical position in time and place, geographic location can have a profound effect on communication. Discourse that is suitable for a bedroom is rarely suitable for the podium at a national convention, from a pulpit unsuitable for a bar on a Friday night. A business phone call made to someone's home number may originate quite properly at ten o'cloc in the morning, yet be most unwelcome arriving in another time zone at 5 am.

Organizational variables refers to a person's position in a job or social hierarchy, and their social position or job category. This often effects what can be said, when and to whom. Discourse occurs more often within organization levels, than among them.

Situational variables are characteristics of the immediate communication context that includes the hierarchical relationships among senders and receivers, the purpose of discourse and the norms and social conventions that apply to the situation. Situational variables can include the demographic characteristics of the communicators: age, gender, race, socio- economic status, residence and such personal characteristics as appearance, dress, accent, tone, mood, size, and attitude. Some norms and social conventions are fairly stable: be polite to your elders, don't talk back to you boss, getting rowdy in a bar on a Friday night is ok within limits, don't reveal personal information to people you don't know well, when in a meeting you attend to the highest status person present and allow them to lead the meeting. Other norms vary depending on the negotiated agreement among the persons present: clapping of hands, shouts of "Amen" during sermons and speaking out 'as moved by the spirit' may be accepted in some worship situations, while in others, worshippers may be expected to remain silent and in their seats when not making unison responses.


Sproull and Kiesler (1986) have an extended discussion on how social context cues affect information exchange between persons through perception, cognitive interpretation and communication behavior in their seminal article on the effect of the reduction of social context cues on electronic mail communication in organizations. Their theoretical framework explains how social context information mediates information exchange.

When a communication opportunity arises the participants must perceive the social context of the communication. Social status, and other social context cues can have no influence if they are not perceived. These cues can be static or dynamic, and may include static elements of the physical setting, like location and furniture, while the dynamic cues are drawn from people's non-verbal behavior in the communication situation. Perceived social context cues can engender cognitive interpretations and give rise to various emotional states. As people interpret a situation they adjust the focus, tone, mood and verbal content of their communications.

"Typically, when social context cues are strong, behavior tends to be relatively other focussed, differentiated and controlled. When social context cues are weak, people's feelings of anonymity tend to produce relatively self-centered and unregulated behavior. That is people are relatively unconcerned with making a good appearance (Cottrell, Wacj, Sekerak, and Rittle 1968). Their behavior becomes more extreme, more impulsive, and less socially differentiated (Diener, Fraser, Beamon, and Kelem 1976; Singer, Brush and Lublin 1965)." (pp. 1495-1496).

Face to face communication is the richest in social context cues and any form of mediated communication lessens the cues available. All that can be perceived over the telephone is the voice of the communicator, most static cues are stripped from the situation. The only cues that remain are, perhaps, the location of the person called (Beverly Hills, Watts), but this is often unknown in the case of a first-time conversation.

With most written communications, all dynamic cues are stripped, as well as all static cues, barring the feel and appearance of the writing materials. And yet norms remain attached to written communications, and their mystique and relative permanence can restrain grossly uninhibited verbal behavior, with the thought that someone else might read or save the correspondence. With e-mail correspondence, it has the illusion of ephemerality, appearing and disappearing from your screen (Sproull and Kiesler, 1991, p. 39). This is often more present in the minds of those who use e-mail than the idea that many persons can have access one's electronic mail, and may contribute to a lessoning of restraint.

Social context cues are almost totally lost when using electronic mail (e-mail) because dynamic cues are eliminated, and static cues are minimal. E-mail is the transmission of text between networked computers regardless of the physical distance between them, which may be a few feet or many thousands of miles. This transmission is very fast and asynchronous. Messages are delivered to the receiver's mailbox to be read at their leisure. However a e-mail reply can be returned within the time it takes to type the message on the screen, plus seconds or minutes of transmission time, across desk or across continents and oceans. Electronic mail is also text-based, without the voice component of the telephone or the visual element to the facsimile transmission. E-mail depends on almost solely on the contents of the text to create meaning between participants, thus favoring those who are literate and whose vocabularies are up to the task (Chesboro and Bonsall, 1989 p. 118).

Text-editing capabilities on most computer systems that facilitate the transmission of e-mail are frequently arcane and difficult to use. An alternative, while time consuming, is to compose your messages using a word-processor and then to upload the document to the machine equipped with the e-mail software and then send it. Most person using e-mail compose on-line with only rudimentary editing tools, so what comes off the fingertips is more likely to go unedited and often unconsidered to the correspondent, whether that be an individual or the many hundreds or thousands of persons reading a subscription-based or usenet discussion groups.

Text-based messages all look more or less the same. The sending machine will put lines of transmission information at the top of the note. Currently the only characters that can be transmitted are those that appear on the traditional type-writer keyboard, and the appearance of text on the screen is unaffected by any physical, social or emotional characteristics of the sender. A note from the CEO of a company looks exactly the same as one from the janitor.


Uninhibited verbal behavior is known, in computer user's jargon, as 'flaming'. This can consist of "swearing, shouting at their terminals, and groups refusing to make a group decision until a group member gave in" (Sproull and Kiesler, 1984, p. 1128); "to abuse, make offensive comments, or criticize sharply" (Kim and Raja, 1991, p 7); "using offensive language and being interpersonally insulting" (Matheson and Zanna, 1990, p. 1); "speaking incessantly, hurling insults, using profanity" (Baron, 1984, p 130). The Hacker's Dictionary defines flaming as "to speak incessantly and/or rabidly on some relatively uninteresting subject or with a patently ridiculous attitude . . Synonym: Rave.". (quoted in Sproull and Kiesler, 1991, p. 49)


As early as 1984 Baron conceived the computer as a force in language change, wondering if CMC "had evolved (or could evolve) some way of compensating for the lack of physical presence or non-linguistic context" (p. 125). She notes that computer conferencing appears to foster a distinctive and homogeneous conversational style, distinguished by the great frequency of arguments and flaming, and says that participants in computer conferencing were consistently rowdier than their face-to-face counterparts. Baron postulates three explanations for this phenomenon: the lack of non-linguistic and visual cues puts added pressure upon the users to use any means possible (such as haranguing) to ensure they are being understood; the fact that CMC is a relatively new form of communication and lacks the established norms of face-to-face conversation; and the lack social context cues masks status differences.

Kiesler, Siegel and McGuire (1984) published a seminal article which firmly established the connection between the absence of social context cues and the presence of uninhibited verbal behavior and is cited by all the following authors. They noted that previous research on CMC had examined the efficiency of computer information transmission systems in terms of their cost and technical capabilities and that it was their intention to examine the social psychological issues.

In a series of experiments designed to explore the impact of computer mediated communication (CMC) on group interaction and decision making, Kiesler et al. used groups of three students who were asked to reach consensus on a choice-dilemma problem in three different contexts: once face to face, once using the computer anonymously, (i.e. not knowing which one of their group was talking/typing) and once using the computer where each member knew when the other was 'talking'. Their data showed, "in all three experiments, that CMC had marked effects on . . . interpersonal behavior..." (p. 1128), in that 'people in CMC groups were more uninhibited than they were in face-to-face groups, as measured by uninhibited verbal behavior, defined as frequency of remarks containing swearing, insults, name calling and hostile comments" (p. 1129).

As part of this same series, an experiment was devised to see if CMC, by its nature, simply encouraged disorderly verbal conduct, or if flaming was a function of lack of constraints on interruptions and distracting remarks. However, the imposition of a strict order of contribution raised the level of uninhibited verbal behavior over the condition when no such restraint was imposed. The conclusion was that having to wait to take turns, when they had previously been able to type their comments in at will, raised levels of frustration which manifested itself in increased 'flaming'.

When allowed to synchronously communicate in computer- conferencing condition rather than by electronic mail (asynchronously), the level of uninhibited behavior rose. Kiesler et al. postulated the following three reasons for their results: "a) difficulties of coordination from lack of informational feedback, b) absence of social influence cues for controlling discussion, and c) depersonalization from lack of nonverbal involvement and absence of norms" (p. 1130). These findings have be elaborated upon and refined in subsequent research.

The Rand Organization (Shapiro and Anderson, 1985), recognizing an emerging need, published a position paper entitled Toward an Ethics and Etiquette for Electronic Mail. The authors noted the possibilities of misinterpretation of meaning inherent in a text-based communication medium which will

"allow casual and formal messages to look superficially the same; that allow near-instantaneous, rather than reasoned, response; that don't permit feedback during the delivery of a message (as in personal conversation); and that require modification to many old traditions of communication. A related phenomenon is "flaming," in which emotions are expressed via electronic mail, sometimes labeled as such, sometimes not." (p vii).

The occupance of flaming is attributed to the absence of cues to social hierarchical position, to the formality or informality of the communication; the absence of a buffering 'time lag' that might moderate response; the messages can attain a permanence that is not possible in verbal interaction (unless recorded); and persons are anonymous, beyond the bare information in their e-mail address, unless they choose to reveal further personal information, and the lack of non-verbal feedback that might moderate and augment the interpretation.

Some highly innovated suggestions were made concerning a future communications system that could imbue the text with some of the missing social content cues:

". . .the boldness of the characters displayed is a function of the force with which keys are hit; the speed at which it is typed is reflected in the character spacing (or color, or size, etc.). Or providing a set of standard forms to be selected, ranging from "Note from the desk of . . ." to "Corporate Memorandum" to give additional cues to the level of formality intended. Perhaps the most informal messages will be displayed in the handwriting of the sender (even though keyboarded for convenience) as an additional cue to its informality. More certainly (because the systems are in prototype form already) there will be systems in which the cold green (or amber, or whatever) characters will be accompanied by voice annotations, so that the humanity and state of the sender will be retained and "read" by the recipient" (p 41-42).

Sproull and Kiesler (1986) in their study of organizational communication determined that "e-mail reduced social context cues, provided information that was relatively self-absorbed, undifferentiated by status, uninhibited, and provided new information" and "people behaved irresponsibly more often on e- mail than they did in face-to-face conversations" (p. 1509) because it 'removed social reminders of norms" (p. 1510) Further they report that respondents who saw flaming in e-mail messages an average of 33 times month, only saw the same kinds of verbal behavior in face-to-face conversations an average of 4 times a month. Besides flaming, Sproull and Kielser also include in their discussion of uninhibited verbal behavior an increased willingness to pass on bad news or negative information and a flouting of social conventions. The particular convention they highlight was the boundary between work and play, noting that 40 percent of all e-mail traffic in the organization studied had nothing at all to do with work, but included movie reviews, recipes, notices of club meetings, etc. Overall, they found "evidence that electronic `mail reduced social context cues, provided information that was relatively self-absorbed, undifferentiated by status, uninhibited, and provided new information" (p. 1509)

The same year, Siegel et al. (1986) published further results from their experimental series at Carnegie-Mellon University. In this series they were examining the effects of CMC on communication efficiency, participation, interpersonal behavior and group choice. They note initially that CMC has only written text for a communication channel and that it lacks aural and visual cues and the social context information that one finds in face-to-face or telephone conversation.

They found that 'submergence in technology, and technologically-induced anonymity and weak social feedback might also lead to feelings of loss of identity and uninhibited behavior" which may cause people to become ". . . 'deindividuated,' leading not just to uninhibited verbal behavior and more equal participation," (p. 183) but to other, as yet uninvestigated effects. Siegel et al. also point out that the incidence of uninhibited verbal behavior may hinge on the direction of the communicator's attention. Perhaps this may be away from group and internal social standards and solely towards the message content, and also away from the behavior of, or even awareness of others as message senders.

Asch's social influence experiment was used as the basis for a study by Smilowitz, Compton and Flint (1988), investigating the effect of the exclusion of contextual cues provided by face to face interaction on individual judgement in CMC contexts. They determined that: "It is easier for a deviant to persist in the CMC environment. Since the effect of the majority opinion is diminished, individuals with deviant opinions are more likely to hold out that to succumb" (p 320). They attribute this to the absence of physical cues focussing attention exclusively on the text, the lack of a sense of others' presence to enforce social norms and the lack of non-verbal informational cues to encourage or discourage particular choices.

Chesebro and Bonsall (1989) indicate that when computers are used for messaging systems that the person dominates and that the computer becomes "merely a kind of elaborate typewriter and delivery system" (p. 97), however it "does affect the users" (p. 116). They note that CMC has a potential for reducing a correspondent's sense of personal responsibility to others, when all that one knows about others is characters appearing a screen. A further form of 'uninhibited behavior' they note is an extension of the concept of 'football widow' to encompass the complaints of wives whose husbands appear so engrossed in their computers as to have no time for social interaction. They also note that in the resolution of conflict in CMC, often engendered by flaming, that "more time and more words must be employed during teleconferencing to eliminate problems and conflicts" (p 123) indicating the difficulties in coordinating meaning in the absence of informational feedback.

Taking as a starting point that CMC users are more likely to display uninhibited verbal behavior, Matheson and Zanna (1990) experimentally examined the extent to which the use of CMC creates a state of deindividuation. The state of 'deindividuation' is characterized by a low sense of public and private self-awareness. In this state, the user loses touch with both their internal behavioral standards, and the awareness of social sanctions from others. When compared with face-to-face groups in problem solving situations, the CMC groups reported significantly higher levels of private self-awareness and marginally lower levels of public self-awareness. This precludes their being deindividuated. The researchers hypothesize that flaming may occur because, with a heightened sense of private self-awareness, users are more reactive to what they perceive as coercive behavior on the part of others, and less sensitive to the thoughts and feelings of others. ". . . this could lead to an escalating cycle of conflict and disagreement, and it could increase the display of affect and uninhibited behavior characteristic of computer users (Siegel et al., 1986). . ."

This study is significant in that it contradicts the position of Kiesler et al., (1984, 1985) and Siegel et al., (1986). who hypothesize that the user becomes so totally caught up in the act of communicating with the computer that they lose not only their private self-awareness, but their public self- awareness also.

Smolensky, Carmody, and Halcomb (1990) examined the extent to which tasks, and the degree to which users are acquainted with one another, will mediate the occupance of uninhibited verbal behavior. They determined that the amount of uninhibited verbal behavior was highest among triads who did know one another prior to the experiment, and those persons who were highly extroverted were likely to exhibit the highest levels of uninhibited verbal behavior. Groups with high levels of uninhibited verbal behavior also showed lower levels of productivity in terms of group decisions made. The authors explain their findings by hypothesizing that a reduction in social presence and social context cues in CMC caused users to perceive their CMC partners "as semi-mechanical objects which can be ignored, insulted, exploited, or hurt with relative impunity (Christie 1976)" (p. 269). They also raise the point that further research needs to be done to determine if limiting the amount of CMC will also limit the amount of creativity in groups.

Flaming is mentioned by Boshier (1990) in the same context as the use of emoticons, ascii symbols intended to add tone to the plain text of e-mail. He suggests that flaming occurs because the sender is far from the receiver and insulated from normal feedback, which, when offered face-to-face would lead to the extinction of the behavior. Boshier finds flaming to be a stimulating feature of e-mail in some if its incarnations, but the heat created by readers who 'viciously correct the spelling or grammar in other contributions' is termed 'tedious'.

"Do You Know Who You Are Talking To?" is the title of a chapter in Sproull and Kiesler's book Connections (1991). This book is a comprehensive discussion of the use of CMC in organizations. They see CMC as creating a new social situations, devoid of social context cues. While people "talk" with others, they do so alone, sitting at their computers, with few reminders of other persons and the social conventions for communication. Computer-based communication, they noted, relies on plain text to convey messages, and is ephemeral in nature, appearing and disappearing from the screen and leaving nothing tangible behind. These two characteristics, in combination, are seen as making it easy for the user to forget their audience, and feel unrestrained by conventional rules and norms of behavior. A blurring of social boundaries also contributes to the occupance of uninhibited verbal behavior, because plain text messages give no indication of status cues, unless the message is signed with a job or position title. With no clue in the messages as to a person's personal characteristics, their competence and ability can be recognized independent of the hierarchical position. This relative anonymity can lead to feelings of isolation and safety from "surveillance and criticism. This feeling of privacy makes them feel less inhibited with others. It also makes it easy for them to disagree with, confront, or take exception to others' opinion" (p. 49).

Sproull and Kiesler suggest several techniques to add context cues to CMC in order to remind users that they are social actors in a social situation. Emoticons, commonly known as "smilies" are an attempt to add mood indicators to CMC. :-) (tilt your head sideways to the left to see the smiley) indicates humor, or a smile to diffuse a possibly hostile reaction. People can change their personas in different communication situations, as professional writers always have. Adding temporal cues, time and date stamps, or setting very clear time lines in which work must be performed might increase social information and add guides to behavior, as will the writing of etiquette manuals. Sproull and Kiesler look forward to the time when sound and pictures will be added to e-mail, thus restoring many of the now missing context cues, and thus significantly lowering the occupance of uninhibited verbal behavior.

The theme of mechanomorphism briefly reference by Smolensky et al., (1990) is elaborated upon by Shamp (1991). With regard to uninhibited verbal behavior, Shamp acknowledges that "There is general agreement in the research that these behaviors and attitudes can be attributed to a lack of social presence of communication partners and an absence of social context cues" (p. 148), but he states such explanations are incomplete. In a study of messages sent to public bulletin boards, he determined that "when little personal information is available, an individual's perception of a computer communication partner becomes similar to his or her perception of the computer" (p. 149). Behavior that is not suitable when dealing with persons may be appropriate when dealing with machines and may be used on people when the people are perceived as machine-like.

Kim and Raja (1991) studied CMC in Usenet which carries over 1500 newsgroups discussing everything from rape to recipes, computers to communication. A large proportion of Usenet newsgroups readers comprise a subculture of computer, engineering, and scientific professionals, many of whom, despite their high levels of education, flagrantly breach conventional communication etiquette. They note that communication via computer terminal apparently allows users to forget they are communicating with other persons. This leads to levels of verbal aggression and self-disclosure that would be unthinkable in other types of communications among "strangers"

Using a sample of 600 messages, they coded them for 'face threatening acts' (FTAs), which they defined as evaluating some aspect of the other negatively and showing indifference towards the others' feelings. Their results show a difference in the level of such behavior among groups but a substantial number of FTAs in all newsgroups sampled.

The first reason they give for the occupance of uninhibited verbal behavior is a diminished desire to maintain the tacit cooperation in face saving usually occurring in face-to-face communication. When user must imagine their audience while sitting at a computer, it can appear that their only audience is the machine itself. The level of sympathetic involvement with others is attenuated and people don't need to be sensitive to other's feelings or messages, nor do the need to avoid impoliteness. The exchanges of rudeness can provoke a rapidly escalating series of aggressive messages.

When an audience must be imagined, the fear of physical retaliation as sanction for verbal aggression is non-existent, especially when the actual audience is geographically widely dispersed. There is a sense that there is very little that others can do to sanction one's behavior in CMC. If one person is rude, often the only remedy appears to be rude in return, again escalating the verbal conflict.

Within Usenet newsgroups, the only thing that can unite apparent enemies is any threat to their freedom to verbally abuse or malign one another. Appropriate verbal behavior on newsgroups is often learned by watching other participants, and where there is an established tradition of grossly uninhibited verbal behavior, a new group member is more likely to be socialized into perpetuating these behaviors, rather than changing them.

While one would expect, they say, that a medium that removes all social context cues, would lead to a undistracted discussion of ideas in a dispassionate, logical and issue-centered debate, the opposite seems to be true of CMC.

McCormick and McCormick (1992) collect samples of undergraduate student's e-mail to test their hypothesis that the computer was a powerful, but neutral technology by looking at the full range of their communication rather than just pinpointing the flames. Less than half of the sample of messages were work related, the rest serving purely social functions. About a quarter of all messages showed high intimate content and "unexpectedly little showing signs of flaming or hostility and social inappropriateness" (p. 379). They noted, following Sproull et al., (1986) that flaming is especially commonplace in the male-dominated, adolescent subculture of the college computer center, "which encourages pranks, idiosyncrasy, and irreverence" (p. 389). Without empirical evidence, the researchers note that the flames appeared to be exchanged by young men who knew each other well. This is consistent with other research reported above. They felt that flaming was akin to the mock physical battles (also a sign of affection and trust) that occur among male adolescents, rather than a sign of dislike or alienation. It appeared to affirm the strength of the relationship that they could call each other names and still remain friends.

This study contrasted sharply with experimental work with unacquainted individuals in which computer-mediated communication, compared to face-to-face discussion, increased the likelihood of disinhibited and socially offensive behavior However the respondents in McCormick and McCormick's were local uses of a system who could expect to continue intensive, future interaction, thus rendering them a little more civil than those who expected to never meet their communication partners.

Over the past decade, there has been consistent support in the literature for the hypothesis that in the absence of social context cues the level of uninhibited verbal behavior in CMC rises. Kiesler and her associates began to articulate the theoretical underpinnings for exploration of social context cues and the effects they have on CMC. Other researchers have furthered that knowledge and attempted to describe some of the variables affecting the phenomenon. In the literature identified to date, all has supported the contention that flaming rises as there are fewer and fewer social context cues within a CMC environment.


Baron, N.S. (1984) Computer mediated communication as a force in language change. Visible Language, 18 (2):118-141.

Boshier, R. (1990) Socio-psychological factors in electric networking. International Journal of Lifelong Education, 9 (1), 49-64.

Chesebro, J.W., & Bonsall, D.G. (1989) Computer-mediated communication: Human relationships in a computerized world. Tuscaloosa, AL: The University of Alabama Press.

Kiesler, S., Siegel, J., and McGuire, T. (1984) Social psychological aspects of computer-mediated communication. American Psychologist, 39 (10): 1123-1134.

Kim, M.S. & Narayan, S. (1991) Verbal aggression and self- disclosure on computer bulletin boards. Paper presented at the Annual Meeting of the International Communication Association, Chicago, IL, May. ED334 620

Matheson, K., & Zanna, M.P. (1990) Computer-mediated communications: the focus is on me. Social Science Computer Review, 8 (1):1-12.

McCormick, N.B., & McCormick, J.W. (1992) Computer friends and foes: Content of undergraduates' electronic mail. Computers in Human Behavior, 8:379-405.

Scovell, P. (1991) Differences between computer and non computer- mediated communication: A preliminary study. Paper presented at the Eastern Communication Association Convention, Pittsburgh, PA, April. ED 333 520.

Shamp, S.A. (1991) Mechanomorphism in perception of computer communication partners. Computers in Human Behavior, 7: 147-161.

Shapiro, N.Z., & Anderson, R.H. (1985) Toward an ethics and etiquette for electronic mail. Santa Monica, CA: The Rand Corp.

Siegel, J., Dubrovsky, V., Kiesler, S., and McGuire, T.W. (1986) Group processes in computer-mediated communication. Organizational Behavior and Human Decision Processes, 37: 157-187.

Smilowitz, M., Compton, D.C., and Flint, L. (1988) The effects of computer mediated communication on an individual's judgement: A study based on the methods of Asch's social influence experiment. Computers in Human Behavior, 4: 311- 321.

Smolensky, M.W., Carmody, M.A. and Halcomb, C.G. (1990) The influence of task type, group structure and extraversion on uninhibited speech in computer-mediated communication. Computers in Human Behavior, 6:261-272.

Sproull, L. & Kiesler, S. (1991) Connections: New ways of working in the networked organization. Cambridge, MA: The MIT Press.

Sproull, L. & Kiesler, S. (1986) Reducing social context cues: electronic mail in organizational communication. Management Science, 32 (11): 1492-1512.

URL: http://cac.psu.edu/~mauri/papers/flames.html