Breaking out of Binaries: Reconceptualizing Gender and its Relationship to Language in Computer-Mediated Communication

Authors

  • Michelle Rodino

    Corresponding author
    1. (MA, Northwestern University; BA University of California at Los Angeles) is a doctoral student at the University of Washington. Her research interests include computer-mediated communication, feminist media studies, organizational communication, and critical theory. She has presented papers in these areas at the annual conferences of the International Communication Association, the National Communication Association, and the Eastern Communication Association.
    Search for more papers by this author

  • The author wishes to thank Professor Diane Gromala at the University of Washington and Professors Roberta Astroff, Janet Skupien, and Carol Stabile at the University of Pittsburgh for their insightful comments. I also thank Professor Gromala for supporting this research. Bill Sucevic at Computing and Information Systems at the University of Pittsburgh and David Brahm provided technical assistance that facilitated this project. Data analyzed in this project was reported in a paper presented at the 1995 convention of the National Communication Association, Feminist and Women Studies Division, San Antonio, TX. Research for this project began at the University of Pittsburgh.

School of Communications, University of Washington, Box 353740, Seattle, WA 98195, USA

Abstract

Virtual environments provide a rich testing ground for theories of gender and language. This paper analyzes interactions in one virtual environment, Internet Relay Chat (IRC), to look at the extent to which research on face-to-face (FTF) talk and computer-mediated communication (CMC) can describe gender and its relationship to language. I find that neither the function of utterances nor the construction of gender adheres to dualistic descriptions, as past research has implied. Reconceptualizing gender as performative helps researchers break out of binary categories that have bound past research. Conceiving of gender as under constant construction also helps demystify and thus disrupt the binary gender system which naturalizes patriarchy.

Introduction

Virtual environments provide a rich testing ground for theories of gender and language. This paper analyzes communication in a text-based virtual environment[i] to look at the extent to which past research can describe gender and its relationship to language. Analysis of interactions on Internet Relay Chat (IRC), a collection of synchronous chat channels, suggests that utterances cannot be adequately described using binary functional categories like “support work” (facilitative) and “controlling” [(Fishman, 1978/1983)] or “personal” and “authoritative” [(Herring, 1993)]. Such designations are grounded in a dichotomous conceptualization of gender that reflects patriarchy's demands rather than an individual's biological composition [(Butler, 1990; Bem, 1993; Bing & Bergvall, 1996; Freed, 1996)].

This project was inspired by feminist linguists who began asking the “question of questions”: why do we ask questions that strengthen the male-female dichotomy? [(Bing & Bergvall, 1996)]. This critical question implies that looking for binary gender differences in language helps recreate them. The dualistic gender system poses several problems for feminist linguists. First of all, this hegemonic system does not accurately describe the array of multiple un/gendered traits that individuals exhibit [(Epstein, 1990; Bem, 1993; Bornstein, 1994/1995)]. Gender constructions in IRC, like those in real life (RL), do not necessarily fit into opposing categories such as male/female, gendered/gender-neutral, or into male/female/gender neutral groupings. Computer-mediated interlocutors may perform masculinity, femininity, gender neutrality, some combination, or none of these [(Curtis, 1992; Bruckman, 1993; Danet, 1996)]. Thus, basing descriptions of gender on a binary system creates representation problems. However, “gender polarization”[1] has more insidious effects; it is used to rationalize oppression [(Bem, 1993; Bing and Bergvall, 1996; Cameron, 1996)]. Those who do not fall neatly into male or female categories face ostracism, discrimination, and repression. Furthermore, those who do fall neatly into the female category may face similar fates because they are viewed as inferior to men [(Lakoff, 1975)].[2] [Bem (1993)] has also argued that, in an androcentric system, those who are “other” to heterosexual, masculine men face oppression. Thus, reproducing the binary gender system sustains rather than weakens patriarchy.

[Bing and Bergvall (1996)], along with several of the authors whose chapters they introduce, urge scholars to use [Butler's (1990)] conceptualization of gender as a series of performances. By viewing gender not as a stable quality but as something that exists only in the works of its production, one can more fully represent the many ways in which gender is experienced and exhibited. Butler draws on Nietzche's claim that the subject does not precede action; “there is no ‘being’ behind doing…the ‘doer’ is merely a fiction added to the deed” [(Nietzsche, 1967/1989, p. 45)]. By looking at gender as “doing,” feminists can avoid the trap of recreating gender as a set of fixed, opposed traits.

However, thinking of gender as “doing” does not guarantee that one will avoid reifying binary gender. Although rethinking gender as performative allows the concept to include a wider array of gender constructions, describing distinct gender productions and ascribing them to differences in biological sex reinforces the binary system.

Furthermore, using a gender-as-performance perspective does not mean that one can ignore gender polarization; it persists on line as it does off. Research suggests that some CMC participants prefer to interact with players who can be distinguished as male or female [(Curtis, 1992; Bruckman, 1993; Reid, 1993; Turkle, 1995)]. Preconceptions linked to gender bifurcation also endure in cyberspace. Studies have found that when users are assumed to be female (whether they are told by researchers in an experimental setting, or infer from user- or nicknames) they may be perceived stereotypically as too talkative [(Herring, Johnson, and Dibenedetto, 1992)], more cooperative [(Matheson, 1991)], in need of technical assistance [(Bruckman, 1993)], or may be sexually harassed [(Kramarae & Taylor, 1993; Petersen, 1994; Spender, 1995)]. The present study confirms that individuals present themselves and are treated on line according to the binary gender system. Analysis of IRC shows an interlocutor who consistently performed masculinity and a female-presenting IRCer who attracted the sort of attention women typically receive on line [(Bruckman, 1993)].

Because the concept helps expose and explain certain forms of prejudice and oppression, I do not propose abandoning gender as an object of study. Instead of discarding gender, I suggest that the concept be retooled. Thinking about gender as continually constructed allows one to look at the various, sometimes inconsistent ways in which a person presents gender. The present project finds that while some IRC players provide stable representations, others give contradictory performances. Similarly, while some IRCers reflect and recreate gender bifurcation, others break out of binary gender categories. In order to enrich analyses of such diverse gender productions without sacrificing the ability to look at women's oppression, the present analysis recommends retaining certain aspects of gender's conceptualization and jettisoning others. This paper also extends the analysis of gender performance in IRC beyond the point at which users select nicknames. Finally, I conclude that conceiving of gender as under constant construction helps demystify and thus disrupt the binary gender system which naturalizes patriarchy.

Let us now look at past research on gender and language in FTF and computermediated modes.

Gender and Language Off Line

Ironically, research on the relationship between gender and language has helped reify gender differences. In her seminal article on gender, language, and power, [Lakoff (1975)] argues that women's conversational strategies are affected by their oppression under patriarchy. What Lakoff calls “women's language” is marked by powerlessness in the forms of “superpoliteness,” qualifiers, exaggerations, and tag questions. These impuissant conversational forms provide “diagnostic evidence” for the inequality between men and women (p. 4). Lakoff further theorizes that women are socialized to speak in ways that are perceived as weak, a process which, in turn, reproduces their oppression. Thus, Lakoff argues that there are important differences between men's and women's speech.

Building on this research is [Fishman's (1978/1983)] work that looks at men's power over women in everyday interactions. Fishman argues that formal features that characterize women's speech (e.g., asking questions) seek to insure response. In contrast, men's speech is marked by features (e.g., statements) that do little to insure further talk. However, interlocutors are more likely to “orient to” topics that men raise; “the definition of what is appropriate or inappropriate conversation becomes the man's choice” (p. 98). In other words, men control talk. Because women work harder than men do to facilitate conversation, Fishman calls women the “‘shitworkers’ of routine interaction” (p. 99). Thus, Fishman uses the binary facilitative/controlling to describe men's and women's speech, respectively.

Subsequent research, however, has criticized [Lakoff's (1975)] and [Fishman's (1978/1983)] claims. Some studies suggest that interactional context affects the extent to which men's and women's speech can be distinguished [(Crosby & Nyquist, 1977; O'Barr & Atkins, 1980; Nichols, 1983; Mulac & Lundell, 1986; Mulac, Wiemann, Widenmann, & Gibson, 1988; Cameron, McAlinden, & O’ Leary, 1989; Graddol & Swann, 1989; Swan, 1989; James & Drakich, 1993; Freed, 1996)]. [2] Lakoff's and Fishman's critics also argue that an utterance's function cannot be placed into a single category or be known in advance. Tag questions, [Cameron et al. (1989) argue, are “characterised by complex multifunctionality and diversity of meaning” (p. 85). [Swann (1989)] faults Fishman's work for implying an a priori form-function relationship in her description of the formal features of women's conversational “shitwork.” [Graddol and Swan (1989)] argue that discrepancies between men's and women's speech are “differences of degree” (p. 89) and that lists of women's and men's speech styles constitute a “gross over-simplification” (p. 89). Thus, representing male and female language as a binary and using other dichotomous categories to describe language overlook complexities in actual speech.

Speech is also described using dualistic categories in later research. In reaction to Lakoff's and Fishman's “dominance” perspectives, which focus on women's oppression in everyday interactions, [3] feminist linguistic scholarship began to attribute discrepancies in male and female speech patterns to “cultural differences” [(Cameron 1985/1992; Cameron, 1996)]. [Tannen's (1990/1991)] work exemplifies the “cultural difference” approach [(Cameron, 1996).] [4] While Tannen does not deny the existence of male dominance in society, her analysis downplays patriarchy's role in producing linguistic differences. Tannen argues that “because boys and girls grow up in what are essentially different cultures…talk between women and men is cross-cultural communication” (p. 18). Cultural differences explain why women use conversation to connect with others and for “negotiating relationships” (p. 77). For men, conversation is for “holding center stage” and for “get[ting] and keep[ing] attention” (p. 77); talk maintains independence and status. Tannen calls women's “rapport-talk” and men's “report-talk.” Thus, speech functions are constructed using the binary rapport/report.

Why have studies used opposing categories to describe language? Dichotomous categories of speech patterns are often grounded in a binary conceptualization of gender [(Freed, 1996)]. Freed contends that communication researchers have looked at utterances through the “lenses of gender” that [Bem (1993)] describes. Gender appears as a binary through these lenses because they link gender to bipolar, “biological” sex. As a result, the lenses of gender influence if not predetermine the sorts of conclusions communication scholars draw.

[Freed's (1996)] and [Bem's (1993)] work also suggests that to eradicate patriarchy feminist scholars must move beyond studying gender differences in communication. Focusing on such distinctions recreates the myth that males and females are discrete, opposing groups; this myth is then mobilized in the service of gender-based oppression. Thus, doing research that recreates such myths works against feminist goals of ending such oppression. [Bing and Bergvall (1996)] contend that “attempts to prove difference are often attempts at gender polarization and [are] one way to rationalize limiting the opportunities of women” (p. 17). [Bem (1993)] argues that in the US, and in other Western societies, gender polarization is accompanied by androcentrism, a system that treats “male” as the neutral standard and “female” as the deviant other. Such treatment is naturalized (made to seem natural and universal) through biological essentialism which claims that gender differences are genetic and ahistorical. Viewing society through the “lenses” of gender polarization, androcentrism, and biological essentialism helps reproduce patriarchy by rationalizing the domination and control of women. [5]

The self-help industry provides an example of using linguistic gender differences in the service of patriarchy and capitalism. This industry profits from descriptions of linguistic gender distinctions by telling women how to improve their speech (make it more like men's) or how to cope with male interlocutors [(Cameron, 1985/1992; Cameron, 1996)]. Women are thus doubly exploited; their own oppression is sold back to them.

To combat such exploitation, specifically at the hands of the self-help industry and more generally, through reproduction of fictions used to oppress women, [Cameron (1996)] and [Bing and Bergvall (1996)] call on feminist linguists to use Butler's (1990) conception of gender. Butler views gender as under constant construction, a series of performances, a work-in-progress and -in-practice. [6] Conceptualizing gender as a sum of productions allows feminists to challenge the binary gender system which “implicitly retains the belief in a mimetic relation of gender to sex whereby gender mirrors sex or is otherwise restricted by it” (p. 6). Butler contends that feminists’ insistence that gender is a cultural construction was meant to counter the biology-as-destiny formulation. Pushing the argument that gender is culturally constructed to its logical limit, as Butler does, suggests that gender is not an expression of sex and can be constructed independently of it. Thus, looking at gender as performative makes the break with biology that the notion of gender has been able to forge but has not yet done.

Kate Bornstein's production of gender illustrates the extent to which gender can be performed independently of one's biological sex. In accounting for her transgender operation, [Bornstein (1994/1995)] says, “I never did feel like a girl or a woman; rather, it was my unshakable conviction that I was not a boy or a man” (p. 24). This conviction and the surgery and hormone therapy Bornstein undergoes show that one's “biological” sex can flow from one's gender instead of vice-versa.

Furthermore, [Bornstein's (1994/1995)] discussion of passing highlights the constant construction of gender. It is not enough that Bornstein underwent surgery or changed her name from Al to Kate. Bornstein must also play the part of a woman to pass as one. She has studied ways of changing her voice, avoiding eye contact, and exiting elevators to hide her male past and communicate femininity. [Garfinkel's (1967)] description of the work done by Agnes, a transsexual and formerly intersexed woman, to pass as a “natural, normal female” also illustrates gender's creation through continual “impression management” [(Goffman, 1959)]. Agnes manages her beach attire, living with a female friend, and dating so that she is able to hide her intersexed past and project being a “natural, normal female.” Like Bornstein's, Agnes’ gender work is never finished but is always under construction.

On the other hand, Bornstein's and Agnes’ gendered productions demonstrate the extent to which gender cannot be performed independently of one's biological sex. Agnes’ sex organs and gender match; Bornstein's surgery reflects her “unshakable conviction” that she is not male. Furthermore, passing as the embodiment of one's new gender involves hiding one's formerly sexed past [(Garfinkel, 1967; Bornstein, 1994/1995)]. Thus, one could read transsexual make-overs as conforming to a system that, as Bornstein puts it, “insists we be one [gender] or the other” (p. 8). Bornstein realizes the irony of this conformity and points out that concealing one's prior gender history to pass upholds the system that transsexuality could radically challenge.

However, intersexed and transsexual individuals are not the only ones who exert efforts to pass. Those born with only male or female genitalia and those who identify with only one gender must also dress, speak, walk, and run the part. [Bem (1993)] suggests that gender polarization expands the notion of biological difference to the point where “men and women [are] vulnerable to the feeling that their maleness or femaleness cannot be taken for granted but must instead be worked at, accomplished, and protected from loss through misbehavior” (p. 194). Thus, the performance of gender is no natural, ahistorical behavior; gender performance is an effect of gender polarization. Gender polarization, however, is an effect of patriarchy. Gender is displayed in reaction to the structural demands of patriarchy, which relies on a marking system that distinguishes women from men. Gender performances are part of this marking system. As I will later suggest, looking at this marking system as a cultural demand disrupts the premises on which patriarchy rests and, thus, can help end women's oppression.

In addition to gender, biological sex is also socially constructed [(Butler, 1990; Bem, 1993; Nicholson, 1994)]. Butler further argues that sex is discursively constructed. However, the discourse of gender hides sex's discursivity by constituting gender as culturally produced and sex as naturally (biologically) produced. In other words, gender as a discursive apparatus constructs sex; “perhaps it was always already gender, with the consequence that the distinction between sex and gender turns out to be no distinction at all” (Butler, p. 7). [7] Thus, claiming that gender is dichotomous relies on a circular argument rather than on biological facts.

To summarize the critiques of gender and language research: asking questions about how men's and women's speech patterns differ creates problems for feminist linguists. Describing speech features as “male” and “female” overlooks the extent to which context influences the ways in which utterances operate. Also, contrasting “male” and “female” language reifies differences between men and women. Such distinctions help rationalize women's oppression. Conceptualizing gender as a sum of performances allows researchers to better represent the ways individuals experience gender and communicate. Rethinking gender along these lines also helps expose biological essentialism, the binary gender system, and patriarchy as cultural constructions.

Now let us look at the ways in which studies of and players on CMC channels have recreated and attempted to shake off such binaries.

Gender and Language On Line

Much of the scholarship on gender and CMC is based on studies of gender and FTF communication. However, gender and CMC research has had to address an additional set of assumptions. During the time that research on and critiques of gender and language were developing, computers were being touted for their democratizing potential. The “equalization view” constitutes some of these arguments. This perspective posits that CMC “extends and equalizes information exchange…[and] can serve to reduce the social barriers to communication, and thus the impact of status differentials, resulting in greater equality of participation” [(Spears & Lea, 1994, p. 428)]. In other words, because CMC reduces social cues, it democratizes communication. [8]

The “equalization view” of CMC challenges “dominance” research's findings that men control conversation. In agreement with the equalization view and against dominance research, [Graddol and Swan (1989)] found that men and women contributed approximately the same number of messages on a university electronic conferencing system. Among the attributes contributing to such democratization were dearth of social cues and lack of a hierarchical system indicating formal rights to talk. [9]

Other research indicates mixed support for the equalization view. [Jaffe, Lee, Huang, & Oshagan (1995)] found that women selected pseudonyms that hid their RL gender more than males did. Such masking “might reflect an effort to maintain a parity of status in the shared activity of conversation” (p. 15). If women's “parity” on line, however, depends on concealing their RL gender, then it seems doubtful that men and women possess equal power in the networks. Support for both equalization and dominance views is also found in [Selfe and Meyer (1991)]. This study reveals that women found pseudonymous electronic talk “liberating.” Pseudonymous communication also seemed to encourage talk about gender and power. On the other hand, Selfe and Meyer report that men talked more than women did in pseudonymous sessions.

While some research attributes CMC's democratizing effects to reduced social cues, other studies find CMC liberatory because such contexts allow users to create gender in ways that defy cultural norms. [Reid (1993)] argues that users construct gender through nicknames they choose. [10] As a result, “[gender's assumed] fixity, and the common equation of gender with sex, becomes problematic when gender reassignment can be effected by a few touches at a keyboard” (p. 63). In other words, gender production in CMC opens up gender conceptually in ways that [Butler (1990)] describes; CMC helps disarticulate gender from “biological” sex. Reid also contends that the “freedom” that gender reconstruction allows may overwhelm or delight users. Either way, however, “the users of IRC show the degree to which the medium challenges and obscures the boundaries between some of our most deeply felt cultural significances” (p. 64). Thus, Reid attributes IRC's liberatory effects to the channel's structural demands; IRCers construct gender (or gender neutrality) through nickname choice.

Other studies on gender and CMC dispute arguments that such contexts are liberating. Critiques of democratizing views show that online gender construction does not deter male domination. [Herring (1993)] reports that a small minority of males led conversation and seemed to think that they should control discussion on an academic list. Women's speech was also censored, an indication that “academic CMC is power-based and hierarchical” [(Herring, 1993, p. 11)]. Additional research reveals that men post twice as many electronic messages as do women [(Selfe & Meyer 1991; Herring, Johnson, and Dibenedetto, 1992; Herring, 1993)]. Men's control of CMC also extends to women-related and women-only bulletin boards [(Kramarae and Taylor, 1993)].

Nor does online gender construction preclude the formation of gender stereotypes. [Matheson (1991)] found that women who were informed that their computer-mediated interlocutors were female expected their partners to act more cooperatively than did women who were unaware of their interlocutors’ gender. Acting cooperatively is a stereotypically feminine behavior. Matheson does not explicitly discuss the relationship between such stereotypes and power in CMC. However, because gender stereotypes bolster patriarchy by helping to rationalize women's oppression [(Bem, 1993; Bing and Bergvall, 1996; Cameron, 1996)], the persistence of gender stereotypes Matheson finds suggests that male power endures in CMC.

However, men's domination of CMC does not mean that women are ignored on line; attention women receive in the networks is often indicative of their sexual objectification in RL. In a study of communication in MUDs (Multi-User Domains/ Dungeons/Dialogues), [11] [Bruckman (1993)] found that female characters had received more sexual attention (wanted or unwanted) and offers of technical assistance than male characters had. Offers of technical assistance were often accompanied by requests for sexual favors. Additional evidence of online sexual harassment can be found in [Kramarae and Taylor (1993), Petersen (1994), and [Spender (1995)].

As does research on FTF communication, studies that provide evidence for or against women's online oppression also help reify men and women as distinct groups [(Graddol and Swan, 1989; Matheson, 1990; Selfe & Meyer 1991; Herring et. al., 1992; Bruckman, 1993; Herring, 1993; Kramarae & Taylor, 1993; Petersen, 1994; Jaffe et al., 1995; Spender, 1995)]. Research that considers the relationship between gender and power in language necessarily confronts binary gender, because looking at this relationship means looking at “men” and “women.” The binary is always already constructed when one considers women's oppression in CMC, because women's oppression has been described in relation to male domination. Projects that look at women's oppression in CMC, however, would be fortified by incorporating a performative view of gender. Conceptualizing gender as performative helps expose the unnaturalness of binary gender and allows binary gender to appear as a discursive apparatus rather than as a natural, taken-for-granted trait.

Studies on gender and CMC also look at differences in interactional styles. Such research supports dichotomies used to describe gender and linguistic features in FTF contexts. Complementing [Tannen's (1990/1991) claim that women tend to use conversation as “rapport” rather than “report” talk is [Kaplan and Farrell's (1994)] finding that for girls, “the sociability of this [online] exchange seems its sole reason for being” (p. 12). Further supporting the existence of women's report and men's rapport CMC styles is evidence that women tend to display textual patterns of social interdependence more than men do in both real-name and pseudonymous conferences [(Jaffe et al., 1995)]. Jaffe et al., however, report that in pseudonymous discussion men exhibited greater social interdependence than they had in real-name sessions. Also in agreement with Tannen's conclusions are [Herring's (1993)] findings that women contributed most to “personal” discussions, and men contributed most to talk about “issues” on an electronic academic bulletin board. In addition, women's speech is described as “personal” and men's as “authoritative.”

Men's and women's speech is also found to differ in other ways. [Witmer and Katzman (1997)] report that while neither men nor women used graphic accents frequently, those who did tended to be women. Female participants also uttered more challenging speech and flamed more than men did. Witmer and Katzman speculate that women's tendency to challenge and flame may be explained by research that describes women's speech as more emotionally expressive than men's. Other research shows that electronic discussion groups with a larger proportion of men used more “male” language patterns such as “fact oriented [sic] language” and calls for action than did groups with smaller proportions of men [(Savicki, Lingengelter, & Kelley, 1996)]. The latter groups exhibited more “female” speech characteristics like “self-disclosure” and “attempts at tension prevention and reduction.”

It is worth noting that studies on gender and CMC vary in the extent to which they associate linguistic differences with power asymmetry. [Witmer and Katzman(1997)] and [Savicki et al. (1996)] describe gender differences in CMC but do not attribute such diversity to power discrepancies between men and women. Witmer and Katzman and Savicki et al. focus on difference rather than dominance. [Herring (1993)] and [Jaffe et al. (1995)], however, argue that men's domination of women accounts for gender differences in CMC; power is thrown into relief in these works.

Like difference and dominance approaches to gender and FTF communication, those applied to gender and CMC recreate the male/female split. My analysis of IRC reveals some of the problems this split causes. One problem involves representing one's object of study. The male/female dualism does not reflect the variety of ways in which gender is presented or experienced in virtual environments. Nor do dichotomous descriptions of speech features, which rest on gender polarization, represent multifunctional utterances in IRC. Furthermore, marking differences between men's and women's speech supports the notion that “male” and “female” are discrete, opposing categories. Making such distinctions helps maintain the oppressive binary gender system.

Research that would be most useful for breaking out of gender polarization would look at the ways in which online gender is produced through performance. Several studies have looked at gender's construction at the moment in which players enter virtual space. [Reid (1993)] argues that IRC users construct gender through their nicknames; nicks may express masculinity, femininity, or gender ambiguity. [Curtis (1992)] and [Bruckman (1993)] discuss the process of gender construction in MUDs via character selection; MUDders may choose gendered, gender-neutral, or gender plural characters when they join a MOO (a type of MUD). In agreement with Reid, Curtis, and Bruckman, [Danet (1996)] argues that nicknames indicate gender (or gender neutrality) in IRC, while character choice determines gender in MUDs.

However, none of these studies considers the constant production of gender. [Curtis (1992), Reid (1993), Bruckman (1993)], and Danet (1996)] provide a starting point for analyzing gender performance but do not look in any great detail at the process of gender making after players have selected nicks and characters. Thus, these studies imply that selecting a nick or character fixes gender. Conceptualizing gender as performative, however, means eschewing notions of gender as a stable, static trait [(Butler, 1990)]. Looking at gender performance requires examining gender's continual production. Furthermore, a performative conception of gender takes seriously Nietzche's argument that “there is no ‘being’ behind doing.” Butler's and Nietzche's claims suggest that gender construction is never done, never finalized. One may knock on a door like a woman and enter a room like a woman, but to continue to appear as a woman, one must continue displaying cues that signify woman. Research that examines gender construction in CMC needs to look at such production beyond the front door. The present project analyzes gender's fabrication through and beyond the moment in which players enter IRC. Thus, this study extends moves that reconceptualize gender in ways that break out of binaries and embrace the performative.

Method

To look at the multiple, sometimes contradictory ways in which users constantly perform gender and in which utterances function in CMC, I qualitatively analyze a continuous, forty-minute stream of conversation on an Internet Relay Chat (IRC) channel. Four-hundred and fourteen lines of text were selected from observations made over a ten-week period, beginning September 27, 1994. Most of these observations were recorded. I made observations by entering chat channels and lurked (entered no text). I watched interactions on chat channels: #boston, #chat, #chatzone, #gaysex, #hottub, #ircbar, #romance, #talk, #teenchat, #texas, #truthdare. [12] The sequence analyzed here comes from #talk.

I selected this conversation for analysis because it provides a rich sample of the ways in which gender is constructed and the ways in which utterances operate. The chat session contains a series of interactions by male, female, and gender-ambiguous characters that developed with minimal interruptions by channel surfers and robots. Interlocutors also performed a range of actions and self-disclosures.

This project's textual analysis of a single conversation is similar to [Danet et al.'s (1997)] analysis of one and one-half hours of continuous interaction in which players simulated smoking dope at a “virtual party.” While these qualitative approaches preclude generalizing conclusions to all IRC or CMC contexts, such methods can be used to explore interactional contexts in greater detail than can statistical studies.

Regardless of methods used, however, generalizing across all IRC and CMC contexts seems nearly impossible. Interactions in IRC channels vary in topic and task. Many of the chat channels are informally organized (#talk, #chat, and #hottub). In contrast, other channels concentrate on topics like computers (#macintosh) or on specific objectives (#truthdare). [13] Thus, scholars may need to limit claims about IRC to “chat channels about computers” or “chat channels that host games.” In addition, differences between talk in #truthdare and talk in electronic academic conferences suggest that conclusions about one CMC context may not be generalizable to others. [14]

The present study does not use interviews or surveys to get at what users intended to say or mean. Analyzing conversation without such data, however, allows me to interpret only those interactions that are accessible to IRCers. Thus, my interpretation reflects players’ performances as they are delivered in synchronous chat and is not influenced by users’ thoughts about such enactments.

It is also important to note that claims made in this study are based on findings from an English-speaking channel. Interpretations of language reflect this specificity. In addition, my Western, American socialization and education probably influence interpretations of language that this project makes.

For ethical purposes, I analyze only publicly displayed messages. Excerpts of IRC interactions are presented as they appeared on the screen, including graphic symbols and typos, but excluding personal information such as email addresses.

Breaking out of Binaries

Gender and speech features

IRC provides a useful testing ground for some of the claims made about speech features in gender and language research. Interactions analyzed here illustrate the incapability of the binaries facilitative/controlling [(Fishman, 1978/1983)], rapport/report [(Tannen, 1990/1991)], and personal/authoritative [(Herring, 1993)] to describe an utterance's functions. These dichotomies represent women's and men's speech, respectively. Ginger's statement in line 754 breaks out of these binaries and out of the female register that [Lakoff (1975)] describes.

Ginger presents herself as a female. She does this initially by selecting a common woman's name for a nick. In addition, ginger refers to herself using a feminine pronoun;

703. * ginger throws gump her sweatshirt.

Ginger's utterance, however, operates in ways that span alleged functional dichotomies and male/female divide. After several rounds of interaction with other users and after complaints of “lag” (delay in receiving lines of text in IRC), ginger exclaims that she is bored. Conversation then develops around this comment.

754. * ginger is VERY BORED!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

759. Why is ginger bored?

762. * grigg asks why ginger is bored.

763. * ginger has no one ot talk to

765. * Gump wants to entertain ginger

766. hi ginger

767. How is ginger tonite?

775. * Gump belly dances for ginger, hehehe

777. * ginger is bored and tired…anyone know of a comfy bed i can nap in?

778. ginger are you in school?

784. wanna nap in mine ginger?

786. hehehe

790. hehe

792. * Gump wonders

794. * Gump thinks where ginger can stay

798. well, at three naps a day and 30 days a month that means i can sleep with

799. all of you….hehe

802. Ginger!

804. *** mrhappy (http://abc@def.edu) has joined channel #talk

805. Gump!

806. really! when will u sleep w/me ginger

808. hehehehehehehe

809. howdy mrhappy can i make you happier?

810. ginger!

811. :)

815. hmmmmm let me get my calendar

819. * ginger gets out her day planner

820. i'll pencil u in

824. * Gump snatches ginger's planner away

831. * Gump writes his name all over gingers planner

832. hehehe

839. * ginger snatches it back…….

841. looks like your goin to be busy ginger

844. * bbob snatches ginger

847. * Gump thinks ginger will be happy

851. * grigg smiles at ginger…

853. * ginger likes the attention….finally not bored

856. * Gump gets ginger back

858. HI GINGER, I PREFER MARY ANNE BUT YOU'LL DO

859. FOR NOW.

865. * Gump knows what ginger's belly look likes

867. * ginger wonders what gump has in mind

Line 754 facilitates conversation; GeneJock and grigg respond to ginger's initial comment, and Gump, hero, and bbob join the discussion in subsequent turns [(Fishman, 1978/1983)]. However, ginger's remark also fits Fishman's definition of interactional control; users orient to the topic line 754 introduces. Ginger's statement initiates a conversation that develops around entertaining her.

Ginger's conversational facilitation also does rapport work by helping maintain personal connections [(Tannen, 1990/1991)]. On the other hand, ginger's statement exemplifies report-talk, which Tannen describes as “holding center stage through verbal performance” and using talk to “get and keep attention,” (p. 77). Ginger gains attention through line 754 and keeps it by asking if anyone knows of a “comfy bed I can nap in?” (line 777). [15] Line 754 demonstrates that utterances operate in ways that overlap the binary categories Tannen's and Fishman's works describe.

The statement “*ginger is VERY BORED!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!” also reveals authoritative and personal orientations [(Herring, 1993)]. Using an action-description [16] makes the utterance appear authoritative; it is as if an omniscient narrator commented on ginger's state of being. However, since ginger describes herself, her comment is also personal. Thus, “authoritative orientation” and “personal orientation” are not mutually exclusive characteristics. [17]

Furthermore, ginger's utterance falls out of the boundaries of “women's language” [(Lakoff, 1975)]. Ginger does not “superpolitely” ask for interaction; she demands it. Ginger expresses her feelings of boredom using all caps and exclamation points, which in IRC and other CMC contexts, signify shouting. Shouting is interpreted much as it would be in real-life environments. In IRC it is considered disruptive and perhaps rude to yell repeatedly. Users may chide someone for shouting indiscriminately as suggested by the following exchange:

858. HI GINGER, I PREFER MARY ANNE BUT YOU'LL DO

859. FOR NOW.

864. quit shouting!

866. NO

869. YES!!!

Although ginger's scream is not, by IRC standards, exceedingly rude or deserving of reprimand, ginger's comment stands in contrast to the “superpoliteness” associated with the female register that [Lakoff (1975)] describes. [18]

[Freed (1996)] has argued that dichotomous descriptions of utterances are grounded in a binary conceptualization of gender. I further suggest that gender bifurcation in foundational research on gender and language stems from the assumption that gender is an identity which is already formed when conversations occur [(Lakoff, 1975; Tannen, 1990/1991)]. [19] Studies on gender differences in CMC styles and research on the relationship between power and gender in CMC have also treated gender as pre-formed, not performed. In these projects, gender becomes an independent variable whose effects on interaction can be studied [(Graddol & Swan, 1989; Kramarae & Taylor, 1993; Herring, Johnson, & Dibenedetto, 1992; Bruckman, 1993; Herring, 1993; Jaffe et al. 1995; Savicki et. al 1996; [Witmer and Katzman, 1997)]. Looking at gender as pre-formed, not performed veils variation in gender construction. As a result, similarities and subtle differences between men's and women's speaking styles are easily neglected. The assumption that one's gender precedes interaction is based on the biological sex-gender link (Butler, 1990; Bem, 1993)]. This link is also one of the lenses of gender that perpetuates myths about men and women.

Perhaps concern with the effects of CMCers’ RL gender encourages researchers to treat gender as a pre-given construction. It makes sense to look at players’ already-constructed-in-RL gender if one wants to understand the nature of women's oppression in conversations between people who know each other's RL identities. Academic lists exemplify such discussions. Studies that look at environments in which creating identity is an end in itself, however, may be more sensitive to gender's construction through online discourse because they are less concerned with RL gender and its effects [(Reid, 1993; Danet, 1996)]. Reid's and Danet's works appear to be more interested in looking at how gender is constructed in virtual environments than in determining gender's effects. Creating gender is part of the “play” in which IRCers and MUDers engage [(Danet, et al. 1997)]. While academic discussion may, at times, allow for such identity play [(Selfe & Meyer, 1991)], academic lists tend to be oriented to discussion and debate of topics rather than to creation of alter-identities [(Selfe & Meyer 1991; Herring, Johnson & DiBenedetto, 1992; Herring, 1993)]. Thus, the range of experimentation is limited in academic lists compared to playful environments like IRC and MUDs.

Studies of playful environments, however, may focus on gender's effects rather than on its creation. [Bruckman's (1993)] analysis of gender swapping in MUDs looks at such effects. While cognizant that “gender becomes malleable in MUDs…[and that] gender becomes a property that can be reset with a line of code” (p. 5), Bruckman also claims that “gender pervades human interactions in such basic ways that its impact is often difficult to observe” (p. 3). The study sets out to explore “the role of gender in human social interaction” (p.1). Thus, Bruckman is concerned with gender's “role” and “impact” and conceptualizes online gender as a “property.” Conceptualizing gender as a trait that has effects de-emphasizes its malleability and constant constructedness.

Thus, IRC allows one to test out dichotomous theorizing about speech features. Analysis of IRC suggests that speech cannot be described using binaries because utterances are multifunctional. Furthermore, making such distinctions is wrong-headed because they are based on the troublesome assumption that gender is pre-formed and bipolar.

Gender construction and coding trouble

IRC also serves as a testing ground for conceptualizations of gender. Analysis of IRC interactions reveals the extent to which gender is flexible and constantly constructed. Identifying such elasticity, however, presents coding problems.

In agreement with other research, the present analysis of IRC finds that gender may be constructed through nickname choice [(Reid, 1993; Bechar, 1995; Danet et al. 1997)]. In IRC common RL male or female names like “john” or “StacyAnn” imply gender. Players with gender-specifying nicks may also confirm their gender, as do Tawyna and ginger in the excerpt below. However, IRC nicks often depart from common real-life male and female names [(Bechar, 1995; Danet et al., 1997)]. In these cases users may refer to themselves in ways that suggest a particular gender [(Danet, 1996)]. Lacking gendered nicknames, IRC participants CaTiger and Flip construct gender through their speech. These players use masculine pronouns “his” and “he” to refer to themselves. Similarly, Tawnya and ginger use feminine self-referencing pronouns.

475. * CaTiger wiggles his ears.

483. * Tawnya applauds CaTiger's genius. fortunately or un she's been able to do that

484 for years.

501. * CaTiger twitches his tail in amusement.

613. * Flip waves to everyone as he tumbles into the channel

703. * ginger throws gump her sweatshirt [emphasis added, lines 475–703]

Gump constructs himself as male by naming himself after the male title character in the film Forrest Gump. At one point a user asks, “gump: forrest is that you?” (line 581). Gump answers by mimicking the character's line in the movie, “I'm forrest gump, you can call me forrest gump” (line 591). Similarly, Danet et al. (1997) observed a player who named himself after a male Star Trek character. Decoding such characters’ gender, however, is possible only for users familiar with Star Trek and Forrest Gump. Thus, for those in the know, Gump implies that he is male. [20]

Coding Gump, CaTiger, Flip, Tawnya, and ginger male and female appears unproblematic. One is able to infer their gender because of the information they share on line. However, IRCers may send contradictory cues. “Mega-D” refers to himself as male but later claims gender neutrality. Such performances confound gender's identification and thus, its coding. [21]

Let us look more closely at Mega-D's gender ambiguity and attendant coding trouble. The nickname “Mega-D” signifies neither masculinity nor femininity. Nor is “Mega-D” a common male or female name in English-speaking cultures. In an attempt to nail it down, one may ground Mega-D's gender in meanings one attaches to reading Mega-D as shorthand for “Megadeath,” a heavy metal rock band. One may further associate Megadeath with masculinity because of heavy metal's popularity with adolescent males or because of traditional masculine roles that metal music and the bands’ behavior reproduce [(Denski & Sholle, 1992)]. However, this method of gender identification seems too unreliable for studies that demand strict coding categories.

Mega-D's speech also compounds coding difficulties that his/its nick presents. [22] The interactions below reveal that Mega-D's speech carries conflictual information about the character's gender.

717. *Mega-D has Iron Maiden going through his head now. I hate it when I get songs

718. stuck in my head.

739. ginger, most people here are going to be guys!

742. mega-d why is that?.

751. ginger something about computers that get guys going. I have no idea…

757. mega-d does that make you male?

769. * Mega-D is just Mega-D *sex doesnt exist!*

780. im sorry to hear that Mega-D

797 JKD well, its only here on the interenet that my sexuality dissapears!

812. Mega-D: gee I couldn't imagine why you'd want that to happen

813. show sarcasim>

826. JKD people leave me alone :) Desperate guys leave me alone, and ugly

827. girls leave me alone too. Only a few know the sexuallity of Mega-D!

854. mega-d: actually its smart.

870. JKD, you catch on! Besides, if people really wanted to know, they

871. could find out :)

Mega-D constructs himself as male by using the masculine pronoun “his” to refer to himself (line 717). Mega-D also appears to speak on men's behalf (line 751). In addition, ginger's response (line 757) suggests that line 751 implies that Mega-D is male. However, Mega-D's references to himself appear to be unintentional revelations, what Goffman (1959) calls “giving off” an impression. When directly asked about his gender (line 757), Mega-D responds that he lacks sex, that “*Mega-D is just Mega-D *sex doesnt exist!*” (line 769). [23]

However, one should not interpret Mega-D's mixed messages as reasons to code the character gender neutral. Such labeling jettisons Mega-D's masculine self-references that help contrive his maleness. Nor can one trace Mega-D's “true” gender to earlier revelations of maleness or to later declarations that it lacks sex; doing so ignores the validity both expressions possess in IRC. In addition, one cannot solve this coding problem by interviewing Mega-D's RL counterpart to assess which gender he/it intended to represent. As I suggested earlier, this method overlooks the ways in which his/its speech is treated in IRC. IRC users tend not to look for reality behind the screen; on-line performances constitute reality for IRC players [(Turkle, 1995)]. Clearly, Mega-D's conflicting performances represent a coding disaster for those who would construct gender as a binary (male/female) or triadic (male/female/gender neutral) category. [24]

Binary gender survives

Despite Mega-D's multiple and conflicting gender performances, however, the binary gender system is alive and well in IRC. One IRCer who presents himself as masculine works to keep this image consistent. As discussed earlier, Gump communicates that he is male through his gendered nickname. However, Gump does additional gender work in answer to Line1's flame that contests Gump's maleness (line 709). Gump's efforts also address bbob's question, “is line1 spreading rumors?” (line 720).

709. ginger and gump are lesbians hehehehhee

711. gump is kising up to ginger

713. gump wants to be the one wtih the strap on dildo

714. * ginger is not a lesbian

715. *** Line1 has left channel #Talk

720. is line1 spreading rumors?

723. * Gump looks in his pants and notices he is still a guy, whew

729. * ginger is glad gump is a guy….

741. * Gump is also glad he is a guy

To answer the challenges posed by Line1's and bbob's utterances, Gump uses an action-description to demonstrate that he possesses male anatomy and confirm that “he is still a guy” (line 723). Ginger and Gump agree that they are glad Gump is male. As Agnes, Kate Bornstein, and Gump illustrate, gender construction is ongoing and does not end with name selection. [25]

Ginger's effectiveness in getting and holding center stage also points to the persistence of the binary gender system. Ginger's success in capturing attention may have more to do with gender performance than with her use of all caps and exclamation points. [26] Ginger attracts the sort of attention (wanted or unwanted) that female-presenting players get in other virtual environments [(Bruckman, 1993). It is difficult to locate the moment in which the conversation becomes sexual. Clearly, however, the interaction grows flirtatious and sexual innuendoes build. Gump “belly dances for ginger” and laughs (line 775). Ginger asks if she can “nap” in a “comfy bed” (line 777), and bbob offers his bed (line 784); bbob and ginger laugh after bbob's offer (line 786, 790). Ginger invites all interlocutors to sleep with her (lines 798–799) and offers to make “mr happy…happier” (line 809). Grigg smiles at ginger (line 851); Gump claims that he knows what her belly looks like (line 865), and with an air of chivalry, declares that he “gets ginger back” (line 856). While ginger's apparent willingness to perform flirtatiously disqualifies the exchange as harassment, the conversation sexually objectifies her. Ginger is further objectified when, late in the rounds of flirting, CYBERPUNK enters and says “HI GINGER, I PREFER MARY ANNE BUT YOU'LL DO FOR NOW (lines 858–859). This statement, in the sexualized context of discussion, implies that ginger will sexually suffice for CYBERPUNK's use. Further CMC research might consider the extent to which women's effectiveness in gaining response is related to their sexual objectification.

Conclusion

This analysis suggests that conceptualizing gender as a dichotomy neglects the variety of gender constructions in IRC. Although some gender performances in IRC conform to dualistic gender categories, others break out of binary categories. Furthermore, because IRC characters may express gender in multiple and contradictory ways, basing descriptions of speech functions on a dualistic conception of gender oversimplifies explanations of the ways in which utterances operate.

Reconceptualizing gender as a series of performances also helps researchers abandon binding binaries without disregarding previous findings about gender and language. Thinking about gender as under constant construction does not contradict studies which suggest that men dominate CMC. It appears that users who present maleness have more power than do those who present femaleness [(Selfe and Meyer, 1991; Herring, Johnson & Dibenedetto, 1992; Herring, 1993; Kramarae and Taylor, 1993)]. Moreover, marking oneself feminine entails vulnerability to harassment and censorship [(Bruckman, 1993; Herring, 1993; Kramarae & Taylor, 1993; Petersen, 1994; Spender, 1995)]. RL women seem to understand the importance of online gender performance, as they are more likely than men to conceal their gender when in cyberspace [(Selfe and Meyer, 1991; Jaffe et al., 1995)].

In addition, thinking about gender as performatively constructed highlights the unnaturalness of the bipolar gender system. Illuminating this system's cultural contingency helps one deconstruct male domination in CMC as an effect of oppressive binary gender discourse [(Butler, 1990; Bem, 1993; Bing & Bergvall, 1996; Cameron, 1985/1992; Cameron, 1996)]. Because defining gender as a series of dramatizations allows researchers to represent the various, sometimes contradictory ways in which individuals express gender, gender-as-performance allows CMC scholars to analyze women's oppression without constructing gender as a biological, stable trait.

However, conceptualizing gender as performative does not guarantee that one will challenge the binary gender system. Detailed analyses of the ways in which gender is produced in computer-mediated or FTF modes may help reproduce gender dichotomies. A study supports gender bifurcation if the work contrasts gender constructions and does so in ways that make differences between constructions seem natural and universal. For example, theorists who adhere to biological essentialism may attribute discrepancies in gender performances to one's allegedly immutable sex. [27] Thus, while thinking about gender as a series of performances helps researchers represent multiple, conflicting ways in which individuals create gender (or lack of), the ability to consider such diversity is lost if expressions are described in ways that reify gender differences.

Gender-as-performance can also be used to reinforce the binary system if CMC is theorized as enabling the re/construction of gender. One may be tempted to conclude that CMC untethers gender from sex or eradicates gender polarization. [Reid (1993)] echoes this thinking by arguing, “The users of IRC show the degree to which the medium challenges and obscures the boundaries between some of our most deeply felt cultural significances” (p. 64). Reid further contends, “the structure of IRC destroys the usually all but insurmountable confines of sex…[and gender's] fixity, and the common equation of gender with sex, becomes problematic when gender reassignment can be effected by a few touches at a keyboard” (p. 63). Similarly, [Dickel (1995)] argues that the “destabilization of gender positions…might spread beyond the Internet into the larger culture” (p. 10). Aside from being a simplistic, technologically determinist view, arguing that CMC “challenges” gender construction, “destroys the usually all but insurmountable confines of sex,” and allows for gender destabilization to “spread beyond the Internet” implies that CMC causes something that otherwise does not occur in RL. Reid's and Dickel's arguments suggest that the Internet causes gender to be performed. Bornstein, Agnes, and other “natural, normal females” in RL, however, must continually give the impression that they are female in order to be perceived as such. Throughout the binary gender system's history, gender has always already been under construction [(Bem, 1993)]. Thus, gender has always already been under construction off and on line. [28]

Arguing that CMC changes gender's constitution also posits a chasm between the real and the virtual. This bifurcation is problematic for two reasons. Others have pointed to the difficulty if not futility in separating these environments [(Rheingold, 1993/1994; Kaplan & Farrell, 1994; Turkle, 1995; Gromala, 1996; Biocca, 1997)]. Rheingold's discussion of community on the WELL suggests that virtual communities have real effects; they influence life off-line. Thus, what appears to be “virtual” is real. But, what appears to be “real” is also virtual. A MUDder Turkle interviews claims, “‘RL is just one more window…and it's not usually my best one’” (p. 13). CMC, therefore, occurs in the overlap of real and virtual worlds.

In addition to misrepresenting the ways in which CMC is experienced, arguments that make sharp distinctions between real and virtual also delegitimize gender's performativity. Such divisions imply that virtual environments constitute fantasy space, a Disneyland of sorts. Thinking of the virtual as pure fantasy makes RL appear as the location of truth and the virtual as the location of fiction [(Baudrillard, 1981/1994)]. Applied to gender, these comparisons suggest; “In cyberspace, gender can be constructed, but such constructions are fantasy. We all know that gender is really based on one's RL biology and can be described using binary categories.” Mega-D expresses this view when he/it says, “Its only here on the internet that my sexuality dissapears!” (line 797, emphasis added). Mega-D's statement suggests that the Internet is anomalous in allowing for deconstruction of his RL gender and its reconstruction as gender neutrality. According to this view, gender's production on line is deemed as exceptional to the off-line rule, and the binary gender system goes undisturbed. Positing impermeability between “real” and “virtual” supports and extends Reid's and Dickel's claims that CMC causes something that otherwise does not occur in RL. Perspectives that divide “real” and “virtual” marginalize CMC as a location of gender performance; CMC appears as a place of art and artifice rather than as a site of reality. Communication scholars should counter such views and continue to look at gender performance in overlapping virtual and real worlds.

Finally, gender should also be recognized as a virtual reality. The present analysis has shown that the binary gender system has effects; it oppresses. In this sense, gender is real. On the other hand, gender is performed and is culturally constructed. In this sense, gender is intangible and is thus, virtual. On line and off, gender is a virtual reality. However, this study also suggests that in some cases, gender is made to seem more “real” (natural, essential) than it really is. In others, gender's performativity is rendered as more “virtual” (flexible, fictive) than it really is. Thus, feminist linguists and CMC researchers should continue exploring the binary system that has real, oppressive effects, while also calling attention to gender's constructedness. In addition, scholars should make clear that gender performances which appear grounded in biology or seem opposed to one another may do because our culture demands that we view the world through gendered glasses. These spectacles are key to male domination's reproduction because the ability to oppress women depends on society's ability to mark women. It is only through exposing the virtual reality of gender that one can destabilize the binary marking system on which patriarchy rests.

Footnotes

  • [i]

    This paper considers text-based computer-mediated environments to be a type of virtual environment (Curtis, 1992).

  • [1]

    “Gender polarization” is what Bem (1993) calls the process or effect of dividing the world into two opposing genders.

  • [2]

    Although Lakoff's (1975) work has been criticized for reinforcing gender dichotomies, Lakoff acknowledges the existence of a binary gender system that punishes girls who speak in ways that signify masculinity and punishes women who speak in ways that signify femininity (p. 5–6).

  • [3]

    Elements considered important to an interaction's context include: characteristics of interlocutors (gender, other status cues, and social group membership), purpose of the interaction, and physical setting.

  • [4]

    Lakoff's (1975) and Fishman's (1978/1983) works exemplify the “dominance approach” to language because they attribute gendered speech patterns to power imbalances between men and women (Cameron, 1985/1992). In a later study Cameron (1996) argues that Lakoff's work represents the “deficit” model of language and gender because it portrays women's speech as weaker than men's. However, because Lakoff traces this weakness to power imbalances, not to women's predispositions (as does Jespersen, 1922/1990), her model is a dominance perspective. For simplicity's sake, I use Cameron's (1992) earlier description of Lakoff's model as a dominance approach.

  • [5]

    Earlier studies also reflect this approach, but Tannen's (1990/1991) is considered the most widely-known. Difference models that precede Tannen's are found in Jones (1980), Maltz and Borker (1982), and Coates (1986).

  • [6]

    Bem recognizes, however, that not all men in patriarchal societies have power. Bem defines “male power” historically, as that which has been held by “rich, white, heterosexual men” (p. 3).

  • [7]

    Following Butler (1990), this paper uses “performance,”“construction,” and “production” interchangeably in conceptualizing gender. Bipolar gender discourse also helps reproduce the hegemonic ideology of “compulsory heterosexuality” (Rich, 1975) by associating gender identity (male or female) with object of desire (female or male, respectively).

  • [8]

    Research which finds that CMC reduces social cues has collectively been called the “cues filtered out” approach (Culnan & Markus, 1987). Arguments that lack of social cues in CMC helps democratize communication can be found in Short, Williams, and Christie (1976), Kielser, Siegel and McGuire (1984), and Sproull and Kielser (1991).

  • [9]

    Herring, Johnson, and DiBenedetto (1992) contend that Graddol and Swan (1989) argue that “the electronic medium is claimed to break down gender barriers” (p. 251). Herring (1993) also claims that Graddol and Swan's work contributes to “a strong a priori case for the democratic nature of CMC” (p. 3). Graddol and Swan, however, discuss contextual factors that upset the balanced communication between men and women in the electronic conference observed. Thus, Graddol and Swan do not portray CMC as optimistically as Herring et al. (1992) and Herring (1993) imply.

  • [10]

    IRCers are expected to select a nickname that differs from their username (in IRC, one's default nickname is one's username). Failing to do so may elicit ridicule from others. When I entered an IRC channel without a handle, other users sarcastically commented on my creativity.

  • [11]

    Kane (1994) offers this translation of “MUDs.” It is outside the scope of this paper to provide criticism of “cues filtered out” and “equalization” theories beyond that which research on gender and CMC suggests. Readers will find more detailed critiques of such perspectives elsewhere (Culnan & Markus, 1987; Walther & Burgoon, 1992; Spears and Lea, 1994). Witmer and Katzman cite the findings of Bate (1988), Quina, Wingard, and Bate (1987), and Mulac and Lundell (1986) which suggest that women's speech is perceived as exhibiting more emotion than men's. However, as a scholar on virtual reality suggests, “Perhaps women flame more in reaction to their oppression” (D. Gromala, personal communication, November 11, 1997). Women may express anger not because they are prone to emotional displays (as biologically essentialist arguments would suggest), but because they are angry at being censored and harassed on and off line.

  • [12]

    No sustained conversations were found in: #buttery, #celtic, #cool, #dog, #gayfr, #McGill, #movies, #Nescafe, #pub, #surfers, #strange, #veggies.

  • [13]

    The channel #truthdare hosts the game “truth or dare” in which participants choose between responding to questions or challenges posed by other players.

  • [14]

    Savicki et. al (1996) also argue that findings cannot be generalized across CMC contexts unproblematically.

  • [15]

    This question also facilitates and controls talk because conversation continues and focuses on ginger's going to bed (and related innuendoes). Thus, questions can do both facilitative and control work, in contrast to Fishman's claim that the feature only enables interaction.

  • [16]

    Action descriptions describe an action, one's surroundings, or one's state of mind in IRC. On the screen action descriptions are denoted by a “*” and the user's nick preceding the statement. Although the user whose nick appears keys in the statement, the comment reads as a third person description of the user or of the user's surroundings.

  • [17]

    Rafaeli and Sudweeks (1993) construct a similar binary between variables “FIRSTPER” and “FACT.”

  • [18]

    While shouting seems to be one of the factors contributing to ginger's gaining response, yelling does not guarantee that one will receive attention.

  • [19]

    Fishman (1978/1983), however, recognizes that gender is an “‘achieved’” rather than “‘ascribed’” characteristic and considers Garfinkel's (1967) description of Agnes’ passing as proof of gender's “continual, routine accomplishment” (p. 99). However, Fishman's construction of men's and women's speech as a binary suggests that she looks through the lenses of gender Bem (1993) and Freed (1996) discuss.

  • [20]

    Of course, reading gender off of a name always requires some local knowledge. One must have a certain amount of knowledge about English-speaking cultures to know that “john” is a common male name.

  • [21]

    Expressing masculinity and gender neutrality is “contradictory” if viewed through the lens of the binary gender system.

  • [22]

    In discussing Mega-D's masculinity and gender neutrality, I use male pronouns for his masculine presentations and gender-neutral pronouns for its gender-ambiguous displays.

  • [23]

    Thus, I disagree with Bechar's (1995) argument that in CMC interlocutors intentionally “give” but do not unintentionally “give off” information about themselves.

  • [24]

    One would be correct to argue that the above analysis of Mega-D conflates sexuality, gender, and sex. This discussion, however, represents IRC players’ conflation of these constructions. When ginger asks Mega-D if it is “male” (one cannot be certain if she means gender or sex), it answers, “* Mega-D is just Mega-D *sex doesnt exist!*” (line 769). JKD interprets “sex” as “having sex,” when it says, “im sorry to hear that Mega-D” (line 780). Mega-D's reply indicates that it was not speaking about “having sex,” but about “sexuality.” Although sex, gender, and sexuality are not naturally linked, the two are linked in hegemonic ideologies within our culture (Rich, 1980; Butler, 1990). However, exposing such conflations as culturally constructed, as I am doing, helps disrupt the notion that such associations are natural or inevitable.

  • [25]

    As I argued earlier, sex, gender, and sexuality are often conflated in various discourses. It may be no coincidence that several lines after Gump asserts that he is “still a guy” and ginger claims that she is “not a lesbian,” Mega-D defends its gender neutrality.

  • [26]

    I am grateful to Margaret McLaughlin for pointing this out to me.

  • [27]

    Of course, individuals may create and experience their gender as grounded in “biological” sex and may view gender as bipolar. As Kate Bornstein's case suggests, however, some do not.

  • [28]

    Perhaps arguments that CMC changes and challenges gender construction can be explained by the access researchers have to gender production on the Net. It may be easier to study gender performances in virtual rather than in real environments because the former feels alien to researchers. As a result, the modes of gender construction may be less naturalized in computer-mediated than in FTF communication and thus, the practices of gender construction in CMC may be more critically approachable. That such environments are more approachable, however, does not mean that they are agents of change.

Ancillary