Communication Institute for Online Scholarship
Communication Institute for Online
Scholarship Continous online service and innovation
since 1986
Site index
 
ComAbstracts Visual Communication Concept Explorer Tables of Contents Electronic Journal of Communication ComVista

Gender and Democracy in Computer-mediated Communication
EJC logo
The Electronic Journal of Communication / La Revue Electronique de Communication
***** HERRING ************** EJC/REC Vol. 3, No. 2, 1993 ***

GENDER AND DEMOCRACY IN COMPUTER-MEDIATED COMMUNICATION


Susan C. Herring
University of Texas at Arlington


        Abstract.  The claim that computers democratize
     communication is evaluated with respect to male
     and female participation in two academic
     electronic discussion lists over a one-year
     period.  A tendency is noted for a minority of
     male participants to effectively dominate
     discussions both in amount of talk, and through
     rhetorical intimidation.  It is argued that these
     circumstances represent a type of censorship, and
     thus that an essential condition for democratic
     discourse is not met.


                 The Democratization Claim

     Despite a substantial body of research demonstrating
sex differences in face-to-face communication (see e.g.
Coates, 1986), the question of sex differences in computer-
mediated communication has only recently begun to be raised.
The lag is due in large part to a climate of general
optimism surrounding the new technology; specifically, the
belief that computer-mediated communication (hereafter, CMC)
is inherently more democratic than other communication
media.  Thus philosophers and social theorists see in CMC a
more equal access to information, empowering those who might
otherwise be denied such information, and leading ultimately
to a greater democratization of society (Ess, to appear;
Landow, 1992; Nelson, 1974).  Educators evoke the potential
of computer networks to foster creativity and cooperation
among students, and to help break down traditional barriers
to communication between students and instructors (Kahn and
Brookshire, 1991; Kiesler, Siegel, & McGuire, 1984;
McCormick & McCormick, 1992).  Even feminists are encouraged
by evidence of more equal communication between women and
men in the absence of status- and gender-marked cues
(Graddol & Swann, 1989), and by the opportunities for women
to establish "grass-roots" electronic communication networks
of their own (Smith & Balka, 1991).

     The notion of democracy as it emerges through these
claims has two essential components: access to a means of
communication, and the right to communicate equally, free
from status constraints.  These components are inherent in
the formal "rules of reason" proposed by the German
philosopher Habermas (1983; discussed in Ess, to appear) as
criteria which must be observed in order for a discourse to
be truly democratic:

     1.  Every subject with the competence to speak and act
         is allowed to take part in the discourse.
     2a. Everyone is allowed to question any assertion
         whatever.
     2b. Everyone is allowed to introduce any assertion
         whatever into the discourse.
     2c. Everyone is allowed to express his [sic] attitudes,
         desires, and needs.
     3.  No speaker may be prevented, by internal or
         external coercion, from exercising his [sic]
         rights as laid down in (1) and (2).  (1983, p.89)

Habermas's third rule provides for an important social
dimension: in a truly democratic discourse, there can be no
censorship.  To the extent that computer technology
facilitates open, egalitarian communication of this sort, it
is held to be democratizing (Ess, to appear).

     A number of specific characteristics of CMC have been
claimed by researchers and users to facilitate communication
that is democratic in nature.  The first of these is
_accessibility_.  Through universities and other
institutions, increasing numbers of people are able to gain
access to computer networks at little or no cost.  This
access in turn makes available to them a variety of
benefits, the most widely-touted of which is information, in
the form of on-line library catalogues, public domain
databases, and the like.  Less commonly mentioned, but
equally if not more important, are the opportunities
provided by electronic networks to connect and communicate,
to express one's views and be recognized in a public forum,
potentially even by large numbers of people (including, now
that President Clinton has a public access e-mail address,
highly influential people) around the world.  In theory,
anyone with access to a network can take equal advantage of
these opportunities.

     A second potentially democratizing characteristic of
CMC is its _social decontextualization_.  As noted by
Graddol & Swann (1989) and Kiesler et al. (1984), the
identity of contributors need not be revealed, especially
given login "names" and return addresses that bear no
transparent relationship to a person's actual name, sex, or
geographical location.[1] Further, CMC neutralizes social
status cues (accent, handwriting/voice quality, sex,
appearance, etc.) that might otherwise be transmitted by the
form of the message.  While on the one hand these
characteristics render the medium less personal, they also
provide for the possibility that traditionally lower-status
individuals can participate on the same terms as others --
that is, more or less anonymously, with the emphasis being
on the content, rather than on the form of the message or
the identity of the sender.  As one member of an academic
discussion list wrote in a recent posting to another:

     One of the greatest strengths of e[lectronic]-mail
     is its ability to break down socio-economic,
     racial, and other traditional barriers to the
     sharing and production of knowledge.  You, for
     example, have no way of knowing if I am a janitor
     or a university president or an illegal alien --
     we can simply communicate on the basis of our
     ideas, not on any preconceived notions of what
     should be expected (or not expected) from one
     another.

While one might question the assumption that the posts of
anonymous janitors, university presidents, and illegal
aliens would not reveal their status (via differences in
e.g. grammatical usage, stylistic register, and familiarity
with conventions of CMC and academic discourse), the
idealism expressed by this writer is typical of that of many
network users.

     Third, as a relatively new discourse type, CMC lacks a
set of consensually agreed-upon and established _conventions
of use_ (Ferrara, Brunner, & Whittemore, 1991; Kiesler et
al., 1984).  As a result, users may be less inhibited,
leading to "flaming" and outrageous behavior on the one
hand, and to greater openness on the other.  This feature
has led hypertext theorists such as Bolter (1991), Landow
(1992), and Nelson (1974) to characterize CMC as "anarchic"
as well as "democratic", with the potential to contribute to
the breakdown of traditional hierarchical patterns of
communication.

     Finally, overt _censorship_ on the electronic networks
is as yet rare; what censorship exists is typically more
concerned with selectively blocking the use of vulgar
language than with blocking message content.[2] Even
moderated discussion lists tend to accept virtually all
contributions and post them in the order in which they are
received.  Thus each and every contributor to a discussion
theoretically has the same opportunity to have his or her
messages read and responded to by other members of the group
(Habermas's third "rule of reason").

     Taken together, these four characteristics would appear
to constitute a strong a priori case for the democratic
nature of CMC.  But how democratic is the communication that
is actually taking place currently via electronic networks?
Specifically, does it show evidence of increased gender
equality, as Graddol and Swann (1989) claim?

             Summary of Investigative Results

     The research reported on in this article is based
primarily on investigations carried out over the past year
on male and female participation in two academic electronic
lists (also known as "bulletin boards" or "discussion
groups"): LINGUIST -- devoted to the discussion of
linguistics-related issues -- and Megabyte University (MBU),
informally organized around the discussion of computers and
writing.  What follows is a summary of findings analyzed in
detail in three recent articles (Herring, 1992; Herring, in
press; Herring, Johnson, & DiBenedetto, in press); much of
the data and analysis on which the earlier studies were
based has of necessity been omitted here.

     Three types of methods were employed in investigating
participation on LINGUIST and MBU. The first was
_ethnographic observation_ of discussions as they occurred:
I subscribed to and saved contributions to both lists over a
period of one year, in the process assimilating information
about contributors, current issues in the field, and other
relevant background information.  (CMC is especially
amenable to data collection of this type, in that observers
can easily remain invisible, thus avoiding the "observer's
paradox" of altering by their presence the nature of the
phenomenon they seek to observe.)  Second, I subjected the
texts of two extended discussions from each list to a
_discourse analysis_ in which patterns of grammatical and
stylistic usage were identified.  Observed patterns of usage
were then correlated with participant sex, which was
determined either from contributors' names when available
(i.e. because they signed their message, or their mailer
program included it in tracing the path of the message), or
else by matching electronic addresses to names from a
publicly-available list of subscribers to each list.[3]
Finally, I prepared and distributed two electronic
_surveys_, one each for LINGUIST and MBU, in which I asked
for participants' reactions to a particular discussion that
had taken place on the list to which they subscribed, as
well as solliciting background information regarding their
sex, professional status, and familiarity/competence with
computers.  (A copy of the LINGUIST survey is appended at
the end of this article.)  The data collected by these three
methods were subjected to quantitative as well as
qualitative analysis.

     The combined results reveal significant differences
between male and female participants.  The principal
differences are discussed below under the headings AMOUNT,
TOPIC, and MANNER of participation.

Amount

     The most striking sex-based disparity in academic CMC
is the extent to which men participate more than women.
Women constitute 36% of LINGUIST and 42% of MBU
subscribers.[4] However, they participate at a rate that is
significantly lower than that corresponding to their
numerical representation.  Two extended discussions were
analyzed from each list, one in which sexism was an issue,
and the other on a broadly theoretical topic.  Although the
'sexism' discussions were more popular with women than
discussions on other topics, women constituted only 30% of
the participants in these discussions on both lists, and in
the 'theoretical' discussions, only 16% of the participants
were women.  Furthermore, the messages contributed by women
are shorter, averaging a single screen or less, while those
of men average one and a half times longer in the 'sexism'
discussions, and twice as long in the 'theoretical'
discussions, with some messages ten screens or more in
length.  Thus while a short message does not necessarily
indicate the sex of the sender, a very long message
invariably indicates that the sender is male.

     What accounts for this disparity?  It does not appear
on the surface as though men are preventing women from
participating -- at least on one of the lists, MBU, male
participants actively encourage more women to contribute.
There is evidence to suggest, however, that women are
discouraged or intimidated from participating on the basis
of the reactions with which their posts are met when they do
contribute.  In a medium which permits multiple contributors
to post messages more or less simultaneously to the group,
gaining the focus of the group's attention or the
"conversational floor" depends entirely on the extent to
which other participants acknowledge and respond to one's
postings.  In the CMC analyzed here, messages posted by
women consistently received fewer average responses than
those posted by men.  In the MBU 'sexism' discussion, 89% of
male postings received an explicit response, as compared
with only 70% of those by women; on LINGUIST, the disparity
is even greater.  Interestingly, it is not only men who
respond more often to men, but women as well; postings from
women acknowledging the postings of other women constitute
the smallest portion of total responses, an implicit
recognition, perhaps, of the more powerful status of men in
the groups.  In keeping with the unequal rate of response,
topics initiated by women are less often taken up as topics
of discussion by the group as a whole, and thus women may
experience difficulty and frustration in getting the group
to talk about topics that are of interest to them.

     On those rare occasions when, out of special interest
or a strong commitment to a particular point of view, women
persist in posting on a given topic despite relative lack of
response, the outcome may be even more discouraging.  During
the period of this investigation, women participated
actively three times: once during the MBU 'sexism' debate,
in which men and women became polarized regarding the
legitimacy of offering a special course on 'Men's
Literature', and twice on LINGUIST, the first in a
discussion of the interpretation of Sister Souljah's remarks
on the Rodney King beating, and the second in a 'sexism'
discussion on the question of whether the label 'dog' refers
primarily to women, or to unattractive persons of either
sex.  In all three discussions, women's rate of posting
increased gradually to where it equalled 50% of the
contributions for a period of one or two days.  The reaction
to this increase was virtually identical in all three cases:
a handful of men wrote in to decry the discussion, and
several threatened to cancel their subscription to the list.
Various reasons were given, none of them citing women's
participation directly: in the MBU discussion, the tone was
too "vituperative";[5] in the LINGUIST discussions, the
topics were "inappropriate".  Although the LINGUIST list
moderators (one male, one female) intervened to defend the
appropriateness of the Sister Souljah thread, the discussion
died out almost immediately thereafter, as did the others.
Of course, the possibility cannot be ruled out that the men
who protested were responding to the content rather than (or
in addition to) to the frequency of women's posts.
Nevertheless, the coincidence is striking, since at no other
time during the period of observation did women participate
as much as men, and at no other time did any subscriber,
male or female, threaten publicly to unsubscribe from either
list.  Reactions such as these are consistent with Spender's
(1979) claim that women cannot contribute half of the talk
in a discussion without making men feel uncomfortable or
threatened.  She found that men (and to a lesser degree,
women) perceive women as talking more than men when women
talk only 30% of the time.  This phenomenon is not limited
to Spender's academic seminar data or to CMC, but rather is
a feature of mixed-sex conversation in public settings more
generally (Holmes, 1992).

     This interpretation is further supported by the results
of a survey conducted on MBU several months after the 'Men's
Literature' discussion.  All available external evidence
points to the conclusion that despite the temporary increase
in women's participation, men were more successful than
women overall in the 'Men's Literature' debate -- men posted
more, were responded to more, and introduced more successful
topics in the discussion; further, the real-world course of
action they advocated was ultimately followed (that is, a
'Men's Literature' course was offered).  However, when MBU
subscribers were surveyed later regarding their reactions to
the discussion, male respondents indicated a higher degree
of dissatisfaction than women, and were more likely to say
that women had "won" the debate; women, in contrast, were
more likely to say that neither side had won (Herring,
Johnson, & DiBenedetto, in press).  When women's attempts at
equal participation are the cause of (male) dissatisfaction
-- even if voiced publicly by only a few -- and disruption
of list functioning, a message is communicated to the effect
that it is more appropriate for women to participate less.
And so they do: the day after the MBU protests, women's
contributions dropped back to 15% of the total, and the
discussion continued apace.  The rather depressing
conclusion to be drawn from this is that is it "normal" for
women to participate less than men, such that an increase in
the direction of true equality is perceived as deviant, even
in liberal, academic settings.

Topic

     The above observations indicate that although women
contribute less than men overall, they contribute relatively
more on certain topics of discussion, specifically those
which involve real-world consequences as opposed to abstract
theorizing.  Herring (in press) describes a ranking of
preferences based on participation in different topic types
during a random two-week period on LINGUIST.  Men were found
to contribute most often to discussions of issues, followed
by information postings (i.e. where they provided
information, solicited or otherwise), followed by queries
and personal discussions.  Women, on the other hand,
contributed most to personal discussions (talk about
linguists, as opposed to talk about linguistics), followed
by queries soliciting advice or information from others,
with issues and information postings least frequent.  The
ranking of preferred topic types is represented
schematically below:


   MEN:     issues > information > queries > personal

   WOMEN:   personal > queries > issues > information


A tendency for women to contribute less to discussions of
theoretical issues than to other types of exchanges is
evident on MBU as well.

     Independent support for these observations comes from
the Women's Studies List (WMST), devoted to issues involved
in the organization and administration of women's studies
programs.  WMST, which is owned by a woman and has a
subscribership that is currently 88% female, constitutes a
context in which women post almost exclusively for and among
themselves.  Personal discussions are avoided on WMST,
presumably in the interest of greater academic
professionalism.  Instead, the overwhelming majority of
messages posted to WMST are queries for advice and/or
information.  Answers to queries, however, are required
(according to posted list protocol) to be sent privately to
the asker -- although summaries of answers thus collected
may be publicly posted -- and issues discussions are
explicitly prohibited by the list owner, who sees the list
as "serving as a source of information" rather than "as a
place to hold discussions about that information".  While on
the one hand participants might simply be following WMST
protocol in formulating their contributions as queries
rather than as messages of other types, as an active list
with a steadily increasing membership (currently approaching
2,000 members), WMST is proof that many women are
comfortable with CMC that consists primarily in asking
advice and information of others.

     At the same time, there are indications that women do
not avoid discussion of issues entirely by choice.  Issues
discussions arise periodically on WMST, only to be cut short
by reminders from the list owner and other self-appointed
list vigilantes; more than a few subscribers have written in
to complain that the most interesting threads are invariably
censored.  The list owner feels, however, that it is
important to avoid such exchanges in the interests of
limiting message volume, and out of a fear that "discussion
about highly-charged societal issues ... would attract all
sorts of unpleasant, acrimonious people who are just looking
for a fight".[6] As the following section shows, this fear
is not entirely unfounded.

Manner

     The stylistic register of all of the CMC analyzed here
is that of academic discourse.  Nevertheless, there are
significant sex-based differences to be noted, such that it
is often possible to tell whether a given message was
written by a man or a woman, solely on the basis of the
rhetorical and linguistic strategies employed.[7]

     In Herring (in press), I identify a set of features
hypothesized to characterize a stylistic variety
conventionally recognizable as 'women's language' as opposed
to 'men's language' on the LINGUIST list.  These features
are summarized in Table 1.


============================================================

Table 1  Features of women's and men's language

     WOMEN'S LANGUAGE             MEN'S LANGUAGE

     attenuated assertions        strong assertions
     apologies                    self-promotion
     explicit justifications      presuppositions
     questions                    rhetorical questions
     personal orientation         authoritative orientation
     supports others              challenges others
                                  humor/sarcasm

============================================================


The examples below, taken from messages posted during the
LINGUIST 'issues' discussion, illustrate some of the
features of each style.

(1)  [female contributor]

     I am intrigued by your comment that work such as that
     represented in WFDT may not be as widely represented in
     LSA as other work because its argumentation style
     doesn't lend itself to falsification a la Popper.  Could
     you say a bit more about what you mean here?  I am
     interested because I think similar mismatches in
     argumentation are at stake in other areas of cognitive
     science, as well as because I study argumentation as a
     key (social and cognitive) tool for human knowledge
     construction.

[personal orientation, attenuation, questions, justification]

(2)  [male contributor]

     It is obvious that there are two (and only two)
     paradigms for the conduct of scientific inquiry into an
     issue on which there is no consensus.  One is [...].
     But, deplorable as that may be, note that either paradigm
     (if pursued honestly) will lead to truth anyway.  That
     is, whichever side is wrong will sooner or later discover
     that fact on its own.  If, God forbid, autonomy and/or
     modularity should turn out to be His truth, then those
     who have other ideas will sooner or later find this out.

[authoritative orientation, strong assertions, sarcasm]

     In order to quantify the distribution of these features
according to sex, I then analyzed 261 messages in two
extended LINGUIST discussions, coding each message for the
occurrence or non-occurrence of each of the features in
Table 1. The results show that 'women's language' features
are indeed used more often by women, and 'men's language'
features more often by men.  Sixty-eight percent of the
messages produced by women contained one or more features of
women's language, as compared with only 31% of those
produced by men.  In contrast, 48% of the messages produced
by men contained features of only men's language, as
compared with 18% of women's messages.  Interestingly, while
the majority of women's messages (46%) combined a mixture of
male and female rhetorical features, the fewest men's
messages (14%) combined features.  This finding supports the
view that it is easier for men to maintain a distinct style
(masculine, feminine, or neutral) than for women, who must
employ some features of 'men's language' in order to be
taken seriously as academics, and some features of 'women's
language' in order not to be considered unpleasant or
aggressive.

     These observations on gender-marked styles lead to a
second finding regarding manner of participation.
Discussion on each of the lists investigated tends to be
dominated by a small minority of participants who abuse
features of 'men's language' to focus attention on
themselves, often at the expense of others.  Such abuse,
which I term 'adversarial' rhetoric, ranges from gratuitous
displays of knowledge to forceful assertions of one's views
to elaborate put-downs of others with whom one disagrees.
In the two LINGUIST discussions analyzed, 4% and 6% of the
participants, respectively (all but one of them male), were
responsible for the majority of adversarial rhetoric.  This
same 4% and 6% also posted the most words (33% and 53% of
the total, respectively, or more than eight times the
participant average), and thus dominated in AMOUNT as well
as in MANNER of participation.[8]

     A similar pattern is found in a very different kind of
CMC -- electronic mail exchanges between undergraduates (75%
male) on a local network, as investigated by McCormick and
McCormick (1992).  The authors report that 4.7% of the
undergraduates used the network "a great deal", and "may
have been responsible for generating most of the electronic
mail".  Although the content and purpose of communication in
this setting is quite different from that on professional
academic discussion lists, the minority also seems to have
imposed its style on the discourse overall, turning the
computer lab into "an adolescent subculture" complete with
crude jokes, threats, and put-downs.

     The extent to which other participants are negatively
affected by the behavior of a dominant minority may depend,
at least partly, on their sex.  A survey of LINGUIST
subscribers distributed after the 'issue' discussion took
place revealed that 73% of respondents of both sexes felt
intimidated and/or irritated by the adversarial tone of the
discussion (Herring, 1992).  Men and women appear to behave
differently on the basis of this reaction, however.  Male
respondents indicated that they take it in stride as part of
academic interaction; as one man remarked: "Actually, the
barbs and arrows were entertaining, because of course they
weren't aimed at me".  Many women, in contrast, expressed a
deep aversion and a concern to avoid interactions of this
type.  Comments included: "I was terribly turned off by this
exchange, which went on and on forever.  I nearly dropped
myself from the list of subscribers," and "I was disgusted.
It's the same old arguments, the same old intentions of
defending theoretical territory, the same old inabilities of
open and creative thinking, all of which make me ambivalent
about academics in general".  The concern expressed by the
owner of the WMST list to avoid acrimonious exchanges is
fully consistent with the comments of the female LINGUIST
survey respondents.  Why do women react with greater
aversion than men to adversarial exchanges?  Sheldon (1992)
suggests that this aversion can be traced to cultural norms
of sex-appropriate behavior with which children are
indoctrinated from an early age: while boys are encouraged
to compete and engage in direct confrontation, girls are
taught to "be nice" and to appease others, a distinction
internalized in the play behavior of children as young as
three years of age.  As a consequence, verbal aggressiveness
comes to have a different significance for women than for
men; as Coates (1986) observes, women are apt to take
personal offense at what men may view as part of the
conventional structure of conversation.

                   Discussion of Results

     The results of this research can be summarized as
follows.  Despite the democratizing potential described in
the first section of this article, male and female academic
professionals do not participate equally in academic CMC.
Rather, a small male minority dominates the discourse both
in terms of amount of talk, and rhetorically, through self-
promotional and adversarial strategies.  Moreover, when
women do attempt to participate on a more equal basis, they
risk being actively censored by the reactions of men who
either ignore them or attempt to delegitimize their
contributions.  Because of social conditioning that makes
women uncomfortable with direct conflict, women tend to be
more intimidated by these practices and to avoid
participating as a result.  Thus Habermas's conditions for a
democratic discourse are not met: although the medium
theoretically allows for everyone with access to a network
to take part and to express their concerns and desires
equally, a very large community of potential participants is
effectively prevented by censorship, both overt and covert,
from availing itself of this possibility.  Rather than being
democratic, academic CMC is power-based and hierarchical.
This state of affairs cannot however be attributed to the
influence of computer communication technology; rather, it
continues pre-existing patterns of hierarchy and male
dominance in academia more generally, and in society as a
whole.

     How can we reconcile these findings with the more
encouraging reports of democratization based on earlier
research?  The claim of status-free communication hinges in
large part on the condition of anonymity (Graddol & Swann,
1989; Kiesler et al., 1984), a condition that is not met in
the discourse analyzed here, since most messages were
signed, or else the sender's identity is transparently
derivable from his or her electronic address.[9] In very few
cases could there have been any doubt upon receipt of a
message as to the sex of the sender, and thus sex-based
discrimination could freely apply.  However, given the
existence of 'genderlects' of the sort identified here, it
is doubtful that such discrimination would disappear even if
everyone were to contribute anonymously.  Just as a
university president or a janitor's social status is
communicated through their unconscious choices of style and
diction, CMC contains subtle indications of participants'
gender.

     Second, CMC is claimed to be more uninhibited
(disorganized, anarchic), due to lack of established
conventions of use (Kiesler et al., 1984; Nelson, 1974).  It
is important, however, to distinguish between the
adversarial behavior observed on academic lists and
'flaming', which is defined as "excessive informality,
insensitivity, the expression of extreme or opinionated
views, and vulgar behavior (including swearing, insults,
name calling, and hostile comments)" by McCormick and
McCormick (1992, p.381).  While 'flaming' may well result
from spontaneously venting one's emotion, adversariality is
a conventionalized and accepted (indeed, rewarded) pattern
of behavior in academic discourse, and characterizes
postings that otherwise show evidence of careful planning
and preparation.  Rather than being at a loss for a set of
discourse conventions, the members of these lists appear to
have simply transferred the conventions of academic
discourse, as they might be observed for example in
face-to-face interaction at a professional conference, to
the electronic medium, with some modifications for the
written nature of the message.

     Another factor claimed to lead to decreased inhibition
is the supposedly depersonalized nature of CMC.  However,
this assumption too can be challenged.  From my
observations, academic list subscribers do not view the
activity of posting as targeted at disembodied strangers.
Their addressees are either people with whom they have a
professional relationship, or could potentially develop such
a relationship in the future.  This is likely to increase
(rather than decrease) inhibition, since one's professional
reputation is at stake.  In this respect, the CMC discussed
here differs from the experimental CMC described by Kiesler
et al., where subjects risked nothing beyond the confines of
the experimental setting.  Three factors in Kiesler et al.'s
(1984, p.1129) experimental design were found to correlate
with less inhibited verbal behavior: anonymity, simultaneity
(as opposed to linear sequencing of messages), and
simultaneous computer conferencing (as opposed to electronic
mail).  None of these conditions obtained in the CMC
investigated in this study, since discussion lists, in
addition to not meeting the anonymity condition, present
postings linearly, and typically after some delay.

     In concluding, we return to the question of censorship,
freedom from which is an essential condition for democracy.
While it is true that no external censorship was exercised
by the moderators or owners of LINGUIST or MBU, women
participating in CMC are nevertheless constrained by
censorship both external and internal.  Externally, they are
censored by male participants who dominate and control the
discourse through intimidation tactics, and who ignore or
undermine women's contributions when they attempt to
participate on a more equal basis.  To a lesser extent, non-
adversarial men suffer the same treatment, and in and of
itself, it need not prevent anyone who is determined to
participate from doing so.  Where adversariality becomes a
devastating form of censorship, however, is in conjunction
with the internalized cultural expectations that we bring to
the formula: that women will talk less, on less
controversial topics, and in a less assertive manner.
Finally, although it was not a focus of the present
investigation, women are further discouraged from
participating in CMC by the expectation -- effectively
internalized as well -- that computer technology is
primarily a male domain (McCormick and McCormick, 1991;
Turkle, 1991).  This expectation is reflected in the
responses of female survey respondents on both LINGUIST and
MBU to the question: "How comfortable/competent do you feel
with computer technology?".  Female respondents
overwhelmingly indicated less confidence in their ability to
use computers, despite the fact that they had had the same
number of years of computer experience as male
respondents.[10] Internalized censorship of this sort
reflects deeper social ills, and it is naive to expect that
technology alone will heal them.


                           Notes

[1]  Examples of opaque electronic addresses drawn from the
     data reported on in this study include
     'f24030@barilvm', 'T52@dhdurz1', 'SNU00169@krsnucc1',
     and the like.

[2]  An example of such censorship is the Defense
     Communications Agency's periodic screening of messages
     on the government-sponsored network ARPANET "to weed
     out those deemed in bad taste" (Kiesler et al., 1984,
     p.1130).

[3]  With sex-neutral first names such as Chris or Robin, or
     with foreign names that I did not recognize, I
     contacted individuals by e-mail, identified the nature
     of my research, and asked whether they were male or
     female.  By employing a combination of methods, I was
     able to determine with a reasonable degree of certainty
     the sex of approximately 95% of contributors on both
     lists.

[4]  These percentages were calculated from lists of
     subscribers as of September 1992, on the basis of names
     from which sex could reliably be inferred.

[5]  While it is difficult to evaluate objectively what
     might count as evidence of a "vituperative" tone, the
     only postings to contain personal criticism of other
     participants (excluding the messages of protest) were
     those contributed by the man who originally proposed
     the 'Men's Literature' course.

[6]  Korenman, Joan (KORENMAN@UMBC.BITNET), Women's Studies
     List, June 9 11:08 PDT, 1992.  Significantly, in the
     only instance I observed where an issues discussion on
     WMST was allowed to take place, the discussion --
     concerned with media bias in reporting research on sex
     differences in the brain -- was one in which all
     participants were essentially in agreement.

[7]  There are also differences in character between the two
     lists.  The overall level of formality is higher, and
     differences in sex-based styles greater, for LINGUIST
     than for MBU. Due perhaps to the rhetorical practices
     that currently characterize the two fields (formal
     argumentation in linguistics, vs. creative
     collaboration in composition), discourse in the former
     tends to be more adversarial, or 'masculine', while
     discourse in the latter is more personal, or
     'feminine'.  (For example, both men and women reveal
     information about their feelings and their non-academic
     lives on MBU, whereas postings of this sort are
     virtually nonexistent on LINGUIST.)

[8]  The tendency for a small minority to dominate the
     discourse is evident on MBU as well.  Eight percent of
     participants (all but one of them male) produced 35% of
     the words in the 'sexism' discussion.  This minority
     dominated rhetorically by posting long-winded and often
     obscure postings, an abuse more common on MBU than
     overt adversarial attacks.

[9]  As far as I was able to ascertain, surprisingly few
     participants took advantage of the anonymity potential
     of the medium.  Fewer than 2% of contributors attempted
     to disguise their identity, and when they did so, it
     was for humorous effect.

[10] In the MBU survey, 30% of female respondents reported
     feeling "somewhat hesitant" about using computers, as
     compared with 5% of the men (the rest of whom rated
     themselves as "competent" or "extremely competent").
     In the LINGUIST survey, 13% of the women responded
     "somewhat hesitant" as compared with none of the men.
     The average length of computer use for both sexes was 9
     years on MBU and 11 years on LINGUIST.


                        References

Bolter, J. D. (1991).  Writing space: The computer,
     hypertext, and the history of writing.  Hillsdale, NJ:
     Lawrence Erlbaum.

Coates, J. (1986).  Women, men, and language.  New York:
     Longman.

Ess, C. (in press).  The political computer: Hypertext,
     democracy, and Habermas.  Forthcoming in George Landow
     (Ed.)  Hypertext and literary theory.  Baltimore: Johns
     Hopkins University Press.

Ferrara, K., Brunner, H., & Whittemore, G. (1991).
     Interactive written discourse as an emergent register.
     Written Communication, 8, 8-34.

Graddol, D., & Swann, J. (1989).  Gender voices.  Oxford:
     Basil Blackwell.

Habermas, J. (1983).  Diskursethik: Notizen zu einem
     Begruendungsprogram.  In Moralbewusstsein und
     kommunikatives Handeln (Frankfurt: Suhrkamp).
     Translated as "Discourse ethics: Notes on philosophical
     justification".  In Christian Lenhardt and Shierry
     Weber Nicholsen (Trans.).  Moral Consciousness and
     Communicative Action (43-115).  Cambridge: MIT Press
     (1990).

Herring, S.C.  (1992).  Gender and participation in
     computer-mediated linguistic discourse.  ERIC
     Clearinghouse on Languages and Linguistics, October
     1992.

Herring, S. C. (in press).  Men's language: A study of the
     discourse of the LINGUIST list.  Proceedings of the
     XVth International Congress of Linguists, Aug. 9-14,
     1992, Quebec.

Herring, S. C., Johnson, D., & DiBenedetto, T. (in press).
     Participation in electronic discourse in a 'feminist'
     field.  In Locating Power: Proceedings of the 1992
     Berkeley Women and Language Conference.  Berkeley
     Linguistic Society.

Holmes.  J. (1992).  Women's talk in public contexts.
     Discourse and Society, 3(2), 131-150.

Kahn, A. S., & Brookshire, R. G.  (1991).  Using a
     computer bulletin board in a social psychology course.
     Teaching of Psychology, 18.4, 245-249.

Kiesler, S., Seigel, J., & McGuire, T. W. (1984).  "Social
     psychological aspects of computer-mediated
     communication".  American Psychologist, 39, 1123-1134.

Landow, G. P. (1992).  Hypertext: The convergence of
     contemporary critical theory and technology.
     Baltimore: Johns Hopkins University Press.

McCormick, N. B., & McCormick, J. W. (1991).  Not for men
     only: Why so few women major in computer science.
     College Student Journal, 85(3), 345-50.

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

Nelson, T. H.  (1974).  Dream machines: New freedoms through
     computer screens -- A minority report.  Computer Lib:
     You Can and Must Understand Computers Now.  Chicago:
     Hugo's Book Service.  (Rev.  Ed.)  Redmond: Microsoft
     Press, 1987.

Sheldon, A. (1992).  Conflict talk: Sociolinguistic
     challenges to self-assertion and how young girls meet
     them.  Merrill-Palmer Quarterly, 38(1), 95-117.

Smith, J., & Balka, E. (1991).  "Chatting on a feminist
     computer network".  In C. Kramerae (Ed.).  Technology
     and women's voices, (pp. 82-97).  New York: Routledge
     and Kegan Paul.

Spender, D. (1979).  Language and sex differences.  In
     Osnabrueker Beitraege zur Sprach-theorie: Sprache und
     Geschlect II, 38-59.

Turkle, S. (1991).  Computational reticence: Why women fear
     the intimate machine.  In C. Kramarae (Ed.)  Technology
     and women's voices, (pp. 41-61).  New York: Routledge
     and Kegan Paul.


                    Appendix A:  Survey

To:  All Linguist List subscribers
Re:  Participation in Linguist List discussions


I am conducting a sociolinguistic study of participation in
the Linguist List discussion group.  The following is a brief
(12 question) survey regarding the debate which took place
during February and March of this past year on the use of
the term 'cognitive linguistics'.  If you read even one
contribution to that debate, please take the time to fill
out the survey below:

-------------------------------------------------------------
                          SURVEY

1)   A total of 48 messages appeared under the heading
     'cognitive linguistics' (or 'language autonomy/
     modularity') in Feb. and March of this year.  What
     percentage (approximate) of these did you read/glance
     through?

2) a. Did you contribute to the discussion, and if so, how
      many times?

   b. If you did not contribute, explain as fully as you
      can why not (not interested in topic; interested but
      too busy; interested but felt intimidated; etc.)

3)   At times the 'cognitive linguistics' discussion became
     heated and even personal.  As best you can recall,
     describe your reactions to the discussion at the time.

4)   Have you contributed to any other discussions on
     LINGUIST?  How often?


Respondent Information (IMPORTANT)
----------------------------------
5)   Your academic position:  (Lecturer [non-tenure track];
     Assist./Assoc./Professor; Emeritus; Grad Student;
     Undergrad; not affiliated with academia)

6)   Male or female?

7)   Number of years in linguistics (break down into student
     years/post-grad):

8) a. If you teach, average number of courses taught per
      year:

   b. Number of courses you were teaching in Feb./March
      1991:

9)   Principal area of specialization within linguistics:
     (If not primarily a linguist, state major field)

10)  If you had to choose, would you describe yourself as
     more of a 'formalist' or a 'functionalist'?

11)  How long have you been using a computer?

12)  How comfortable/competent do you feel with computer
     technology?

------------------------------------------------------------

(Note: Roughly 5% (N=64) of LINGUIST subscribers responded
to this survey. 11% (N=28) of MBU subscribers responded to a
similar survey distributed on MBU.)

------------------------------------------------------------
                      Copyright 1993
   Communication Institute for Online Scholarship, Inc.