Download Effects of Gendered Language on Gender

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
Human Communication Research ISSN 0360-3989
ORIGINAL ARTICLE
Effects of Gendered Language on Gender
Stereotyping in Computer-Mediated
Communication: The Moderating Role
of Depersonalization and Gender-Role
Orientation
Eun-Ju Lee
Department of Communication, University of California—Davis, Davis, CA 95616
This experiment examined what situational and dispositional features moderate the
effects of linguistic gender cues on gender stereotyping in anonymous, text-based
computer-mediated communication. Participants played a trivia game with an ostensible partner via computer, whose comments represented either prototypically masculine
or feminine language styles. Consistent with the social identity model of deindividuation
effects, those who did not exchange brief personal profiles with their partner (i.e., depersonalization) were more likely to infer their partner’s gender from the language cues
than those who did. Depersonalization, however, facilitated stereotype-consistent conformity behaviors only among gender-typed individuals; that is, participants conformed
more to their masculine- than feminine-comment partners, and men were less conforming than were women, only when they were both gender-typed and depersonalized.
doi:10.1111/j.1468-2958.2007.00310.x
Although recent technological advancements have substantially expanded the capacity of computer networks to transmit various auditory and visual cues, the most
common form of computer-mediated communication (CMC) still remains text
based. Thus, when compared with face-to-face interaction, CMC is normally characterized by a greater (perceived) anonymity and a reduced sense of social presence
(Short, Williams, & Christie, 1976). According to this ‘‘cues-filtered-out’’ model,
which dominated early CMC research, the paucity of sociocontextual cues has the
potential of liberating individuals from normative concerns and social constraints
that typically regulate everyday interpersonal encounter, leading to uninhibited
behaviors (Joinson, 1998; Walther & Parks, 2002). Disinhibition might manifest
itself in antinormative behaviors, such as flaming, but it can also promote positive
Corresponding author: Eun-Ju Lee; e-mail: [email protected]
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
515
Gender Stereotyping in CMC
E.-J. Lee
outcomes. For example, anonymity could mask status differentials and thus equalize
the participation and perceived influence of individual members; that is, by reducing
or eliminating sociocontextual cues, CMC might lower evaluation apprehension,
especially among those with lower status, and thereby enable people to interact more
fully and equally (e.g., Dubrovsky, Kiesler, & Sethna, 1991; Kiesler & Sproull, 1992).
Similarly, one might expect that the elimination of visual cues will suppress social
stereotypes, which are often linked to group memberships readily discernible from
physical appearance, such as gender and race (Herschel, Cooper, Smith, & Arrington,
1994; Sproull & Kiesler, 1991).
Contrary to the optimistic notion that this cue lean medium will trivialize group
membership differences, however, more recent studies have documented the persistence of social stereotypes in CMC. For example, Epley and Kruger (2005) reported
that after interviewing a target over e-mail, participants rated Asian American
women as more introverted than African American women, although no such differences were found after telephone interviews (Experiment 2). Likewise, even though
the interaction partner’s gender was not explicitly identified, when a gender-marked
cartoon character represented their partner, participants exhibited greater conformity to the male- than female-charactered partners (Lee, 2005; Experiment 3).
Collectively, these findings seem to support the social identity model of deindividuation effects (SIDE), which posits that the lack of individuating information (i.e.,
depersonalization) in CMC can amplify, rather than attenuate, social stereotyping
by heightening the salience of simple category cues in the immediate environment
(Postmes, Spears, & Lea, 1998, 2002; Spears & Lea, 1992, 1994).
The present research tested the notion that depersonalization fosters categorybased person perceptions and group-oriented behaviors in CMC (Postmes et al.,
1998; Spears, Postmes, Lea, & Wolbert, 2002). Specifically, by varying the amount
of personal information about anonymous interactants, this study investigated how
depersonalization moderates the extent to which people spontaneously utilize
gender-linked language cues (Mulac, 1998) to make inferences about their interactants
and respond to informational social influence. In doing so, participants’ gender-role
orientation (Bem, 1981b; Frable & Bem, 1985) was also measured to probe how
depersonalizing context interacts with stable individual differences, which has often
been overlooked in previous CMC studies.
Gendered language use and its consequences
Despite the absence of physical indicators, some category membership cues might
still remain in CMC. For example, mirroring gender differences in language use
observed in face-to-face communication (Eakins & Eakins, 1978; Lakoff, 1975;
Mulac, 1998), researchers have found that men and women exhibit different stylistic
features and communication patterns in CMC. Specifically, men’s postings to newsgroups showed self-promotion, sarcasm, insults, and strong assertions, whereas
women’s messages revealed supportiveness and personal orientation (Herring,
516
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
E.-J. Lee
Gender Stereotyping in CMC
1993). Likewise, female-only discussion groups displayed more self-disclosure and
coalition language than did male-only or mixed-gender groups (Savicki, Kelley, &
Oesterreich, 1998). Even when individuals consciously sought to deceive others
about their gender, ‘‘their use of stylistic features at the word and sentence level
remained largely true to their real-life genders’’ (Herring & Martinson, 2004, p. 428).
If men and women tend to adopt different communication styles, albeit to
different degrees depending on contextual and dispositional factors (Palomares,
2004), how do such language differences influence individuals’ inferences about
and behavioral reactions to their CMC partners? The current research investigated
if and when gender-typed language use evokes corresponding gender inferences and
activates associated stereotypes in synchronous, text-only CMC. In doing so, unlike
previous studies that examined how accurately people detect the communicator’s
gender in CMC (e.g., Nowak, 2003; Savicki, Kelley, & Oesterreich, 1999), the present
study focused on the effects of an individual difference (gender-role orientation) and
its interaction with a situational factor (depersonalization) on language-based gender inferences. In addition, although previous research has mostly defined genderlinked language effect in terms of the respondents’ perceptions of the speakers, such
as aesthetic quality and dynamism (Mulac, Incantro, & James, 1985; Mulac &
Lundell, 1980), this study also examined how individuals’ behavioral responses to
their anonymous interactant vary as a function of the gendered language style. Last,
to remove immediate gender indicators, such as voice and physical appearance,
respondents in previous studies were presented with transcribed speeches, rather
than engaging in actual real-time interactions. By detaching the message from the
source, however, this procedure might have lowered the salience of the speaker,
thereby discouraging spontaneous gender inferences; that is, the respondents might
not have thought about the speaker until asked to evaluate the imaginary person
on various dimensions. If so, gender-linked language effects might have been exaggerated due to the experimental demands. To address this limitation, the current
experiment examined the effects of gendered language styles in a synchronous
CMC context.
Moderators of gendered person perception: Depersonalization and
gender-role orientation
According to the SIDE model, the lack of individuating information in CMC can
foster social stereotyping by elevating the salience of social category cues in the
otherwise cue-deprived environment, which in turn prompts category-based perceptions and group-oriented behavior (Postmes et al., 1998, 2002). Although they
did not directly examine social stereotyping, Tanis and Postmes (2003) found that
preference for in-group (same university) over out-group (different university)
members for future collaboration dissipated with the provision of portrait pictures
and biographic information (Study 2). Likewise, when gender stereotypes were activated prior to discussion, depersonalized participants were more likely to exhibit
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
517
Gender Stereotyping in CMC
E.-J. Lee
stereotype-consistent communication behavior than their personalized counterparts
(Postmes & Spears, 2002); that is, men were more likely than women to dominate the
group discussion and asked fewer questions when the topic was masculine, whereas
the reverse was true for a feminine topic.
If the paucity of personalizing information in CMC leads people to think and
behave in terms of social categories, it might as well make them more vigilant to the
gender-linked language styles their interactants display. In fact, participants in previous research were typically induced to focus on a particular social identity; that is,
they were directly informed of the target’s racial identity (Epley & Kruger, 2005) or
primed with gender stereotypes (Postmes & Spears, 2002). Therefore, it was impossible to evaluate how depersonalization might moderate the extent to which people
utilize available cues to make spontaneous inferences about their interactant’s category membership. By merely exposing participants to short comments reflecting
either prototypically masculine or feminine language styles during dyadic CMC, this
experiment directly tested if the lack of personal information increases individuals’
sensitivity to relatively unobtrusive group identity cues, even when they are not
explicitly induced to think in group terms.
In doing so, the present experiment also investigated the role of individual
differences in shaping individuals’ reactions to the information restrictions of the
medium, by focusing on traditional gender-role orientation (i.e., gender-typedness).
Defined as the extent to which individuals’ self-perceptions ‘‘conform to the culture’s
definitions of maleness and femaleness’’ (Bem, 1981b, p. 355), gender-role orientation differentiates those who tend to attribute to themselves the traits traditionally
associated with their biological gender (gender-typed) from those who do not
(nongender-typed). If, as self-categorization theory posits, those who see themselves
as more prototypical of their own group have a stronger proclivity to act in terms of
group norms and display stronger in-group bias (Spears, 2001; Turner, 1987),
gender-typed individuals might also show greater vigilance to linguistic gender
markers in the absence of explicit gender indicators. In fact, researchers have shown
that gender-typed individuals were more likely to confuse the members of the opposite gender with one another (Frable & Bem, 1985) and actively avoid crossgender
behavior even at monetary costs (Bem & Lenney, 1976), suggesting that gendertyped individuals are more prone to spontaneously use gender categories in person
perception and engage in stereotype-consistent behavior.
H1a-b: The effects of linguistic gender cues on gender inferences will be more
pronounced (a) among gender-typed (vs. nongender-typed) individuals and
(b) under depersonalization (vs. personalization).
Gender and social influence
In addition to gender inferences, this study also examined gender stereotyping as
potential behavioral consequences of the gender-linked language use. Specifically,
518
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
E.-J. Lee
Gender Stereotyping in CMC
gender stereotyping was operationalized as differential attribution of competence
and conformity to male and female influence agents (see Carli, 1999, for a review on
gender differences in social influence). According to expectation states theory
(Berger, Fisek, Norman, & Zelditch, 1977; Wagner & Berger, 1997), men enjoy
greater advantage over women in exerting social influence due to the differential
performance expectations. In a group setting, absent information to the contrary, the
higher status person is assumed to be more competent at the group’s task. Because
men tend to hold higher positions than do women within most groups and organizations (Eagly, 1983), there are external status differences between men and women.
This sex-based status differential becomes internalized, as people develop greater
performance expectations for men than for women (Meeker & Weitzel-O’Neill,
1977).
Advancing this idea, status characteristic theory further argues that gender functions as a ‘‘diffuse status characteristic,’’ such that even with a task irrelevant to
gender, people tend to ‘‘implicitly treat it as relevant unless challenged’’ and ‘‘form
higher performance expectations for the men than for the women’’ (Ridgeway, 1988,
pp. 191–192). Consequently, people more often overtly agree with men’s contribution
than with women’s and are more likely to defer to the opinions of men than those of
women (Berger et al., 1977; Wagner & Berger, 1997). Likewise, although there was no
objective difference in performance, participants rated men to have performed more
competently than women after face-to-face interaction (Carli, 1991; Wood & Karten,
1986). Despite the fact that more women are now in management positions than
before, recent studies suggest that gender still operates as a diffuse status characteristic. For example, people believed that men had advantages in negotiations, and
moreover, men actually outperformed women when the negotiation was perceived
as diagnostic of ability, whereas no such difference was found when the task was
framed as a learning tool (Kray, Thompson, & Galinsky, 2001). Based on previous
findings, gender stereotyping was examined in terms of the differential competence
attribution and conformity to masculine and feminine interaction partners.
H2a-b: Gender-typed individuals will be more likely than their less gender-typed
counterparts to (a) attribute greater task competence and (b) exhibit conformity to
their partner whose comments reflect masculine than feminine language styles.
Less straightforward, however, was the potential interaction between depersonalization and gender-role orientation. First, depersonalization might override the effects
of gender-role orientation by heightening the salience of group membership cues,
and thereby triggering gender-stereotypical reactions even from nongender-typed
individuals, who would not exhibit such responses otherwise. For example, Onorato
and Turner (2001) reported that when female participants were induced to compare
themselves with men (intergroup comparison), they came to describe themselves as
more dependent, regardless of their prior self-ratings of dependence as a personal
trait. In contrast, self-reported dependence continued to serve as a good predictor of
self-descriptions when personal identity was made salient in intragroup comparison
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
519
Gender Stereotyping in CMC
E.-J. Lee
with other women. Similarly, Mulac, Jansma, and Linz (2002) found that a sexually
explicit film prompted nongender-typed men to display more dominance and anxiety in subsequent interaction with a female partner, whereas gender-typed men
behaved in a consistently gender-typed manner regardless of the materials shown.
Although they did not manipulate the level of self-identity (personal vs. social), such
findings suggest that certain contextual cues might override relatively stable dispositional differences in gender-role orientation. If so, the removal of interpersonal
cues might temporarily elevate the salience of the category cues and subsequently,
‘‘eliminate or, at least attenuate individual differences’’ (Onorato & Turner, 2001,
p. 165).
Alternatively, depersonalization might amplify the effects of gender-role orientation. According to the SIDE model, depersonalization does not automatically
prompt people to focus on their group identity and behave in group terms. Rather,
it accentuates the influence of group norms ‘‘provided that group identity is salient’’
(Spears et al., 2002, p. 96). In such a case, depersonalization might increase the
salience of gender identity and evoke gender stereotyping only among those who
tend to identify with their gender group. Although they did not deal with enduring
individual differences in gender-role orientation, Postmes and Spears (2002) showed
that depersonalization promoted self-stereotyping only when gender stereotypes
were activated prior to the computer-mediated discussion (Study 2). Similarly,
depersonalization was also found to increase in-group favoritism only among high
identifiers, not among low identifiers (Tanis & Postmes, 2003; Study 3); that is,
individual differences in group identification predicted the preference for in-group
members only in the absence of interpersonal cues. Given that gender-role orientation captures (self-assessed) prototypicality of an individual as a member of his or
her gender group, depersonalization might amplify gender stereotyping only for
those whose self-perceptions are congruent with traditional gender roles. Just as
pornography led only gender-typed men to focus on the women’s sexuality and
disregard their intellectual competence (McKenzie-Mohr & Zanna, 1990), depersonalization might accentuate individual differences in gender-typed person perceptions
and behavioral reactions.
RQ1: Does the deficiency of personal information moderate the ways in which
gender-typed and nongender-typed individuals (a) attribute task competence and (b)
conform to their partner, in response to gender-linked language use? If so, how?
Although research has indicated that women tend to be more susceptible to social
influence than are men (Burgoon & Klingle, 1998; Eagly & Carli, 1981), such a difference is more or less likely to emerge depending on dispositional and situational
variables. First, considering that gender-role orientation not only biases the way
people process information about the social world but also limits their behavior to
gender-appropriate domains (Bem & Lenney, 1976), and that there are cultural
expectations prescribing men to be more independent and women to be more
520
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
E.-J. Lee
Gender Stereotyping in CMC
supportive (Eagly & Wood, 1985), gender-role orientation was expected to amplify
gender differences in conformity.
H3: Women will be more likely than men to conform to their partner. Such a tendency,
however, will be more pronounced among gender-typed than nongender-typed
individuals.
Second, depersonalization might have different effects on men’s and women’s conformity behavior. Although there are several explanations for why men tend to show
less susceptibility to social influence (Burgoon & Klingle, 1998), if gender differences
stem from the adherence to cultural norms valuing independence for men and not
for women (Eagly & Wood, 1985), depersonalization might encourage people to
comply with such gender norms, thereby accentuating gender differences. That is, if
group behavior can be understood as individuals acting in terms of a shared identity,
rather than as individual persons (Turner, 1987), the removal of individuating cues
might prompt people to exhibit stereotype-consistent conformity behavior by discouraging the activation of personal identity. Consistent with this notion, Hogg and
Turner (1987) found that both men and women categorized themselves as more
typical of their own gender and stereotyped themselves more, following an interaction in which gender identity was salient. As reasoned above (RQ1), however, such
depersonalization effects might take different forms depending on individuals’
chronic proclivity for self-stereotyping, as indicated by the gender-role orientation.
RQ2: How does the deficiency of personal information affect men’s and women’s
conformity, independently and in conjunction with the gender-role orientation?
Method
Participants
A total of 193 undergraduate students (96 women and 97 men) participated in the 2
(participant gender: male vs. female) 3 2 (gendered language style: masculine vs.
feminine) 3 2 (personalizing information: depersonalization vs. personalization)
3 2 (gender-role orientation: high vs. low) between-subjects design experiment
for extra course credit. Due to the uneven distribution of gender-role orientation
across conditions, cell sizes ranged from 8 to 16.
Pretest of gender-linked language style
To test if the partner’s comments revealed the intended gender, 44 participants
(22 women and 22 men) took part in a pretest. Specifically, the comments were
constructed to reflect previously documented gender differences in language style
(e.g., Herring, 1993; Mulac, 1998). First, I created short, gender-neutral comments
for each question (e.g., ‘‘I have no clue’’ ‘‘It’s either a or b’’ ‘‘Peace and love sound
more feminine’’ ‘‘Vertigo was before Psycho, I don’t know about others’’). Then,
feminine comments were constructed by adding prototypically feminine stylistic
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
521
Gender Stereotyping in CMC
E.-J. Lee
features, such as graphical accents to express emotions (i.e., emoticons), tag questions, and apologies (e.g., ‘‘Sorry, no clue . any ideas?’’ ‘‘Peace and love sound more
feminine, don’t they?’’ ‘‘I think it’s either a or b :-)’’ ‘‘Vertigo was before Psycho . I
don’t know about others’’). In contrast, masculine comments were more direct,
made references to quantity, and included elliptical sentences (e.g., ‘‘not a clue,’’
‘‘peace and love sound more feminine,’’ ‘‘either a or b, 50% confident,’’ ‘‘vertigo was
before psycho, dunno about others’’). Participants were presented either masculine
or feminine comments and rated how well each of the following adjectives described
the author on a 10-point scale (1—describes very poorly, 10—describes very well):
trustworthy, masculine, intelligent, confident, informed, friendly, competent, likable, feminine, and honest. A series of 2 3 2 (Participant Gender 3 Gendered
Language Style) analyses of variance (ANOVAs) showed that the two sets of comments were perceived differently only in gender-related terms. That is, participants
rated the feminine comments (M = 7.91, SD = 2.35) as significantly more feminine
than the masculine comments (M = 4.23, SD = 1.95), F(1, 40) = 31.54, p , .001, h2 =
.44. Likewise, they thought the masculine comments (M = 5.91, SD = 1.97) were
significantly more masculine than the feminine comments (M = 2.64, SD = 1.40),
F(1, 40) = 39.10, p , .001, h2 = .49. In addition, to obtain the relative masculinity–
femininity index, femininity score was subtracted from the masculinity score
(r = 2.85, p , .001). One sample t-tests showed that both comments were perceived
as significantly different from 0.00 (i.e., the same degrees of masculinity and femininity attribution) in the intended direction, t(21) = 26.79, p , .001 and t(21) = 2.22,
p = .04, for feminine and masculine comments, respectively. There were no significant
main or interaction effects on other dimensions, all Fs , 1.53, all ps . .23, suggesting
that the masculine and feminine comments were equivalent in other respects. Furthermore, when the participants were asked to guess the author’s gender, male or
female, they assigned the gender category as intended, x2(1, N = 44) = 9.17, p = .002.
Procedure
Upon arrival at the laboratory, participants were told that they would play a trivia
game with another person as a team. To ensure rigorous experimental control over
the content of interaction, however, the entire interaction was preprogrammed.
When asked about the purpose of the study after the experiment, 18 of 211 participants (8.5%) expressed suspicion regarding the existence of the interaction partner.
Their data were not analyzed.
Before they started the interaction, those in the personalization condition had the
opportunity to introduce themselves to their ostensible partner by sharing brief
personal profiles, such as major, age, hobby, favorite TV show, favorite color, and
favorite music genre, without disclosing identifying information. For example, one
profile represented a 21-year-old sociology major who likes hiking and watching the
TV show ‘‘Lost,’’ whose favorite color is green, and who frequently listens to jazz and
alternative. After providing this information, they saw their interactants’ input.
Those in the depersonalization condition skipped this part and thus had no personal
522
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
E.-J. Lee
Gender Stereotyping in CMC
information about their supposed partner. To avoid the possibility that some idiosyncratic features of the personalizing information significantly affect participants’
impression about their partner, four different sets of personal profiles were created
and one of them was randomly presented to each participant.
Participants were then asked to select a number, ranging from 1 to 10, to determine trivia questions, although the computer presented identical questions to every
participant, irrespective of the number chosen. For each multiple-choice question,
participants picked their initial answer and typed a short comment to communicate
with their partner. They also indicated their confidence before seeing their partner’s
choice. On the following page, the participants’ and their partners’ answers and
comments were displayed. For 12 of 15 questions, the partner picked a different
answer from the participants’, thereby creating an opportunity to exert conformity
pressure. Because masculine language tends to convey greater confidence of the
communicator, to rule out the possibility that participants become more receptive
to the masculine-comment partner due to the inflated perception of the person’s
confidence, the partner’s self-confidence was also presented in numeric form, which
was held constant across conditions. After reading the partner’s comments and the
confidence score, participants submitted their final answer. Hence, the participants’
main task was to determine how likely their partner would know the correct answer
and decide whether or not to change their answer accordingly. To reduce suspicion,
the partner presented the same answer as the participants’ on three filler questions.
To enhance participants’ involvement in the study and more importantly, to
discourage blind rejection of their partner’s answers, a cash prize was offered for the
team with the highest score. Moreover, to reduce strategic nonconformity (i.e.,
choosing a different answer from the partner’s to maximize their team score when
neither of them was certain about the correct answer), the experimenter told the
participants that the team would earn points only when both members answered
correctly. Thus, they needed to accept the partner’s answer if it seemed like a correct
one. Once participants submitted the final answer, the computer presented the next
question without revealing either the correct answer or the final answer chosen by
the partner, so that the partner’s (non)conformity on prior questions would not
affect their conformity decisions on subsequent questions. This procedure was
repeated for 15 questions.
Index construction
Following the interaction, participants were instructed to indicate ‘‘how well each of
the following words describes your partner you interacted with’’ on a 10-point scale
(1—describes very poorly, 10—describes very well). For gender inferences, participants
evaluated how feminine and masculine their interactant was, respectively. The femininity score was then subtracted from the masculinity rating, resulting in an index
whose positive scores indicated greater masculinity relative to femininity attribution
and whose negative scores represented the opposite (a = .69, M = 2.26, SD = 3.83).
Similarly, perceived competence of the partner was measured by the following three
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
523
Gender Stereotyping in CMC
E.-J. Lee
items: intelligent, informed, and competent. The ratings were summed across the
three items (a = .89, M = 16.77, SD = 4.76).
Conformity was operationalized as how many times, out of 12 critical questions,
participants switched their initial answer to their partner’s (M = 6.30, SD = 2.38).
Participants’ self-confidence in their own answer before seeing their partner’s choice
was measured on an 11-point scale, ranging from 0% to 100% in 10% increments,
and then averaged (a = .87, M = 24.03, SD = 14.66).
Gender-role orientation was measured by the Bem Sex-Role Inventory (Bem,
1981a), which consists of 60 items asking how self-descriptive certain adjectives are
(1—almost never true, 7—almost always true). Twenty of the items measure femininity (e.g., tender, understanding), 20 measure masculinity (e.g., assertive, independent), and the remaining items serve as filler (a = .85, M = 4.80, SD = .67 for
femininity score and a = .90, M = 4.82, SD = .82 for masculinity score). Gender-role
orientation scores were dichotomized based on the difference between a participant’s
masculinity and femininity scores, for it is a more sensitive index of gender-role
orientation, less susceptible to social desirability, and of higher predictive utility for
gender-role attitudes than the simple median split approach (Orlofsky, Aslin, &
Ginsburg, 1977). Specifically, participants’ femininity score was subtracted from
their masculinity score and women’s scores were then reverse-coded to yield a gender
typicality scale. The median (10) split classified the participants as either gendertyped (above the median) or nongender-typed. Among nongender-typed participants, 48 participants’ self-perceptions were congruent with their gender identity
(above zero), but not as strongly gender biased as those of gender-typed participants;
7 participants attributed the same degrees of masculinity and femininity to themselves; and 48 participants described themselves in a counterstereotypical way
(e.g., men depicting themselves as more feminine than masculine).
Given that the feminine language style, characterized by greater self-disclosure,
expression of emotions and we-references, tends to foster group development
(Savicki et al., 1998), gender-linked language differences might also have affected
the extent to which participants identified with their partner as a team, and subsequently, conformity decisions. To assess this possibility, participants were asked
how much they agreed with each of the following statements (1—strongly disagree,
10—strongly agree) and the scores were summed across the three items: ‘‘I felt I had
a lot in common with my partner’’; ‘‘I found it easy to identify with my partner’’;
‘‘I felt as though we were a team’’ (a = .84, M = 15.34, SD = 5.81).
Results
Hypothesis tests
To test H1a-b, which predicted that (a) traditional gender-role orientation and (b)
depersonalization will make individuals more prone to utilize linguistic gender cues
for gender inferences, a 2 3 2 3 2 3 2 (Personalizing Information 3 Gendered
Language Style 3 Gender-Role Orientation 3 Participant Gender) ANOVA was
524
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
E.-J. Lee
Gender Stereotyping in CMC
performed on gender inferences. First, although the interaction between gender-role
orientation and gendered language style failed to reach a significance level, F(1, 177) =
3.37, p = .07, hp2 = .02, simple effects tests (Page, Braver, & MacKinnon, 2003) showed
that traditional gender-role orientation indeed amplified the individuals’ tendency to
infer their partner’s gender-related traits from language characteristics. Specifically,
gender-typed participants ascribed greater masculinity to their partner whose comments reflected the masculine (M = 1.14, SD = 3.12) than feminine language styles
(M = 21.87, SD = 4.13), F(1, 189) = 14.89, p , .001, hp2 = .07. For nongender-typed
participants, however, linguistic gender cues had no significant effects on gender
inferences, F(1, 189) = 2.25, p = .14.
With respect to H1b, there emerged a significant interaction between personalizing information and gendered language style, F(1, 177) = 7.94, p = .005, hp2 = .04.
Decomposition of the interaction showed that when participants were not given any
personal information about their partner, gender-linked language styles elicited
corresponding gender inferences; that is, depersonalized participants attributed
greater femininity to the feminine- (M = 22.60, SD = 4.06) than masculinecomment partners (M = .83, SD = 2.88), F(1, 189) = 22.40, p , .001, hp2 = .10.
By contrast, when participants had read their partner’s personal profile prior to the
interaction, linguistic gender markers did not significantly influence their gender
inferences, F , 1. Significant main effects for gendered language style, F(1, 177) =
15.83, p , .001, hp2 = .08, and for personalizing information, F(1, 177) = 5.48,
p = .02, hp2 = .03, should be interpreted in the context of this disordinal interaction.
To ensure that the results were not biased by (a) those who did not assign
a particular gender to their partner or (b) those who ascribed vastly different degrees
of masculinity and femininity to their partner (i.e., outliers), an additional analysis
was performed on the dichotomized gender inference scores. Specifically, those who
perceived their partner as more masculine than feminine were classified as ‘‘male
inference’’; those who perceived their partner as more feminine than masculine were
categorized as ‘‘female inference’’; those who attributed the same degrees of masculinity and femininity to their partner were excluded from the analysis. A series of chisquare tests confirmed the aforementioned interactions. First, when depersonalized,
those exposed to the feminine comments (N = 37) were more likely to perceive
their partner as more feminine than masculine (N = 29), whereas those presented
with the masculine comments (N = 28) were more likely to think of the partner in
masculine terms (N = 19), x2(1, N = 65) = 14.03, p , .001. By contrast, personalized
participants did not show the corresponding tendency, x2(1, N = 69) , 1. Second,
only gender-typed individuals took into account gender-typed comments to infer
their partner’s gender-related traits. That is, among gender-typed individuals, those
who saw the masculine comments (N = 29) were more likely to make male inference
(N = 20), whereas those who received the feminine comments (N = 35) were
more likely to make female inference (N = 25), x2(1, N = 64) = 10.39, p , .001.
No such effects were found for nongender-typed individuals, x2(1, N = 70) = 1.45,
p = .23.
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
525
Gender Stereotyping in CMC
E.-J. Lee
H2a-b concerned whether gender-typed individuals would be more likely to
exhibit gender stereotyping in terms of competence attribution and conformity to
their partner. First, a 2 3 2 3 2 3 2 (Personalizing Information 3 Gendered
Language Style 3 Gender-Role Orientation 3 Participant Gender) ANOVA on
perceived overall competence showed that gender-role orientation significantly
moderated the ways in which linguistic gender cues affect individuals’ perceptions
of their partners’ competence, F(1, 177) = 10.18, p = .002, hp2 = .05. As H2a
predicted, gender-typed individuals attributed greater competence to their partner
whose comments represented the masculine (M = 18.30, SD = 5.14) than feminine
language styles (M = 16.11, SD = 4.18), F(1, 189) = 4.85, p = .03, hp2 = .03. However,
nongender-typed individuals showed the opposite tendency, perceiving their
feminine-comment partners (M = 17.42, SD = 5.28) as more competent than their
masculine-comment counterparts (M = 15.57, SD = 4.62), F(1, 189) = 4.45, p = .04,
hp2 = .02.
Second, H2b predicted that greater conformity to masculine- than femininecomment partners would be more characteristic of gender-typed individuals.
A 2 3 2 3 2 3 2 ANOVA on conformity, however, yielded no significant interaction
between gendered language style and gender-role orientation, F(1, 177) = 1.14,
p = .29. Instead, there was a significant higher order interaction involving personalizing information, which addressed RQ1, F(1, 177) = 4.73, p = .03, hp2 = .03.
Decomposition of the interaction revealed that the predicted interaction between
gendered language style and gender-role orientation occurred only under depersonalization, F(1, 90) = 5.47, p = .02, hp2 = .06. Specifically, when personal information
about the partner was absent, gender-typed individuals tended to conform more to
their masculine (M = 6.96, SD = 1.51) than feminine partners (M = 5.71, SD = 2.90),
F(1, 90) = 3.08, p = .08, hp2 = .03, whereas nongender-typed individuals showed an
opposite tendency, F(1, 90) = 2.46, p = .12. When they exchanged personal profiles,
however, gendered language style had no significant main or interaction effects for
both gender-typed and nongender-typed individuals, all Fs , 1.
H3 raised an issue of whether traditional gender-role orientation would amplify
gender differences in influenceability. Consistent with the hypothesis, women were
more conforming to their partner than were men, but such a difference was found
only among gender-typed individuals, F(1, 177) = 9.82, p = .002, hp2 = .05. This
interaction, however, was also qualified by the significant three-way interaction
involving personalizing information, F(1, 161) = 4.30, p = .04, hp2 = .02. Simple
effects tests revealed that women (M = 7.91, SD = 1.68) were more susceptible to
social influence than were men (M = 4.87, SD = 1.82), only when they were both
gender-typed and depersonalized, F(1, 90) = 20.79, p , .001, hp2 = .19. When they
viewed their partner’s brief biographic information prior to the interaction or did
not have gender-stereotypical self-perceptions, men and women did not significantly
differ in overall conformity, all Fs , 1. Interestingly, however, nongender-typed men
(M = 6.92, SD = 1.69) exhibited greater conformity than nongender-typed women
(M = 5.95, SD = 2.63) under personalization, F(1, 95) = 5.06, p = .03,
526
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
E.-J. Lee
Gender Stereotyping in CMC
hp2 = .05. Although there was a significant interaction between personalizing information and participant gender, F(1, 177) = 14.96, p , .001, hp2 = .08, it should be
understood in the context of this three-way interaction, which addressed RQ2.
In addition, there was a near-significant three-way interaction involving gendered language style, participant gender, and personalizing information, F(1, 177) =
3.59, p = .06, hp2 = .02. Specifically, the interaction between participant gender and
gendered language style was significant only under depersonalization, F(1, 95) =
3.95, p = .05, hp2 = .04; that is, men (M = 5.09, SD = 2.56) were less likely than
women (M = 7.39, SD = 2.13) to switch to their partner’s answers, but only when
their partner displayed the feminine language style, F(1, 90) = 10.82, p = .001,
hp2 = .11. When presented with the masculine language style, men (M = 6.09,
SD = 2.14) and women (M = 6.44, SD = 2.58) did not show different levels of
conformity, F , 1. In contrast, when participants had an opportunity to share their
personal profiles, only the main effect for participant gender was statistically significant, with men (M = 6.83, SD = 2.02) conforming more than women (M = 5.74,
SD = 2.48), F(1, 95) = 5.73, p = .02, hp2 = .06. No other main or interaction effects
were statistically significant, all Fs , 1.1
To control for the potentially contaminating effects of the participants’ subjective
uncertainty on conformity, a 2 3 2 3 2 3 2 analysis of covariance was performed on
conformity with the participants’ self-confidence as a covariate. Although participants’ self-confidence turned out to be a significant covariate, F(1, 177) = 25.21,
p , .001, hp2 = .13, covarying it out did not substantially alter the findings. In
addition, a 2 3 2 3 2 3 2 ANOVA on self-confidence yielded no significant main
or interaction effects, all Fs , 2.35, all ps . .12, except that men were in general more
confident than were women, F(1, 161) = 13.88, p , .001, hp2 = .07. Taken together,
these results suggest that the variations in subjective uncertainty do not explain away
the conformity findings.
If feminine language style fosters group development (Savicki et al., 1998), it
might have evoked stronger team identification with the partner and subsequently,
elicited greater conformity. Indeed, feminine comments induced stronger wefeelings than did masculine comments, F(1, 177) = 4.19, p = .04, hp2 = .02. In
addition, team identification was positively correlated with conformity (r = .19,
p = .008). Nonetheless, conformity results remained virtually identical after team
identification was covaried out, which indicates that team identification does not
account for the effects of the gendered language style on conformity.
Additional analyses
To elucidate further how gender-role orientation moderates individuals’ reactions to
masculine and feminine language styles, bivariate correlations between gender inferences (higher scores indicating greater masculinity than femininity attribution) and
perceived competence were calculated for gender-typed and nongender-typed individuals, separately. Results showed that gender inferences were significantly correlated with perceived competence only for gender-typed individuals (r = .33, p = .002
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
527
Gender Stereotyping in CMC
E.-J. Lee
vs. r = 2.01, p = .94 for nongender-typed individuals, Z = 2.41, p = .008). Likewise,
the bivariate correlation between gender inferences and conformity was significant
for gender-typed individuals (r = .21, p = .05) but close to zero for nongender-typed
individuals (r = 2.06, p = .55), Z = 1.86, p = .06. Moreover, the correlation between
gender inferences and conformity for gender-typed individuals became nonsignificant when perceived competence was partialled out (r = .10, p = .34). On the one
hand, these findings seem to support expectation states theory (Berger et al., 1977);
that is, it is through differential performance expectations that people become more
receptive to male than female influence agents’ opinions. At the same time, the
results also suggest that there are significant individual differences in the extent to
which gender evokes differential performance expectations, casting doubt on the
general notion that gender serves as a diffuse status characteristic that invites differential competence attributions (Ridgeway, 1988).
Because gender-role orientation was measured after the experiment, it was possible that the experimental procedure might have affected self-assessed masculinity
and femininity. A 2 3 2 3 2 (Personalizing Information 3 Gendered Language Style
3 Participant Gender) ANOVA on gender-role orientation, however, yielded no
significant main or interaction effects, all Fs , 2.04, all ps . .15. This suggests that
experimental manipulations did not significantly affect the extent to which participants attributed gender-stereotypical traits to themselves.
Discussion
Consistent with the SIDE model, the present research demonstrated that the lack of
personalizing information, independently or in conjunction with the gender-role
orientation, facilitated gender-typed person perceptions and conformity behavior.
Unlike previous studies, participants were not specifically induced to think of gender
as the identity-defining dimension. Even so, when no personal information about
the partner was available, they became more vigilant to the implicit gender category
cues embedded in short comments and utilized such information to infer their
anonymous partner’s gender identity. Moreover, depersonalization also served as
a precipitating condition for gender-typed individuals to behave in accordance
with gender stereotypes; that is, gender-typed individuals conformed more to the
masculine- than feminine-comment partners, and gender-typed men were more
resistant to social influence than gender-typed women, only when they were deprived
of personal information about their partner.
Interestingly, although depersonalization amplified the effects of linguistic gender cues on gender inferences, it fostered gender stereotyping only among those
whose self-concepts are congruent with traditional gender roles. Such findings seem
to underscore the distinction between the two facets of social stereotyping: categorization (i.e., assignment of a group membership) and characterization (i.e., attribution of stereotypic traits to individual members of the group) of the target (Gilbert,
528
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
E.-J. Lee
Gender Stereotyping in CMC
Pelham, & Krull, 1988).2 That is, although depersonalization increased a person
perceiver’s cognitive sensitivity to category membership cues available in the immediate environment (categorization), it did not automatically trigger stereotypical
beliefs about the target (characterization). Instead, depersonalization prompted people to behave in line with their gender-role self-concepts, such that only those who
view themselves as a prototypical member of their gender group exhibited gender
stereotyping. Collectively, these results suggest that depersonalization not only
affects the level of identity activated (social vs. personal: self-categorization) but
also moderates the impact of that self-definition (accentuation) (Spears, 2001,
pp. 175–176).
The accentuation effect of depersonalization might not be generalizable to different personality traits, though. Unlike other individual difference variables that
define one’s ‘‘personal’’ or ‘‘I’’ self as a unique and distinctive entity, which are likely
to be undermined by depersonalization, gender-role orientation entails perceived
prototypicality of oneself as a member of his or her gender group. In other words, it
expresses one’s ‘‘collective’’ or ‘‘we’’ self (Onorato & Turner, 2001; Spears, 2001). If
so, it is not surprising that gender-role orientation had greater influence when the
lack of personal information substantially obscured interpersonal differences, in
much the same way that an experimentally induced group identity exerts stronger
effects under depersonalization (Postmes et al., 1998, 2002). Future research should
address this possibility and explore when depersonalization might override or
amplify individual differences, depending on whether they represent personal versus
collective self.
The present study also extends the previous work on gendered language use and
its effects in CMC. First, unlike previous research in which participants were explicitly asked to read transcribed messages and guess the author’s gender (e.g., Koch,
Mueller, Kruse, & Zumbach, 2005; Nowak, 2003), participants in this study were not
instructed to pay close attention to linguistic gender indicators. Instead, they were
merely exposed to gender-typed language styles as part of (simulated) synchronous
interaction and their impressions were measured afterward. By lowering experimental demands, such a procedure allowed some interesting findings to emerge; for
example, results suggest that gender-linked language differences are a relatively elusive and underutilized form of gender identity cue. Although the comments were
constructed to represent either prototypically masculine or feminine language styles,
they elicited corresponding gender inferences only among gender-typed individuals
or in the complete absence of personal information. Second, instead of assessing
overall accuracy of gender guess, this study elucidated what factors moderate the
extent to which people take into account gender-linked language differences to
identify the communicator’s gender; that is, in addition to documenting the variability in gender-linked language effects, this study accounted for such variability by
employing gender-role orientation and depersonalization. Last, the effects of gendered language were explored well beyond person perception. In support of expectation states theory and status characteristic theory, participants showed differential
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
529
Gender Stereotyping in CMC
E.-J. Lee
behavioral reactions to masculine than feminine influence agents, albeit only under
certain conditions.
One limitation of the present study concerns limited ecological validity due to
the peculiar format of CMC. Specifically, participants were made to play a trivia
game with a complete stranger, who in fact did not even exist. Although preprogramming the partner’s answers and comments was necessary to systematically vary
the gender cues conveyed through different channels and hold the interaction content constant, such manipulation might reduce the generalizability of the findings.
For example, Palomares (2004) showed that people used gender-linked language
styles in their e-mail only when they were gender schematic and when their conscious
attention was specifically drawn to their gender identity. Although it is still unclear
what particular attributes of CMC might inhibit or facilitate gender-based language
use, if people are unlikely to exhibit gender-typed language styles in CMC, the effects
of gendered language observed in this study might not be replicated in naturally
occurring interactions.
The fact that this study used a one-shot CMC interaction could also have affected
participants’ motivation to relate to their supposed partners (Walther, 1996). As
social information processing theory posits, ‘‘when time is short, communication is
significantly less interpersonally oriented’’ and thus, ‘‘one-time-only, time-limited
CMC groups are bound to appear more task oriented than are parallel FtF groups’’
(Walther, 1996, p. 11). If so, it remains to be seen how the current findings might
apply to ongoing, CMC-based relationships. Quite possibly, people might choose to
reduce uncertainty, for example, by asking personally relevant questions (Tidwell &
Walther, 2002), rather than rely solely upon the subtle language cues and draw overly
crude stereotype-based inferences. At the same time, the finding that the exchange of
personal information significantly discouraged language-based gender categorization and associated stereotyping comports very well with the social information
processing theory: It is the amount of social information, not some technical attributes of the medium per se, that shapes the nature of the relationships formed and
developed within the online context.
Last, it remains unclear through what psychological processes the depersonalizing context facilitated categorical person perception and behavior. Given that depersonalization was operationalized as the relative deficiency of personal information, it
might have fostered language-based gender inferences by removing distractions and
drawing closer attention to the verbal message as the only available cue to the
interactant’s identity. Alternatively, both personalized and depersonalized participants might have heeded the messages equally well, but personalized participants
might have corrected their impression in light of the individuating information they
had received earlier, discounting relatively subtle gender cues embedded in the
language style. To shed light on the exact mediating processes underlying depersonalization effects on person perception and self-categorization, future research will
need to incorporate additional measures, such as message recall (attention) and
reaction time to words associated with their personal versus social identities.
530
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
E.-J. Lee
Gender Stereotyping in CMC
In sum, the present research demonstrated that even though the target person
was not prelabeled (e.g., Asian American, a man, University
C Student), depersonalization prompted category-based person perceptions and behavioral reactions,
especially among those whose self-concepts are congruent with traditional gender
roles. Not only did the findings establish that ‘‘computer-mediated social perceptions are indeed sensitive to social cues and these cues may invoke stereotypes or
a priori expectations’’ (Matheson, 1991, p. 137) but they also highlight that the
sensitivity to social cues varies as a function of situational and dispositional moderators. In fact, more recent studies have begun to show that depersonalization
effects are qualified by certain personality traits. For example, Lee (2006) showed
that when depersonalized, those who are chronically motivated to differentiate
themselves from others in a public setting (i.e., high need for public individuation)
became less resistant to conformity pressure and were equally likely to conform to
the majority as those with lower need for public individuation. Although the
SIDE model was not originally proposed as a theory of individual differences, identifying how dispositional differences interact with the lack of cues in CMC to
shape person perception and group dynamics will surely enrich our understanding
of the impact of this relatively new medium. As McLuhan (1964) claimed, ‘‘the
medium is the message,’’ but the message appears to mean different things to different individuals.
Notes
1
A series of hierarchical multiple regression analyses was conducted to examine if
dichotomization of gender-role orientation substantially altered the results. First, participant gender (0 = male, 1 = female), personalizing information (0 = depersonalization,
1 = personalization), gendered language style (0 = masculine, 1 = feminine), and genderrole orientation (0 = low, 1 = high) were entered in the first block. Then, the two-way
interaction terms were entered in the second block, followed by three-way and then
four-way interaction terms in the third and fourth blocks, respectively. To reduce
potential problems with multicollinearity, gender-role orientation was transformed into
z scores prior to the formation of the interaction terms (Cohen & Cohen, 1983). Overall,
results were consistent with those based on ANOVA, except for the following differences. First, the lack of interaction between gendered language style and gender-role
orientation (b = .13, t = 1.37, p = .17) suggested that the tendency to make languagebased gender inferences does not vary as a linear function of traditional gender-role
orientation (H1a). Possibly, among nongender-typed participants, those holding
counterstereotypical self-concept might be in fact more sensitive to linguistic gender
cues than those with relatively undifferentiated gender-role self-concept. Second, the
three-way interaction among participant gender, personalizing information and genderrole orientation was not statistically significant (b = 2.12, t , 1). Instead, significant
two-way interactions emerged between participant gender and gender-role orientation
(b = .45, t = 3.18, p , .001) (H3) and between participant gender and personalizing
information (b = 2.60, t = 23.61, p , .001) (RQ2), which indicate that gender
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
531
Gender Stereotyping in CMC
2
E.-J. Lee
differences in influenceability are, at least in part, due to situational and dispositional
variables.
In fact, Gilbert et al. (1988) defined these concepts in a slightly different way from the
current usage. That is, categorization in their study denoted the identification of the
actor’s behavior (‘‘What is the actor doing?’’) and characterization referred to the dispositional attribution of the actor’s behavior (‘‘What trait does the action imply?’’)
(p. 733). Replacing the actor’s behavior with the target’s group identity (i.e., ‘‘What is
the person’s group identity?’’ ‘‘What trait does the group identity imply?’’), the conceptual distinction between the two nonetheless seems to offer a valid analogy, especially
in the context where no explicit category membership cues are available, making
categorization a variable.
References
Bem, S. L. (1981a). Bem Sex Role Inventory professional manual. Palo Alto, CA: Consulting
Psychologists Press.
Bem, S. L. (1981b). Gender schema theory: A cognitive account of sex typing. Psychological
Review, 88, 354–364.
Bem, S. L., & Lenney, E. (1976). Sex typing and the avoidance of cross-sex behavior. Journal
of Personality and Social Psychology, 33, 48–54.
Berger, J., Fisek, H., Norman, R., & Zelditch, M. (1977). Status characteristics and social
interaction. New York: Elsevier.
Burgoon, M., & Klingle, R. S. (1998). Gender differences in being influential and/or
influenced: A challenge to prior explanations. In D. J. Canary & K. Dindia (Eds.), Sex
differences and similarities in communication: Critical essays and empirical investigations
of sex and gender in interaction (pp. 257–285). Mahwah, NJ: Erlbaum.
Carli, L. L. (1991). Gender, status, and influence. In E. J. Lawler, B. Markovski,
C. L. Ridgeway, & H. Walker (Eds.), Advances in group processes (pp. 89–113). Greenwich,
CT: JAI.
Carli, L. L. (1999). Gender, interpersonal power, and social influence. Journal of Social Issues,
55, 81–93.
Cohen, J., & Cohen, P. (1983). Applied multiple regression/correlation analysis for the
behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum.
Dubrovsky, V. J., Kiesler, S., & Sethna, B. N. (1991). The equalization phenomenon: Status
effects in computer-mediated and face-to-face decision making groups.
Human-Computer Interaction, 6, 119–146.
Eagly, A. H. (1983). Sex and social influence: A social psychological analysis. American
Psychologist, 38, 971–981.
Eagly, A. H., & Carli, L. L. (1981). Sex of researchers and sex-typed communications as
determinants of sex differences in influenceability: A meta-analysis of social influence
studies. Psychological Bulletin, 90, 1–20.
Eagly, A. H., & Wood, W. (1985). Gender and influenceability: Stereotype versus behavior.
In V. E. O’Leary, R. K. Unger, & B. S. Wallston (Eds.), Women, gender, and social
psychology (pp. 225–256). Hillsdale, NJ: Erlbaum.
Eakins, B. W., & Eakins, R. G. (1978). Sex differences in human communication. Boston:
Houghton Mifflin.
532
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
E.-J. Lee
Gender Stereotyping in CMC
Epley, N., & Kruger, J. (2005). When what you type isn’t what they read: The perseverance of
stereotypes and expectancies over email. Journal of Experimental Social Psychology, 41,
414–422.
Frable, D. E. S., & Bem, S. L. (1985). If you are gender schematic, all members of the opposite
sex look alike. Journal of Personality and Social Psychology, 49, 459–468.
Gilbert, D. T., Pelham, B. W., & Krull, D. S. (1988). On cognitive busyness: When person
perceivers meet persons perceived. Journal of Personality and Social Psychology, 54,
733–740.
Herring, S. C. (1993). Gender and democracy in computer-mediated communication.
Electronic Journal of Communication, 3(2). Retrieved January 4, 2005, from http://
ella.slis.indiana.edu/Eherring/ejc.txt
Herring, S. C., & Martinson, A. (2004). Assessing gender authenticity in computer-mediated
language use: Evidence from an identity game. Journal of Language and Social Psychology,
23, 424–446.
Herschel, R. T., Cooper, T. R., Smith, L. F., & Arrington, L. (1994). Exploring numerical
proportions in a unique context: The group support systems meeting environment.
Sex Roles, 31, 99–123.
Hogg, M. A., & Turner, J. C. (1987). Intergroup behavior, self-stereotyping and the salience
of social categories. British Journal of Social Psychology, 26, 325–340.
Joinson, A. (1998). Causes and implications of disinhibited behavior on the Internet.
In J. Gackenbach (Ed.), Psychology and the Internet: Intrapersonal, interpersonal, and
transpersonal implications (pp. 43–60). San Diego, CA: Academic Press.
Kiesler, S., & Sproull, L. (1992). Group decision making and communication technology.
Organizational Behavior and Human Decision Processes, 52, 96–123.
Koch, S. C., Mueller, B., Kruse, L., & Zumbach, J. (2005). Constructing gender in chat groups.
Sex Roles, 53, 29–41.
Kray, L. J., Thompson, L., & Galinsky, A. (2001). Battle of the sexes: Gender stereotype
confirmation and reactance in negotiations. Journal of Personality and Social Psychology,
80, 942–958.
Lakoff, R. (1975). Language and women’s place. New York: Harper & Row.
Lee, E.-J. (2005). Effects of the influence agent’s sex and self-confidence on informational
influence in computer-mediated communication: Quantitative vs. verbal presentation.
Communication Research, 32, 29–58.
Lee, E.-J. (2006). When and how does depersonalization increase conformity to group norms
in computer-mediated communication? Communication Research, 33, 423–447.
Matheson, K. (1991). Social cues in computer-mediated negotiations: Gender makes
a difference. Computers in Human Behavior, 7, 137–145.
McKenzie-Mohr, D., & Zanna, M. P. (1990). Treating women as sexual objects: Look to the
gender schematic male who has viewed pornography. Personality and Social Psychology
Bulletin, 16, 296–308.
McLuhan, M. (1964). Understanding media. The extensions of man. London: Routledge.
Meeker, B. F., & Weitzel-O’Neill (1977). Sex roles and interpersonal behavior in task-oriented
groups. American Sociological Review, 42, 91–105.
Mulac, A. (1998). The gender-linked language effect: Do language differences really make
a difference? In Canary, D. J. & Dindia, K. (Eds.), Gender differences and similarities in
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
533
Gender Stereotyping in CMC
E.-J. Lee
communication: Critical essays and empirical investigations of gender and gender in
interaction (pp. 127–153). Mahwah, NJ: Lawrence Erlbaum Associates.
Mulac, A., Incantro, C. R., & James, M. R. (1985). Comparison of the gender-linked
language effect and sex role stereotypes. Journal of Personality and Social Psychology,
49, 1099–1110.
Mulac, A., Jansma, L. L., & Linz, D. G. (2002). Men’s behavior toward women after viewing
sexually-explicit films: Degradation makes a difference. Communication Monographs,
69, 311–328.
Mulac, A., & Lundell, T. L. (1980). Differences in perceptions created by syntactic-semantic
productions of male and female speakers. Communication Monographs, 47, 111–118.
Nowak, K. L. (2003). Gender categorization in computer-mediated communication:
Exploring the utopian promise. Media Psychology, 5, 83–103.
Onorato, R. S., & Turner, J. C. (2001). The ‘‘I,’’ the ‘‘me,’’ and the ‘‘us’’: The psychological
group and self-concept maintenance and change. In C. Sedikides & M. B. Brewer (Eds.),
Individual self, relational self, collective self (pp. 147–170). Philadelphia: Psychology Press.
Orlofsky, J. L., Aslin, A. L., & Ginsburg, S. D. (1977). Differential effectiveness of two
classification procedures on the Bem sex role inventory. Journal of Personality Assessment,
41, 414–416.
Page, M. C., Braver, S. L., & MacKinnon, D. P. (2003). Levine’s guide to SPSS for Analysis of
Variance. Mahwah, NJ: Erlbaum.
Palomares, N. A. (2004). Gender schematicity, gender identity salience, and gender-linked
language use. Human Communication Research, 30, 556–588.
Postmes, T., & Spears, R. (2002). Behavior online: Does anonymous computer
communication reduce gender inequality? Personality and Social Psychology Bulletin,
28, 1073–1083.
Postmes, T., Spears, R., & Lea, M. (1998). Breaching or building social boundaries?
SIDE-effects of computer-mediated communication. Communication Research, 25,
689–715.
Postmes, T., Spears, R., & Lea, M. (2002). Intergroup differentiation in computer-mediated
communication: Effects of depersonalization. Group Dynamics, 6, 3–16.
Ridgeway, C. (1988). Gender differences in task groups: A status and legitimacy account.
In M. Welster Jr. & M. Foschi (Eds.), Status generalization: New theory and research
(pp. 188–206). Stanford, CA: Stanford University Press.
Savicki, V., Kelley, M., & Oesterreich, E. (1998). Effects of instructions on computer-mediated
communication in single- or mixed-gender small task groups. Computers in Human
Behavior, 14, 163–180.
Savicki, V., Kelley, M., & Oesterreich, E. (1999). Judgments of gender in computer-mediated
communication. Computers in Human Behavior, 15, 185–194.
Short, J., Williams, E., & Christie, B. (1976). The social psychology of telecommunications.
London: John Wiley.
Spears, R. (2001). The interaction between the individual and the collective self.
In C. Sedikides & M. B. Brewer (Eds.), Individual self, relational self, collective self
(pp. 171–198). Philadelphia: Psychology Press.
Spears, R., & Lea, M. (1992). Social influence and the influence of the ‘‘social’’ in
computer-mediated communication. In M. Lea (Ed.), Context of computer-mediated
communication (pp. 31–65). New York: Harvester Wheatsheaf.
534
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
E.-J. Lee
Gender Stereotyping in CMC
Spears, R., & Lea, M. (1994). Panacea or panopticon? The hidden power in
computer-mediated communication. Communication Research, 21, 427–259.
Spears, R., Postmes, T., Lea, M., & Wolbert, A. (2002). When are net effects gross products?
The power of influence and the influence of power in computer-mediated
communication. Journal of Social Issues, 58, 91–107.
Sproull, L., & Kiesler, S. (1991). Connections: New ways of working in the networked
organization. Cambridge, MA: MIT Press.
Tanis, M., & Postmes, T. (2003). Social cues and impression formation in CMC. Journal of
Communication, 53, 676–693.
Tidwell, L. C., & Walther, J. B. (2002). Computer-mediated communication effects on
disclosure, impressions, and interpersonal evaluations: Getting to know one another
a bit at a time. Human Communication Research, 28, 317–348.
Turner, J. C. (1987). A self-categorization theory. In J. C. Turner, M. A. Hogg, P. J. Oakes,
S. D. Reicher, & M. S. Wetherell (Eds.), Rediscovering the social group: A self-categorization
theory (pp. 42–67). Oxford, England: Basil Blackwell.
Wagner, D. G., & Berger, J. (1997). Gender and interpersonal task behaviors: Status
expectation accounts. Sociological Perspectives, 40, 1–32.
Walther, J. B. (1996). Computer-mediated communication: Impersonal, interpersonal, and
hyperpersonal interaction. Communication Research, 23, 3–43.
Walther, J., & Parks, M. B. (2002). Cues filtered out, cues filtered in: Computer-mediated
communication and relationships. In M. L. Knapp & J. A. Daly (Eds.), Handbook of
interpersonal communication (pp. 529–563). Thousand Oaks, CA: Sage.
Wood, W., & Karten, S. J. (1986). Gender differences in interaction style as a product of
perceived gender differences in competence. Journal of Personality and Social Psychology,
50, 341–347.
Human Communication Research 33 (2007) 515–535 ª 2007 International Communication Association
535
Les effets du langage genré sur les stéréotypes de genre dans la
communication assistée par ordinateur : Le rôle modérateur de la
dépersonnalisation et des orientations de rôles sexuels
Eun-Ju Lee
Department of Communication One Shields Avenue
University of California-Davis
Résumé
Cette étude a cherché à connaître les caractéristiques circonstancielles et de disposition qui
modèrent les effets sur les stéréotypes de genre des signaux linguistiques genrés dans une
communication assistée par ordinateur, anonyme et textuelle. Les participants ont joué à un jeuquestionnaire général avec un partenaire visible par ordinateur, dont les commentaires
représentaient un style de langage prototypiquement masculin ou féminin. Conformément au
Social Identity model of Deindividuation Effects (SIDE, modèle d’identité sociale des effets de
désindividualisation), ceux qui n’ont pas échangé de brefs profils personnels avec leur partenaire
(c.-à-d. dépersonnalisation) étaient plus susceptibles d’inférer le genre de leur partenaire à partir
des signaux linguistiques que ceux qui en ont échangé. Toutefois, la dépersonnalisation n’a
contribué à des comportements de conformité fidèles aux stéréotypes que chez les personnes
catégorisées selon le sexe (gender-typed); c’est-à-dire que les participants se conformaient aux
partenaires aux commentaires masculins plus qu’aux partenaires aux commentaires féminins et
que les hommes se conformaient moins que les femmes, seulement lorsqu’ils étaient à la fois
catégorisés selon le sexe et dépersonnalisés.
Die Wirkung von geschlechtsspezifischer Sprache auf
Geschlechtstereotypisierung in computervermittelter Kommunikation: Die
moderierende Rolle von Entpersonalisierung und GeschlechtsrollenOrientierung
Dieses Experiment untersucht, welche situationalen und dispositionellen Eigenschaften die
Wirkung von linguistischen Geschlechtshinweisen auf Geschlechtsstereotypisierung in
anonymer,
text-basierter,
computervermittelter
Kommunikation
moderieren.
Die
Studienteilnehmer spielten per Computer ein Spiel mit einem fiktiven Partner, dessen
Kommentare entweder prototypisch maskuline oder feminine Sprachstile repräsentierten. Im
Einklang mit dem Social Identity Model of Deindividuation Effects (SIDE) war es bei
denjenigen, die keine kurzen Persönlichkeitsprofile mit ihrem Partner austauschten
(Entpersonalisierung) wahrscheinlicher, dass sie das Geschlecht des Partners auf Basis der
Sprachhinweise erschlossen, als bei denjenigen, die Profile austauschten. Entpersonalisierung
förderte stereotyp-konsistentes Konformitätsverhalten jedoch nur unter geschlechtstypischen
Individuen; das heißt, Teilnehmer passten sich mehr dem männlichen im Vergleich zum
weiblichen Kommentarpartner an. Männer passten sich generell weniger an als Frauen, und
wenn, dann nur, wenn sie sowohl geschlechtstypisiert und entpersonalisiert waren.
Los Efectos del Lenguaje de Género sobre el Estereotipo de Género en
la Comunicación Mediada por la Computadora: El Rol Moderador de
la Despersonalización y la Orientación de Rol de Género
Eun-Ju Lee
Department of Communication
University of California-Davis
Resumen
Este experimento examinó qué características situacionales y de disponibilidad moderan los
efectos de indicadores lingüísticas de género sobre el estereotipo de género en las
comunicaciones anónimas, y de texto mediadas por la computadora. Los participantes jugaron un
juego trivial con un compañero aparente por medio de la computadora, cuyos comentarios
representaban estilos de lenguaje prototípico masculino ó femenino. Consistente con el modelo de
identidad social de efectos de supresión del individuo, aquellos que no intercambiaron perfiles
personales breves con sus compañeros (a saber, despersonalización) infirieron con más
probabilidad el género de sus compañeros basados en las claves del lenguaje, que aquellos que lo
hicieron.
La
despersonalización,
no
obstante,
facilitó
comportamientos
conformistas
estereotípicos consistentes solo entre los individuos de un género-tipo; esto es, los participantes se
ajustaron más a los comentarios masculinos- que a los femeninos de sus compañeros, y los
hombres se ajustaron menos que las mujeres, solo cuando ellos eran del mismo género y
despersonalizados.
电脑中介传播中性别语言对性别形象化的作用:
非个体化和性别角色取向的中和作用
Eun-Ju Lee
加州大学戴维斯分校
[摘要]
本实验检测在匿名和以文本为基础的电脑介面传播中情景和个
性特点如何左右语言性别暗示对性别形象化的作用。参与者通过电脑
与一较真实的伙伴参加一个无关重要的游戏,该伙伴的评论代表了典
型的男性和女性语言风格。与社会身份非个性化效果模式相一致,那
些不和同伴介绍个人资料的参与者更有可能以他们同伴的语言暗示
推断对方的性别。然而,非个性化只帮助那些有性别分辨倾向的参与
者进行与形象化相一致的行为,也就是说,参与者更加趋附男性或女
性评语的对方。女性比男性更趋附,但这种现象只在有性别分辨和非
个性化的情况下出现。
컴퓨터 매개 커뮤니케이션에서의 젠더 전형화에 있어 젠더화된 언어의 효과에 대한
연구: 탈개성화의 중재적 역할과 젠더-역할 지향성
Eun-Ju Lee
Department of Communication
One Shields Avenue
University of California-Davis
Davis, CA 95616
Email: [email protected]
Fax: 530-752-6705
Phone: 530-754-8624
요약
본 실험은 상황적 그리고 의향적 특성들이 익명의 텍스트에 근거한 컴퓨터 매개
커뮤니케이션내 젠더 정형화에 있어 언어적 젠더 단서들의 효과를 어떻게 중재하는가를
연구한 것이다. 참여자들은 컴퓨터를 통하여 그들의 말들이 전형적인 남성 또는 여성
형태를 대표하는 형식적인 파트너들과 트리비아 게임을 하였다. 탈개인화 효과의
사회적 동질화 모델과 일치하게, 그들의 파트너들과 짧은 개인적 프로파일을 교환하지
않은 사람들은 (탈개인화) 그들 파트너들의 젠더를 개인적 프로파일을 교환한
사람들보다 언어 단서들에 의해 추론하려는 경향을 보였다. 탈개인화는 그러나 오직
젠더 형태 개인들 사이에서만 전형적으로 일치하는 행위들을 촉진하였다.
즉,
참여자들은 여성적 언어를 사용하는 파트너들보다 남성적 언어를 사용하는 파트너들을
더욱 잘 확인하는 형태를 보였으며, 남성들은 여성들보다 덜 확인적이었는데, 오로지
그들이 젠더 형태와 탈개인화일때였다.