INTERPERSONAL RELATIONS AND GROUP PROCESSES Gender, Language, and Influence

College of the Holy Cross

Mixed- and same-sex dyads were observed to examine effects of gender composition on language and of language on gender differences in influence. Ss discussed a topic on which they disagreed. Women were more tentative than men, but only in mixed-sex dyads. Women who spoke tentatively were more influential with men and less influential with women. Language had no effect on how influential men were. In a second study, 120 Ss listened to an audiotape of identical persuasive messages presented either by a man or a woman, half of whom spoke tentatively. Female speakers who spoke tentatively were more influential with male Ss and less influential with female Ss than those who spoke assertively. Male speakers were equally influential in each condition.

Linda L. Carli

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

the absence of specific information, to assess his or her competence, ability, or value (Berger & Fisek, 1974; Berger, Rosenholtz, & Zelditch, 1980; Eagly, 1983). People with relatively high status are expected to be more competent, to perform better, and to have more desirable attributes than low status individuals; they are also given more opportunities to perform well and are, consequently, more influential (Berger et al, 1977, 1980). In addition, in interactions among people who differ in relative status, it is considered illegitimate for the individual possessing lower status to behave too assertively, as such behavior could be construed as an attempt to gain status at the expense of other members of the group (Meeker & Gender and Status Weitzel-O'Neill, 1977). Therefore, low status individuals who According to expectation states theory, inequalities in face-to- behave assertively risk the rejection of others (Berger et al, 1980; Meeker & Weitzel-O'Neill, 1977). face interactions are a function of the relative status of participants (Berger, Fisek, Norman, & Zelditch, 1977). In this model, Women generally have lower status than men, as is evidenced status is culture specific and situation dependent. That is, indiby thefindingsthat stereotypical feminine traits are evaluated viduals may possess characteristics that reflect relatively low less favorably than stereotypical masculine traits (Broverman, status in one culture or situation but reflect high status or conVogel, Broverman, Clarkson, & Rosenkrantz, 1972) and that vey no status information in another culture or situation. In women are considered to be less competent than men (LockAmerican culture, race, class, education, age, occupation, physi- heed & Hall, 1976; Meeker & Weitzel-O'Neill, 1977). In interaccal attractiveness, and gender can act as diffuse status character- tions between men and women, in the absence of specific inforistics, characteristics of a person that are used, particularly in mation about group members' ability or competence, women would possess relatively lower status than men. However, in same-sex interactions, gender would not act as a diffuse status characteristic because it would convey no information about Preliminary analyses were presented at the 97th Annual Convention of the American Psychological Association, August 1989, and at the the relative ability, competence, or value of different members 2nd Annual Convention of the American Psychological Society, June of the group. This suggests that in mixed-sex but not same-sex 1990. groups, women would be given fewer opportunities to make I thank Michael Dorsey and three anonymous reviewers for their task contributions, would receive less support for their contrihelpful comments on previous drafts of this article. I also thank Mibutions, and would be less influential than men.
chael Carmen, Jane Kravitz, James Levy, Mary Ellen Mackesy, and Debbie Rearick for their assistance in collecting, coding, and analyzing the data and for their thoughtful contributions to discussions about the research. Correspondence concerning this article should be addressed to Linda L. Carli, Department of Psychology, College of the Holy Cross, Worcester, Massachusetts 01610.

Researchers have reported a wide variety of language differences between the sexes (e.g, Dabbs & Ruback, 1984; Haas, 1979; Kimble, Yoshikawa, & Zehr, 1981; Mulac, Lundell, & Bradac, 1986; Steckler & Rosenthal, 1985), as well as gender differences in influenceability (Eagly & Carli, 1981). This article focuses on a number of the language differences that have been linked to gender differences in status and power. In particular, it examines the effect of sex composition of dyads on gender differences in language and, in turn, examines how these differences affect social influence.

Language, Status, and Gender Lakoff(1975) proposed that assertive speech is one domain of power denied to women, but available to men. That is, because women are relatively powerless and marginal compared

Journal of Personality and Social Psychology 1990, Vol. 39, No. 3,941-931 Copyright 1990 by the American Psychological Association, Inc. 0022-3514/90/$00.75

941

942

LINDA L. CARLI

with men, they presumably are not given the opportunity to express themselves as forcefully and directly as men are. Lakoff (1975) argued that women's less powerful speech would be manifested in their tendency to swear less, speak more politely, and use more tag questions, intensifiers, and hedges. Tag questions refer to declarative statements that are followed by a question concerning the statement (Lakoff, 1975), for example, "Teenagers have more car accidents than older people, don't they?" Statements are presumably less confident and assertive when tag questions are added to them (Lakoff, 1975). Intensifiers are adverbs, such as the word so in the sentence, "Drinking and driving is so dangerous," that are used to provide emphasis but are considered by some researchers (Key, 1972; Lakoff, 1975) to be less powerful than more absolute superlatives. Hedges are adverbs or adverb phrases, such as sort of, perhaps, and maybe, that weaken the strength of a statement (Hewitt & Stokes, 1975; Lakoff, 1975). Other researchers have also proposed links between status and language, arguing that the lower status of women causes them to interrupt others less than men do (Thorne & Henley, 1975); to qualify and weaken their statements with disclaimers such as / may be wrong but, I mean, and / don't know but (Pearson, 1985); and to verbally reinforce the speech of others more than men do by interjecting yeah, mm-hmm, and right while others are speaking (Thorne & Henley, 1975). Studies testing these hypotheses have yielded mixed results, but overall provide support that gender differences in language do exist. Although there is no documented evidence that men swear more than women, and one study found no gender difference in the reported use of expletives (Staley, 1978), women are more polite (Hartman, 1976), less likely to interrupt (Argyle, Lalljee, & Cook, 1968; Eakins & Eakins, 1976; McCarrick, Manderscheid, & Silbergeld, 1981; Mulac, Wiemann, Widenmann, & Gibson, 1988; Natale, Entin, & Jaffe, 1979; Octigan & Niederman, 1979; West & Zimmerman, 1983; Willis & Williams, 1976; Zimmerman & West, 1975), and less successful at gaining the floor after interrupting (Zimmerman & West, 1975) than are men. In addition, some evidence indicates that women are also more likely than men to hedge (Crosby & Nyquist, 1977, Studies 1 and 3; Fishman, 1978, 1980; Mulac & Lundell, 1986; Mulac et al, 1986), ask tag questions (Crosby & Nyquist, 1977, Studies 1 and 3; McMillan, Clifton, McGrath, & Gale, 1977; Zimmerman & West, 1975), verbally reinforce others (Fishman, 1978; Hirschman, 1974), and use disclaimers (Hartman, 1976; Hirschman, 1973) and intensifiers (Key, 1972; Mulac & Lundell, 1986; Mulac et al, 1986, 1988). Other researchers have found no gender differences in language (Bauman, 1976; Beattie, 1981; Crosby, Jose, & Wong-McCarthy, 1981; Crosby & Nyquist, 1977, Study 2; Dubois & Crouch, 1975; Moore, Shaffer, Goodsell, & Baringoldz, 1983; Roger & Nesshoever, 1987). There are several possible explanations for the inconsistency in findings. First, it is possible that the gender differences are real, but small, because small differences can be expected to sometimes lead to null results (Eagly, 1983).' Another possibility is that gender differences occur primarily when men and women are together and are less likely to occur in same-sex interactions. In mixed-sex interactions, gender can act as a diffuse status characteristic. Consequently,

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

dominant or assertive behavior among women would be least appropriate when they are interacting with men (Berger et al, 1980; Meeker & Weitzel-O'Neill, 1977). It is likely that subjects in past research on language have used gender to infer status, because in much of this research subjects have been strangers who had little specific information about one another. Diffuse status characteristics are more likely to be used under such conditions. If gender differences in language are related to status differences between the sexes, then status characteristics other than gender should also affect language. In fact, people,regardlessof gender, may use tentative language when interacting with someone possessing higher status or power, but not when among equals. There is evidence that this is the case. For example, among romantic couples (Courtright, Millar, & Rogers-Millar, 1979; Kollock, Blumstein, & Schwartz, 1985), parents and children (West & Zimmerman, 1977), and strangers (O'Barr, 1982; Roger & Nesshoever, 1987; Rogers & Jones, 1975), the more powerful or dominant person of either sex is more likely to interrupt and to be more successful at it, whereas the less powerful person tends to use more tag questions (Kollock et al, 1985). In addition, high status or high dominance men and women display a greater amount of verbal and nonverbal power, as measured by the amount that subjects look at their partners while speaking and look away while listening (Dovidio, Ellyson, Keating, Heltman, & Brown, 1988), and in mixed-sex discussions of gender neutral topics, men display a greater amount of verbal and nonverbal power, except when the topic discussed was sex typed; in that condition the gender that possessed greater knowledge of it exhibited more verbal and nonverbal dominance (Dovidio, Brown, Heltman, Ellyson, & Keating, 1988). A third possible explanation for the inconsistency in findings is that not all of the reported gender differences in language may reflect the greater tentativeness of women. In fact, two of the gender differences, the use of intensifiers and verbal reinforcers, appear to be less a reflection of women's greater tentativeness than oftheir greater emotional expressiveness and sociability.2 Verbalreinforcersserve to encourage others to con1 In some instances, unpredicted or nonsignificant gender differences in language may be a function of the amount that subjects spoke. For example, Bilous and Krauss (1988) reported a greater number of interruptions among pairs of women than among mixed-sex pairs or pairs of men. However, they also reported that female pairs talked more than any other type of dyad. Clearly, the absolute number of interruptions (as well as tag questions, disclaimers, hedges, and other types of speech) should increase as the amount of speaking increases. Unfortunately, a number of researchers reporting unpredicted or nonsignificant results failed to report a test for the amount of speech (see Dubois & Crouch, 1975; Lapadat & Seesahai, 1978). 2 Some researchers have suggested that tag questions can reflect a social-emotional orientation rather than tentativeness, because, for example, tag questions can be used to encourage others to speak (Fishman, 1980; Johnson, 1980). Tag questions used in this way function much more as questions than as statements and are sometimes coded as a type of question (Johnson, 1980). If questions do serve a socialemotional function, then gender differences in asking questions should be reported in studies of interaction processes, particularly among subjects interacting in same-sex groups. However, gender dif-

GENDER, LANGUAGE, AND INFLUENCE

943

tinue speaking. As such, their function may be similar to that of positive social behaviors and agreements in task-oriented groupscreating a more social or group-oriented interaction (Meeker & Weitzel-O'Neill, 1977). Although intensifies, such as vastly, so, and awfully, have been considered tentative, they are probably used by speakers to provide emphasis rather than to express uncertainty (McMillan et al, 1977). It is possible, then, that gender differences in the use of intensifies and verbal reinforcers may not be due to the status difference between men and women, but rather to a gender difference in the orientation that men and women have toward others. A number of researchers have noted that, in general, women tend to exhibit a social-emotional or relational orientation in interactions with others, whereas men tend to exhibit a more independent and unemotional orientation (Chodorow, 1978; Dinnerstein, 1977; Eagly, 1987; Gilligan, 1982; Miller, 1976). This difference is often attributed to stable personality characteristics acquired through the different socialization of men and women. However, if this gender difference is intrinsic to male and female personalities, women should consistently exhibit a greater social-emotional orientation toward both men and women across a wide variety of situations. However, research has revealed that this is not the case; this gender difference occurs primarily for same-sex interactions. That is, women tend to be particularly social and emotional in interactions with other women, and men the least so in interactions with other men. For example, the friendships of women emphasize intimacy and emotional expressiveness, the friendships of men emphasize shared activity (Aries & Johnson, 1983; Aukett, Ritchie, & Mill, 1988; Barth & Kinder, 1988; Bell, 1981; Rubin, 1985), and cross-sex friendships tend to be less sex stereotyped than same-sex friendships (Buhrke & Fuqua, 1987). Women exhibit more positive social-emotional behavior than men in same-sex groups; in mixed-sex groups, both men and women behave more like members of the opposite sex (Carli, 1989; Piliavin & Martin, 1978). In a study examining vocal cues in the conversations of same- and mixed-sex dyads, interactions were rated as most pleasant and least businesslike among women, least pleasant and most businesslike among men, and between these two extremes in mixed-sex interactions (Hall & Braunwald, 1981). This same pattern offindingshas been also revealed in reviews of research on nonverbal behaviors, such as smiling and touching (Hall, 1984), and cooperativeness (Carli, 1982). It appears, then, that different norms do operate in same-sex interactions than in mixed-sex interactions, and that these norms lead to a pattern of behavior that is most social in groups of women and least social in groups of men. A reasonable conclusion based on the literature is that the
ferences in asking questions have not been found in studies of interaction style in same- or mixed-sex groups (Carli, 1989; Piliavin & Martin, 1978), and asking questions has not been considered a social-emotional behavior in Bales's (19S0) model of group interaction. People may ask questions for many different reasons, for example, to obtain information, to test the knowledge of others, to assist others in formulating their ideas, or even to express disagreement (see Carli, 1989). For the purpose of this study, questions added to the end of statements were considered tag questions when the statements that they followed were known to reflect the subject's opinion.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

gender difference in social-emotional orientation is a function of expectancies and behavior norms that depend, in part, on the gender composition of the group in which subjects interact, and not on gender differences in personality. The sex composition of groups may affect the salience of gender as a social category; this, in turn, may trigger gender-linked schemas leading to different gender-related expectancies and behaviors (Deaux & Major, 1987). It is likely that different schemas and expectancies are associated with same-sex interactions than with mixed-sex interactions. Because gender belief systems (Deaux & Kite, 1987), which are the set of beliefs that people hold about the characteristics and behaviors of men and women, typically include the stereotype that women are more social and expressive than men (Broverman et al, 1972; Williams & Best, 1982), subjects may expect a high amount of social-emotional behavior in interactions among women and very little in interactions among men; these expectancies may be self-fulfilling. Subjects may also enter mixed-sex interactions expecting sex-typed behavior from those of the opposite sex. Because norms governing their behavior may be less clear than in same-sex interactions, individuals may, as a result, exhibit behavior that is more similar to that which they expect from the opposite sex. Gender-related expectancies may be clearer for same-sex groups because sex segregation is common in American culture. From preschool on, children interact primarily in samesex groups (Lockheed & Klein, 1985; Maccobx 1988; Maccoby & Jacklin, 1987). Although this pattern weakens somewhat during adolescence and adulthood, sex segregation during these developmental stages is characteristic of friendships (Verbrugge, 1977), occupations (Reskin, 1984; Reskin & Hartmann, 1986; Roos, 1985), and activities (Berk, 1985; Hartmann, 1981). If the use of intensifies and verbal reinforcers does reflect a social-emotional orientation to a group interaction, then gender differences in their use should occur in same-sex interactions, but to a lesser extent, ifat all, in mixed-sex interactions. On the other hand, if the gender difference in the use of tentative versus assertive language is a function of the status difference between the sexes, then gender differences in the use of interruptions, tag questions, hedges, and disclaimers should occur in mixed-sex interactions but not in same-sex interactions. Few of the previous studies that have obtained gender differences have included comparisons of the gender differences in mixed-sex interactions with those in same-sex interactions. However, those that have, have reported that the gender difference in interruptions (McMillan et al, 1977; Octigan & Niederman, 1979; Zimmerman & West, 1975) was larger in mixed-sex interactions and the gender difference in use of verbal reinforcers (Hirschman, 1974) was marginally larger in same-sex interactions, providing some support for the argument developed above.3
One study (Crosby & Nyquist, 1977) found no effect of sex composition on language. However, in this studyresearcherscombined politeness, tag questions, and hedges to create an index of tentative language rather than testing each behavior separately. A second (Mulac, Wiemann, Widenmann, & Gibson, 1988), found larger gender differences
3

944

LINDA L. CARLI

Hypothesis 1. Gender differences in the use of tag questions, hedges, and disclaimers, with a higher amount of each behavior exhibited by women, will be larger in mixed-sex than same-sex groups. Hypothesis 2. Gender differences in interruptions, with a higher amount of interruptions and successful interruptions exhibited by men, will be larger in mixed-sex than same-sex groups. Hypothesis 3. Gender differences in the use of intensifies and verbal reinforcers, with a higher amount of each behavior exhibited by women, will be larger in same-sex than mixed-sex groups.
This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Gender, Language, and Influence According to Lakoff (1975), the use of uncertain or tentative language limits women's ability to express themselves and puts them at a disadvantage when interacting with others. In fact, individuals who speak tentatively are evaluated less favorably than those who speak assertively (Wiley & Eskilson, 1985) and are considered less credible and attractive (Erickson, Lind, Johnson, & O'Barr, 1978). Moreover, women who speak tentatively are considered less intelligent and knowledgeable than men who speak tentatively (Bradley, 1981), so the use of tentative speech would appear to interfere with a woman's ability to influence others more than a man's. However, it is possible that tentative language may also be functional for women, particularly when they are interacting with men. In interactions with men, women are not only expected to be less competent, but they are also expected to show relatively little competitiveness or dominance (Meeker & Weitzel-O'Neill, 1977). Exhibiting competitive or dominant behavior can be construed as an attempt to gain status or influence, and such attempts are considered inappropriate in people who are low in external status, regardless of their level of competence (Meeker & Weitzel-O'Neill, 1977). Such people must demonstrate an altruistic desire to help the group before their contributions will be accepted by high status members; moreover, for low status individuals, displaying behavior that is group oriented and not self-enhancing may be even more important than appearing competent (Hollander & Julian, 1970; Meeker & Weitzel-O'Neill, 1977). Because women may find it difficult to influence men if they behave too assertively (Lockheed & Hall, 1976; Meeker & Weitzel-O'Neill, 1977), they may instead have to rely on more subtle and less direct strategies to induce influence (Johnson, 1976). One subtle approach to influence may be the use of uncertain or tentative language. Women, and people who possess little power, do report using more indirect influence strategies than men and more powerful individuals (Cowan, Drinkard, & MacGavin, 1984; Falbo & Peplau, 1980; Gruber & White, 1986; Howard, Blumstein, & Schwartz, 1986; Offermann & Kearney, 1988; Raven, Centers, & Rodrigues, 1975). For example, women report that they rely on being likable or pleasant (Falbo & Peplau, 1980; Offermann & Kearney, 1988; Raven et al, 1975) and cry or hint to get their

way (Falbo & Peplau, 1980; Howard et al, 1986). Although there is no evidence that an indirect strategy is more^ effective for women than being direct, women interacting with men would probably be more influential by speaking tentatively, even though they would probably be perceived as less competent. However, there may be little benefit for women to speak tentatively to other women. In same-sex interactions, gender does not act as a diffuse status characteristic (Berger & Fisek, 1974), so there would be no need for a woman to behave deferentially under such conditions. Moreover, women are generally assumed to be less competent than men (Lockheed & Hall, 1976; Meeker & Weitzel-O'Neill, 1977), and their use of tentative language further reduces their perceived competence (Bradley, 1981). Consequently, a woman using tentative speech with another woman may not be seen as competent enough to be persuasive and may actually be less influential than a woman speaking assertively. Hypothesis 4. Women will be more influential with men when speaking tentatively than when speaking assertively, and more influential with women when speaking assertively.

Study 1 Method
Pretest. A pretest questionnaire was administered to 229 undergraduates in introductory psychology classes at a state university. They were contacted at the beginning of the semester and asked to complete a questionnaire on which they would indicate their opinion on a variety of topics. Subjects were informed that completing the questionnaire would provide them with an opportunity to participate in a study later in the semester. The questionnaire included 27 topics, 2 of which have been found to be sex neutral in previous research (Carli, 1989): "The drinking age should be lowered to 18 in this state (Massachusetts)" and "The federal government should provide free day care for working parents." Subjects indicated their agreement with each item on a scale ranging from completely disagree (1) to completely agree (10), their interest in each item on a scale ranging from no interest (1) to extremely high interest (10), and their knowledge of each item on a scale ranging from no knowledge (1) to extremely high knowledge (10). The purpose of the questionnaire was to identify sex-neutral topics, those for which there are no sex differences in interest, knowledge, or opinion. The use of sex-biased topics could lead to artifactual gender differences; the use of topics that favor one sex results in greater verbal assertiveness and power by members of that sex (Dovidio et al, 1988; Kelly, Wildman, & Urey, 1982) and greater influenceability among those of the opposite sex (Sistrunk & McDavid, 1971). The drinking age and day-care topics were again found to be sex neutral and, in addition, opinions on these topics were found to be quite variable, ensuring that subjects could be paired with partners with whom they disagreed. Subjects. Respondents expressing relatively neutral opinions, reporting scores of 5 or 6 on the opinion scale, were eliminated from the sample of 229. The subjects, 59 men and 59 women, were selected at random from the remaining pretest subjects. Procedure. Subjects were recruited by telephone about 6 weeks after completing the pretest and asked to schedule an appointment to participate in the study. They were scheduled in pairs, half with samesex partners and half with opposite-sex partners, resulting in 58 subjects in mixed-sex pairs, 30 in female pairs, and 30 in male pairs. Pairings were made randomly, with the exception that partners always

in same-sex groups, but, again, combined all behaviors to create an overall measure of gender differences in language.

GENDER, LANGUAGE, AND INFLUENCE were paired so that they disagreed with one another on both topics. The amount of disagreement varied randomly among the pairs. Because subjects had been assigned partners at random, the difference in opinion between partners was not expected to vary across the three types of sex composition (mixed-sex, male, or female). One-way analyses of variance (ANOVAS) on the difference scores revealed no effect of sex composition for either topic, F < 1. Levene's (1960) test to detect heterogeneity of variance revealed no difference between the groups, F <\. Before the experiment began, partners were informed that they would be discussing a controversial topic with one another and that they would be videotaped during the discussion. Subjects were then given an opportunity to withdraw. None did. A random selection of approximately half of the pairs was then presented with the drinking age topic and half with the day-care topic,4 and these subjects were asked to discuss the topic for 10 min. The experimenter then left and videotaped the discussion from an adjacent room through a one-way mirror. After the discussion, subjects were separated from their partners and asked to indicate their opinion on the topic they had just discussed on a scale ranging from completely disagree (1) to completely agree (10). Subjects were then fully debriefed and excused.

945

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Results
The topic of the discussion had no effect on any of the results and was eliminated from the analyses. Because gender was both a within-group variable for the mixed-sex dyads and a between-group variable for the same-sex dyads, separate ANO VAS were required for these two types of dyads.' For the mixedsex dyads, a 2 X 29 (Gender X Dyad) repeated-measures ANOVA

subject. Pearson correlation coefficients were used to test the reliability of the raters. Because subjects' responses could not be considered independent of their partners', the 59 pairs of subjects were randomly divided into two groups of 59 subjects and the analyses were performed separately for the two groups. The average of the correlation coefficients for the two groups, computed separately for each type of behavior and for the total number of words, ranged from r(57) = .82, p < .001, to r(57) = .96, p < .001. The judgments of the two raters were highly reliable. Effect of gender and sex composition on language. A total measure of subjects' use of tentative language was computed by summing each subject's number of qualifiers, hedges, and tag questions. Analyses were conducted on each of the three behaviors separately, as well as on the total measure and the number of words spoken by each subject. There were no main effects or interactions for the number of words spoken by each subject, F< 1; there were no differences in the amount that men and women spoke in either the mixed- or same-sex dyads. A main effect of sex composition was found for the number of hedges, ?(56) = 2.38, p < .05. There were more hedges in mixed-sex than in same-sex dyads (M = 15.64 vs. 12.15, respectively). A main

was conducted because each dyad contained both a male and a female subject. For the same-sex dyads, a 2 X 15 (Gender X Dyad) ANOVA was conducted, with dyads nested within gender because some dyads were exclusively male and some exclusively female.61 followed a procedure that has been used previously (Carli, 1989) to combine the two data analyses; this involved computing a linear combination of the means as well as a linear combination of the between- and within-groups error terms. The analysis yielded t tests of the following effects: main effect of gender, main effect of sex composition, and the interaction of gender and composition. For example, to test the hypothesis that the size of the gender difference is larger in mixed- than in same-sex dyads, the interaction term, I performed the following contrast: - Mf, +
(l)

in which M t o is the mean for women in mixed-sex groups, A4,m is the mean for men in mixed-sex groups, Mb is the mean for women in same-sex groups, and A4, is the mean for men in same-sex groups. MS^ is the pooled error term, which combines the error term for gender from the repeated-measures analysis with that from the between-groups analysis; n{ is the number of observations on which each mean was based. Coding ofvideotapes. A male and a female rater, unaware of the hypotheses or purpose of the study, analyzed all ofthe videotapes and recorded for each subject the frequency of the following behaviors: disclaimers, hedges, tag questions, interruptions, successful interruptions, intensifiers, and verbal reinforcers.7 A record was also made of the number of arguments presented by each subject and the total number of words spoken by each

Fifteen of the mixed-sex pairs, eight of the female pairs, and eight of the male pairs were assigned the drinking-age topic. 5 The sources of variance and degrees of freedom for the mixed-sex dyads were as follows: dyads on 28 degrees of freedom, gender on 1 degree of freedom, and the interaction of gender with dyads on 28 degrees of freedom. The sources of variance and degrees of freedom for the same-sex dyads were as follows: gender on 1 degree of freedom, dyads nested within gender on 28 degrees of freedom, and subjects nested within dyads on 30 degrees of freedom. 6 A 2 (Sex ofSubject) X 2 (Sex of Partner) analysis of variance (ANOVA) would require arbitrarily assigning one member of each dyad to be the subject and the other to be the partner. The means for each gender would then be based on only one half of the actual subjects in each condition. For example, although there were a total of 30 men in the same-sex dyads, the mean would be based on the 1S men assigned to be subjects. The analysis used in this study allows all subjects to be included in the analysis. 7 Examples of each of the behaviors are given later. Disclaimers were recorded when they immediately preceded a statement or opinion and included I'm no expert, I may be wrong, I'm not sure, I don't know, I suppose, I mean, and / guess. Adverbs or adverb phrases used in the middle of statements were coded as hedges when they conveyed either moderation or no particular meaning at all, as in "Drinking and driving is like dangerous," and included kind of, sort of, you know, maybe, or whatever, and like. Questions such as isn't it?aren't they?don't you think? wouldn't you say? you knew? and right? were coded as tag questions when they were added to the end of statements that were consistent with a subject's original attitude, for example, "It's unfair to prevent 18-year-olds from drinking when they can be drafted and killed in wars, isn't it?" Interruptions were recorded whenever a subject attempted to make a statement while his or her partner was speaking. They were coded as successful when the person doing the interrupting gained the floor immediately after interrupting. Adverbs used in the middle of statements were coded as intensifiers when they were used to convey emphasis or intensity, such as so, very, really, awfully, and truly. Adverbs indicating agreement were coded as verbal reinforcers when subjects used them while or immediately after their partner spoke, for example, right, yeah, yes, mm-hmmm, sure, and uh-huh.

946

LINDA L. CARLI

Table 1 Effect of Sex Composition and Gender on Use of Tentative Language and Interruptions in Mixed-Sex Dyads
Sex
Women Men '(56) Disclaimers 7.55 2.41 4.36** Tag questions 0.83 0.31 2.38* Hedges 18.59 12.69 2.83* Total 26.97 15.41 4.42** Interruptions 6.28 7.79 0.91

Note. The numbers reflect the mean for each behavior. Total refers to the total amount of tentative language. Tests are one-tailed. */><.05. **/><.001. This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

effect of gender was obtained for the number of disclaimers, f(56) = 2.73, p < .05, and the total amount of tentative language, t(56) = 2.79, p < .05. Women used more disclaimers than men (M = 6.25 vs. 4.04) and spoke more tentatively overall (M = 21.99 vs. 16.98). No other main effects were obtained. To test Hypothesis 1, that women's speech would contain more tentative language and that this gender difference would be more pronounced in mixed-sex dyads, I compared the gender differences in mixed-sex pairs with those in same-sex pairs. As predicted, significant one-tailed interactions were obtained for the number of disclaimers, t{56) - 3.49, p < .001; tag questions, *(56) = 1.70, p < .05; hedges, ?(56) = 2.25, p < .05; and for the total amount of tentative language, f(56) = 3.52, p < .001. In each case, the gender difference was larger in mixed-sex dyads. Contrasts revealed that in mixed-sex dyads, women exhibited more of each type of tentative behavior than men (see Table 1). There were no gender differences for the same-sex dyads, p > .25. Analyses conducted on the number of interruptions and the number of successful interruptions revealed no main effect of gender or sex composition and no interactions for either variable. The number of times men interrupted was larger than the number of times women interrupted (Af = 7.10 vs. 5.81), but this difference was not significant, {(56) = 1.48. In addition, as a further test of Hypothesis 2, a contrast testing the predicted gender difference in interruptions in mixed-sex dyads was conducted. As is shown in Table 1, there was no gender difference in interruptions in mixed-sex dyads. Analyses were conducted for the number of intensifiers and the number of verbalreinforcers.No main effects of gender or sex composition were obtained. The third hypothesis, that women would use a greater number of intensifiers and verbal reinforcers and that these gender differences would be more pronounced in same-sex dyads, was then tested by comparing the gender differences in mixed- and same-sex dyads. As predicted, significant one-tailed interactions were obtained for the number of intensifiers, t(56) = 1.75, p < .05, and verbal reinforcers, f(56) = 1.79, p < .05. In both cases, the gender difference was larger for same- than for mixed-sex dyads. One-tailed comparisons revealed that in same-sex dyads, women used more intensifiers (M = 2.83 vs. 1.52), t(56) = 2.21, p < .05, and verbalreinforcers(M = 23.90 vs. 19.76), <(56) = 2.57, p < .05, than men did. No gender differences were found for mixed-sex dyads, p > .25. Analysis ofthe opinion measure. To compute the amount of

attitude change, I subtracted subjects' opinion ratings on the questionnaire from their initial pretest opinions. Attitude change scores in the direction of the partners' opinion were assigned a positive value. Scores ranged from - 2 to 9. Levene's (1960) test of heterogeneity of variance revealed no differences between men and women in variability of influence scores, F < 1. Pearson correlation coefficients were computed to test the predictedrelationshipbetween the use of tentative language by women and their ability to influence male partners. In support of Hypothesis 4, women were more influential with men when speaking tentatively, r(27) = .32, p = .05, one-tailed. However, contrary to prediction, there was no relationship between women's use of tentative speech and their ability to influence other women. In addition, the use of tentative speech had no effect on how influential men were with either men or women.8 Discussion The results suggest that the lower status of women, compared with men, does affect women's speech. When interacting with men, women spoke more tentatively than when interacting with women. Moreover, the use of tentative speech appears to be functional for women. Men were influenced to a greater degree by women who spoke tentatively than by those who spoke assertively. It may be important for a woman not to behave too competitively or assertively when interacting with men in order for her to wield any influence, even if she may risk appearing incompetent. Contrary to prediction, men did not interrupt more than women. Nor were men more successful than women at gaining the floor after interrupting. Previously, gender differences in interruptions have been reported for conversations between people who knew each other (Eakins & Eakins, 1976; Zimmerman & West, 1975) and informal 30-min conversations between strangers (Natale et al, 1979). Perhaps gender differences in interruptions did not emerge in the present study because
It is possible that the relation between the use of tentative language and influence was due to a third variable, the number of arguments presented. Tentative women may have been more influential with men because they presented more arguments to support their position when interacting with men than when interacting with women. An analysis of the number of arguments presented by men and women in mixed- and same-sex dyads revealed no effect of gender or interaction of gender with sex composition, p > .25.
8

GENDER, LANGUAGE, AND INFLUENCE

947

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

subjects did not know each other well enough or have enough time to get to know one another. It is possible that when interactions are more formal, as may have been the case in the present study, the overall amount of interrupting is reduced, creating a floor effect and obscuring any gender difference.9 Finally, support was obtained for the third hypothesis. In same-sex dyads, women were more likely to use intensifiers and verbal reinforcers than men, whereas no gender differences emerged in mixed-sex dyads. Women's greater use of these two forms of speech may reflect their tendency to exhibit more social and emotional behavior when interacting with other women and the tendency of male dyads to emphasize taskoriented behavior (Carli, 1989). These results indicate that stereotypical gender differences in language may occur for reasons unrelated to status. Consequently, a careful examination of other such differences would be needed before concluding that a particular form of speech is less powerful simply because it is favored by women. Although the results of this study demonstrate a relationship between the use of tentative language on the part of women and their ability to influence men, this relationship may or may not be causal. It is possible, for example, that the women who spoke most tentatively to their male partners were also more pleasant or friendly than the more assertive women; their ability to influence their partners may have been a function of their friendliness rather than their tentativeness. A second study was conducted to test whether the relationship between language and influence was, indeed, causal.

men may be assumed to be competent and knowledgeable, regardless of their speech. Hypothesis 1. Women will be more influential with men when speaking tentatively and more influential with women when speaking assertively. Hypothesis 2. Men will be equally influential, whether they speak assertively or tentatively. Method
Pretest. A sample of 34 male and 67 female introductory psychology students indicated their interest, knowledge, and opinion concerning each of 28 topics. Ratings were made on 11-point scales ranging from no interest (1) to extremely high interest (11), no knowledge (1) to extremely high knowledge (11), and completely disagree (1) to completely agree (11), respectively. The purpose of the pretest was to identify topics that revealed no sex differences and for which there was little variability in opinion. The topic selected to be used in the study was "The college bus system should charge a fare each time someone uses a bus." All students in the pretest felt strongly that the bus system should remain free. Subjects. Sixty male and 60 female undergraduates were recruited from psychology classes. Subjects received extra credit for participating in the study. Procedure. A persuasive message of approximately 500 words was written in support of charging a fare. Some of the arguments included in the message were as follows: More buses could be purchased with the money, which would allow buses to run more frequently; the money could be spent on additional training foi drivers, which would increase bus safety; the money could be used to help maintain the buses and prevent breakdowns; students are already paying for the buses through student fees, but not all students who pay student fees actually use the buses; people who are not students can take the buses even though they pay no student fees; and additional routes could be added. This message without added tag questions, hedges, or disclaimers constituted the assertive version of the message. A second, tentative version was created by adding tag questions, hedges, and disclaimers. The language used in this version was constructed to be comparable to that used by highly tentative subjects in Study 1. The two versions were identical in every other respect. Two male and two female confederates rehearsed both versions of the message until they were able to present them in arelaxed,informal manner. Separate audiotapes were made of the four speakers presenting each of the two versions of the message,resultingin a total of eight different speeches. Subjects were randomly assigned to one version of the speech. They were told that the tape had been made of another student who had been asked whether a fare should be charged for the use of the college bus system. The experiment was conducted in a language laboratory, which made it possible for subjects to participate in groups and for different versions of the speech to be presented simultaneously over headphones. After listening to one version of the speech, subjects rated their opinion on the topic on an 11-point scale ranging from complete disagreement (1) to complete agreement (11). They then rated the speaker on 11 -point scales indicating whether he or she was not knowledgeable
(1) or very knowledgeable (11), not interested (1) or very interested (11), high in confidence (1) or low in confidence (11), powerless (1) or powerful

Study 2
In the first study, the use of tentative speech had no effect on how influential women were when interacting with other women. It is possible that in interactions between individuals of relatively equal status, women interacting with women or men interacting with men, language is irrelevant. On the other hand, it is more probable that some other characteristic of the interactions between the women in the present study eliminated the effect of tentative language. For example, influence may have been affected by subjects' interaction style, which has already been shown to affect influence, and which is, in turn, affected by the sex composition of groups (Carli, 1989). Finally, the lack of an effect may have been due to the small number of female pairs and the few degrees of freedom, resulting in too little power to detect the effect. As stated earlier, women in general are considered to be less competent than men, and the use of tentative language is likely to further reduce perceived competence. In interactions with men, this may not be a disadvantage because a woman's tentativeness would then be consistent with her relative status. In interactions with other women, who are status equals, tentativeness may have no particular advantage, but would still probably reduce perceived competence. Consequently, the use of tentative language may reduce a woman's ability to influence another woman. In the first study, the use of tentative language had no effect on how influential men were. Being male may make one a legitimate leader (Meeker & Weitzel-O'Neill, 1977). As a result,

9 Cramer's (1946) test of skewness revealed that the distribution of scores was highly skewed to the right, sk = 5.87. The majority of subjects interrupted approximately 6 or 7 times; a small number of them showed a higher amount of interruptions.

948
Table 2 Measure of Agreement With Speaker

LINDA L. CARLI

p < .05, and marginally for ratings of how competent the speaker was, F(l, 112) = 3.10, p < .10. Contrasts revealed no effect of language on the perceived competence or knowledge Gender of subject of male speakers, p > .25. However, women speaking assertively were judged to be more competent, F(l, 112) = 13.97, p < Speaker Male Female .001 (M= 8.00 vs. 6.20), and more knowledgeable, F(l, 112) = Male 6.65, p < .05 (M = 8.77 vs. 7.13), than those speaking tenta4.07 Tentative language 4.13 tively. No other effects were found for these variables, and no Assertive language 3.80 5.20 effects were obtained for subjects' judgments of how interested Female the speaker was in the topic. Tentative language 5.00 3.13 Gender of Subject X Type of Language interactions were obAssertive language 2.93 5.93 tained for subjects' ratings of how trustworthy, F(l, 112) = 9.71, Note. Higher scores reflect greater agreement with the persuasive mesp < .01, and likable, F(l, 112) = 4.70, p < .05, the speaker was. sage. Female subjects considered the assertive speaker to be more trustworthy, F(l, 112) = 4.97, p < .05 (M = 7.13 vs. 6.10), and more likable, F(l, 112) = 4.56, p < .05 (M= 7.33 vs. 6.17), than (11), low in competence (1) or high in competence (11), trustworthy (1) or the tentative speaker, whereas male subjects considered the asuntrustworthy (11), very likable (1) or not likable (11), low in intelligence sertive speaker to be less trustworthy, F(l, 112) = 4.69, p < .05 (1) or high in intelligence (11), and not tentative (1) or very tentative (11). (M = 6.57 vs. 7.57), but not significantly less likable (M= 6.47 After completing the questionnaire, subjects were debriefed and exvs. 6.97), than the tentative speaker. cused. Three-way interactions were also obtained for how trustworthy, F(l, 112) = 7.89, p < .01, and likable, F(l, 112) = 5.09, Results p < .05, speakers were judged to be. These results paralleled those found for the opinion measure. Planned contrasts reThere were no effects due to the individual speakers, so this vealed that judgments of female speakers depended on both variable was eliminated from the analysis. Levene's (1960) test their language and the gender of the subjects, whereas judgto detect heterogeneity revealed no difference in variability for the male and female speaker conditions, F < 1. A 2 (Gender of ments of male speakers did not. Female speakers were judged Subject) X 2 (Gender of Speaker) X 2 (Type of Language) ANOVA to be more trustworthy by men, F(l, 112) = 10.04, p < .01, and less trustworthy by women, F(l, 112) = 7.59, p < .01, when they was conducted on the dependent variables. For the opinion were tentative than when they were assertive (see Table 4). They measure, a Gender of Subject X Type of Language interaction were also judged to be more likable by men, F(l, 112) = 3.66, was obtained, F(l, 112)= 10.34, p < .01. Women were inp < .05, and less likable by women, F(l, 112) = 6.31, p < .05 (see fluenced more by assertive than tentative language (M = 5.57 Table 4). vs. 3.60), F(l, 112) = 8.04, p < .01, and men were marginally more influenced by tentative than assertive language (A/= 4.57 vs. 3.37), F(l, 112) = 2.98, p < .10. In addition, as predicted, a three-way interaction was obtained revealing larger language General Discussion effects for female speakers, t(l 12) = 1.71, p < .05, one-tailed. Planned contrasts performed on the means, which are preThe results indicate that the use of tentative speech enhances sented in Table 2, provided support for Hypothesis 1. Female a woman's ability to influence a man but reduces her ability to speakers were more influential with men when they spoke tentainfluence a woman. Moreover, both male and female speakers tively than when they spoke assertively, t(\ 12) = 2.11, p < .05, judged a woman who spoke tentatively to be less competent and one-tailed, and more influential with women when they spoke knowledgeable than a woman who spoke assertively, but did not assertively than when they spoke tentatively, r(l 12) = 2.85, p < .01, one-tailed. As predicted by Hypothesis 2, male speakers were equally influential, regardless of their language or the gender of the subject, p > ,25. Table 3 Judgments of Tentative Versus Assertive Speakers Subjects' ratings of the speaker were coded so that high scores reflected a high amount of each characteristic, ANOVAS revealed Speaker main effects of type of language for how tentative, confident, F(l, 112) Characteristic Tentative Assertive powerful, competent, intelligent, and knowledgeable the speaker was perceived to be. As Table 3 shows, speakers who 7.12 13.68** 5.90 Tentative spoke tentatively were judged to be more tentative and less conConfident 6.14 8.40 32.43** fident, powerful, competent, intelligent, and knowledgeable. A Powerful 5.55 6.90 13.54** gender of speaker effect was also obtained for how knowledgeCompetent 6.68 7.88 12.42** Intelligent 5.83 7.36 14.04** able the speaker was perceived to be, F(l, 112) = 6.57, p < .05. Knowledgeable 7.95 8.83 6.57* Male speakers were judged to be more knowledgeable than female speakers (M = 8.75 vs. 8.03, respectively). In addition, a Note. Higher scores reflect higher ratings of confidence, powerfulness, Type of Language X Gender of Speaker interaction was obcompetence, intelligence, knowledge, and tentativeness. tained for ratings of the speaker's knowledge, F(l, 112) = 4.36, *p<.05. **p<.001.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER, LANGUAGE, AND INFLUENCE

949

Table 4 Ratings of Trustworthiness and Likableness

Gender of subject Trustworthiness Speaker Male Tentative language Assertive language Female Tentative language Assertive langauge
This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Likableness Male 5.93 6.40 8.00 6.53 Female 6.53 6.93 5.80 7.73

Male 6.73 6.80 8.40 6.33

Female 6.73 7.00 5.47 7.27

Note. Higher scores reflect greater trustworthiness and likableness.

consider language when rating the competence and knowledge of male speakers. Although it may be commonly assumed that speakers who are judged to be incompetent and lacking in knowledge are not likely to be very influential, a female speaker was actually more influential when speaking tentatively to a male audience, even though she was also considered less knowledgeable and competent than a woman speaking assertively. These results provide supporting evidence that gender differences in the use of tag questions, disclaimers, and hedges are probably a function of status differences between men and women. The results of both studies indicate that status may be an important determinant of gender differences in language. However, status is clearly not the only determinant. It is likely that the gender differences in use of intensifiers and verbal reinforcers are due to different norms that have been established for male and female groups. In particular, such behavior may reflect subjects' expectation that interactions between women are often highly social. In the first experiment, women spoke more tentatively when interacting with men than with women. The results of the second experiment indicate that such behavior may represent an attempt by women to deal with status inequities when in mixed-sex interactions. They also support the contention of Meeker and Weitzel-CNeill (1977) that low status persons must first demonstrate that they have no desire to compete for status before their ideas will be considered by higher status individuals. Tentative language may serve the function of communicating that the speaker has no desire to enhance his or her own status. Although the results of these studies are consistent with the status explanation developed in the introduction, it is possible that women may use tentative language when interacting with men for more than one reason. Tentative language may be used as a subtle influence strategy, as I have already suggested. It may also be used by women because they may expect men to be too assertive and overconfident in their speech. That is, the presence of men may make women conscious of their speech style and concerned that they not behave in the same overconfident manner that they expect from men. Finally, women may speak more assertively to women than to men because they may feel that women are generally more careful and correct in their

speech than men are and, consequently, it may be more important to speak correctly to women. In this study, men perceived a tentative woman to be more trustworthy and likable than an assertive woman, whereas women judged her to be less likable and trustworthy. A woman who spoke tentatively may not have been perceived as particularly competent or knowledgeable, yet she was still more influential with a male audience, perhaps because they have found her behavior more acceptable than that of a more assertive woman and, consequently, may have been less resistant to her arguments. Moreover, if women generally do speak tentatively to men, as Study 1 suggests, then male subjects may have expected the female speaker to be tentative. They may have been less influenced by a woman who spoke assertively because her behavior violated their expectations. Women were less influenced by a female speaker when she spoke tentatively than when she spoke assertively. This may be because women found tentative female speakers to be less believable, likable, and competent. In addition, if women are accustomed to hearing other women speak assertively, they may expect such behavior and may be less influenced when their expectations are violated. Unfortunately, for women, the use of tentative language as a subtle influence strategy either compromises their perceived competence or makes it difficult for them to be persuasive to an audience of both men and women.

References
Argyle, M., Lalljee, M., & Cook, M. (1968). The effects of visibility on interaction in a dyad. Human Relations, 21, 3-17. Aries, E. J., & Johnson, F. L. (1983). Close friendship in adulthood: Conversational content between same-sex friends. Sex Roles, 9, 1183-1196. Aukett, R, Ritchie, J, & Mill, K. (1988). Gender differences in friendship patterns. Sex Roles, 19, 57-65. Bales, R. F. (1950). Interaction process analysis: A method for the study of small groups. Cambridge, MA: Addison-Wesley. Barth, R. J, & Kinder, B. N. (1988). A theoretical analysis of sex differences in same-sex friendships. Sex Roles, 19, 349-363. Bauman, M. (1976). Two features of"women's speech?" In B. L. Dubois & I. Crouch (Eds.), The sociology of the languages of American women (pp. 33-40). San Antonio, TX: Trinity University Press. Beattie, G. W (1981). Interruption in conversational interaction, and its relation to the sex and status of the interactants. Linguistics, 19, 15-35. Bell, R. R. (1981). Friendships of women and of men. Psychology of Women Quarterly, 5, 402-417. Berger, J, & Fisek, M. H. (1974). A generalization of the theory of status characteristics and expectation states. In J. Berger, T. L. Conner, & M. H. Fisek (Eds.), Expectation states theory: A theoretical research program (pp. 163-205). Cambridge, MA: Winthrop. Berger, J, Fisek, M. H., Norman, R. Z, & Zelditch, M, Jr. (1977). Status characteristics and social interaction: An expectation-states approach. New York: Elsevier, Science. Berger, J, Rosenholtz, S. J, & Zelditch, M, Jr. (1980). Status organizing processes. In A. Inkeles, N. J. Smelser, & R. H. Turner (Eds), Annual review of sociology (Vol. 6, pp. 479-508). Palo Alto, CA: Annual Reviews. Berk, S. F. (1985). The gender factory: The apportionment of work in American households. New York: Plenum Press. Bilous, F. R, & Krauss, R. M. (1988). Dominance and accommodation

950

LINDA L. CARLI

in the conversational behaviors of same- and mixed-gender dyads. the languages of American women (pp. 53-62). San Antonio, TX: Language & Communication, 8,183-194. Trinity University Press. Bradley, P. H. (1981). The folk-linguistics of women's speech: An empir- Erickson, B., Lind, E. A., Johnson, B. C, & O'Barr, W M. (1978). Speech style and impression formation in a court setting: The effects ical examination. Communication Monographs, 48, 73-90. of "powerful" and "powerless" speech. Journal ofExperimental SoBroverman, I. K, Vogel, S. R, Broverman, D. M, Clarkson, F. E, & cial Psychology, 14, 266-279. Rosenkrantz, E S. (1972). Sex role stereotypes: A current appraisal. Journal ofSocial Issues, 28, 59-78. Falbo, T, & Peplau, L. A. (1980). Power strategies in intimate relationships. Journal ofPersonality and Social Psychology, 38, 618-628. Buhrke, R. A, & Fuqua, D. R. (1987). Sex differences in same- and cross-sex supportive friendships. Sex Roles, 17, 339-352. Fishman, P. M. (1978). Interaction: The work women do. Social Problems, 25, 397-406. Carli, L. L. (1982). Are women more social and men more task oriented? A meta-analytic review ofsex differences in group interaction, reward Fishman, P. M. (1980). Conversational insecurity. In H. Giles, W P. allocation, coalition formation, and cooperation in the Prisoners Di- Robinson, & P. M. Smith (Eds.), Language: Social psychological perlemma Game. Unpublished manuscript. spectives (pp. 127-132). New York: Pergamon Press. Carli, L. L. (1989). Gender differences in interaction style and influGilligan, C. (1982). In a different voice: Psychological theory and ence. Journal ofPersonality and Social Psychology, 56, 565-576. womens development. Cambridge, MA: Harvard University Press. Chodorow, N. (1978). The reproduction of mothering: Psychoanalysis Gruber, K. J, & White, J. W (1986). Gender differences in perceptions and sociology ofgender. Berkeley: University of California Press. of self Is and others' use of power strategies. Sex Roles, 15,109-118. Haas, A. (1979). Male and female spoken language differences: StereoCourtright, J. A, Millar, F. E, & Rogers-Millar, L. E. (1979). Domitypes and evidence. Psychological Bulletin, 86,616-626. neeringness and dominance: Replication and expansion. Communication Monographs, 46,179-192. Hall, J. A. (1984). Nonverbal sex differences: Communication accuracy and expressive style. Baltimore: Johns Hopkins University Press. Cowan, G, Drinkard, J, & MacGavin, L. (1984). The effects of target, age, and gender on use of power strategies. Journal of Personality and Hall, J. A., & Braunwald, K. G. (1981). Gender cues in conversations. Social Psychology, 47,1391-1398. Journal ofPersonality and Social Psychology, 40, 99-110. Cramer, H. (1946). Mathematical models of statistics. Princeton, NJ: Hartman, M. (1976). A descriptive study of the language of men and Princeton University Press. women born in Maine around 1900 as it reflects the Lakoff hypothesis in "Language and women's place." In B. L. Dubois & I. Crouch Crosby, F, Jose, P, & Wong-McCarthy, W (1981). Gender, androgyny, (Eds.), The sociology of the languages of American women (pp. 81and conversational assertiveness. In C. Mayo & N. M. Henley (Eds.), Gender and nonverbal behavior (x>x>. 151-169). New York: Springer- 90). San Antonio, TX: Trinity University Press. Verlag. Hartmann, H. (1981). The family as the locus of gender, class, and political struggle: The example of housework. Signs: Journal of Crosby, F, & Nyquist, L. (1977). The female register: An empirical Women in Culture and Society, 6, 366-394. study of Lakoff's hypothesis. Language in Society, 6, 313-322. Hewitt, J. P, & Stokes, R. (1975). Disclaimers. American Sociological Dabbs, J. M, Jr, & Ruback, R. B. (1984). Vocal patterns in male and Review, 40,1-11. female groups. Personality and Social Psychology Bulletin, 70, 518 525. Hirschman, L. (1973, December). Female-male differences in conversational interaction. Paper presented at the meeting of the Linguistic Deaux, K, & Kite, M. E. (1987). Thinking about gender. In B. B. Hess Society of America, San Diego, CA. & M. M. Ferree (Eds.), Analyzing gender: A handbook of social science research (pp. 92-117). Newbury Park, CA: Sage. Hirschman, L. (1974, July). Analysis oj"supportive and assertive behavior in conversations. Paper presented at the meeting of the Linguistic Deaux, K., & Major, B. (1987). Putting gender into context: An interacSociety of America, San Francisco. tive model of gender-related behavior. Psychological Review, 94, 369-389. Hollander, E. P., & Julian, J. W (1970). Studies in leader legitimacy, influence, and innovation. In L. Berkowitz (Ed.), Advances in experiDinnerstein, D. (1977). Themermaidandtheminotaur: Sexualarrangemental social psychology (Vol. 5, pp. 33-69). San Diego, CA: Acaments and human malaise. New York: Harper & Row. demic Press. Dovidio, J. F, Ellyson, S. L, Keating, C. F, Heltman, K, & Brown, C. E. Howard, J. A, Blumstein, P, & Schwartz, P. (1986). Sex, power, and (1988). The relationship of social power to visual displays of domiinfluence tactics in intimate relationships. Journal of Personality nance between men and women. Journal of Personality and Social and Social Psychology, 51,102-109. Psychology, 54, 233-242. Dovidio, J. F, Brown, C. E, Heltman, K, Ellyson, S. L, & Keating, C. F. Johnson, J. L. (1980). Questions and role responsibility in four professional meetings. Anthropological Linguistics, 22, 66-76. (1988). Power displays between women and men in discussions of Johnson, P. (1976). Women and power: Toward a theory of effectivegender-linked tasks: A multichannel study. Journal of Personality ness. Journal of Social Issues, 32, 99-110. and Social Psychology, 55, 580-587. Kelly, J. A., Wildman, H. E, & Urey, J. R. (1982). Gender and sex role Dubois, B. L., & Crouch, I. (1975). The question of tag questions in differences in group decision-making social interactions: A behavwomen's speech: They don't really use more of them, do they? Lanioral analysis. Journal ofApplied Social Psychology, 12,112-127. guage in Society, 4, 289-294. Key, M. R. (1972). Linguistic behavior of male and female. Linguistics, Eagly, A. H. (1983). Gender and social influence: A social psychologi88,15-31. cal analysis. American Psychologist, 38, 971-981. Eagly, A. H. (1987). Sex differences in social behavior: A social-role Kimble, C. E., Yoshikawa, J. C, & Zehr, H. D. (1981). Vocal and verbal assertiveness in same-sex and mixed-sex groups. Journal ofPersonalinterpretation. Hillsdale, NJ: Erlbaum. ity and Social Psychology, 40,1047-1054. Eagly, A. H., & Carli, L. L. (1981). Sex of researchers and sex-typed Kollock, P., Blumstein, P, & Schwartz, P. (1985). Sex and power in communications as determinants of sex differences in influenceinteraction: Conversational privileges and duties. American Socioability: A meta-analysis of social influence studies. Psychological logical Review, 50, 34-46. Bulletin, 90,1-20. Lakoff, R. T. (1975). Language and womans place. New York: Harper Eakins, B, & Eakins, R. (1976). Verbal turn-taking and exchanges in &Row. faculty dialogue. In B. L. Dubois & I. Crouch (Eds.), The sociology of

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

GENDER, LANGUAGE, AND INFLUENCE

951

Piliavin, J. A, & Martin, R. R. (1978). The effects of sex composition of Lapadat, J, & Seesahai, M. (1978). Male versus female codes in inforgroups on style of social interaction. Sex Roles, 4, 281-296. mal contexts. Sociolinguistics Newsletter, 8, 7-8. Raven, B. H, Centers, R, & Rodrigues, A. (1975). The bases of conjuLevene, H. (1960). Robust tests for equality of variances. In I. Olkin gal power. In R. E. Cromwell & D. H. Olson (Eds.), Power in families (Ed.), Contributions to probability and statistics (pp. 278-292). Palo (pp. 217-232). New York: Wiley. Alto, CA: Stanford University Press. Reskin, B. F. (Ed.). (1984). Sex segregation in the workplace: Trends, Lockheed, M. E., & Hall, K. R (1976). Conceptualizing sex as a status explanations, remedies. Washington, DC: National Academy Press. characteristic: Applications to leadership training strategies. JourReskin, B. F, & Hartmann, H. I. (Eds}. (1986). Womeris work, men's nal ofSocial Issues, 32,111-124. work: Sex segregation on the job. Washington, DC: National AcadLockheed, M. E, & Klein, S. S. (1985). Sex equity in classroom organiemy Press. zation and climate. In S. S. Klein (Ed.), Handbook for achieving sex equity through education (pp. 189-217). Baltimore: Johns Hopkins Roger, D, & Nesshoever, W (1987). Individual differences in dyadic conversational strategies: A further study. British Journal of Social University Press. Psychology, 26, 247-255. Maccoby, E. E. (1988). Gender as a social category. Developmental PsyRogers, W T , 4 Jones, S. E. (1975). Effects of dominance tendencies on chology, 24, 755-765. floor holding and interruption behavior in dyadic interaction. ComMaccoby, E. E., & Jacklin, C. N. (1987). Gender segregation in childhood. In H. W Reese (Ed), Advances in child development and behav- munication Research, 1,113-122. Roos, P. A. (1985). Gender and work: A comparative analysis ofindusior (Vol. 20, pp. 239-287). San Diego, CA: Academic Press. trial societies. Albany, NY: State University of New York Press. McCarrick, A. K., Manderscheid, R. W, & Silbergeld, S. (1981). Rubin, L. B. (1985). Just friends: The role of friendship in our lives. New Gender differences in competition and dominance during marriedYork: Harper & Row. couples group therapy. Social Psychology Quarterly, 44,164-177. Sistrunk, F, & McDavid, J. W (1971). Sex variable in conformity behavMcMillan, J. R, Clifton, A. K, McGrath, D, & Gale, W S. (1977). ior. Journal ofPersonality and Social Psychology, 17, 200-207. Women's language: Uncertainty or interpersonal sensitivity and Staley, C. M. (1978). Male-female use of expletives: A heck of a differemotionality. Sex Roles, 3, 545-559. ence in expectations. Anthropological Linguistics, 20, 367-380. Meeker, B. F, & Weitzel-O'Neill, P. A. (1977). Sex roles and interperSteckler, N. A., & Rosenthal, R. (1985). Sex differences in nonverbal sonal behavior in task-oriented groups. American Sociological Reand verbal communication with bosses, peers, and subordinates. view, 42, 91-105. Journal ofApplied Psychology, 70,157-163. Miller, J. B. (1976). Toward a new psychology ofwomen. Boston: Beacon Thorne, B, & Henley, N. (1975). Difference and dominance: An overPress. view of language, gender, and society. In B. Thorne & N. Henley Moore, S. F, Shaffer, L, Goodsell, D. A, & Baringoldz, G. (1983). (Eds.), Language and sex: Difference and dominance (pp. 5-42). Gender or situationally determined spoken language differences? Rowley, MA: Newbury House Publishers. The case of the leadership situation. International Journal ofWomeris Verbrugge, L. M. (1977). The structure of adult friendship choices. Studies, 6, 44-53. Social Forces, 56, 576-597. Mulac, A., & Lundell, T. L. (1986). Linguistic contributors to the West, C, & Zimmerman, D. H. (1977). Women's place in everyday talk: gender-linked language effect. Journal ofLanguage and Social PsyReflections on parent-child interaction. Social Problems, 24, 521chology, 5, 81-101. 529. Mulac, A, Lundell, T. L, & Bradac, J. J. (1986). Male/female language West, C, & Zimmerman, D. H. (1983). Small insults: A study of indifferences and attributional consequences in a public speaking situterruptions in cross-sex conversations between unaquainted peration: Toward an explanation of the gender-linked language effect. sons. In B. Thorne, C. Kramarae, & N. Henley (Eds.), Language, Communication Monographs, 53,115-129. gender and society (pp. 102-117). Rowley, MA: Newbury House Mulac, A., Wiemann, J. M, Widenmann, S. J., & Gibson, T. W (1988). Publishers. Male/female language differences in same-sex and mixed-sex dyads: Wiley, M. G, & Eskilson, A. (1985). Speech style, gender stereotypes, The gender-linked language effect. Communication Monographs, 55, and corporate success: What if women talk more like men? Sex 315-335. Roles, 12, 993-1007. Natale, M., Entin, E, & Jaffe, J. (1979). Vocal interruptions in dyadic Williams, J. E., & Best, D. L. (1982). Measuring sex stereotypes: A thirty communication as a function ofspeech and social anxiety. Journal of nation study. Beverly Hills, CA: Sage. Personality and Social Psychology, 37, 865-878. Willis, EN,& Williams, S. J. (1976). Simultaneous talking in conversaO'Barr, W (1982). Linguistic evidence: Language, power, and strategy tion and sex ofspeakers. Perceptual and Motor Skills, 43,1067-1070. Zimmerman, D. H, & West, C. (1975). Sex roles, interruptions, and in the courtroom. New \brk: Academic Press. silences in conversation. In B. Thorne & N. Henley (Eds.), Language Octigan, M, & Niederman, S. (1979). Male dominance in conversaand sex: Difference and dominance (pp. 105-129). Rowley, MA: tions. Frontiers, 4, 50-54. Newbury House Publishers. Offermann, L. R., & Kearney, C. T. (1988). Supervisor sex and subordinate influence strategies. Personality and Social Psychology Bulletin, Received October 17,1989 14, 360-367. Revision received May 14,1990 Pearson, J. C. (1985). Gender and communication. Dubuque, I A: William C. Brown. Accepted June 27,1990

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.