Submitted March 12, 1996
Resubmitted April 1, 1996
Resubmitted April 26, 1996
Accepted April 29, 1996
Publication Date: May 14, 1996
"FEMININE SPEECH" IN HOMOGENEOUS GENDER GROUPS*
Robert K. Shelly
Ohio University
ABSTRACT
A characteristic "feminine speech" style has been identified as an
ideal type, either using tag questions or rising inflections at the end
of declarative sentences or actually stating task contributions as
questions. This style has been characterized as part of the feminine
role repertoire socialized in American society. Data from seventy
three-person open interaction task groups are used to test the idea
that this behavior pattern is a role style, with women enacting the
behavior more than men. Results indicate that men and women initiate
this behavior at a similar rate in the homogeneous gender groups
studied, suggesting that this behavior reflects status patterns of the
larger society enacted in heterogeneous task settings, but not in
homogeneous gender settings.
[page 50]
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[page 51]
INTRODUCTION
Lakoff (1975) and Tannen (1994) describe a feminine interaction
style with verbal characteristics distinct from a masculine interaction
style. These characteristics include a greater use of verbal tags such
as "...don't you think?" or "That is a good idea, isn't it?" Women are
also thought to be more likely to employ verbalizations that imply a
question, with rising inflection placed on the end of a declarative
statement. But how and under what conditions is this set of behavior
observed? Is it truly a feminine pattern of behavior, or is it
observable in particular situations in both genders? This paper
examines interaction in homogeneous task groups to answer the question
of whether the behavior is a "feminine style" or is observable for
both females and males under different circumstances.
Researchers report a variety of effects of gender in interaction in
task groups. These effects, involving both verbal and nonverbal
behavior, have been reported in heterogeneous and homogeneous groups.
In general, men are reported to be more task-oriented than women
(Strodtbeck and Mann, 1956; Nemeth, Endicott, and Wachtler, 1976;
Aries, 1976; Borgatta and Stimson, 1963; Craig and Sheriff, 1986;
Kelley, Wildman, and Urey, 1982; Lockheed and Hall, 1976; Mabry, 1985;
Piliavin and Martin, 1978; Wood and Karten, 1986), and more likely to
display high status interaction cues such as chin thrusts and looking
while speaking than women (Dovidio, Brown, Heltman, Ellyson, and
Keating, 1988; Dovidio, Ellyson, Keating, Heltman, and Brown, 1988;
Carli, 1990). Women are more likely to engage in socioemotional
behavior, be more concerned with social process (Piliavin and Martin,
1978; Lakoff, 1975; Tannen, 1995), smile more, and display low status
cues such as looking while listening and withdrawn posture (Dovidio, et
al., 1988; Dovidio, et al., 1988).
EXPLANATIONS
Two explanations have been advanced for the research results
reviewed above. One emphasizes the feminine role, with an associated
interaction style (Lakoff, 1975; Eagly and Karan, 1991; Maccoby, 1990;
Tannen, 1995). This explanation asserts men and women are socialized to
play particular gender roles in society, that this socialization
creates trans-situational behavior repertoires, and that these
repertoires are activated in most interaction situations, regardless of
the situation or gender composition of the groups in the situation.
These behavior repertoires include verbal behaviors, such as the use of
tag questions and the demonstration of a willingness to agree with
assertions of others, and nonverbal behaviors, such as smiling, gaze,
and posture.
[page 51]
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[page 52]
The other explanation asserts that observed differences in both
interpersonal behavior and interaction cues such as verbal style,
posture, and gaze are due to the instantiation of status differences
from the larger society into the task group. Thus, the relative
advantages of males and females in society are reflected in task
groups, with females enacting a low status role and males enacting a
high status role in situations in heterogeneous groups. In homogeneous
groups, males and females should exhibit no differences in behaviors or
interaction style. This approach has been developed by Berger and
others in expectation states theory (Wagner and Berger, 1995; Ridgeway,
1988; 1991; Ridgeway and Diekema, 1992; Shelly and Munroe, 1994).
According to this explanation, behavioral differences are observed
when status differences are activated in task groups. One situation in
which such differences are activated occurs when males and females are
in heterogeneous task groups. Because of their higher social standing,
males exhibit more high-status verbal and nonverbal behaviors than
females in heterogeneous groups. Males talk more, exert influence, look
more while speaking, and sit forward in their chairs. Similarly,
females talk less, accept influence, look while listening, and sit back
in their chairs.
In homogeneous groups, there should be no difference in verbal or
nonverbal behavior due to the gender of the participants in the groups.
Males and females should talk the same amount, be equally likely to
exercise influence or be influenced, look while speaking, or look while
listening. Other status organizing structures may produce these
effects, but gender should not lead to behavioral differences in
homogeneous groups.
According to expectation states theory, the distinct interaction
style associated with females in heterogeneous groups is a status
effect, due to a lower standing of women in society and not a
socialized pattern of role behavior. But, if gender is not activated in
task groups, then males and females should employ verbal acts at the
same rate in homogeneous task groups.
Hypothesis: In task groups, males and females should employ verbal
tags, such as questions at the end of declaratives and rising tone
of voice at the end of declaratives, at different rates. That is,
males should employ this behavior less than females.
[page 52]
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[page 53]
METHOD
The data used to test the hypothesis consist of observations of
interaction in seventy gender-homogeneous, three-person task groups
(thirty-four male groups, thirty-six female groups). The groups were
composed of undergraduate students at a state university who met to
solve the NASA "Lost on the Moon Exercise" (Johnson and Johnson,
1994). The groups were distributed equally across seven conditions
designed to test how different imposed social structures affect
interaction. The conditions included a control condition in which no
manipulation was carried out, a condition in which liking for others
was manipulated, a condition in which an authority position was
created, a condition in which skills of participants were varied, a
condition in which the authority position was occupied by a highly
skilled person, a condition in which the authority position was
occupied by a well liked person, and a condition in which the skilled
person was well liked. A more detailed discussion of the techniques
employed in the experiment appears in Shelly and Munroe (1994).
The independent variables are the sex of the group members, the
condition of the experiment as an organizing structure, and the
position of the actor (advantage with respect to others in the group).
Measures of interaction were coded from video tapes of each group.
Discussion lasted from ten to twenty minutes per group. All discussions
concluded when consensus had been reached about the best solution to
the problem. The coding identified statements with the following
properties as verbal tags (TAG):(1) a declarative content, but ending
with a tag question (e.g., "I think we should put the rope next, don't
you?"); (2) statements that are declarative in content, but ending with
a rising inflection (e.g., "I think the gun goes next (rise in tone of
voice)."); (3) a response to a question that contains a suggestion for
a task solution, but phrased as a question (e.g., (Q) "What should we
put next?" (A) "What about the rope?"); and (4) a directly stated
question that offers a solution to a task problem (e.g., "Should we put
the rope next?").
Two individuals coded independently of one another. The reliability
analysis compared the constituent parts of the coding to one another
and the total tagged activity, arriving at a Cronbach's alpha of .775.
This is a moderate reliability compared to other studies of face-to-
face interaction that report values above .90.
RESULTS
The individual measures of tagged interaction were summed to arrive
at a total number of tags per actor (TAG). This ranged from zero to
fifteen acts, with a mean of 3.64 and a median of 3.0.
[page 53]
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[page 54]
Two other measures may be created from this measure and its place in
the total interaction in the group. One is the percent of each actor's
own initiations that are tagged (PCTTAG), which measures the extent to
which a person is consistently deferential to others in a task group.
The range for this measure is zero to twenty seven percent, with a mean
of 4.66 and a median of 3.74.
The second measure is the percent of total activity in a group
accounted for by tagged activity by an individual, which indicates how
much an individual might stand out by using tagged speech in
interaction (TAGPCTTOT). This variable has a range from zero to thirty
one percent, with a mean of 2.29 and median of 1.44.
The simplest test of the hypothesis is to compare means for males
and females for tagged speech (TAG). Comparing means for the percent of
own speech that is tagged (PCTTAG) provides a test normed on the actor.
Comparing means for the percent t of total group speech tagged
(TAGPCTTOT) allows a test of the hypothesis normed on the group. The
results of these tests are shown in Table 1.
Table 1. Means of tagged speech measures by sex.
Sex TAG PCTTAG TAGPCTTOT
Male 3.43 4.92 2.73
(N=102)
Female 3.84 4.40 1.87
(N=108)
T-test t=-1.02 t=.95 t=1.85
p=.308 p=.342 p=.066
It is clear that in these homogeneous groups, females do not generate
significantly more verbal tags than males. This is true for the
frequency with which this behavior is generated (TAG). Females initiate
slightly more behavior of this sort, but a goodness of fit
interpretation of the t-test would suggest that this behavior is
initiated at the same rate for each gender (p=.308).
For both percentage measures, males initiate more activity, though
the results are not significantly different for either measure. A
goodness of fit interpretations for the percent of a person's activity
generated (PCTTAG) would suggest no difference between males and
females on this behavioral measure (p=.342). Males initiate a higher
percentage of total interaction than females as tagged activity
(TAGPCTTOT), though this difference is not significant at the usual
level.
[page 54]
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[page 55]
If a directional hypothesis is formulated for tagged activity as a
percentage of total interaction, with males expected to generate more
of this activity than females, the result (males=2.73% versus
females=1.87%) is significant at the .05 level. This is a direct
reversal of the gender hypothesis I set out to test.
The possibility of interactions between the sex of the group and the
imposed structure is examined to determine if females and males behave
differently in some situations but not others. An analysis of variance
tests this hypothesis. The ANOVA identifies two statistically
significant main effects and one significant interaction between gender
and condition. One main effect is for position: being in the relative
advantaged position versus the disadvantaged position makes it more
likely the actor will engage in tagged speech (X(a)=4.17 versus
X(d)=3.38, F=5.494; 1 d.f., p=.02). There is also a statistically
significant effect for condition of the experiment, with tagged
behavior more likely in Skill at task and Authority and Skill
conditions (F=7.062, 6 d.f., p=.00). Table 2 presents the results for
TAG for such an analysis. Analysis of the PCTTAG and TAGPCTTOT produce
similar results.
Table 2. Means of tagged speech (TAG) by sex and imposed structure.
Condition Total Tags Male Female
Control 3.93 3.40 4.47
(N=15) (N=15)
Formal 2.37 2.33 2.42
Authority (N=18) (N=12)
Liking 3.13 2.58 3.50
(N=12) (N=18)
Skill 5.30*** 2.53 8.07*,**
at task (N=15) (N=15)
Authority 2.50 2.00 3.00
and Liking (N=15) (N=15)
Skill 3.37 3.73 3.00
and Liking (N=15) (N=15)
Authority 4.90*** 8.50*,** 2.50
and Skill (N=12) (N=18)
* Significantly different from the opposite sex actor in this
condition. This comparison is by Tukey test for significant
differences.
** Significantly different from the other conditions in this sex. This
comparison is by Tukey test for significant differences.
*** The mean for the Skill condition is significantly different than
the means for the Authority, Authority and Liking, and Liking
Condition. The mean for the Authority and Skill condition is
significantly different from the means for the Authority and Authority
and Liking Conditions. This comparison is by Tukey test for differences.
[page 55]
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[page 56]
The introduction of manipulations for structures in these groups
does not alter the result that males and females do not differ in the
rate at which they initiate tagged speech. But there are two conditions
for which this is not true. Females in the Skill condition initiate
more tagged speech than either males in this condition or females in
other structures. Males in the Authority and Skill condition initiate
more tagged speech in this condition than males in other conditions as
well.
Controlling for the rate of initiation by examining the percentage
of tagged speech does not alter these conclusions for PCTTAG. There is
an interaction effect observable in the Authority condition for
TAGPCTTOT. Females assigned to the Authority condition initiate more
tagged speech than do males as a percentage of their total interaction
(.85 percent of total activity for males versus 1.59 percent of total
activity for females).
Overall, males and females initiate tagged speech at the same rate
in these homogeneous gender task groups. Effects observed in imposed
social structures show advantaged actors employ the speech style more
than disadvantaged actors. Interactions with gender and imposed
structures show complementary results with females engaging in this
behavior more in one condition of the experiment and males in another.
Differences between males and females are not significant in the other
five conditions of the study. Females initiate more tagged behavior as
a percentage of total interaction, but in only one condition of the
experiment. The task of explaining these results remains.
SUMMARY AND CONCLUSIONS
The question I set out to answer was whether tagged speech can be
thought of as a distinct interaction style, characteristic of one
gender. The alternative interpretation is that such speech
distinguishes one gender from another only when gender is activated as
a status characteristic. This alternative suggests males and females in
task groups will exhibit such interaction patterns differentially only
in heterogeneous groups, and not in homogeneous groups. This is the
overall finding for the seventy groups studied for this report: males
and females behave the same in this study.
Males and females initiate tagged speech, the "feminine style," at
the same rate in homogeneous groups, no matter how the activity is
measured (simple frequency, normed on the actor, or normed on the
group). Some interaction effects are observed in various imposed social
structures, but these effects are observed in male groups in one
structure and in female groups in another structure. There are no
differences in rate of initiation of tagged speech in the other five
types of groups. Thus the conclusion is that tagged speech is not a
"feminine speech" style socialized in gender roles.
[page 56]
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[page 57]
Two explanations for these results are possible. One is that actors
seek to "sugarcoat" task activity if legitimation of task action is
suspect. This would account for the results cited by Tannen (1994) and
others, and is consistent with the results described above. The
activity would thus have differential meaning as suggested by Ridgeway
and Diekema (1992): high status actors employ the behavior under some
conditions, and low status actors employ it under other circumstances.
Another possibility is that tagged behavior is a status cue (Berger,
Webster, Ridgeway, and Rosenholtz, 1986). Task cues, verbal and
nonverbal acts, provide information about an actor's capacity to
perform a task. Tagged behavior communicates such information (Newcombe
and Arnkoff, 1979). A well designed experiment in which male and female
actors reproduce tagged and untagged speech patterns in attempting to
influence subjects would provide a strong test of tagged speech as a
status cue. Such a study would advance our understanding of how men
and women interact with members of the same sex and members of the
opposite sex and the meanings they attach to verbal and nonverbal
interaction behavior.
ENDNOTE
*Research reported here was supported in part by grants from Ohio
University Research Challenge Funds and from DAAL03-86-D-0001 from the
Naval Training Systems Center. Views, opinions, and findings contained
in this report are those of the author and should not be construed as
an official Department of the Army position, policy or decision.
REFERENCES
Aries, Elizabeth J. 1976. Interaction patterns and themes of male and
female,and mixed sex groups. Small Group Behavior 7: 7-18.
Berger, Joseph, Murray Webster, Jr., Cecilia L. Ridgeway, and Susan
Rosenholtz. 1986. Status Cues, Expectations, and Behavior. In Edward J.
Lawler (Editor) Advances in Group Processes. Greenwich, CT: JAI Press,
Inc.
Borgatta, Edgar F. and John Stimson. 1963. Sex differences in
interaction characteristics. Journal of Social Psychology 60: 89-100.
Craig, Jane M. and Carolyn W. Sherif. 1986. The effectiveness of men
and women in problem-solving groups as a function of group gender
composition. Sex Roles 14: 453-66.
Dovidio, John F., Clifford E. Brown, Karen Heltman, Steve L. Ellyson,
and Caroline F. Keating. 1988. Power displays between women and men in
discussions of gender-linked tasks: A multichannel study. Journal of
Personality and Social Psychology 55: 580-587.
[page 57]
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[page 58]
Dovidio, John F., Steve L. Ellyson, Caroline F. Keating, Karen Heltman,
and Clifford E. Brown. 1988. The relationship of social power to visual
displays of dominance between men and women. Journal of Personality and
Social Psychology 54: 233-242.
Eagly, Alice H. and Steven J. Karan. 1991. Gender and the emergence of
leaders. Journal of Personality and Social Psychology 60: 685-710.
Johnson, David W. and Frank P. Johnson. 1994. Joining Together: Group
Theory and Group Skills. Boston, MA: Allyn and Bacon.
Kelley, Jeffrey A., Hal E. Wildman, and Jon R. Urey. 1982. Gender and
sex role differences in group decision-making social interactions: a
behavioral analysis. Journal of Applied Social Psychology 12: 112-127.
Lakoff, Robin. 1975. Language and Women's Place. New York: Harper and
Row.
Lockheed, Marlaine E. and Katherine P. Hall. 1976. Conceptualizing sex
as a status characteristic: Applications to leadership training
strategies. Journal of Social Issues 32: 111- 124.
Mabry, Edward. 1985. The effects of gender composition and task
structure on small group interaction. Small Group Behavior 16: 75-96.
Maccoby, Eleanor E. 1990. Gender and relationships: A developmental
account. American Psychologist 45: 513-520.
Nemeth, Charlan, Jeffrey Endicott, and Joel Wachtler. 1976. From the
'50's to the '70's: Women in jury deliberations. Sociometry 39: 293-
304.
Newcombe, Nora and Diane Arnkoff. 1979. Effects of speech style and sex
of speaker on person perception. Journal of Personality and Social
Psychology 37: 1293-1303.
Piliavin, Jane A. and Rachel R. Martin. 1978. The effects of sex
composition of groups on style of social interaction. Sex Roles 4: 281-
96.
Ridgeway, Cecilia L. 1988. Gender differences in task groups: A status
and legitimacy account. In Murray A. Webster, Jr. and Martha Foschi
(Editors) Status Generalization: New Theory and Research. Stanford, CA:
Stanford University Press.
-----. 1991. The social construction of status value: Gender and other
nominal characteristics. Social Forces 70: 367-386.
Ridgeway, Cecilia L. and David Diekema. 1992. Are gender differences
status differences? In Cecilia L. Ridgeway (Editor) Gender,
Interaction, and Inequality. New York: Springer-Verlag.
[page 58]
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
[page 59]
Shelly, Robert K. and Paul T. Munroe. 1994. Do women engage in less
task behavior than men? Paper presented at the Annual Group Processes
Conference. Los Angeles, California.
Strodtbeck, Fred L. and Richard D. Mann. 1956. Sex role differentiation
in jury deliberations. Sociometry 19: 3-11.
Tannen, Deborah. 1994. Talking 9 to 5. New York: William Morrow and
Company.
Wagner, David G. and Joseph Berger. 1995. Gender and interpersonal
behavior: Status expectation accounts. Working paper 100-10. Department
of Sociology. Stanford University: Stanford, CA.
Wood, Wendy and Steven J. Karten. 1986. Sex differences in interaction
style as a product of perceived sex differences in competence. Journal
of Personality and Social Psychology 50:341-347.
AUTHOR BIOGRAPHY
Robert K. Shelly is Professor of Sociology at Ohio University. He is
continuing studies of interaction and the formation of expectations,
and is currently investigating how social structures without socially
defined status-value organize interaction. Address correspondence to:
rshelly1@ohiou.edu