Documente Academic
Documente Profesional
Documente Cultură
Hillary Clinton is arguably the most prominent woman in American politics today. Past research suggests female politicians conform
to masculine communication styles in an attempt to evade the double bind. Clintons long and varied career thus provides an
important and useful case study for investigating how female politicians present themselves strategically. Drawing on research in
political psychology, political communication, social psychology, and linguistics I examine whether Clinton talked like a man as she
navigated a path toward political leadership by conducting a quantitative textual analysis of 567 interview transcripts and candidate
debates between 19922013. Results on Clintons linguistic style suggest her language grew increasingly masculine over time, as her
involvement and power in politics expanded. I also consider Clintons language in the context of her 20072008 presidential
campaign. In 2007, Clintons linguistic style was consistently masculine, supporting widespread accounts of Clintons campaign
strategy. Beginning in late 2007, however, Clintons language became more feminine, reecting a shift in the self-presentational
strategies advised by her campaign staff. Throughout the 2008 campaign period, Clintons language uctuated dramatically from one
interview to the next, reecting a candidateand campaignin crisis. This study reveals hidden insight into the strategies Clinton
used as she navigated through the labyrinth toward leadership. Changes in Clintons linguistic style reect the performance of gendered
roles, expectations of political leaders, and the masculine norms of behavior that permeate political institutions.
1
992 was the year of the woman. Fifty-three president or vice president, and most scholars agree that
women were elected to the United States Con- there has only been one truly viable female candidate for
gress, twenty-four of them for the rst time. 1 president: Hillary Clinton.
Despite continued progress for women in politics, Women pursuing leadership positions are not simply
however, the promise of 1992 remains largely un- halted by a glass ceiling, but by a labyrinth of obstacles
fullled. Today women hold 19 percent of U.S. they must navigate along the way.3 These obstacles, both
Congressional seats, 25 percent of statewide executive implicit and overt, do not pose concrete barriers, but rather
ofces, and 24 percent of state legislative seats. 2 circuitous routes toward attaining leadership positions.4
Under-representation is even more apparent at the Expectations of leadership and institutional arrangements
highest levels of government. Worldwide, women advanced have implications for the types of individuals who run for
to key executive ofces in a number of countries, including public ofce as well as the self-presentational strategies that
Chile, Germany, Jamaica, Lithuania, and South Korea. politically ambitious women use to advance through the
In the United States, however, there has never been a female labyrinth of leadership. To be successful, they must cultivate
A list of supplementary materials provided by the author precedes the references section.
Jennifer J. Jones is a Doctoral Candidate in the Department of Political Science at the University of CaliforniaIrvine
(jonesjj@uci.edu). Her research is focused at the intersection of American politics, political psychology, and political
communication, and her dissertation builds on this article to examine the linguistic styles of U.S. party leaders, governors and
presidential candidates and to explore whether such language acts as an implicit cue informing candidate evaluations. She has
published in Political Psychology, Proceedings of the ACM Conference on Computer Supported Cooperative Work
(CSCW2012) and in the Routledge Handbook of Interpretive Political Science. Jennifer is grateful to Peter Suedfeld and
members of the American Politics Working Group at UC Irvine for their insight and critiques on an earlier draft and to Valerie
ORegan and fellow presenters at the 2015 Meeting of the Western Political Science Association and the 2015 Meeting of the
International Society of Political Psychology for feedback and suggestions on an earlier version of this paper. Jennifer would like to
thank James Pennebaker for his inspiration and guidance on gendered language, members of her dissertation committee, Kristen
Monroe, Marty Wattenberg, and Michael Tesler, for their invaluable advice and support as well as the anonymous reviewers and
Perspectives on Politics editor, Jeffrey C. Isaac, for their constructive criticism and helpful suggestions.
doi:10.1017/S1537592716001092
American Political Science Association 2016 September 2016 | Vol. 14/No. 3 625
3676 :DD C 53 4 697 9 5 7 2 C1 7 CD .4 3 / 3D C 4 75D D D:7 ,3 4 697 , 7 D7 C C7 3 3 34 7 3D
:DD C 53 4 697 95 7 D7 C :DD C 6 9 0
Articles | Talk Like a Man
an appropriate and effective self-presentationone that psychology, and linguistics, I conceptualize feminine and
reconciles symbolic attitudes toward gender with masculine masculine styles of communication in an original way.
prototypes of political leaders. Despite the difference that I then analyze these gendered linguistic styles in Clintons
women make for the political agenda and for the outcome of natural language using a quantitative textual analysis of
legislation, womens minority status in decision-making 567 interview and debate transcripts between 19922013.
bodies often results in their conformity to a normative, In doing so, this study reveals how Clintons linguistic
masculine style of communication, one that restricts the full style changed over time as she transitioned between roles
expression of their ideas.5 As the former prime minister of and climbed up the political ladder. Ultimately I nd that
Canada, Kim Campbell, describes it: Clintons linguistic style grew increasingly masculine
over time, as her involvement and power in the political
I dont have a traditionally female way of speaking . . . Im quite world expanded. I argue that changes in her linguistic
assertive. If I didnt speak the way I do, I wouldnt have been seen
as a leader. But my way of speaking may have grated on people style reected the performance of gendered roles, expect-
who were not used to hearing it from a woman. It was the right ations of political leaders, as well as the masculine norms
way for a leader to speak, but it wasnt the right way for a woman of communication that permeate political institutions.
to speak. It goes against type.6
Gender and Self-Presentation in Politics
Former Press Secretary for the Clinton administration,
Dee Dee Myers, captures this conundrum atly: If male The relationship between gender and democracy is well
behavior is the norm, and women are always expected to grounded in broader theories of substantive, descriptive,
act like men, we will never be as good at being men as men and symbolic representation.10 Over the past two decades,
are.7 The tension confronted by women pursuing power a number of studies have examined whether and to what
extent women legislators represent womens substantive
within male-dominated political institutions thus raises
concerns. In general, this research suggests that when
several important questions. How do female politicians
women are involved in the decision-making process there
present themselves as viable leaders given the power
are substantive differences in the issues discussed on the
imbalances that persist within political institutions? What
agenda as well as in the policy outcomes that result.11
strategies do they use to navigate through the political
Despite this, however, womens substantive interests cannot
labyrinth? Must they talk like men?
be advanced simply by increasing the sheer numbers of
1992 also marked Hillary Clintons debut onto the
women in public ofce.12 Representation and the advance-
national political scene. In the years since, Clinton
ment of women in society takes place in non-political
transitioned from rst lady of Arkansas to rst lady of the
contexts tooon the boards of multinational companies,
United States to an important politician in her own right, in news media, blockbuster lms, social movements,
winning election for U.S. Senate in 2000, and again in and more. The realm of electoral politics is onealbeit
2006. She campaigned for president in 2008, served as crucialarena where womens substantive representation
secretary of state from 20092013, and today stands as occurs, but it is mutually dependent on womens represen-
a frontrunner in the 2016 presidential contest. Clinton is tation in other areas of civil society. Still, the dispropor-
one of the most prominent and well-known politicians tionate number of women in public ofce and positions of
alivenine out of ten Americans recognize her name and leadership has implications beyond representation. It has
have an opinion of her.8 Moreover, attitudes toward gender consequences for the salience of gender in politics, the types
have been projected onto opinions of Clinton throughout of individuals who run for public ofce, as well as the
much of her public career.9 Clintons career thus provides behaviors and decisions that women express in these roles.
a valuable and instructive case for exploring the strategies
that women use to achieve power and inuence in politics. Gender Identity and Performance
Her example also raises broader questions about how male- Drawing from social identity theory13 and self-
dominated political institutions affect women who aspire categorization theory,14 much research has been dedicated
to move up the political ladder. Does Clinton talk more to understanding how social identities are manifest in a
like a man the more her political power grows? given context and how they inuence perceptions of
Language provides a valuable lens for understanding political actors and events. A well-established body of
how political life affects the self-presentation of women in research in political psychology demonstrates that social
politics. By examining Clintons linguistic style, this study identities including gender, race, religion, and partisanship
reveals hidden insight into the strategies Clinton used as fuel group-based attachments, and consequently shape
she navigated a path toward leadership. Linguistic style perceptions, attitudes, and judgments of the political
does not refer to the content or substance of Clintons world.15 However, the availability or salience of a particular
speech, but rather, to the way she communicates and how social identity largely depends on the context or situation.
she conveys meaningful content. Drawing from research In the context of an election, for example, partisanship is
in political psychology, political communication, social a highly salient identity that inuences the way partisan
gender traits and abilities and believe they are more research that suggests when women adhere to feminine
competent when dealing with issues related to social welfare, styles of conduct and communication, their views are
but less competent on issues of crime, defense, and the considered subordinate and are often challenged by men
economy, in which men are assumed to be more compe- in the group.41 In a revealing anecdote, Deborah Cameron
tent.32 In contrast, in a recent study by Deborah Brooks, describes how Margaret Thatcher prepared herself for
survey respondents rated male and female candidates the United Kingdoms top post by undergoing a linguistic
similarly on traits such as competence, empathy, and the makeover, which required her to lower the pitch of her
ability to handle an international crisis.33 In the same study, voice, atten her accent, and slow her delivery.42 To be
inexperienced female candidates were rated as stronger, successful in these institutions, then, women must negotiate
more honest, and more compassionate than inexperienced their authority among their male colleagues, which tends to
male candidates.34 Although the implications of these result in their conformity to a dominant, masculine style of
studies are mixed, they nevertheless indicate that gender communication.43
factors signicantly into public perceptions of politicians Communication in government institutions is often
and candidates for ofce and is thus an important consid- biased toward a masculine style of interaction, which can
eration for womens self-presentation. The work by Brooks, be seen in assertive, adversarial, hierarchical, and rule-
among others, reects a growing trend toward data-driven dominated legislative bodies like the U.S. Congress and
approaches to the double bind that, in time, may paint British Parliament. Regardless of gender, communication
a clearer picture of the obstacles female politicians face. styles within these institutions reect a masculine style.44
Therefore, in addition to looking toward voters (and self- As minority members, women are perceived (and often
report measures) to understand how gendered power perceive themselves) to be interlopers and as such,
dynamics manifest in the self-presentation of women in they adjust their behavior according to the norms of
politics, it is also important to consider the institutional, the group. 45 For example, female members of the
procedural, and implicit pressures that shape interactions British Parliament are just as likely as their male colleagues
within the political arena. to engage in a competitive and self-assertive style of
speaking and even more likely to adhere to the ofcial
Masculine Norms of Interaction in Institutional rules of the chamber.46 As interlopers to the political arena,
Settings their linguistic behaviour reects their understanding that
The self-presentation of women in politics is also affected to be judged as good community members they must
by the institutional procedures, interpersonal interac- put special effort into displaying their adherence to
tions, and norms of communication that govern political behavioural norms that carry particular symbolic weight.47
institutions. In The Silent Sex, Christopher Karpowitz and This suggests that institutional norms of behavior and
Tali Mendelberg examine how womens behavior is interaction embody and thus reward masculine styles of
impacted by procedural rules as well as the ratio of men communication. Instead of defying entrenched norms of
to women within deliberative groups.35 They nd that behavior, women appear to internalize their social
women have greater inuence when collective decisions environments, consciously and unconsciously conforming
are bound by unanimous consent, but less inuence when their interaction to align with the established, masculine
decisions are bound by majority rulethe dominant status quo. Such pressures illustrate the circuitous routes
procedure for democratic decision-making.36 They also women must navigate when pursuing power, inuence, and
nd that when women are minority members, they speak leadership in the political arena.
less often, have less inuence on the group outcome, and
align their speech patterns with the men in the group even Do Women Have to Talk Like Men to Be Considered
when they care about the topic of conversation and have Viable Leaders?
preferences distinct from men (e.g., generosity towards Altogether, research into the self-presentation of female
the poor).37 The nding that women speak less often, politicians suggests that expectations of leadership as well
however, is disputed elsewhere.38 Together, these ndings as institutional arrangements have signicant conse-
suggest that norms of interaction and institutional proce- quences for the communication strategies women adopt.
dures are both consequential for womens self-presentation. These factors can be summarized briey. First, gender is
Karpowitz and Mendelberg suggest that elite women, who a performance and particular notions of how women are
usually work in highly masculine environments, may be supposed to act encourage particular types of perform-
predisposed or socialized in ways that make them more ances. At the same time, however, particular notions of
inclined toward the views and interaction styles that how leaders are supposed to act encourage different, and
characterize the male central tendency.39 However, they sometimes conicting performances. Simply put, the
also point to evidence from interviews with female prototypical political leader looks, acts, and talks like a
politicians who believe they cannot get far with the man and a woman simply does not t into this prototype.
feminine style.40 This latter view is supported by Additionally, norms of behavior and interpersonal
Clintons increased involvement and power within the self-presentational strategies that female politicians use to
male-dominated institutions of the Senate and State achieve power and inuence in a male-dominated pro-
Department, suggests that her language became increas- fession. One approach to studying languagecontent
ingly masculine over time. This expectation is consistent analysishas been used extensively in political science
with the broader literature on women in politics, which to identify, for example, the integrative complexity of
suggests that female politicians adopt masculine commu- statements by members of the British House of
nication styles when it is the dominant style of interaction Commons,63 the issues legislators emphasize when com-
within the institutions they serve.56 municating with constituents,64 the policy positions of
In a thorough analysis of Clintons 2008 presidential political parties over time,65 and the differences in com-
campaign, Regina Lawrence and Melody Rose write that munication strategies in mixed-gender political debates.66
Clinton more often than not avoided calling attention to Despite substantial variation in the conceptualization and
her gender and instead focused on demonstrating her measurement of variables, such research typically ignores
policy expertise and toughness (though occasionally with or altogether removes common style or function words
some subtly gendered ourishes).57 Despite the historic (e.g., I, you, the, it, and, from) becauseat least on the
nature of her candidacy, Clinton explicitly intended to run surfacethese words contain little lexical or semantic
as a candidate, not as a woman. During a debate hosted by meaning. However, research in social psychology and
CNN in July 2007, Clinton was asked how she would linguistics demonstrate that function words do contain
respond to critics who say she is not authentically value.
feminine. She responded, Well, I couldnt run as Function wordsarticles, prepositions, pronouns, and
anything other than a woman . . . but, obviously, Im auxiliary verbsshape and connect the content of our
not running because Im a woman. Im running because thoughts into meaningful forms of communication.67
I think Im the most qualied and experienced person to While function words are the most commonly written
hit the ground running in January 2009.58 Rather than and spoken words in the English language, they have little
exposing the question as sexist and irrelevant or acting semantic meaning by themselves and are often implicit
ladylike and expressing herself as authentically feminine, in speech and not always consciously evaluated when
Clinton instead presented herself as an experienced pol- speaking.68 Linguistic style thus refers to the way an
itician with strong leadership abilities. Indeed, she suc- individual communicates and how she conveys meaningful
cessfully conveyed this image to the public. A survey by content to others.69 Linguistic style can provide insight
Pew in September 2007 found that among Democratic into a number of psychological and social processes.
voters 67 percent said Clinton rst came to mind when In prior research, linguistic style has been linked to per-
they heard the word tough, compared to 14 percent sonality traits, levels of depression, relationship quality,
for Obama and 7 percent for Edwards.59 Only 22 percent status and social hierarchy, gender, and more.70 By analyz-
said Clinton came to mind when they heard the word ing function words, researchers can gain insight into the
friendly, compared to 31 percent for Obama and 28 implicit, micro processes by which individuals weave
percent for Edwards.60 Clintons likability among voters disparate thoughts into meaningful narratives that organize
was a growing concern among her advisors and from and shape experience. Therefore, rather than ignoring or
late 2007 into January 2008, Clinton deviated from removing function words, my analysis focuses heavily on
her dominant, experienced-based and gender neutral Clintons use of function words and investigates her style of
strategy and attempted to present herself as a warmer, speaking.
more feminine candidate.61 However, this strategy Work by James Pennebaker and colleagues nd that
was short-lived. Once Clinton began to lose key con- language encodes gender in very subtle ways. Reliable and
tests to Obama, she returned to an aggressive, mascu- consistent gender differences in linguistic style have been
line strategy. found in studies analyzing tens of thousands of speech
The literature surrounding Clintons 2008 bid over- samples from both men and women.71 In general and
whelmingly suggests that her self-presentation was highly on average, women tend to use pronouns (especially
masculine over the course of her campaign, a strategy that rst-person singular pronouns), verbs and auxiliary
is consistent with the ndings from broader research into verbs, social, emotional, cognitive, and tentative words
the self-presentational strategies female candidates use to more frequently than men.72 In general and on
win.62 Consequently, I expect Clintons language was average, men tend to use nouns, big words (words greater
particularly masculine during her own campaignsin than six letters), articles, prepositions, anger, and swear
2000, 2006, and 2008. words more frequently than women.73 Utilizing this
insight, I constructed two indices and refer to them as
Feminine and Masculine Linguistic Styles feminine linguistic style and masculine linguistic
Language is a key site where gender is routinely performed, style, respectively. Table 1 describes the linguistic
and it thus provides a valuable lens for understanding the markers that comprise these contrasting styles.
This appears, at least on the surface, to conceptualize men, as it conforms to the expectation that male leaders
feminine and masculine styles quite differently than pre- are aggressive.
vious studies in the politics and gender literature.74 In much My approach also shares some similarity with prior
of this research, the coding schemes for feminine style studies. As referenced earlier, common coding schemes in
include factors such as using a personal tone, addressing the politics and gender literature suggest that female
viewers as peers, identifying with the experiences of others, politicians rely more on personal and social references.
inviting viewer participation, discussing family relationships, Talking about oneself in a personal way and talking to
inviting the audience to trust their experiences/perceptions and about other people implies the use of pronouns and
in making political judgments, and using personal social references, both of which are included in the
experiences/anecdotes.75 In contrast, coding schemes feminine linguistic style. References to external objects
for masculine style often include factors such as using like statistics, expert reports, and policy issues tend to rely
statistics, emphasizing ones own accomplishments, and on the use of articles (object references), prepositions
referencing expert authorities or sources. 76 (spatial and temporal hierarchies), and big words, which
By analyzing function words, which are often discarded are similarly included in the masculine linguistic style.
or ignored in coding schemes, my approach picks up on Hence, the variables examined in this study (derived from
less overt, more implicit expressions of gender than is empirical observations by Pennebaker among others) are
typical of many studies in the politics and gender literature. not as different from prior studies as they may appear.
This study also differs in that codes are well dened. As a case study, several critical factors are not taken
In general, a pronoun is a pronoun regardless of the data into account, including how partisanship or the interac-
source one analyzes. Another notable difference is the tion between party and gender79 might affect Clintons
inclusion of emotion into feminine and masculine linguistic style. Similarly, it is not clear from this study how
linguistic styles. Emotion has important implications age, race, ethnicity, education, and socioeconomic back-
for gendered self-presentationas recently as 2010, thirty ground impact the linguistic styles of political leaders.
percent of Americans believed that men were better suited Future research is needed to examine these factors and to
emotionally for politics than women.77 Consider Clintons explore the linguistic styles of both male and female
emotional response during a campaign event the day politicians more systematically. In addition, Clinton has
before the New Hampshire primary, when momentarily, experienced a unique trajectory into politics and, arguably,
her voice waivered and it appeared that she might cry. In an her career is not a typical case. It is, however, an
article titled, Can Hillary Clinton Cry Herself Back to the exceptionally important one. Clinton has been a well-
White House?, published the day after the primary, known gure in U.S. politics for nearly 25 years, through-
Maureen Dowd of the New York Times likens Clinton to out which she has taken on a variety of gendered roles.
the heroine of a Lifetime movie, a woman in peril who Very few, if any, women in U.S. politics have come close to
manages to triumph.78 Such depictions serve to re- reaching the level of prominence that Clinton has achieved
inforce the stereotype that tears and visible emotions are and sustained. Her example is a rare and worthy one for
feminine traits and signs of weakness, which can be studying the strategies female politicians use to navigate
consequential especially for female leaders. On the other a path toward leadership and for building on the limited
hand, anger is an acceptable emotional expression by body of existing research on this topic. Although Clintons
case cannot be generalized to understand broader trends, social, swear, and emotion words and a high rate of articles
my approach offers a promising direction for research into and big words in congressional speeches regardless of
gendered communication styles. gender, indicating that a formal, masculine linguistic style
is indeed pervasive in the chambers of the U.S. Congress.85
Methods and Data Finally, for each transcript I calculated a feminine to
I investigate Clintons linguistic style using an original masculine ratio by taking the sum of feminine linguistic
corpus80 of 567 interview and debate transcripts from markers and dividing by the sum of masculine linguistic
19922013. All interview transcripts with Hillary Clinton markers described earlier in table 1.86
available on the Clinton Presidential Librarys website
were included in this analysis and constitute the majority Linguistic Trends in Context:
of data analyzed from 19921999.81 All interview tran- How Clintons Language Reveals
scripts (including newspaper, magazine, broadcast, and a Gendered Self-Presentation
cable TV) and debate transcripts featuring Clinton be- Since 1992, Clintons self-presentation has been affected
tween 19922013 available through archived databases by gendered expectations of her various roles as well as
and on the Department of States website were also the norms of communication within the institutions she
included.82 This corpus thus represents a comprehensive has served. Before turning to a more detailed discussion
collection of interview and debate transcripts featuring of Clintons language and what it says about her self-
Clinton between 19922013. I then analyzed the femi- presentation within these roles, gure 1 presents a broad
nine and masculine linguistic markers within these texts overview of Clintons feminine/masculine linguistic style
using Linguistic Inquiry and Word Count (LIWC), a text and how it changed over time.87
analysis program.83 LIWC has been used to examine the In 1992 and 1996the years she campaigned for
linguistic patterns of political texts in a number of studies. BillClinton used a higher rate of feminine relative to
One, for example, found that candidates running for masculine linguistic markers, which is consistent with
president and vice president in 2004 used high rates of her expected role as a supportive wife and rst lady.
articles, prepositions, positive emotions, and big words,84 The feminine/masculine ratio declined abruptly in 1993
markers that are more consistent with a masculine lin- 1994, however, indicating that Clintons language became
guistic style. Another study found a low rate of pronouns, more masculine. This coincides with Clintons role on
Figure 1
Ratio of feminine to masculine styles over time
Note: Figure 1 gives a yearly time series plot of the ratio of feminine to masculine linguistic markers. The dotted lines represent election years
in which Clinton actively campaigned for herself (2000, 2006, 2008) or Bill (1992, 1996). The light grey line represents a smoothed
generalized linear estimate (with shaded confidence intervals) from the ratio model presented in table 2.
Table 3
Weighted average for all linguistic markers (%)
Examples 19921999 2000 20012006 20072008 20092013
Feminine style
Pronouns I, you, she, it 18.6 17.4 17.3 17.7 16.3
1st person singular I, me, my 4.3 5.3 4.2 4.6 2.8
Verbs Went, walk, listen 18.0 16.9 17.3 18.0 16.6
Auxiliary verbs Have, is, will, Im 11.5 10.8 11.1 11.4 10.9
Social references Friend, they, talk 12.0 9.8 10.3 10.1 10.8
Positive emotion Enjoy, nice, thank 3.6 4.2 3.7 3.6 3.9
Negative emotion Worry, nasty, cried 1.1 0.7 1.7 1.3 1.6
Tentative words Maybe, perhaps, guess 3.2 2.2 2.5 2.2 2.2
Cognitive mechanisms Cause, think, believe 20.9 19.0 20.6 19.3 20.1
Masculine style
Words . 6 letters 16.4 17.4 18.8 17.9 19.4
1st person plural We, our, lets 2.5 2.3 3.1 3.1 3.4
Articles A, an, the 5.7 7.0 6.8 6.5 7.1
Prepositions After, to, for, of, by 13.7 14.2 13.5 14.0 14.1
Anger words Hate, kill, annoyed 0.3 0.2 0.4 0.5 0.6
Swear words Ass, bastard, crap 0.0 0.0 0.0 0.0 0.0
Feminine/Masculine ratio 2.42 2.10 2.09 2.10 1.91
Word count 465,848 31,515 70,563 129,781 389,128
No. documents 156 19 56 65 271
Total word count 1,086,835
Total No. documents 567
Note: Raw values for each transcript were weighted by word count in calculating yearly averages. Values are expressed as
a percentage of total words per year and for multiple years, yearly values were averaged.
traditional boundaries of the rst ladys appropriate lady, Clintons use of feminine linguistic markers declined
sphere of inuence on policy matters.89 Following the during her run for Senate. Simultaneously, Clintons use of
failure of health reform, Clinton tried to soften her masculine linguistic markers, particularly big words, articles,
image to better fulll her role as rst lady and to lessen her and prepositions, sharply increased. This explains the sizable
perceived liability to the Clinton administration.90 Table 3 drop in the feminine/masculine ratio seen in gure 1 around
indicates that on average Clintons linguistic style was the year 2000. Table 3 also indicates that Clinton used an
more feminine during her time as rst lady than at any unusually high rate of positive emotion words and a corre-
other point in her public career. Her use of tentative words spondingly low rate of negative emotion words during this
(e.g., almost, probably, kind of, sort of) was particularly time. Indeed, this positive self-presentation is apparent
high during this time. While this nding suggests that when reading these transcripts. Clinton was enthusiastic
Clinton was relatively uncertain or insecure when discus- about the possibility of serving in the Senate and bringing
sing topics with journalists, tentative language is also positive changes to New York. This may have been a strategy
common with individuals who have not fully processed she used to combat perceptions of her as a carpetbagger
and formed a reliable narrative about an event or topic.91 and re-breathing dragon among New Yorkers.92
In reviewing transcripts with a high rate of tentative words, During a campaign, it is reasonable to expect a candidate
I found both factorsuncertainty and lack of a consistent to discuss him or herself more frequently than usual since
narrativewere at play. She often used tentative words as the purpose of a campaign is to educate voters about their
a buffer against potential criticism or to express cautious ideology, experience, and policy goals. Indeed, table 3
certainty when making factual assertions or statements shows an increase in Clintons use of rst-person singular
that implicated her husbands administration. pronouns during her 2000 and 2008 campaigns, which
indicates that Clinton talked in a personal way about her
Clinton for Senate (2000) beliefs, experiences, and plans. Interestingly, pronouns are
The most dramatic and sustained shift in Clintons not only a marker of gender but also of social status.
language was in her transition from rst lady to Senate Contrary to a widely held assumption, lower status
candidate. Table 3 reports that the feminine/masculine ratio individuals are more likely to use rst-person singular
declines from 2.42 during her time as rst lady to 2.10 pronouns especially when talking up to higher status
during her Senate race in 2000 when Clinton campaigned individuals, who are more likely to talk down to
for herself for the rst time. Compared to her tenure as rst you, or for the generalized, all-assuming we, which
Figure 2
Ratio of feminine to masculine styles for all interviews and debates in 20072008
After Super Tuesday, February 5, Obama had accumu- over time, yet are often hidden from view. This study
lated a sizable advantage over Clinton, and the Clinton reveals how these forces manifest in Hillary Clintons
campaign responded with an aggressive messaging cam- self-presentation by tracking her subtle linguistic behav-
paign attacking Obama, what Lawrence and Rose describe ior over time. Overall, my ndings show that when
as a testosterone blitzkrieg.99 This masculinized messag- Clinton occupied a political ofce or took on a major
ing proved successful in Texas and Ohio, which encouraged policy initiative (as in 19931994), her language con-
Clinton to maintain this strategy in subsequent state formed to a masculine style. Indeed, Clintons language
contests.100 Figure 2 does not reect this strategy, however. grew increasingly masculine over time, as her involve-
Figure 2 shows Clintons language became more feminine ment and power in politics expanded. This result sup-
starting in late 2007, but it does not indicate a noticeable ports prior research suggesting that women adopt
shift toward a more masculine style after January 2008. As masculine communication styles when seeking inuence
gure 2 illustrates, Clintons linguistic style was scattered in male-dominated settings.101
and uctuated much more dramatically from one interview Clintons linguistic style changed according to the
to the next throughout the 2008 campaign period, which gendered expectations of the roles she performed as well
ended once Clinton conceded the race to Obama in June. as the masculine norms of communication within the
This volatility in Clintons linguistic style may reect a institutions she served. These ndings can be summarized
candidateand campaignin crisis without a clear strat- succinctly. In 1992 and 1996, Clintons linguistic style
egy on her self-presentation as a female candidate for was consistent with her expected performance as the wife
president. It is also possible that internal disagreements of a presidential nominee. When she led the administra-
and confusion over Clintons gendered self-presentation tions health reform policy in 19931994 however,
seeped into her linguistic behavior. Clintons linguistic style changed in response to the
political environment, reecting the masculine norms of
Conclusion: Power Speaks with communication that dominate the policymaking arena.
a Masculine Voice After 1994, Clinton performed more traditional duties of
The self-presentation of female politicians is affected by the rst lady and her language followed suit, turning more
the salience of their gender, perceptions and expectations feminine. As a candidate for Senate, her language shifted
of leaders, and their interpersonal interactions within pro- toward a masculine style, a performance she sustained
fessional and institutional contexts. These complex dy- throughout her time in Congress. As a candidate for
namics reinforce certain behavioral norms and expectations president in 20072008, Clintons self-presentation was
contexts will provide the comparisons and controls neces- 25 Jamieson 1995.
sary to isolate the effect of gender on linguistic style. 26 Carlin and Winfrey 2009; Kellerman and Rhode
For politically ambitious women like Clinton, self- 2007.
presentation is consequential and thereby strategic. Gen- 27 Haste 1993.
der encourages a particular type of self-presentation, yet 28 Bystrom et al. 2004; Banwart and McKinney 2005.
for female politicians the prototypes of political leaders 29 Bystrom et al. 2004.
encourage a differentand sometimes conictingself- 30 Herrnson, Lay and Stokes 2003.
presentation. I nd that these performances play out 31 King and Matland 2003; Winter 2010.
within even the shortest and most forgettable words we 32 Huddy and Terkildsen 1993; Winter 2008.
speak. This has important implications for the strategies 33 Brooks 2013.
women use to navigate a path toward leadership and offers 34 Ibid.
valuable insight for future research. To that end, this study 35 Karpowitz and Mendelberg 2014.
contributes an original approach to studying gender in 36 Ibid.
political communication, one that unveils some of the 37 Ibid.
more complex and subtle mechanisms that undermine 38 Pearson and Dancey 2011a.
womens representation and authority in politics. Such 39 Karpowitz and Mendelberg 2014, 334.
research contributes to the challenging and extraordinarily 40 Ibid., 336.
important task of uncovering the power of identity in 41 Kathlene 1994.
politics. 42 Cameron 2005.
43 Gertzog 1995; Cameron 2005; Dodson 2006;
Notes Karpowitz and Mendelberg 2014.
1 Manning and Brudnick 2014. 44 Yu 2014.
2 Center for American Women in Politics 2015. 45 Eckert 2000.
3 Eagly and Carli 2007. 46 Shaw 2000.
4 Ibid. 47 Cameron 2005, 498.
5 Gertzog 1995; Dodson 2006; Karpowitz and 48 Winter 2008; Tesler and Sears 2010.
Mendelberg 2014. 49 Clinton 1992a; Clinton 1992b.
6 Quoted in Eagly and Carli 2007, 102. 50 Clinton 1992b.
7 Quoted in Krum 2008. 51 Burrell 2001.
8 Jones 2015. 52 Watson 1999.
9 Tesler and Sears 2010. 53 Burns 2009; Burrell 2001.
10 See, e.g., Mansbridge 1999; Lovenduski 2005. 54 Edwards 2009.
11 Swers 2002; Dodson 2006; Pearson and Dancey 55 McThomas and Tesler 2016.
2011b. 56 Gertzog 1995; Cameron 2005; Dodson 2006;
12 Beckwith and Cowell-Meyers 2007. Karpowitz and Mendelberg 2014.
13 Tajfel and Turner 1979. 57 Lawrence and Rose 2010, 122.
14 Turner et al. 1987. 58 The American Presidency Project., Democratic
15 Kinder and Sears 1981; Winter 2008; Tesler and Presidential Candidates Debate at The Citadel in
Sears 2010; Iyengar, Sood, and Lelkes 2012; Tesler Charleston, South Carolina Transcript of televised
2014. debate, July 23, 2007 compiled by Gerhard Peters and
16 See e.g. Hawkesworth 2003; Htun 2004. John T. Woolley. http://www.presidency.ucsb.edu/
17 Tesler and Sears 2010. ws/index.php?pid575575, accessed October 2014.
18 Butler 2013. 59 Pew Research Center for the People & the Press
19 Butler 1999. 2007.
20 Lawless and Fox 2010. 60 Ibid.
21 Ibid. 61 Lawrence and Rose 2010; Kornblut 2011.
22 Ibid. 62 Carroll 2009; McKinney, Davis, and Delbert 2009;
23 Kinder et al. 1980. Lawrence and Rose 2010.
24 Huddy and Terkildsen 1993; Duerst-Lahti and Kelly 63 Tetlock 1984.
1995; Eagly and Carli 2007; Kellerman and Rhode 64 Grimmer 2010.
2007; although see Brooks 2013. For an extensive 65 Lowe et al. 2011.
meta-review of studies that nd women are associated 66 Banwart and McKinney 2005.
with communal traits, whereas men and leaders are 67 Pennebaker 2011.
associated with with agentic traits see Eagly and Carli 68 Pinker 1994; Tausczik and Pennebaker 2010.
2007. 69 Pennebaker 2011.
Pew Research Center for the People and the Press. 2007. Tajfel, Henri and John C. Turner. 1979. An Integrative
Voter Impressions of Leading Candidates: Clinton Theory of Intergroup Conict. Social Psychology of
Seen as Tough and SmartGiuliani as Energetic. Intergroup Relations 33(47): 74.
September 20. www.people-press.org/2007/09/20/ Tausczik, Yla R. and James W. Pennebaker. 2010. The
clinton-seen-as-tough-and-smart-giuliani-as-energetic/, Psychological Meaning of Words: LIWC and Com-
accessed April 12, 2015. puterized Text Analysis Methods. Journal of Language
Pinker, Steven. 1994. The Language Instinct: The New and Social Psychology 29(1): 2454.
Science of Language and Mind. New York: HarperCollins. Tesler, Michael. 2014. Priming Predispositions and
Schultheiss, Oliver C. 2013. Are Implicit Motives Changing Policy Positions: An Account of When Mass
Revealed in Mere Words? Testing the Marker-Word Opinion Is Primed or Changed. American Journal of
Hypothesis with Computer-Based Text Analysis. Political Science 59(4): 80624.
Frontiers in Psychology 4(748). Tesler, Michael and David O. Sears. 2010. Obamas Race:
Schwartz, Andrew, Johannes C. Eichstaedt, Margaret L. The 2008 Election and the Dream of a Post-Racial
Kern, Lukasz Dziurzynski, Stephanie M. Ramones, America. Chicago, IL: University of Chicago Press.
Megha Agrawal, Achal Shah, Michal Kosinski, David Tetlock, Philip E. 1984. Cognitive Style and Political
Stillwell, and Martin E. P. Seligman. 2013. Personal- Belief Systems in the British House of Commons.
ity, Gender, and Age in the Language of Social Media: Journal of Personality and Social Psychology 46(2):
The Open-Vocabulary Approach. PLoS ONE 8(9). 365.
Sears, David O. and Carolyn L. Funk. 1999. Evidence of Turner, John C., Michael A. Hogg, Penelope J. Oakes,
the Long-Term Persistence of Adults Political Predis- Stephen D. Reicher, and Margaret S. Wetherell. 1987.
positions. Journal of Politics 61(1): 128. Rediscovering the Social Group: A Self-Categorization
Shaw, Sylvia. 2000. Language, Gender and Floor Ap- Perspective. New York: Blackwell.
portionment in Political Debates. Discourse & Society Watson, Robert P. 1999. The Presidents Wives: Reassessing
11(3): 40118. the Ofce of First Lady. Boulder, CO: Lynne Rienner.
Slatcher, Richard B., Cindy K. Chung, James W. Penne- Winter, Nicholas J. G. 2008. Dangerous Frames: How
baker, and Lori D. Stone. 2007. Winning Words: Ideas About Race and Gender Shape Public Opinion.
Individual Differences in Linguistic Style among U.S. Chicago, IL: University of Chicago Press.
Presidential and Vice Presidential Candidates. Journal . 2010. Masculine Republicans and Feminine
of Research in Personality 41(1): 6375. Democrats: Gender and Americans Explicit and Im-
Spiker, Julia A. 2009. It Takes a Village to Win: A plicit Images of the Political Parties. Political Behavior
Rhetorical Analysis of Hillary for President. In 32(4): 587618.
Cracked But Not Shattered: Hillary Rodham Clintons Woolley, John and Peters Gerhard. 2014. The American
Unsuccessful Campaign for the Presidency, ed. Theodore Presidency Project. University of California. Available
F. Sheckels. Plymouth, UK: Lexington Books. from http://www.presidency.ucsb.edu/debates.php.
Swers, Michele L. 2002. The Difference Women Make: The Yu, Bei. 2014. Language and Gender in Congressional
Policy Impact of Women in Congress. Chicago, IL: Speech. Literary and Linguistic Computing 29(1):
University of Chicago Press. 11832.