“Talk “Like a Man”: The Linguistic Styles of Hillary Clinton, 1992–2013”. Perspectives on Politics 14:1-18.
“Political Psychology as an Interpretive Field” with Helen Haste and Kristen Monroe. In Routledge Handbook of Interpretive Political Science, edited by Mark Bevir, and R.A.W. Rhodes, 2015. New York: Routledge.
“The Egyptian Blogosphere: A Counter-narrative of the Revolution” with Ban Al-Ani, Gloria Mark, and Justin Chung. In Proceedings of the Association for Computing Machinery (ACM) Conference on Computer Supported Cooperative Work (CSCW 2012), Seattle, WA. pp. 17-26. Honorable Mention Award.
“The Rationalizing Voter by Milton Lodge and Charles S. Taber” Political Psychology, 2015. 36(1): 137–40. (Book review).
Select Conference Papers
2016 American Political Science Association: Implicit Cues in Candidate Communication
2014 Midwest Political Science Association: Framing moral intuitions? The evolution of same-sex marriage in news media
Focused at the intersection of American politics, political psychology and political communication, my research asks about the effects of latent and dynamic communication between political elites and the American public —in policy debates, electoral campaigns, news and social media— on citizens’ emotional, moral, ideological and partisan attachments. My research is informed by work on political predispositions, the formation and updating of individual preferences, the cognitive, emotional, and sub-conscious levels of information processing, continuity and change in American public opinion and the ordinary and strategic use of language in politics.
My dissertation and most recent conference papers utilize a computational approach to language and discourse (machine learning, topic modeling, sentiment and linguistic analysis) performed on large text corpora that I curated by extracting text from the web (from blogs & news sites). I am interested in natural language processing, latent semantic analysis and large-scale text analysis and their applications more broadly. I primarily work in R but occasionally use Stata, LIWC (Linguistic Inquiry and Word Count), zTree (for experiments) and Inquisit (for psychological testing/experiments).
Talk “Like a Man:” Feminine Style in the Pursuit of Political Power
“I get it that some people just don’t know what to make of me.”
– Hillary Clinton, 2016 acceptance speech at the Democratic National Convention
Like many of her past speeches, reactions to Hillary Clinton’s historic acceptance speech at the Democratic National Convention focused not on what she said, but how she said it. Fox News commentator Greg Gutfeld tweeted “even when she says, ‘you know,’ it’s recited like a wind up doll” and New York Times columnist David Brooks told PBS News anchor Judy Woodruff that Clinton failed to “emotionally connect” in her speech. When then-Senator Obama spoke at the 2008 convention, it aligned with our expectations about how a leader should talk. When Clinton talks, it comes across as “unrelaxed”, “hair-raising”, “hectoring”, “nagging” and “grating”—at least to some people. It conflicts with expectations about how she should talk. Such comments illustrate a paradox that women in leadership roles confront. As a leader, Clinton is compared against traditionally masculine qualities that have long been associated with leadership—strength, determination, self-confidence, and more. She is criticized when she fails to display masculine leadership qualities and she is criticized and disliked when she fails to display feminine warmth. Despite her critics, however, Clinton has successfully navigated a path toward leadership in a profession dominated by men and by a male model.
Given the state of women’s representation in U.S. politics, how do female politicians like Clinton position themselves for success in key leadership positions? This question is the foundation for my dissertation research, which examines the gendered linguistic patterns in the language of political leaders and candidates for office. I conceptualize gendered linguistic styles based on empirical work in social psychology and linguistics, which shows that men and women tend to speak differently—not necessarily in the content or topics of their conversations, but in the use of seemingly unremarkable “function words,” such as pronouns (I, we, their), articles (a, an the) and prepositions (to, from) (Argamon et al., 2007; Mulac, 2006; Newman et al., 2008; Pennebaker, 2011; Schwartz et al., 2013). Function words are essential to communication in any language, but they are often implicit and not consciously evaluated in speech (Pinker 1994). They serve to structure and connect the content of our thoughts when we communicate with others and for this reason, they reflect both the deeply social nature of communication as well as how individuals organize and orient themselves within the world. They are the smallest, most frequent words in the English language and together reflect our linguistic style.
This project began with a case study examining whether Hillary Clinton talks more “like a man” (linguistically-speaking) the more her power and prominence in the political arena grows. I analyzed the gendered linguistic patterns in Clinton’s natural language using a computational text analysis of 567 interview and debate transcripts from 1992–2013. My findings reveal that as Clinton transitioned from first lady to US senator to secretary of state, she spoke in an increasingly masculine way. In talking more “like a man,” Clinton conformed to prominent gender norms in American politics (Sanbonmatsu, 2002; Bystrom et al., 2004; Banwart and McKinney, 2005; Dittmar, 2015).
To follow up on these findings, my second empirical chapter examines a broader sample of male and female political leaders in the United States—Congressional and party leaders, governors, and presidential candidates. In a computational text analysis of 2,484 interview and debate transcripts featuring 126 unique political leaders, my findings show that female leaders adopt masculine styles of communication as they move into leadership roles. This suggests that Hillary Clinton is not a particularly unique case. Among male leaders, Democrats and Republicans speak in ways that are consistent with gender stereotypes commonly mapped onto the two parties. Thus, male Democratic leaders are more likely to use a feminine style, whereas male Republican leaders are more likely to use a masculine style. These findings are consistent with public perceptions that view the Republican Party as masculine and more competent at dealing with “masculine issues” like foreign policy, and the Democratic Party as feminine and more competent at dealing with “feminine issues” like education and healthcare (Hayes, 2011; Winter, 2010).
My final empirical chapter features two exploratory studies designed to investigate (1) whether individuals reliably associate gendered linguistic styles with a politician’s gender and/or party, and (2) the extent that gender-linked language matters for male and female candidate evaluations. Study 1 was a survey (N=448) that investigated whether gendered linguistic patterns in fictitious candidate statements act as an implicit cue by asking participants to identify the gender and partisan affiliation of the statement’s author. Results show that participants overwhelmingly associated masculine statements with male candidates and, to a lesser degree, feminine statements with female candidates. Participants displayed strong associations between masculine statements and the Republican party and between feminine statements and the Democratic party, again lending support to Hayes (2011) and Winter (2010). Study 2 was an experiment (N= 557) that explored whether individuals perceive male and female candidates differently depending on whether they conform to or deviate from gender-linked language by randomizing candidate gender and measuring evaluations of candidates with gender conforming and non-conforming statements. Participants rated candidates with a feminine statement to be significantly more warm than candidates with a masculine statement, regardless of candidate gender. Such perceptions had an overall positive effect on evaluations since participants were also more likely to vote for candidates with feminine statements. There was very little difference, particularly on competence ratings, across male, female, and control candidate groups, which supports recent work by Brooks (2013), Dolan (2014) and Hayes and Lawless (2015).
 Greg Gutfeld Twitter, https://twitter.com/greggutfeld/status/758854585565917185
 Bob Woodward, http://www.cnn.com/TRANSCRIPTS/1602/06/smer.01.html
 Maureen Dowd, http://www.nytimes.com/2007/09/30/opinion/30dowd.html