The Secret Life of Pronouns: What Our Words Say About Us

by James W. Pennebaker

  With all best regards,

  Famous Professor

  All of a sudden, the formerly distant, high-status Pennebaker is writing like the lower-status, slightly groveling student. If you look a little more closely at the e-mails you will also see that all the e-mails are friendly and enthusiastic. Not using I-words does not make the writer appear cold or arrogant, just slightly less accessible and more distant.

  Many people ponder these findings and think the differences in I-word usage simply go along with making a request of some kind. If you want something, maybe you use I. When you are in the position to grant the request, perhaps you don’t. Some other studies indicate this isn’t the case. A good example are some e-mails I sent a few years ago in my capacity as an administrator in my department. We were running low on office space and I had to ask a few people to move to different offices. Here are some edited versions of an e-mail I sent to a Very Important Professor and another one to a Humble Graduate Student. The results were painfully predictable:

  Dear (Very Important Professor):

  I’ve been trying to avoid this but I think I may need to ask you if you would be willing to give up your office … I can find you something significantly less grand should worse come to worse.

  Dear (Humble Graduate Student):

  As you probably know, office space in the department is at a premium. We are doing our best to accommodate all the students. However, would you be willing to move your office …? Thanks so much for your help with this.

  As with the earlier examples, the tones of the e-mails are honest, warm, and constructive. Even though the content is essentially the same, they convey slightly different messages. The request to the Very Important Professor is written as though I have my hat in my hand. You can see me stooping over a little and speaking in a quiet voice. In the second e-mail, I’m not speaking at all. I’m merely conveying information from “us”—the department. “We really don’t want to bother you but we want you to move. Nothing personal, mind you.”

  Did I consciously adjust the language of these e-mails? No. I have no memory of specifically writing them. My brain must have kicked in on its own and adjusted the writing style knowing who would read the e-mail later. Don’t blame me. It was my brain that was so sensitive to status differences.

  I love the e-mail study because it demonstrated an important effect that held up across several people and, more important, revealed my own hypocrisy. Status hierarchies are everywhere yet they go on right below our noses. We are generally oblivious to them, especially when corresponding with friends whom we may have known for years.

  THE SPEEDY UNFOLDING OF SOCIAL HIERARCHY IN CONVERSATIONS WITH STRANGERS

  Think back to the scene from The Godfather at the beginning of the chapter. Do we have the ability to detect pronoun use in natural conversations so, like Vito Corleone, we can determine who is the powerful person in any given situation? In a word, no. People talk too quickly and we simply can’t detect small differences in pronoun use with our ears. Once the conversations are transcribed, however, computer programs have little problem.

  Unlike e-mails or letters, when talking with others, the words fly out of our mouths with very little forethought. Despite these very different forms of social interaction, the same general word categories can identify social status in conversations as in written communications. Several studies have now been conducted with college students and, in a manner of speaking, in the White House with top administration officials.

  In one of the first studies linking pronoun use to status in natural conversations, my graduate student Ewa Kacewicz invited students to participate in a get-to-know-you task. Pairs of strangers entered a room and talked about anything that interested them for ten minutes while they were videotaped. The conversations were fairly standard—where are you from? What’s your major? How do you like the university? At the end of the conversation, the two people went to separate rooms where they each evaluated the interaction. Both tended to agree on which person had more status and who tended to control the interaction more.

  The findings from Ewa’s project and others like it have been clean and consistent. High-status people use we-words and you-words at high rates and use I-words at low rates. In fact, this same pattern emerges when people are chatting with each other on the Internet. What is most striking, however, is how fast the social hierarchy emerges in the conversation. In Ewa’s first study, the two people who were going to have an interaction met just as they were going into the laboratory. They were asked not to talk until both were seated and the video cameras were turned on.

  Within the first minute of the conversation, the social hierarchy was established. I suspect that the two people had a rough sense of the hierarchy within seconds of seeing one another. Perhaps they noticed each other’s height, weight, attractiveness, clothing, or general bearing. Whatever it was, their relative use of pronouns was fixed almost immediately.

  When two strangers meet and chat on the Internet, you would think that it would take a little longer for the hierarchy to emerge. Surprisingly, it doesn’t take much longer. The Internet project was part of a large classroom exercise that I conducted with my colleague Sam Gosling and graduate student Yla Tausczik. Hundreds of students were asked to go online at particular times where they were randomly connected with another person. They were simply asked to chat about anything they wanted for fifteen minutes. At the end of the fifteen minutes, they completed a questionnaire about the conversation.

  Just like the face-to-face natural conversation project, the status differences were apparent in the first three minutes of the interaction. This amazed us because there were no physical appearance cues, voice tone, or anything else that could prejudge status for the two participants. How was it possible for the two people to tacitly agree who was more dominant so quickly?

  Reading some of the online transcripts helped to understand the process. Early in the conversation, both participants tossed out cues about themselves and, at the same time, fished for cues about the other person. Some participants would also make subtle moves that diminished the other person’s stature. Here is an example of two females at the beginning of their conversation. Person B (Brittany), whose comments are in bold, is the one who both agree was more dominant when they completed questionnaires at the end of the conversation:

  A: hello, is anyone here?

  B: I am. Hi.

  A: oh hey

  B: I’m Brittany

  A: what are we supposed to talk about? I’m Chris, nice to meet you.

  B: dont take this the wrong way, are you a boy or a girl?

  A: haha girl

  B: Chris is kind of ambiguous, lol, nice to meet you. what year are you

  A: freshman. What about you?

  B: cool. im a junior. Major?

  A: studio art but i am going to be transferring to communications to do photo journalism. You?

  B: history

  A: are you involved in any organizations or anything like that here

  B: um, im getting certified to teach high school but other than that, no. you?

  A: i am thinking about being a photographer. I don’t want to draw for a living

  B: photography is great

  A: i’m not a serious art student … i did art in high school so i could get into the art school here

  Notice at the beginning the two people appear to be approaching the interaction in the same way. By the third time she writes, however, Brittany makes an underhanded power play, “don’t take this the wrong way, are you a boy or a girl?” I ask you, dear reader, can you imagine any sentence that starts with “Don’t take this the wrong way …” that can end well for the listener? Brittany has already put Chris on notice that she has the potential to be a bully.

  Afterward, both look for information about the other that can help peg them on some kind of social hierarchy. However, once Brittany discovers that Chris is younger and in a less prestigious major, she psychologically takes
control. Once Brittany has established her dominance, Chris begins to focus on herself in a more nervous, submissive way. Much later in the conversation (not included here), the higher-status Brittany works to give Chris advice about how she can (and should) get teacher certification.

  Interestingly, both Chris and Brittany reported that they enjoyed the conversation and liked one another. At the end of the time period, the two exchanged e-mail addresses and promised to get in touch with one another in the future. And it is safe to say should they meet for coffee, Brittany will still be the person with the higher status in both of their eyes.

  In reading over hundreds of interactions such as this, it is fascinating to watch the unfolding of the status hierarchy. “Innocent” status-related questions include asking the other person about their parents’ occupation, last vacation, exercise habits, ethnicity, living location on campus, grade in the psychology class, and others. Not surprisingly, the person with high status in a particular domain usually is the one who asks a question about it. Notice how the higher-status Brittany is the one who asks about the other person’s year in school and major.

  Assessing each other’s status is not unique to English-speaking college students. It is something that happens around the world. In fact, some societies have much simpler rules in assessing status. In South Korea, for example, relative age is one of the most strict determinants of who is higher in the social hierarchy. If the age is the same, wealth or income is assessed next. In these societies, it is common for people to directly ask one another about their lives, which, in the West, might be viewed as impolite. On a recent trip to South Korea, for example, I sat next to a Korean man roughly my age. Within the first ten minutes of the flight, he started a conversation asking me about my age. When it was clear that we were exactly the same age, he asked what my yearly income was. When I answered, he smiled warmly and simply said, “We make about the same.” I got the impression that his income was far higher, which made him somewhat more comfortable in talking with me afterward.

  PEGGING STATUS HIERARCHIES IN THE NIXON WHITE HOUSE

  Laboratory experiments to study status are pale imitations of the natural social processes that often take months or years to develop. In real-world organizations, the person who has emerged as the boss is likely to be someone who has been groomed for years and has the respect of others within the group. On occasion, an opportunity arises to capture natural conversation among well-known individuals with known levels of status. When this happens, we can begin to see the overall social hierarchy of the group. One of the best natural experiments of relative status emerged through tape recordings made a generation ago. The recordings were released as part of the famous Watergate scandal that shook the United States and resulted in the resignation of a president.

  Two years after his 1968 election as president, Richard M. Nixon had a secret recording system installed in his office in the White House. Only a couple of his staff knew about it, and based on the recordings, he himself was rarely inhibited by the presence of the hidden microphones. After serving for four years, Nixon ran for reelection against a Democratic candidate who hopelessly trailed in the polls from the very beginning. Partly because of the paranoia that ran rampant in the Nixon administration, some of his high-level aides approved a number of illegal activities to help ensure Nixon’s reelection. One scheme was a late-night break-in of the Democratic National Headquarters in the Watergate office building in Washington, D.C. The burglars, some of whom worked in the White House, installed listening devices on the phones of the election strategists. Because of the vigilance of a single night watchman, the burglars were caught by the local police as they left the building.

  Because the burglary was so peculiar, very few people could imagine that it had been initiated by responsible people in the White House. In the first months after the break-in, few took it seriously and its occurrence had no discernible impact on the election four months later. However, two young reporters from the Washington Post newspaper, Bob Woodward and Carl Bernstein, doggedly pursued the story until Nixon’s closest aides began to be implicated. The Watergate scandal made headlines from early 1973 until Nixon’s resignation in August 1974.

  The turning point in the investigation occurred when it was revealed by a midlevel White House staffer, Alexander Butterfield, that the White House taping system always recorded conversations when Nixon was present. In the subsequent months as their legal problems mounted, the White House released several of the transcribed recordings to the public. When the final one (also known as the “smoking gun” tape) was made public in the summer of 1974, Nixon was forced to resign.

  The most pertinent tapes involved Nixon’s conversations with his lawyer John Dean, his chief domestic adviser John Ehrlichman, and his chief of staff H. R. Haldeman. The recordings, which were published in the spring of 1974, have been a treasure trove for historians and for language analysis researchers like me. Of the initial transcripts, we were able to analyze fifteen conversations where Nixon had one-on-one conversations with Dean, Ehrlichman, or Haldeman. For each conversation, we compared Nixon’s words with those of his aides.

  Consistent with the lab studies, the high-status Nixon used far fewer I-words than his aides. Overall, 3.9 percent of Nixon’s words were first-person singular pronouns, compared to his aides’ 5.4 percent. Nixon also used more we-words (1.4 percent versus 0.8 percent) and you-words (3.4 percent versus 1.8 percent) than his aides.

  Closer inspection of the transcripts suggested that Nixon had very different relationships with the three men. In their conversations, Nixon’s use of the first-person singular was significantly lower when talking to Dean and Ehrlichman than in his interactions with Haldeman. These pronoun patterns suggested that Nixon distanced himself much more from Dean and Ehrlichman, but that Nixon and Haldeman spoke to each other as equals. This is a great hypothesis, but is it true?

  Of the four men, only John Dean was still alive when we began analyzing the Watergate transcripts in the early 2000s. He consented to an e-mail interview with me about Nixon’s relationships with the three aides. Haldeman and Nixon had known each other since the early 1950s and, in Dean’s words, were peers in running the White House. “They were not friends however. [Haldeman] once said that [Nixon] had no idea how many children he had, and they only shook hands once—the day [Nixon] fired him.” But ultimately “it was a partnership.”

  Dean’s own relationship with Nixon was formal and respectful. Interestingly, Dean characterized Ehrlichman, who was often in “over his head” with respect to Washington politics, as arrogant yet insecure. In listening to the Watergate tapes himself, Dean was impressed with the degree to which Ehrlichman was making a power play in the hopes of getting Haldeman’s job. In his interactions with Nixon, Ehrlichman was overly solicitous, almost groveling. Nixon’s reaction was that of even greater psychological distance than with Dean, with whom he had a more formal relationship.

  One final analysis with the I-words is noteworthy. The Watergate tapes that were released in 1974 were recorded from June 1972 (soon after the break-in) through July 1973, when the scandal was front-page news most every day. Remember that people who enjoy very high status and high self-esteem and who tend to be overconfident generally use I-words at very low levels. And in his interaction with his aides in June through about November 1972, Nixon’s rate of I-word usage ranged between 2 and 4 percent. However, as the scandal continued to develop and Nixon’s status began to erode, his use of I-words increased month by month. By the last recordings in July 1973, his average I-word usage with his aides hovered around 7 to 8 percent of all his words. In other words, as Nixon’s political world began to crumble, he became less dominant and powerful in all his dealings with others.

  The Watergate transcripts point to our ability to distinguish the social hierarchy of Nixon’s aides through Nixon’s own eyes (or, perhaps, mouth). We can infer that he was most responsive and open to what Haldeman may have said compared to others such
as Ehrlichman. Interestingly, the pronoun analyses can’t tell us about the relative hierarchy among the aides themselves. Nixon may have conferred more status on Dean than Ehrlichman inside the Oval Office. However, when Dean and Ehrlichman were together in other settings, it is possible that Ehrlichman could have had more status than Dean. Only through analyses of different groups of people can we begin to uncover the complexities of a group’s status hierarchy.

  Finally, there is a large difference between status and liking. As Dean pointed out, there is no evidence that Nixon and Haldeman particularly liked one another. Respecting another person and liking them is not the same.

  THE LANGUAGE OF LEADERSHIP

  Can the research on status be extended to our understanding of leadership? Is it possible to identify good leaders by the ways they use words? Can leaders become more effective if they change the ways they use words? Surprisingly few studies have examined the language of leaders. Even fewer have studied the words of people in everyday life in order to predict what kinds of leaders they might be in the future.

  ASSUMING A LEADERSHIP ROLE AFFECTS THE LEADER’S LANGUAGE

  Within small groups, leaders do, in fact, talk differently than followers. Ethan Burris from the Red McCombs School of Business at the University of Texas and his colleagues conducted a project where about forty groups of business students worked together on a group task. Each of the four-person work groups were told that they represented a small consulting firm that was to offer advice to a fictitious company to improve its customer service division. The task was complex and required each small group to work together to achieve a solution.