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

by James W. Pennebaker

  THE GROWING SENSE OF GROUPNESS: FROM ME TO US

  The longer people talk with others, the more they use we-words and the less they use I-words. As we get to know others, we let down our guard and start to accept them. The pattern of increasing we-words and decreasing I-words emerges across a wide array of groups.

  Speed-dating is the perfect place to start. Think back to the speed-dating project in the last chapter where strangers met for ten consecutive four-minute “dates” back-to-back. Even in these ridiculously short meetings, people’s use of we- and I-words changed quickly and predictably. Minute-by-minute in the brief speed-dating sessions, I-words for both people dropped and we-words increased.

  The I-drop/we-jump effect can be seen far outside the dating context. In some studies, students visit a psychology laboratory and end up talking to a complete stranger for fifteen minutes in a get-to-know-you setting. In the first five minutes, both participants typically talk a little about their backgrounds, their majors, and where they live. Because both people are describing themselves, they tend to use I-words at relatively high rates. During the next five minutes, they talk less about themselves as they begin to establish common ground between them. By the final five minutes, their use of I-words drops between 10 and 50 percent compared to their starting levels. At the same time, their use of we-words increases anywhere from 20 to 200 percent. The patterns are even stronger in online get-to-know-you chats between two strangers.

  The same shifts occur in larger groups. In a business school experiment with four-person groups, people worked on a complex group-decision task for thirty minutes. Same results: People in the groups decreased their I-word usage by 19 percent and increased their we-words by 39 percent from the first ten minutes to the last ten minutes.

  More interesting are shifts in we- and I-words in real-world groups that last over hours, days, months, and even years. The airline cockpit project discussed earlier found similar patterns. The longer a crew is together, the more the group uses we-words.

  And the I-drop/we-jump effect extends to bigger groups over much larger time frames. One project involved eighteen engineers, economists, and computer experts working on a complex online national defense project over almost two years. Another included around twenty professional therapists who met twice a year for three years as part of their professional training. Another project tracked the lyrics of the Beatles over their ten years of singing together. For all these groups, I-word usage dropped month by month, year by year, just as we-word use increased.

  What does all this mean? The more time we spend with other people, the more our identity becomes fused with them. We may not necessarily like or trust them but as our history becomes intertwined, we see ourselves as part of the same group. An interesting corollary of this phenomenon happens as we age. By and large, the older we get, the more time we have spent with virtually everyone around us. One might even predict that older people would use fewer I-words and more we-words than when they were younger. And it’s true. In a study of language use among almost three thousand people, those over the age of seventy used 54 percent more we-words and 79 percent fewer I-words than adolescents.

  BRINGING “US” TOGETHER: WHEN EVENTS CREATE GROUP IDENTITY

  Gradually adopting the identity of a group is a natural process that we rarely notice. One minute in a conversation our opinions are sprinkled with I and the next with we. It just happens. There are other times, however, when the ways we identify with a group change quickly and dramatically.

  The simplest examples can be seen with our allegiances to sports teams. One of the most clever social psychology experiments to demonstrate this was run in the mid-1970s by Robert Cialdini and his colleagues. Students who were attending universities with top-ranked football teams were called in the middle of football season as part of a survey purportedly dealing with campus issues. In the previous weeks, their home football teams had won a major game but also had lost another. After a few preliminary questions, the interviewer asked about one of the two pivotal games in the season, “Can you tell me the outcome of that game?”

  If their team had won, they usually answered, “We won.” But if their team had failed to win the game, their answer was more likely “They lost.” Taking partial credit for your team’s winning is a form of basking in reflected glory. Basically, we want to be close to groups that are successful and distance ourselves from losers.

  The sense of wanting to belong to a powerful group may have its roots in evolution. Most social animals seek the protection of a group when they are threatened. Even the appearance of an outsider can make people more aware of their own social network. In the same football project, Cialdini’s interviewers sometimes claimed that they lived in the same town as the respondents. Other times, the interviewers reported that they were from out of state. In other words, half the time the participants thought they were talking to someone like them whereas the other half they believed that they were being interviewed by a football foreigner. The us-them effect was much stronger when talking with the outsider. When reminded of competing groups, our own membership in a successful group becomes more important.

  The need to be part of a group is most likely to occur when an outside group threatens the very existence of our own group. Nothing galvanizes people more quickly than a physical attack against them. In the United States, the September 11, 2001, attacks provided a powerful example. In the first days of September, George W. Bush had been president for less than eight months and had an approval rating of about 54 percent. Within a week of the 9/11 attacks, his ratings skyrocketed to almost 90 percent. In every city in the country, people flew American flags and all the news stations discussed the outpouring of patriotism and national pride.

  You may recall the large-scale project that we conducted on thousands of blog entries that spanned a four-month period starting two months before 9/11. Within minutes of the attacks, LiveJournal.com bloggers started switching from the use of I-words to we-words. In fact, this pattern continued for several weeks afterward. Coinciding with the elevation in we-words was the brief drop but then long-term increase in positive emotion words. A horrible trauma such as 9/11 has the unintended effect of bringing people together, making them less self-focused, and within a few days, making them more happy.

  That natural and man-made disasters can bring people together is not a new idea. When residents of London were enduring an extended period of nightly bombings by the Germans in World War II, several indicators pointed to the much tighter social bonds among the people. One frequently cited statistic is that suicide rates dropped dramatically during this period. More recent data found large drops in suicide rates in the weeks after September 11, 2001, and, in the United Kingdom, after the 2006 subway bombings. You will recall similar findings discussed earlier surrounding the eruption of the Mount St. Helens volcano in 1980. Those from cities that had suffered the most damage later reported they were glad that it had happened in their lifetime and that the eruption had brought the community together.

  Study after study has found the same pattern of effects, many of which have been mentioned elsewhere in the book. The we-jump and I-drop phenomenon is as reliable as, well, taxes and death. Examples include:

  • Newspaper articles, letters to the editors of newspapers, and online tributes following the A&M bonfire tragedy that killed twelve students in 1999. The disaster galvanized the student body, resulting in tighter social connections than ever before. Illness rates over the next six months dropped 40 percent from the year before, unlike all other large universities in the region.

  • Online chat groups talking about the death of Princess Diana in a car accident in 1997. The effect lasted about a week.

  • Online bulletin boards discussing the Oklahoma City bombing that killed 168 people in 1995. Even in the earliest days of shared online communication, people rushed to talk about their feelings.

  • Hundreds of essays of college students in New Orleans four months after
Hurricane Katrina destroyed the city.

  • Hundreds of interviews and surveys of people across Spain in the weeks after the March 11, 2004, Madrid subway bombings.

  • Google Internet searches in England following the London underground bombings in 2005, Pakistan after the assassination of Benazir Bhutto in 2007, India after the attacks in Mumbai in 2008, and Poland after the airline crash that killed the country’s president and other leaders in 2010, which all showed comparable trends.

  THE USE OF we-words and the sense of group solidarity is not something that just occurs when people are threatened. Much like the football example, people embrace their group identity when it succeeds or behaves admirably. Google Labs has developed a marvelous application called Google Trends (www.google.com/trends) that allows users to track the words people use when searching the Internet. With it, anyone can see how particular cities, states, or countries use words over time. Using Google Trends, it is easy to come up with examples of increased we-word use during times of cultural pride. Nationwide, use of we-words increased in the United States following the election of Barack Obama in 2008 and in Britain after the success of the Conservatives in their 2010 national election. In Canada, use of we-words spiked in February 2010, during the Winter Olympics, which it hosted.

  THE IMPORTANCE AND LIMITS OF GROUP IDENTITY

  We-words tell us about group identity. By analyzing them closely, it is possible to get a sense of the speakers’ connections with others. Knowing if someone feels that they are an integral part of a marriage, a family, their company, their community, or their country is valuable information. In fact, many psychologists and sociologists have devoted their careers to understanding group identity. People who identify with a group are simply more loyal to it. They tend to rely on people in their group and distrust people outside it. Group identity has frequently been associated with stereotyping, prejudice, and discrimination. Many family feuds, regional battles, instances of genocide, and world wars have been fought where group identities served as the rallying cries.

  If you are interested in groups and how they work, knowing about the people’s identities is a good first step. But it’s only part of the story. Just because I am a member of various types of groups doesn’t mean that I necessarily like all the people in each group, would work well with them, or would even like to eat lunch with them. To really understand how groups work, it is important to track how they communicate with each other.

  BEYOND GROUP IDENTITY: CAPTURING GROUP DYNAMICS

  It’s easy to forget that the reason language evolved was so we could communicate with other people. To appreciate how a group functions, look at how members of the group interact with one another. When people in the group are talking, are other members of the group listening? When they are working on problems together, do they tend to talk in similar ways?

  In the chapter dealing with close relationships, the topic of language style matching, or LSM, was discussed. LSM measures if two people are in synch with one another by looking at how similarly they use stealth words—personal pronouns, articles, auxiliary verbs, and the like. The more similarly two members of a couple use stealth words when talking, e-mailing, or IMing, the more the couple is “clicking” and the more likely the couple will still be together a few months later.

  As with couples, all groups can vary in the degree to which they are cohesive, in synch, or clicking. You may have recently had dinner with a group of friends and marveled at how your group was in perfect harmony. Meeting again a week later, the same group may have been completely out of step for no apparent reason. Most of us have had similar ups and downs with group meetings at work, at home, or in online chats. By measuring each person’s relative use of stealth words in these interactions, we can begin to get a sense of how well the group is working and even identify where the problems might be.

  SMALL WORKING GROUPS: SOLIDARITY AND PRODUCTIVITY

  Imagine that you are thrown into a work group that must come up with a solution to a complex task. As part of a business school exercise, your group must decide if it would be a good investment to buy LuProds Manufacturing Company. LuProds, you learn, has designed a new gear system for bicycles that could revolutionize the bicycle business. No one in your five-person work group knows anything about LuProds or the bicycle business. Over the next four hours, however, your group must write a report that lays out your recommendations backed up with solid information. To add to the pressure, there are ten other groups doing the same task and it is imperative that your group’s performance is among the best.

  To succeed on a task like this, each group member must take on an independent task. Someone must learn something about the bicycle business, the gear business, the economic history of LuProds, its competitors, the quality of LuProds’s management team, and potential other LuProds buyers. The team members must work closely together in a coordinated way to come up with an overall recommendation based on all the information available.

  What makes for a good working group? Even within the first few minutes of your group being together, we can begin to predict how it will do on the task. This may not surprise you but the ways the group members use function words with each other is one of the best predictors. Variations of this project have been run in business schools and psychology departments around the world. In two studies, my colleagues and I calculated style matching scores for each person in the group. So, for example, if there were five people in the group, we could determine the degree to which each of the five people matched with the average of everyone else in the group. The average style matching, or LSM, score for each group tapped the degree to which the entire group was talking in similar ways. The more group members talked alike, the more cohesive the group. In other words, higher LSM scores reflected more tightly knit groups.

  More important, style matching was related to how well the group performed. If people in a group are using similar function words at similar rates, the team performs better. Even before the groups turned in their answers, our language analyses could make reasonable predictions about the ultimate success of the different groups. At least in controlled laboratory settings, the use of stealth words such as pronouns and prepositions among the people in the group showed that the participants were all on the same page—they all shared common assumptions about the task and what needed to be done.

  LARGER REAL-WORLD GROUPS: WIKIPEDIA

  Lab studies of group behavior tell us how people might behave in more natural settings. How many times in your life will you be in a group with four twenty-year-olds making a major business decision on a topic you know nothing about over a four-hour period? Does language style matching predict success in actual real-world groups? The answer is in Wikipedia.

  Created in 2001, Wikipedia is an online encyclopedia-like information source that has several million articles. Many of the articles are expertly written and have been carefully edited by dozens, sometimes hundreds of people. For the most commonly read articles, an elaborate informal review takes place. Often, a single person will begin an article on a particular topic. If it is a topic of interest, others will visit the site and frequently make changes to the original article. In many ways, a Wikipedia article is really two articles. The casual visitor sees only the final product. However, by clicking on the “discussion” tab, it is possible to find conversations among the various contributors. These interactions are often quite professional, sometimes detailed and thorough, but occasionally downright rude and nasty.

  What makes Wikipedia amazing is that it reflects a form of intellectual democracy that actually works. Although many scholars would be loath to admit it, Wikipedia is often the first professional site they visit to learn about a new topic. It is also a rich source of language data for someone interested in how groups work. In fact, it so intrigued one of my graduate students, Yla Tausczik, that she enlisted the help of the University of Texas at Austin’s supercomputer team to download all of Wikipedia, including the behind-the-sce
nes conversations—a nontrivial task, I might add.

  For starters, Yla identified about a hundred American cities that had relatively extensive Wikipedia entries. The cities were all midsized, ranging in population from about 500,000 to 1.5 million people. Each city’s site had been edited multiple times by at least fifty different people over several years, so that there had been lengthy discussions concerning the entries. In addition to the articles themselves, each entry had gone through a review process. Indeed, all articles are categorized by Wikipedia along a continuum from “stub” (meaning not worthy of even being called an entry) to exemplary.

  The degree to which the various editors used similar language in communicating with each other reflected higher-quality articles. Just as in the lab studies, teams of Wikipedia authors and editors who are in synch with each other—as measured by their similar use of stealth words—produce the best, most authoritative articles. Language style matching in the real world reflects better real-world products.

  EVEN LARGER REAL-WORLD GROUPS: COMMUNITIES AND CRAIGSLIST

  If we can examine the cohesiveness and interconnections of a changing group of editors on Wikipedia, why not look at real communities? Some towns are more tightly knit than others. Most of us have occasionally visited towns or neighborhoods where people appeared to be remarkably similar in their opinions, food preferences, and social behaviors. In others, virtually no one seems to know their neighbors or has a sense of the history or values of their community.