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

by James W. Pennebaker

  Around the year 2000, I was contacted by Professor James Dabbs, a world-renowned expert on the psychological correlates of testosterone. A tall, distinguished southerner with a polite, regal bearing, Jim had measured testosterone levels of murderers, lawyers, actors, priests, academics, and others—always with a twinkle in his eye. Much of his work examined the links between aggressive behaviors and testosterone levels. (During one serious conference on hormones and behavior, Jim started his talk by looking around the packed lecture hall and snorting, “There’s not enough testosterone in this entire room to rob a single liquor store.”)

  Jim had received a letter from a biological female who was undergoing therapy to become male, a female-to-male (FTM) sex change, or, more accurately, gender reassignment. The person, GH, was twenty-eight years old and had been involved in gender reassignment for three years. In addition to a double mastectomy, GH received injections of testosterone every two to four weeks. GH had read about Jim’s research and wanted to know if he would like to study GH’s diaries, which he (GH) had been keeping for several years. Knowing that I studied language, Jim felt I could examine the effects of testosterone on function words.

  GH was articulate and a prolific writer. He provided my research team with two years of his diaries along with a detailed record of his testosterone injections. While we were transcribing GH’s diaries, I was invited to give a series of talks in Boston. One night, I found myself in a bar talking to a gentleman who happened to be an anthropologist. In the midst of our conversation, I mentioned the GH project. My new anthropologist friend was silent for a minute and finally said that he, too, was taking testosterone injections on the advice of his physician. He was sixty years old and had been taking testosterone for four years in an attempt to restore his upper-body strength. “Really?” I asked innocently. “Do you have a record of your injections?” Yes. “By any chance do you keep a diary?” No, but he was willing to let me analyze all of his outgoing e-mails for the previous year.

  The analyses were straightforward. For both GH and the anthropologist, the words that they wrote each day were analyzed using the LIWC computer program. The word categories were then compared against the number of days since each person’s last testosterone injection. The idea was that as time passed since the last injection, each person’s testosterone levels would drop.

  The similarity of effects for the two people was striking and the overall results promising. Whether their testosterone went up or down, no predictable changes were seen in their use of articles, prepositions, nouns, verbs, and negative emotion words. However, there was one fascinating and reliable difference—in social pronouns (including words like we, us, he, she, they, and them). As testosterone levels dropped, they used more social pronouns. Think what this means: Both GH and the anthropologist inject themselves with testosterone and they now focus on tasks, goals, events, and the occasional object—but not people. And then, as the days wear on, they awake each morning and notice that there are other human beings around. And these human beings are interesting and worth talking about and talking to. And then, a few weeks later and another injection, and voilà! They’re cured. No more worrying about others.

  This study gives us a glimpse of how testosterone may psychologically affect people. Prior to receiving their results, I asked both participants how they felt that testosterone affected their language. GH was certain that the injections gave him more energy, made his mood more positive, and stimulated more thoughts of sex, aggression, sports, and traditional masculine things. The anthropologist, on the other hand, was adamant that testosterone had absolutely no psychological effect on him whatsoever. Both were wrong.

  What does a study like this tell us about sex differences and the biology of function words? Only a little, I’m afraid. Injections of testosterone did not change GH’s sex or the anthropologist’s masculinity. There is no evidence that this hormonal assault had any effects on the ways either person understood, thought about, or categorized their world. It is significant, for example, that not a hint of changes in articles, nouns, verbs, cognitive words, or prepositions occurred. Finally, this was a study on only two people in unusual circumstances, meaning we should be cautious in generalizing the findings.

  On a broader level, however, the testosterone study, along with the various gender-language projects, tells us a little about the ways men and women see and understand their worlds. Men are less interested in thinking and talking about other people than are women. The effects may be influenced by testosterone, but quite frankly, the testosterone findings are too preliminary to make grand pronouncements just yet.

  That there are differences in social interests between women and men isn’t exactly news. However, the fact that men consistently use more articles, nouns, and prepositions is news. OK, maybe it’s not going to be on CNN’s Headline News. But it is news because these language differences signal that men tend to talk and think about concrete objects and things in highly specific ways. They are naturally categorizing things. Further, the use of prepositions signals that the categorization process is being done in hierarchical and spatial ways. Think about when we use prepositions:

  The keys to the trunk of the car from Maria are next to the lamp under the picture of the boat painted by your mother.

  The words to, of, from, next to, under, of, and by specify which and whose keys and exactly where they are in space. There is a hierarchical structure in the sense that these are not just any keys. Rather, they are part of the category of trunks, which are part of the category of cars from Maria. Similarly, the location of the keys is specified in both horizontal (next to) and vertical (under) planes. Statistically, it is much more likely that a sentence such as this would be uttered by a man than a woman. It’s not that the man has a penchant for prepositions and nouns—rather a man naturally categorizes and assigns objects to spatial locations at rates higher than women.

  WORDS OF WISDOM: LANGUAGE USE OVER THE LIFE SPAN

  In most large surveys, researchers ask three questions: sex, age, and social class. Social class, which we will discuss soon, is usually asked indirectly with questions linked to race, education, or income. Why do survey makers want to know your sex, age, and social class if studying your political attitudes or buying behaviors? Because these three variables predict a stunning amount about people. If I know your sex, age, and social class, I can make surprisingly accurate guesses about your movie and music preferences, your religious and political views, your physical and mental health, and even your life expectancy.

  These three demographic factors are also linked to language use. In many ways, the connections between language and age are more interesting than the connections to gender. If you are born a woman, the odds are very, very high that you will remain a woman your entire life. However, if you are born a baby, the odds are astronomically high that you won’t remain one very long.

  That our language—more specifically our use of function words—changes over the life span is, in some ways, surprising. But less so if you consider how our goals and life situations evolve, as do our bodies. As we get older, we change in the ways we orient to our friends and family, sex, money, health, death, and dozens of other life concerns.

  Our personalities shift as well. Studies with thousands of people indicate that how we feel about ourselves is generally quite positive up until about age twelve. From about thirteen to twenty, our self-esteem levels drop to some of the lowest levels in our lives. Thereafter, to around the age of seventy, most people’s self-esteem gradually increases. In fact, by around the age of sixty-five, many of us feel as good about ourselves as when we were nine years old—which is as good as it ever gets. And then, self-esteem tends to drop during the last few years of life. Other studies, based on 5,400 pairs of twins, find that as people get older, they become less outgoing, more emotionally stable, and a bit more impulsive.

  The personality research is at odds with the bleak stereotypes of older people as being lonely, selfish, ri
gid, and bitter. Some of the more promising research that combats these stereotypes is being conducted by Laura Carstensen and her colleagues at Stanford University. She finds that as people age, emotion becomes a more important part of life. With greater attention to emotional states, people learn to regulate their emotions more effectively, resulting in greater happiness and fewer negative feelings. Although people over the age of seventy tend to have fewer friends, their social networks become stronger.

  A fitting observation by the main character, Lloyd, in the Farrelly brothers’ classic film Dumb and Dumber says it all. After asking an elderly woman to guard his possessions for a few minutes, Lloyd remarks, “Thanks. Hey, I guess they’re right. Senior citizens, although slow and dangerous behind the wheel, can still serve a purpose. I’ll be right back. Don’t you go dying on me!” The elderly woman cheerfully snatches everything Lloyd owns.

  How does language change as people age? We have tested this in several ways. The first was to go back to the giant blog project where we analyzed the collected posts of over nineteen thousand bloggers. The blog site that we “harvested” in 2004 was made up of people who were, on average, about twenty years old. We split the sample into three age groups: adolescent (ages thirteen to seventeen), young adult (ages twenty-three to twenty-seven), and adult (ages thirty-three to forty-seven). Despite the restricted age range, the three groups used emotion and function words very differently. Teens used personal pronouns (I, you), short words, and auxiliary verbs at very high rates. The older the people became, the more they used bigger words, prepositions, and articles.

  We also conducted a systematic analysis on people who took part in one of several dozen expressive writing studies. As you recall from the first chapter, I’ve long been involved in studies where we ask people to write about deeply personal and oftentimes traumatic experiences. Over the years, researchers from labs around the world have sent me the writing samples from studies they have conducted. For the aging project, I teamed up with Lori Stone Handelman, a brilliant graduate student who later became a highly respected book editor in New York. Lori and I analyzed data from over 3,200 people who participated in one of 45 writing studies from 17 different universities. Although the average person in the studies was about twenty-four years old, the ages ranged from eight to eighty years old.

  As you might imagine, people wrote about a wide array of topics—from sexual abuse and drug addiction to the death of a pet to not making the high school football team or cheerleading squad. Many of the essays were heartbreaking. For most, there was the sense that people put their soul into writing their stories.

  Computer analyses of the texts yielded large and sometimes unexpected findings. When writing about emotional topics, the younger and older participants used strikingly different words.

  Perhaps more impressive was the emotional tone of the young versus older writers. Consistent with the survey research, older people used more positive emotion words and younger people expressed more negative feelings. The differences became apparent by the age of forty but exploded among the oldest age group.

  YOUNGER WRITERS

  OLDER WRITERS

  Personal pronouns (especially I)

  Articles, nouns, prepositions

  Time references

  Big words

  Past-tense verbs

  Future-tense verbs

  Cognitive words (insight words)

  What makes these findings so intriguing is that all people were asked to write about some of the most troubling experiences of their entire lives. Younger people could dredge up an impressive number of dark words to express their pain. As writers got older and older, their negative emotion vocabulary diminished and their positive emotion word count skyrocketed. As you can see on the next page, the very youngest writer in this sample, who was eight years old, had a very different approach to the subject of his emotional story (Randy) than the oldest participant, who was dealing with cancer:

  Eight-year old elementary school student: My enemy is Randy he makes me mad. Outside he calls me names, he ignores me, he does not stop bothering me, so I call him names when he starts. He does not stop so I start and get him back he makes me mad so I make him mad so I try to ignore him but he keeps making me mad … My mom says he is a bad influence.

  Eighty-year old retiree. I am 80 years old, but I try to keep busy and go about life as though there is no ending. Sure I am not as fast or move as quick as I once did, but I do not feel it was the fault of the cancer and I try to do the best I can with what I’ve got. When negative thoughts try to creep in my thoughts I immediately try to replace them with positive thoughts of how lucky I am to still be here. For the 38 treatments I drove the car myself every morning alone and had a good chance of meditation and this is where I found my positive attitude from seeing the beauty of nature and changing of the seasons.

  Would you rather be eighty years old or eight? After reading these essays, the answer is not as simple as you might have thought. You get a greater appreciation of Laura Carstensen’s argument that the older we get, the better we are able to regulate our emotions. At the same time, aging allows us to look at the world in a more detached way.

  One concern that many life-span scientists have is that differences in language between, say, a group of seventy-year-olds and a group of forty-year-olds may not reflect the effects of age. Rather, all the seventy-year-olds have gone through shared life experiences that the forty-year-olds haven’t. For example, in some of the studies that were run in the early 1990s, all those who were in their seventies had grown up without television and had lived through World War II. Those in their forties generally had lived with television all their lives and were part of the baby boom generation. Maybe seventy-year-olds are just happy because they didn’t have television growing up. Probably not, however.

  One way that we bypassed this problem was to study the collected works of ten novelists, poets, and playwrights over the last four centuries who wrote extensively over the course of their lives. With this group of authors, we simply tracked their language use as they got older. Overall, eight of the ten authors showed the same age-related language patterns that we found in our other projects (the two exceptions were Louisa May Alcott and Charles Dickens).

  A nice example is from the writings of the British novelist Jane Austen. Austen was born in 1775 and died at the age of forty-two. Her first work was written when she was twelve and she continued writing until her death. Her original manuscripts, including short novels, letters, and poems, were later published in a book called Juvenilia; her last book, Sanditon, was not quite finished when she died. Compare the first paragraphs of her first and last novels.

  Jack and Alice: A Novel (from Juvenilia)

  Mr Johnson was once upon a time about 53; in a twelve-month afterwards he was 54, which so much delighted him that he was determined to celebrate his next Birth day by giving a Masquerade to his Children and Friends. Accordingly on the Day he attained his 55th year tickets were dispatched to all his Neighbours to that purpose. His acquaintance indeed in that part of the World were not very numerous as they consisted only of Lady Williams, Mr and Mrs Jones, Charles Adams and the 3 Miss Simpsons, who composed the neighbourhood of Pammydiddle and formed the Masquerade.

  Sanditon

  A Gentleman & Lady travelling from Tunbridge towards that part of the Sussex Coast which lies between Hastings & E. Bourne, being induced by Business to quit the high road, & attempt a very rough Lane, were overturned in toiling up its long ascent half rock, half sand. - - - The accident happened just beyond the only Gentleman’s House near the Lane - - - a House, which their Driver on being first required to take that direction, had conceived to be necessarily their object, & had with most unwilling Looks been constrained to pass by.

  Even at the age of twelve (and she may have been as old as fifteen), Austen was precocious. Nevertheless, the writing samples betray clues of the author’s age. In Sanditon, Austen uses far more prepositions, nouns, and c
ognitive words (e.g., induced, quit, overturned, constrained). The young Austen uses more personal pronouns and references to time (e.g., time, month, day). Although the young Austen uses an inordinate number of large words for a person of any age, it’s clear that her thinking is far less complex than the older Austen’s.

  Austen’s age-related language changes map those of the poets Wordsworth, Yeats, Robert Graves, and Edna St. Vincent Millay, as well as fellow novelist George Eliot and playwright Joanna Baillie. Shakespeare’s profile is a bit complicated but even he shows the same general patterns.

  One final note of interest about the language of age and sex: You may have noticed that older people often use function words like men and younger people tend to use them like women. This isn’t some kind of statistical fluke. These patterns hold up across cultures, languages, and centuries. Interestingly, it’s not that women start to talk like men and men just stay the same. Rather, men and women usually have parallel changes. For example, at ages eight to fourteen, about 19 percent of girls’ words are pronouns, compared with 17 percent for boys. By the time they reach seventy, the rates drop to 15 percent for women and 12 percent for men.

  We will return to the possible reasons for these patterns after discussing social class differences, which, as you will see, overlap with some sex and age effects.

  SOCIAL CLASS AND LANGUAGE

  From the very beginning of my education, I was always taught that in the United States we do not have social classes. England and India did, of course. But not the U.S. In graduate school in social psychology, there were often discussions of racial differences and inequality in the United States but never anything about social class. That’s right, the U.S. didn’t have social classes. By the time I was a young faculty member studying physical health, most conferences I attended in the U.S. would include presentations demonstrating large racial differences in blood pressure and other diseases, smoking and obesity and other health behaviors, and life expectancy. Nothing about social class. Well, you know why.