Hidden Brain - What's Hidden in Your Words

Episode Date: October 7, 2024

I. Me. My. You. He. She. They. It. To. Of. For. These are all words we use without a second thought. But psychologist James Pennebaker says if we pay close attention to the patterns in speech and writ...ing, we can understand profound things about others, and even ourselves.For more on the relationship between language and our minds, check out this classic Hidden Brain episode: https://hiddenbrain.org/podcast/watch-your-mouth/

Transcript
Discussion (0)
Starting point is 00:00:00 This is Hidden Brain. I'm Shankar Vedantam. There was a time when most detective stories featured the brilliant insights of a genius. Sherlock Holmes and Miss Marple noticed clues the rest of us didn't and solved mysteries using little more than their eyes and their intuition. In the story, The Adventure of the Blue Carbuncle, Sherlock Holmes inspects a hat that was found next to a stolen gem. He tells his friend, John Watson, that the hat's owner is a smart, formerly wealthy man who recently turned evil. He is middle-aged, out of shape, recently cut his hair, and his wife no longer loves him.
Starting point is 00:00:47 All of this from simply looking at the hat. In the real world, crime solving today relies less on sharp eyes and intuitive leaps, and more on the science of forensics. more on the science of forensics. Suspects leave fingerprints at scenes of crime, and tiny strands of DNA can identify one person out of billions. In recent years, psychologists have gotten into the game. They have discovered another set of clues
Starting point is 00:01:21 that can give us a window into a suspect's mind. The source of this hidden information? Language. The words we use in both speech and writing reveal a lot about us. They can give us insight into a potential criminal, yes, but also whether someone is depressed, when a person is lying, even how long a lover will remain in a romantic relationship. What your pronouns and prepositions really say about you. This week on Hidden Brain. Have you ever gotten a confusing message from a friend or family member? Maybe a text or email.
Starting point is 00:02:30 You read it and realize, I have no idea what this person is actually feeling. If only there was a way to decipher what she really means. In recent years, scientists have tried to do exactly that. They find that if we pay close attention to the patterns in speech and writing, we can understand profound things about others and even ourselves. At the University of Texas at Austin, psychologist James Pennebaker has spent decades studying how the words we use can be an X-ray into our minds. Jamie Pennebaker, welcome to Hidden Brain. It's nice to be here,
Starting point is 00:03:05 thank you. Jamie, early on in your career you were studying how writing can help people get over traumatic experiences, but as part of that research you noticed something. The language people use changed as their mental states changed. Can you tell me how this discovery unfolded? I was trying to figure out why is it that writing about an upsetting experience can improve people's health? And it occurred to me, I needed a more objective way to look at essays.
Starting point is 00:03:34 And so I looked for a computer program to analyze the text. And I ended up asking one of my graduate students to help put together a computer program that can analyze language. And it was that program that allowed us to all of a sudden start to see writing samples in ways that I had never imagined we'd be able to do. The computer program that Jamie developed could analyze a text, breaking it down by word choice, parts of speech, and meaning. Take the sentence, it was a dark and stormy night.
Starting point is 00:04:12 The program would categorize the word it as a pronoun. The word was would be counted as a verb. Stormy is an adjective. After developing this program, Jamie then brought volunteers into the lab and asked them to give him writing samples. He analyzed their writing through the computer program and then compared the results to the personalities of the writers. He came to several astonishing conclusions. One of the first was that men and women employ language differently.
Starting point is 00:04:51 Jamie assumed that men are more likely than women to act in boastful ways. So he expected they would refer a lot to themselves. I. They would use a lot of words like I. Me. Me and my. My. Jamie figured women would use a lot of words like I, me, and my.
Starting point is 00:05:06 Jamie figured women would use words like we more often. The most striking early discovery was that looking at men and women, they used language in ways that made no sense to me. I found that women used I words, I, me, and my, much higher rates than men. Me, my. I expected big differences in the use of we, but in fact men and women use we words at the same rates. We.
Starting point is 00:05:42 Us. Our. And there were these other parts of speech that were just very different and I had no idea what they meant. And over the years we did more and more studies and part of it was men and women are paying attention to the world differently. Women are looking and are more interested in other human beings and by definition definition, if you're interested in human beings, you have to use pronouns. He.
Starting point is 00:06:09 Words like he, she, they, we, I, etc. She. They. And if you're interested in objects and things, which men tend to be, you use nouns and articles. A. A. N. V. V. use nouns and articles and prepositions to a four. And so all of a sudden I realized the world was way more complex than I ever thought.
Starting point is 00:06:42 I confess I was skeptical about these conclusions. There are all kinds of pop psychology claims about gender differences that mostly rely on stereotypes about men and women. But then Jamie told me about another dimension of this research. It had an experimental quality to it. It tracked individuals getting supplements of testosterone, the primary male sex hormone. Testosterone seems to affect the way we see our world. And several years ago, working with Jim Dabbs and others who had been doing work on testosterone,
Starting point is 00:07:16 we were able to study a person who was transitioning from female to male and was taking periodic injections of testosterone. I was studying this person's diaries over time, and trying to find out when this person took the testosterone injections, did their language change? I was telling a friend of this,
Starting point is 00:07:40 and this friend was an academic, and he was saying, wow, that's weird, I've been taking testosterone as well because he had moved to a farm and he was older by this time and he started taking testosterone to increase his upper body strength. And I asked him, would he mind sharing when he had taken these shots and also sharing his outgoing emails for the previous year.
Starting point is 00:08:06 So for both of them, I could look at when they took shots, which was usually about once every three weeks, and how their language changed from before taking the testosterone to the three weeks afterwards. And I found the same kind of changes. And it was not what either of them expected at all. And that was when the individuals took the testosterone shots, they started using pronouns,
Starting point is 00:08:32 he, she, they, etc. at lower rates. It's almost as though they became less interested in other human beings. Testosterone then didn't make people talk more about power or aggression or anger or anxiety or anything. There was no differences there. Only this lack of interest in humans after taking a testosterone shot. Jamie had discovered that the language people use reveals something about what is happening inside their heads. Words reflect not just what we are trying to communicate, but how we see the world, how we see each other. Fascinated, Jamie started using his language analysis tool to study other facets of human behavior. He quickly noticed a group of words that was extremely important, but these are not words
Starting point is 00:09:30 most people would think are important. I focus a lot on a group of words called function words. These are the words you learned in high school. They're the most boring words in language. They include pronouns. He. She. They. Negations, no, not, never, prepositions, auxiliary verbs, am, is. All of these are words that don't have any solid meaning like a noun does or a verb or adjective does. What I started to discover was it
Starting point is 00:10:04 was these function words, which are telling us how people are seeing in the world, how they're negotiating with others, how they're thinking about themselves and others. And with this, we can start to get a picture of their personality, if you like. So for example, I could go out to your listening audience, ask them to send me their, you know, their last two or three emails that they sent out. And people who use a high number of pronouns like he, she, they, you, etc., those people are more socially connected.
Starting point is 00:10:40 They're more interested in other human beings. She is one of those people who will maintain composure in the situation. I can't believe he did that. You're super smart, you're funny, you know what you're doing. I trust you. Or another interesting one is people who are anxious. Now, a person who's anxious will use anxiety related words, words like worry, upset, nervous, but they're also very much self-focused
Starting point is 00:11:15 and they tend to use pronouns like I, I, me, and my, I refer to as I words. I at much higher rate than people who are not anxious. And in fact, all of us when we're anxious, we start using more I words. And the same thing is true of depression. Any emotion that makes us look inward, we will start using I words more. And even if we're physically sick or in pain, we use I words more. In other
Starting point is 00:11:46 words, I words are reflecting self-focus. Hmm. Hmm. I mean, what's striking here, Jamie, is that I think when most people think about language, they think about words that have, you know, an affective quality. I'm excited, I'm happy, I'm sad, I'm I'm moody. But what you're really finding is that it's almost the connective tissue between words that actually reveals the inner workings of our minds. That's exactly the way I think about it is these words, what I love about them is they're invisible. We can't hear them. We can't control them very well, but they are revealing parts of ourselves that we just didn't know that we were, you know, that we were spilling those beans.
Starting point is 00:12:35 What is striking about these function words is how few there are and how often we use them. They're not that many in the English language. There are only about 200 common ones. But we use these function words at very high rates. For example, these pronouns, prepositions and articles, etc. account for about 60% of all the words we say, we hear. Wow. The majority of words you and I have spoken have been function words in this conversation and we don't even know it. I confess I was skeptical about these conclusions. They're invisible.
Starting point is 00:13:09 We can't hear them. We can understand profound things about others and even ourselves. All of these function words together are so powerful because they are telling us about our social orientation if we're looking inward. A lot of it is telling us where we're paying attention and how we're thinking and what our connections with others are. So one thing that I want to point out and flag is that the differences in the rate at which these words are used can be subtle. So it's not as if you can listen to someone's conversation intuitively and pick up that
Starting point is 00:13:52 the fact that they're using more of these kinds of function words or fewer of these function words. In some ways, these have to be analyzed almost at a statistical level. That's right. A good example is people who are depressed use I words at higher rate than when they're not depressed. And when they're depressed in, let's say, a regular conversation, they might use maybe six percent of their words might be I words.
Starting point is 00:14:20 If they're not depressed, it could be four percent or five percent., it sounds like the difference between 4% and 6% is trivial, and on one level it is, but as it happens, it's a huge statistical effect because people don't vary a lot in terms of their use of these. So it is a really powerful marker that we just can't hear. On one level, given that language is a product of the mind, it's not surprising that our words should reflect the patterns in our thinking. But Jamie is reverse engineering the process. He is looking at our words and inferring what is happening in our minds. Used this way, words become a sort of x-ray into our thinking.
Starting point is 00:15:06 Jimmy cautions that this is a diagnostic tool, not a magic wand. You're probably wondering, well, if I could get people to, let's say, if we know that I words are associated with depression, can we get a depressed person to stop using I words and they'd be happy? LAUGHS
Starting point is 00:15:24 By the way the answer is no. It does not work that way. Yeah so it's not as if you can actually manipulate the words you're using temporarily and suddenly dramatically change your mind states. It's really the words that you're using automatically almost unconsciously that are really telling you something about what's happening inside your head. That's exactly right. So for example, one thing that happens is when people assume a leadership role, they use I words less, but they tend to use we words, we, us, and our more.
Starting point is 00:15:58 In other words, you take on the mantle of leadership and you are now looking out at others. You're not paying attention to yourself, which by the way is opposite of what most of us think. But years ago we did some studies where we were able to induce people to use we words more. Did they become leaders? No. Did they feel more connected to others?
Starting point is 00:16:19 No. It made no difference. But if we induced them to become leaders, they then started using I words less and we words more. So about a decade ago, Jamie, you became fascinated by the story of an explorer in rural Australia. Tell me the story of Henry Hellyer. So Henry Hellyer was a British explorer. He was born in the late 1700s in England and
Starting point is 00:16:52 then emigrated to Australia and he was interested in discovery and he went to Tasmania. Tasmania is an island about the size of England in southern Australia, about 150 miles from from the Australia mainland. And he had written in root on this trip, he had written several long journals, he had written lots of letters and so forth. So Henry Hellyer was known as a very clever and intelligent man, and he also wrote about his adventures as he was having them. One entry he noted on August 9th 1827 went, We have had a very stormy night. Just before dark, the men saw a tree falling directly towards my tent. I had just time to run out being apprised
Starting point is 00:17:38 of the danger by their shouting, and the next instant my poor tent and porch of bark were laid prostrate. Had it been dark, I could not have escaped." So it sounds like he was having a real adventure here, Jamie. He was, and his relatives who got these letters picked that up as well. But then things, two or three years after that, started to go oddly. There seemed to be some tension between him and the other men on his party. He then died. And in the years afterwards,
Starting point is 00:18:12 there was increasing questions about this. Was it suicide? Was he murdered? And I'll be honest, I had never heard of Henry Hellyer, but I often get emails from people all over the world who have some interesting case they're trying to work on. And one of the people that I heard from was a woman by the name of Gwyneth Daniel, who was interested in Hellyer. And she wanted to know what I thought of his letters and where his letters consisted with suicide.
Starting point is 00:18:54 Jamie analyzed Henry Hellyer's letters. He put them through his computer program and counted the function words and non-function words. He also drew on some earlier work he had done comparing two groups of poets. The first group was writers who had died by suicide. The writers in the second group were roughly comparable to the first group, but had not died by suicide. Did the first group refer more often
Starting point is 00:19:19 to depression and death in their poems? In fact, there was virtually no difference in their content. The big differences surrounded pronouns, specifically I words, that suicidal poets used I words at much higher rates in their poetry than non-suicidal. That was what I was interested in looking at Hellyer's writing. If he was depressive in his writing over time, did his use of I words start to increase? And that's what we looked at. And indeed, they increased dramatically.
Starting point is 00:19:51 In the first two or three years of his explorations, his average number of I words was about 1% of all of his words. He was so immersed in the world around him. And in the last year, in the last maybe three or four writings, they averaged a much higher rate between eight and ten percent of all of his words. He was deeply, deeply self-focused, which tells me there was a very good chance he died of suicide.
Starting point is 00:20:27 side. Jamie had discovered that the words people use can give us insights into their minds. When we come back, Jamie applies his research to a complicated murder case. First though, a quick note to say that if you or someone you love is struggling with suicidal thoughts, there are people who can help. You can call, text or chat with the Suicide and Crisis Lifeline at 988. That number again is 988. You're listening to Hidden Brain. I'm Shankar Vedantam.
Starting point is 00:21:10 This is Hidden Brain. I'm Shankar Vedantam. At the University of Texas at Austin, James Pennebaker has spent many years studying what our choice of words can reveal about us. He's found subtle differences in how we use words when we are happy, depressed, angry or friendly. Often, words we think of as insignificant provide important clues about our inner states of mind. Jimmy, a few years ago you received a request asking for your expertise in a murder case in Australia that involved a woman named Kathleen Folbig. Now, Kathleen and
Starting point is 00:21:45 her husband Craig welcomed their first child in 1989. Obviously, this was a happy moment in the life of any family, but what came next was a tragedy. Can you tell me the story of what happened, Jamie? So their first child, Caleb, was born and the child was about three weeks old and one night Craig woke up in the middle of the night hearing her screaming and Kathleen was screaming that something was the matter, the baby wasn't moving. And Craig ran into the room and saw her standing over the child.
Starting point is 00:22:19 And the child wasn't breathing and they called an ambulance and the child who was only just 20 days old was dead. And the official cause of death was Sid's sudden infant death. And of course that has been known to happen to children, I think especially if they're placed on their stomachs. We know much more about Sid's today than perhaps they did in the late 80s. And I'm imagining that Kathleen and Craig experienced this as a great tragedy. Tell me what happened next. Did they recover? Did they have more children, Jamie? They did, and ultimately
Starting point is 00:22:55 they had three more children, and each child died usually within a few months of birth. The last one survived for one year, about a year and a half. So this was over essentially a 10-year period. It's difficult to imagine what Craig and Kathleen Folbig endured during those 10 years, as they brought children into the world and lost them again and again. Eventually, the weight of these tragedies tore the couple apart and they divorced. Kathleen moved out. Sometime later, as he was cleaning the house, Craig stumbled upon several of Kathleen's
Starting point is 00:23:42 diaries. They revealed a conflicted woman. She was excited by motherhood and loved her children. But she also felt frustrated, resentful, fearful, and jealous. Reading the entries, Craig became alarmed. Some entries implied that Kathleen was struggling with unresolved mental health issues. In one entry she wrote, Another, referring to her daughter Sarah, said simply, "...with Sarah, all I wanted was her to shut up." And one day she did.
Starting point is 00:24:36 Craig took the diaries to the police. Kathleen Folbig was arrested and tried for the murder of her own children. She maintained she was innocent, but she was found guilty, based in part on the diary evidence. Years passed. Kathleen appealed her conviction, but was turned down. That's when Jamie got involved. I was contacted by her lawyer.
Starting point is 00:25:06 She'd already been in prison for 20 years. And her lawyer felt she had been railroaded in terms of her conviction. And I should tell you, I knew nothing about this case. But the lawyer asked, would I be willing to look at her diaries? As it happened, she had kept diaries and there was a huge amount of entries. There were like 250 entries, tens of thousands of words that she had written. So they were quite extensive. And the lawyer wanted to know, is it possible to get a sense of her psychological state?
Starting point is 00:25:42 And so what happened was they ended up sending me all of her diaries that had been transcribed. And I went about analyzing them, trying to get a sense of just psychologically who she was, what was going on in her life. Was there any signs of instability? Were there any signs of hostility? Any signs of kind of premeditation
Starting point is 00:26:06 or anything like that. So that's the way I approached it. Now again, we talked earlier about how sometimes when lay people think about language and words, we're often very focused on is someone saying that they're happy, is someone saying they're sad, is someone saying they're anxious, is someone saying they're depressed.
Starting point is 00:26:23 Your approach has always been to sort of analyze more of these function words that are happening in some ways invisibly. And I'm imagining there probably were a lot of these function words if you had tens of thousands of words to work with. That's right. And I can also do a reasonably good job
Starting point is 00:26:40 identifying big changes in psychological state. So for example, we have done some research on presidents who go to war or the Boston bomber in terms of his decision to blow things up. When people are planning to do something nefarious, what they often do is they start trying to act innocently, which is a hard thing to do. And what they do is they become more deceptive, they become less self-reflective, they are holding things back.
Starting point is 00:27:15 Or if somebody is showing signs of deep depression, there are also markers of language. In other words, language can pick up if a person is becoming deceptive, if they're becoming depressed, if they're becoming manic, if there are major changes in their social behavior. All of these are available to us through the way they're using words. In one of your earlier studies, Jamie, you asked volunteers to come into the lab and make both true and false statements about their views on a hot button topic, in this case, abortion. And your computer program was able to detect people's true opinions 67% of the time, whereas human
Starting point is 00:27:54 judges got it right only 52% of the time. What were the language clues you were examining in this study? There were several, but the one that had the most potency were I words, I, me, and my. When somebody is telling the truth, they use I words. When a person is holding back or lying, they are hiding themselves and they suppress their use of eyes. So they're trying to get people to pay attention to something else or pay attention to some other event or object.
Starting point is 00:28:40 In other words, we change the way we're paying attention and hopefully the way other people are paying attention when we're being deceptive. So again if language and words are reflecting what's happening inside our minds and I am trying to hide something, I'm actually trying to pay less attention to myself to change the focus and in some ways that comes out in the words and language we're using. I'm imagining that being able to tell whether people are being honest or they're lying is kind of useful when it comes to a murder investigation. It does. The big problem is there is no good method that's wildly reliable that can identify
Starting point is 00:29:20 deception versus telling the truth. We can do better than chance, but we can't, there's no such thing as a great lie detector, and if anybody says they have one, be very, very careful. Right. I mean, in the study that we just talked about, the computer program was better than humans, but it was not perfect. It was not detecting dishonesty 100% of the time. That's correct. O, did she change in her language as it got closer to the child's death? Because if she was actually planning to kill the child, you would pick up a signal. There's no question about that. Her language was exactly the same as it had been for several weeks and months before.
Starting point is 00:30:33 And you would expect afterward changes there that might be associated with sadness or grief or maybe relief, but you know there wasn't much of a signal there either. Hmm. One of the things that you noticed is that after the death of her first son Caleb, Kathleen's diaries seem to repeat words at an extremely high rate. What was the conclusion you drew from this, Jamie? When people repeat, what they are doing is they are not paying attention to what they're saying. They're tripping over their words. They are consumed by other thoughts. And that can tell us all sorts of things. We'd have to ask ourselves,
Starting point is 00:31:27 why are they not focusing on the topic? My guess was that she was, you know, really upset by the death of her child and then later her children. You could imagine having a child and then a second child and then a third die this way and no clear explanation for it. The wording was really sparse and there was this sense that she was just emotionally disengaged
Starting point is 00:31:58 and my sense was this was clearly a mark of postpartum depression, that she just wasn't psychologically connecting. In one of your studies that used tens of thousands of words as inputs, you found that about 10% of the words were what you called cognitive words, and in Kathleen's writings the average was much higher about 15%. What did this tell you about her mental state, Jimmy? So these words, cognitive process words, these are words like think, believe, wonder. And if you are nervous or concerned you hedge a lot more.
Starting point is 00:32:42 You don't say it's cold outside. You might say I You don't say, it's cold outside. You might say, I think it's perhaps, maybe it's cold outside. These words are common when people are ruminating or trying to work through issues. They're in a situation that they don't understand very well. They tend to be insecure. And all of us increase in those words
Starting point is 00:33:03 during periods like that. Now isn't it possible that rumination could have been a sign of guilt? It could be but I would say it more likely it's just a sign of uncertainty and confusion about life and it's interesting during her pregnancies and actually in her diaries throughout her diaries, her use of these cognitive words were high. And I think this was a marker of who she is. Hmm. We talked earlier about the use of I-words, words like I, me, and my. What did Kathleen's writings reveal about those words,
Starting point is 00:33:45 if any? Not much. Her I words were a little bit above average, but not much higher. And there weren't periods where they dropped a lot, which I would have expected had there been deception going on. And you know, at some points they were a little bit higher,
Starting point is 00:34:03 but that I think probably goes along with, I'm gonna guess kind of a residual anxiety and depression. But again, it wasn't so striking that I found that to be out of a normal range. So, taking all of this together, what were your conclusions about what actually happened in the Kathleen Folby case, Jamie? So, I felt the only thing that I could say was I didn't see high rates of anything that
Starting point is 00:34:31 would suggest levels of anger or hostility. That was not true in anything that I saw. Her anxiety levels and negative emotion words were a little above average, but they weren't off the charts. I saw no signs of deception. She was remarkably stable across her her writing. I think what you saw is what you get with her. From my perspective there was nothing there that would raise any red flags. To be clear, Jamie did not conclude that Kathleen was innocent. He only concluded that if she had killed her children, her writing style would probably have reflected her inner storms.
Starting point is 00:35:21 His analysis did not find evidence for such storms. Jamie's report was one of two new pieces of scientific evidence in the case. Besides getting a language analysis from Jamie, Kathleen Folbig's lawyers enlisted scientists to sequence her DNA and the DNA of her children. They discovered an extremely rare genetic mutation that is linked to a heightened risk of sudden death in infancy. The appeals court judges considered that evidence, along with Jamie's findings in Kathleen's diaries. Two decades after Kathleen's original conviction,
Starting point is 00:36:00 they reached a new conclusion about the case. Kathleen's lawyers gave Jamie a link to watch the court proceedings. And so I watched this very formal proceeding, everybody in these very fanciful robes. The Chief Justice gets up and he reads this brief discussion of the case. In relation to the diary entries, it may readily be understood how certain entries viewed in isolation had a powerful influence on the original
Starting point is 00:36:29 jury in a manner adverse to Miss Folbig. Viewed in their full context however as they must be and informed by the expert psychological and psychiatric expert evidence referred to extensively in the report, the diary entries were not reliable admissions of guilt. Thus while the verdicts at trial were reasonably open on the evidence then available, there is now reasonable doubt as to Ms. Fulvig's guilt. And then he says the court unanimously agrees that she should be exonerated. It is appropriate that Ms. Fulvig's conviction be quashed. The court so orders and directs the entry of verdicts of acquittal.
Starting point is 00:37:08 The court will now adjourn. And I remember watching that and tears coming down my eyes. And even as I'm telling you this, I get really, I get really choked up. Hmm. I mean, in some ways, it really points to the power of this research. I mean, it's obviously fascinating in its own right at an intellectual level, but really, this is an example of where the real world effects of this research are potentially transformative. It really is, and it just makes me proud to be able to make some kind of contribution to something like this. Most of us are not asked to investigate a string of strange and potentially sinister
Starting point is 00:38:06 deaths, but all of us constantly evaluate the mind states of others. When we come back, what language can reveal about students' performance in college and what it can tell us about whether our romantic relationships are going to last? You're listening to Hidden Brain. I'm Shankar Vedanta. This is Hidden Brain. I'm Shankar Vedanta. Every day, the words we use reflect who we are. Language is the medium through which we communicate our innermost thoughts and feelings. Psychologist James Pennebaker has studied how the words we use can reveal things about our love lives. Jamie, you once studied people who were on speed dates.
Starting point is 00:38:59 You recorded their conversations with potential mates and then transcribed them. What were you looking for and what did you find? What had happened was I had never been interested in studying speed dates but there had been a wonderful study done out of Northwestern. Eli Finkel and his colleagues, Paul Eastwick and others had done some speed dates where they brought these were heterosexual speed dates, they brought men and women in and each person would talk to multiple other people and then afterwards they would rate each person and they were trying to find out which people were likely to go on a date subsequently. So the study was published and one of my graduate students, Molly Ireland, told me about the study
Starting point is 00:39:45 and she and I were studying something that we called language style matching, a degree which two people's language connect with each other, that they use pronouns at similar levels, articles, prepositions, et cetera. And we wondered, would this be related to the speed dating? So Molly contacted the Northwestern group, asking, would it be possible for us to analyze their data to see if the language of these people in the speed dates
Starting point is 00:40:15 could predict whether or not they'd go out on a subsequent date? And what we discovered was that if you looked at how much the two people matched in their language style, we did a pretty decent job at predicting who would subsequently go out on a date. Specifically, we did better than the people themselves in predicting if they'd go out on a date. And of course, one reason is, is very often you get one person who said,
Starting point is 00:40:49 wow, that date went great. And the other person said, not so much. And what happens is this style matching is a marker of the degree to which the two of them really are on the same page, that they're clicking. In some ways you told me earlier, Jamie, that there were differences between men and women in terms of the language they used, the words they used, especially the function words they used. In some ways, especially I think with heterosexual speed dating, it does raise questions, doesn't it? Because in some ways, this research is suggesting that the closer people are, the more likely they are to go out on dates. And the earlier research had found
Starting point is 00:41:33 that in general, men and women often use different kinds of words. It doesn't portend well, I suppose, for most speed daters. You know, I think we have to think about how conversations work. That sometimes I'll speak like a professor, sometimes I speak like a lover, sometimes I speak like a father. And my language changes in all those situations. And if I'm going on a speed date and I'm really interested in the other person, I will start to talk like them. And if they're really interested in me,
Starting point is 00:42:05 they'll start talking like me, and the two of us will come together in a way that we might not do under different circumstances. Interesting. You've also done a follow-up study featuring college freshman couples, and you look to see the text exchanges
Starting point is 00:42:23 they had with one another as a potential predictor of whether the couples would still be together three months down the road. So this would be using language in some ways as a predictor for how long relationships might last. Tell me about the study and what you found. So this was a study that one of my former students, Rich Slatcher, did, and it was a wonderful study where he was able to find young couples that were dating. They had to have been together a certain amount of time, and they had to agree to give 10
Starting point is 00:42:54 days of their instant messages or their text messages to us to analyze. We went through and we analyzed the degree to which the two people in their text messages were matching in terms of this language style matching. And then we used that to find out would this predict how long the relationship lasted. What we found was that we could split the couples in terms of their language style matching scores were above average or below average. And those that were above average, almost 80 percent were still together three months later. And those that were in the bottom half of style matching, only about half of them were still together three months later. In other words, here is an interesting marker of the quality of their relationship, but
Starting point is 00:43:48 the thing that was most interesting was it was completely unrelated to their self-reports. We had asked them beforehand, how strong is your relationship, how likely is it you think you'll be together in the next six months, etc. And it turns out their judgment to their relationship was completely unrelated to whether or not they were still together three months later. This language style matching, however, was a really good predictor of it. You've also studied couples who have broken up,
Starting point is 00:44:24 and this was a study that involved the social media site Reddit. What did you do and what did you find, Jamie? This was the first study that really introduced me to the power of social media. Reddit is a particularly interesting source because tens of thousands of people go there and they will often write about how they have recently broken up. The majority have been dumped by somebody else and sometimes they get you know emotional support from others sometimes they just want to reveal it. The average person actually had posted
Starting point is 00:44:58 between one and a few times in the breakup subreddit and they'd posted over a hundred times and all these other subreddits in the breakup subreddit, and they'd posted over 100 times in all these other subreddits in the months before and the months after. And we got all of their posts in all of their subreddits from the year before the breakup to the year afterwards. And what we found was that in the months before, starting about three months before the breakup, their language starts to change. They become more anxious, they start using more I words, their cognitive process words
Starting point is 00:45:28 go up, they become less logical and formal in their language, less structured, almost as though they're under increasing amounts of stress. And then they have their breakup and they post about it. And at this point, they are low in logical thinking, their anxiety and negative emotions are high, their I-words are high. And then the after effects, we can track lasting six months. People going through a breakup are profoundly affected. They are not good at work. They even talk about how they're making errors,
Starting point is 00:46:05 they're coming in late, they're drinking too much, et cetera. It really gave me a different perspective. But in some ways, Jamie, what you're finding is that language is both a predictor of when a relationship might come apart, and then it also in some ways is an X-ray into what's happening in people's minds after a relationship comes apart. That's exactly
Starting point is 00:46:30 right and one thing that was interesting was how many people when they were writing about their breakup they would say this came out of nowhere I had no expectation. Yeah. But I think they kind of did. You know, their numbers showed that they were thinking differently. They were orienting their lives in a different way in those months beforehand. You've also done some really unusual work predicting how successful students are going to be in college based on their college admission essays. Tell me about that work, Jamie. When people apply to universities, they write an admissions essay. And I was curious, would it be possible to analyze their admissions
Starting point is 00:47:15 essays and see if it would predict how well they would do at the university? So working with our admissions office, I was able to analyze about 50,000 admissions essays from 25,000 students who enrolled at the university. And we were curious, were there features of language in their admissions essays that could tell us something? We discovered that there was this group of words, these are all function words, that naturally hang together in an interesting way. We now call it analytic thinking. If you are thinking in an analytic way, you're thinking in a logical, formal,
Starting point is 00:47:59 hierarchical way. You're using lots of articles and nouns and prepositions, but you're not using many pronouns or auxiliary verbs. And what we found was that the more analytic a person's essay was, the better they did at the university. So four years later, people who were high in their analytic essays, they made better grades Whether they were in the business school philosophy nursing fine arts. It didn't matter I'm wondering Jamie after doing all this work. Do you find yourself? You know perusing people's language as work, do you find yourself, you know,
Starting point is 00:48:45 perusing people's language as they talk to you or write to you? Do you find that as you're reviewing emails from friends, you can't help but say, you know, huh, that's an unusual number of pronouns and prepositions? Usually not. And it's always funny, but I give a talk on this, you know, I can tell people are so self-conscious. Here's the good news. People can't tell unless you talk in a weird way. And you can't believe how many emails that I've gotten over the years.
Starting point is 00:49:18 Dear Dr. Pennebaker, just wanted to say the work you do is good. Thinking about such and such, and they'll have no first person singular pronouns. And I'm thinking, why? And then sometimes someone at the end will say, it took me an hour to write this email so I didn't use any I words. And what it does is it makes them sound deranged.
Starting point is 00:49:49 Psychologist James Pennebaker is the author of the secret life of pronouns, What Our Words Say About Us. Jamie, thank you so much for joining me today on Hidden Brain. It's been a joy. Thank you. My My I Me My Be my I
Starting point is 00:51:02 Me Me My One of the unusual things about Jamie Pennebaker is that he is responsible for two major discoveries in the course of his psychology research career. We've talked about the first in today's show. In our companion story, we explore the other dimension of Jamie's work on the power of what he calls expressive writing to help us think through difficult challenges in our lives. If you have ever
Starting point is 00:52:05 experienced a breakup or lost a job or suffered a serious illness, that episode is for you. Jimmy has found surprising and unusual effects that come from expressive writing, but he has also found that some kinds of writing have psychological benefits, while other forms have no effects or can even be harmful. I learned a lot from that conversation. I think you will too. To listen to it now, please go to the episode titled Dear Diary in our subscription feed, Hidden Brain Plus. If you are not yet a subscriber, please go to apple.co. slash hiddenbrain on your iOS device.
Starting point is 00:52:42 If you're using another kind of phone or tablet, you can sign up via our Patreon page at support.hiddenbrain.org. In either case, you can get a free 7-day trial. Your support helps us build more episodes like this. If you're already a supporter, thank you. You have immediate access to the Dear Diary episode right now. Again, if you'd like to give the subscription a try,
Starting point is 00:53:06 please visit apple.co. slash hiddenbrain or support.hiddenbrain.org. Hidden Brain is produced by Hidden Brain Media. Our audio production team includes Annie Murphy-Paul, Kristen Wong, Laura Quirell, Ryan Katz, Autumn Barnes, Andrew Chadwick, and Nick Woodbury. Tara Boyle is our executive producer. I'm Hidden Brain's executive editor.
Starting point is 00:53:36 Special thanks to Liza Gutstein Katz for voice acting and singing in this week's episode. I'm Shankar Vedantam. See you soon.

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.