CREATING A LANGUAGE LOVE DETECTOR
Wouldn’t it be nice if there existed a device that could alert people if they were compatible? You could take it on dates with you, and at the end of the evening, the two of you would read the love detector’s “date compatibility score” on the front of the meter and decide if you should meet again tomorrow or never see each other again.
Good news. We may have a prototype for your startup company. It’s not a device, really. Instead, it just requires that you use a digital recorder to transcribe your interaction and then analyze the words to assess LSM. Oh, and for it to work, you should probably plan on having at least ten dates that are virtually identical in format. Hmmm. Perhaps a bit like speed-dating.
THE DATING DANCE
Indeed, Molly Ireland, several other colleagues, and I have found that the LSM between couples on a brief four-minute date predicts if they want to meet again. As you may know, in a typical speed-dating session, you meet and talk with eight to twelve different “dates” for a few minutes each. At the end of each date, each person rates the other person before moving to the next date. Usually the following day, the daters contact the speed-dating organizers and tell them which of the ten people they would be interested in meeting in a more relaxed setting. Most people report that the conversations range from superficial to passionate to exhausting. Sometimes people meet the loves of their lives; most times they don’t.
Lonely hearts may not always like speed-dating but researchers love it. In a recent project, about eighty daters gave permission for us to record their four-minute conversations so that our research team could analyze their word use. Does LSM in the brief interaction predict whether the couple gets together in the future? Yes, to some degree. Those in speed dates characterized by above-average LSM were almost twice as likely to want future contact as those with below-average LSM.
More interesting, however, is that we could predict which couples would get together afterward better than the individuals themselves. Immediately after each brief date, participants completed a short questionnaire about the desirability of the person they just met. The individual ratings of desirability were, of course, related to their eventually meeting, but LSM was more strongly related. Why? Whether or not two people eventually meet up is dependent on both parties. A guy might find a woman attractive but she might find him repulsive. The two must tango together. And LSM is capturing the dance—whereas the questionnaire is just assessing the dancers separately.
PREDICTING YOUNG LOVE
Let’s assume that a relationship advances beyond the four-minute speed-date and the young couple starts to date seriously. Would style matching between the two of them predict the long-term prospects of their relationship? Preliminary findings suggest yes.
Most people in passionate relationships are highly attentive to their partners. They detect subtle shifts in the other person’s moods and behaviors. Linguistically, their levels of style matching are quite high. After a few weeks or months together, some couples begin to realize that their relationship may have a limited shelf life. As their attention to their partner wanes, their degree of language style matching drops as well.
By tracking the language style matching of young dating couples, it is often possible to predict which couples are most likely to survive. For example, my former graduate student Richard Slatcher and I worked on a project with young dating couples. Rich, who is now on the faculty at Wayne State University, wanted to see how the couples talked with each other using instant messaging, or IM. He recruited eighty-six couples who reported that they IMed with each other on a daily basis and who agreed to let us analyze several days of their messages. We then compared what happened to couples who had high LSM in their instant messages with each other and those who had low LSM.
Of the forty-three couples with the highest LSM scores, 77 percent were still dating three months later, compared with only 52 percent of the couples with low scores. In other words, those couples who naturally synchronized their function words with each other were more likely to maintain their relationship over time.
Reading the IMs of the various couples painted a clearer picture of how the relationships were succeeding or failing. The more successful high-LSM relationships revealed how both members of the couple were truly interested in each other. In addition, the overall tone of the IMs was positive and supportive. The relationships with lower LSM scores, which were more destined to fail, often revealed striking patterns of detachment between the two people. For example, the following couple had low LSM scores although they claimed that their relationship was passionate and satisfying. Nevertheless, their failure to connect with each other is obvious:
HIM: hey you! how ya doing?!
HER: fine you
HIM: great. me and jim are gonna go out on the town
HER: sweet
HIM: i have to find an oil painting of a cow for my mom’s birthday and i’m gonna get my hair cut hopefully. what are you up to?
HER: not much
HIM: bout to jump in the shower too. we had a sweet christmas party last night. i got pics. pretty funny. listening to christmas music, its great
HER: good
HIM: ha ha, not very chatty today
HER: eh well … guess not
HIM: so.… you have a good week? what are you doing this weekend?
HER: yeah it was okay. working
HIM: ah … oh well at least youll have $$$$$$$$! that’s good. Im gonna jump in the shower. you think of somehting to talk about! be back in 10 min if you’re still here. jim is waiting on me to get ready. bye bye
HER: then go!
HIM: miss ya! ha ha. fine bye
The male is chatty and, I would guess, falsely upbeat. His girlfriend is cool, detached, and unresponsive. His language is personal, with high rates of pronouns, and hers is noncommittal. By her relative silence, she is conveying her annoyance, which he avoids confronting.
Compare their interaction with another where the female is trying to emotionally engage her soon-to-be-ex-boyfriend. Over the previous two days of their occasional IMs, the female has tried to talk about their relationship and the male has constantly been busy with homework, exercising, even watching television. On the third day, she finally catches him when he might have some time to chat:
HER: are you there? are you gonna be able to talk for 5 minutes
HIM: yes. go
HER: what are you doing at the same time you’re chatting with me?
HIM: cleaning
HER: i feel too mad to say anything to you
HIM: fine
HER: did you not even feel one bit of bad when you read the card?
HIM: of course i did
HER: are you sure
HIM: hang on for 1 min please … Yes.
HER: or are you just saying … please hurry … hello?… what is the problem
HIM: alright … sorry
HER: what were you doing
HIM: my friends got here. Go
HER: but tell me what made you feel bad about the card
HIM: that you were being so nice. but i still dont feel that im wrong
HER: so why did you feel bad …
HIM: it did, but the fact that you told me not to expect it
HER: i don’t understand. the fact that i told you not to expect it made you feel bad?
HIM: look, i will call you @ nine. i cant speak like this
In this second case, the style matching numbers are also low. Although both use personal pronouns at similar levels, the computer analyses reveal that she is more concrete (using more articles) and more complex (e.g., conjunctions, prepositions) in her thinking than he is. She is also more personal and emotional, and focuses all of her attention on her boyfriend. He, on the other hand, is distracted and avoidant. Whenever she tries to get closer or to connect with him, he slips out of reach. This pattern of interaction was consistent over several days of instant messaging. Nevertheless, when both members of the couple
completed questionnaires about their relationship, both rated their levels of intimacy and relationship satisfaction as very high (!). As is typical, the language style matching numbers did a far better job at predicting the couple’s later separation than their self-reports.
The LSM method allows us to make reasonable guesses about the success of relationships by tracking a few interactions between couples. Similar ideas have been tested with recently married couples. John Gottman and his colleagues at the University of Washington report that by listening to the ways a young couple fights during a laboratory exercise, he can predict the likely success of the marriage. Gottman brings the couples into his lab and has them discuss issues that they have conflicts about—usually topics dealing with money, sex, or household chores. If, during tense discussions, both members of the couple show respect, try to reduce tension, avoid accusations, and inject positive emotions, their marriage is more likely to last. On the other hand, if one or both spouses are dismissive of the other, actively avoid discussing an emotional topic, or use the task as a way to launch personal attacks on the other, the marriage is in trouble.
A happy marriage is more than the egalitarian sharing of pronouns and prepositions. When two people use function words in similar ways and at similar rates, they are seeing their worlds in parallel ways. Having a similar worldview, though, does not ensure marital bliss. Gottman’s work reminds us that the strongest relationships are also characterized by shared positive emotions. Similar findings emerged in our dating couples—those with the highest style matching scores and the highest rates of shared positive emotions were the couples most likely to stay together. Interestingly, shared positive emotions on their own do not predict the long-term happiness or potential for survival of a relationship. The members of the couple must be both positive in their outlooks and engaged with each other.
BUILDING AN LSM DETECTOR
Imagine having a portable LSM detector that could track the quality of all your conversations. Perhaps you could point the LSM detector at your e-mails, text messages, or IMs and get a sense of the quality of your connection with your correspondents. Or, if you have been married for many years, you could keep tabs on your relationship and help point out to your partner when he or she isn’t holding up their end of the conversation.
As discussed earlier, my research team and I have built a crude working version of an LSM detector. If you go to www.SecretLifeOfPronouns.com/synch, you can enter text that you have sent and received from a friend, lover, or enemy. With the click of a button, you will receive feedback about the degree to which the two of you are in synch in your use of function words. You can then compare your LSM numbers with average LSM scores generated by others who have used the website. Even better, you can try out the LSM detector several times, comparing your interactions with different people. This should give you a sense of your general skill at matching your language with others.
The bad news is that a real-world LSM detector would be of only limited value. Yes, it could tell us when relationships were in synch and out of synch but not what the synchrony meant. Recall that LSM tends to be elevated when both people are passionate about each other and when they truly hate one another. Synchrony also increases when one of the two people is lying to the other. So, if your LSM detector registers a high number, you and your conversational partner are paying close attention to each other.
Think of the LSM detector as a device that can evaluate the synchrony of the conversational dance. Good dancing requires each of the two people to anticipate the other’s next move and, in a sense, briefly inhabit their thoughts. The dancing may last only as long as the music is playing or could reflect the two people’s interest in each other and ability to synchronize their thoughts and actions for years.
USING LSM TO UNDERSTAND PAST RELATIONSHIPS
One advantage of using computerized language analysis is that we can explore relationships over history wherever some kind of written record exists. For example, if we have access to old letters, poems, or lyrics that two people shared, we can get a glimpse of their emotional connection with each other. Computerized text analysis can illuminate our understanding of marriages, friendships, or alliances in history. Molly Ireland, who has been immersed in the language style matching research, has tracked down several enlightening cases, two of which include the poems of poets from one well-known happy marriage and from one equally famous unhappy one. By analyzing their works, we get better insight into each pair’s relationship.
ELIZABETH BARRETT AND ROBERT BROWNING
The idea of measuring language style matching was initially developed to track word use in natural conversations. Conversations should be especially sensitive to the immediate back-and-forth of ideas and shared references implicated in the ways mirror neurons work. By standing back, it is possible to imagine that language style matching can occur on a much broader level. If I hear a rival politician make a speech on a given topic, I might write my own speech in reaction to it several days or weeks later that would linguistically match the original speech. If my spouse is an elegant writer of novels, her writing style could influence the way I write professional journal articles. It’s not much of a stretch to imagine that two professional poets who read each other’s work might begin to use function words in similar ways—especially if the two people were passionately attracted to one another. A wonderful example is the case of the mid-nineteenth-century British poets Elizabeth Barrett and Robert Browning.
The two poets already had sterling reputations by the time they first corresponded when she was thirty-eight and he was thirty-two. Elizabeth, whose health had been poor since the age of fourteen, had never left home and continued to live with her despotic father while she wrote about emotions, love, and loss. Robert was a vibrant man-about-town whose work was more psychological and analytic. After Elizabeth published a poem that referenced one of his poems, Robert wrote her and professed his love for her work and, more directly, for her—even though the two had never met.
Over the next several months, they corresponded and he proposed marriage—something that she initially resisted by trying to pull away from him. After Robert relentlessly pursued Elizabeth, the two eventually eloped, moving to Italy, where they spent most of their lives together. By all accounts, the marriage was supremely happy, with both continuing to publish their work. The only dark time in their fifteen-year marriage came during the last two to three years, when Elizabeth’s health failed, resulting in her death at the age of fifty-five.
Over the course of their careers together, both Elizabeth and Robert were prolific, producing hundreds of poems. Even before they met, the two tended to use function words similarly. This may explain why the two were so attracted to one another long before they ever corresponded. As you can see in the graph, the LSM statistics indicate that their use of function words was quite similar before they met and for most of their marriage. During their one-and-a-half-year courtship, however, the poetry they wrote diverged. This was a period when Elizabeth struggled with the decision of breaking away from her father and, in her own mind, burdening Robert with her health problems.
As you ponder the figure, you might be curious to know who is responsible for the increasing and decreasing LSM over time. After all, LSM has been likened to a dance between two partners. Can we determine if one of the dancers is more responsible for the occasional faltering in the dance? In fact, we can. Closer inspection of both poets’ use of function words suggests that Robert was impressively steady in his writing style over the course of his career, whereas Elizabeth tended to alter the ways she used function words from one period to the next. For example, during most of their time together, Elizabeth’s use of function words tended to mirror Robert’s. However, during her difficult courtship year and as her health declined toward the end of her life, she tended to withdraw into her own world, writing less like Robert and less similarly to her own writing at other times of her life.
Language style matching (
LSM) scores between the works of Elizabeth Barrett and Robert Browning and of Sylvia Plath and Ted Hughes. The higher the LSM score, the more similarly the two poets used function words.
SYLVIA PLATH AND TED HUGHES
Whereas the Brownings’ marriage is often idealized by modern writers, the relationship between Sylvia Plath and Ted Hughes is not. Plath was born in Boston and met the British-born Hughes while she was on a Fulbright scholarship at Cambridge University. Both were considered rising young stars in the poetry world. Her writing was often emotional, confessional, and occasionally whimsical. His was more cerebral, often dealing with nature and myth. Four months after meeting at a party, they married—she was twenty-four; he was twenty-six.
The Secret Life of Pronouns: What Our Words Say About Us Page 22