Q&A from “Junk Words: What the, above, he, and I say about your lovelife”

Wise Readers, 

Could word-counting as a lie-detector measure work?  How accurate is it, anyway?  Could reliance on computer programs as lie detectors ruin your relationship?   How can our beliefs about who says ‘I’ reveal a lot about our politics and stereotypes—yet almost nothing about our conscious processing of ‘I’? 

Read on!

From Dan Knipe: —Word-Counting As Lie Detector = Palm-Reading?— 

You know, this sort of thing seems like palm-reading to me. Just sayin,’ trying to tell someone’s truthfulness via syntax is a gamble at best. It worries me that people already predisposed to superstition might cause their relationships real harm because they think they know enough about their significant others’ pronouns to tell if they’d been cheating or not.

Duana’s response:  —A Brief History Of Lie Detection—

Dear Dan, thanks for writing in. When I teach Intro Psych each semester, we go over various methods of lie detection, such as the polygraph and ‘brain fingerprinting’. The polygraph is well-known —most students have heard of that—and it’s inadmissible in American courts today precisely because it’s so imprecise. It only differentiates guilt from innocence around 65% of the time.

The brain fingerprinting method presumes that if a person recognizes images from a crime scene, as shown by brain-wave changes, then that person is guilty. A version of brain fingerprinting has been used to exonerate at least one man charged with murder in America.  And it’s been used to convict people of murder in India…with no other evidence brought to bear. The PhD-level psychologist who created the method, Lawrence Farwell, seems supremely confident of its efficacy—both in this article and this interview.

Problem? For good reason, it would seem nearly all other social scientists whose opinions I’ve found, including great thinkers in neuroscience such as Michael Gazzaniga, share your skepticism of lie-detection methods—whatever methods are used. And so does Dr. Pennebaker. And for what it’s worth, so do my students and I.

In the interview Dr. Pennebaker gave me, he noted that while his language word-count method detects lies at better-than-chance levels, it’s not strong enough to use as stand-alone evidence. That’s because life is messy: Many factors are brought to bear in any one person’s life, and no one lie detection method satisfactorily measures and accounts for all of those~not yet, and perhaps not ever.

Dr. Pennebaker elaborates the point in his book The Secret Life Of Pronouns, which expounds on his research in word counting and what it tells us about many things, including lie detection. Basically, accuracy of his method has been experimentally tested by having the program analyze word use in persons who are randomly assigned to lie or tell the truth; then, the program’s accuracy is compared to the accuracy of bystanders who are asked to say whether others are lying.

Results? The LIWC program has occasionally detected lies at the 75% accuracy rate, but much oftener it’s about as accurate as other lie-detection methods—around 65%. That’s a 15% gain over human beings, who only perform at 50/50, aka chance. But it’s not nearly sufficient as a stand-alone method of conviction—either in court, or in a personal relationship.

That said, Dr. Pennebaker implies a necessary question in his book: Since lie detection methods are about as accurate as eyewitnesses, why should we exclude polygraph evidence but keep eyewitness accounts?

“If polygraph, nonverbal, eyewitness, brain scan, and any other type of evidence can help classify the guilt or innocence of a witness, it should be introduced in court. However, it should be introduced in a way that calibrates its accuracy to the jury. Each type of evidence is simply something else for the jury to weigh, knowing that there are problems with each type. Life is probabilistic—courtroom evidence is no different” (p. 155).

In other words, let’s say someone really did use the LIWC to test their sweetie’s honesty. At best, it’d be a piece of all the information they already knew.

At worst, it’d be just what you indicated: a dangerous misuse of technology that could seriously undermine relationships.

Thanks again for the great question.

From Paula: —‘I’ And Arrogance—

Fascinating! I’m surprised about the usage of “I,” or, rather that the lack of using “I” indicates arrogance. I would have thought that the more people use “I,” then the more arrogant they are.

Duana’s response: —Stereotypes and The Unconscious—

Hi, Paula,

That struck me as odd, too. In the interview and in the book, Dr. Pennebaker also spoke/wrote about how ‘I’ can be a test of one’s political leanings. To wit, how often do you think President Obama says ‘I’?

(pauses while people guess)

As it happens, people’s answer depends more on their party than on fact. Republicans —including some prominent pundits— routinely posit that Obama says ‘I’ excessively. Democrats guess he says ‘I’ less. In reality, LIWC analyses of existing speeches from all presidents since speeches have been recorded showed Obama says ‘I’ less than any former president.

Dr. Pennebaker doesn’t construe these findings as indicative of lying, by the way; ‘I’ associated with lying deals with being directly queried about a falsehood, not how often one says ‘I’ in general. Nor did he say Obama was arrogant. Instead, he conceptualized this as aloofness and cool detachment.

At any rate, what ‘I’ really shows is where people’s attention is focused. Men focus more on things than people, usually—which the articles ‘a’, ‘an’, and ‘the’ show. People who are dominant in a given social interaction use more articles and fewer first-person pronouns too. People who are more relationship-oriented (usually women) and those more depression-prone (also usually women) use ‘I’ more in general. And, of course, when a person is lying, ‘I’ is the single most-predictive word indicating truthfulness. Or lack thereof.

Ultimately, since we’re all largely unconscious of how often anyone says ‘I’, our perceptions tend to be based more on stereotypes and personal beliefs than reality~hence the ‘I’ political test and hence researchers thinking men (and perhaps the arrogant) use ‘I’ more than women, even when studies began finding the reverse.

Cheers,
Duana

Do you have a question for Duana? Email her atDuana@LoveScienceMedia.com, and a free, confidential response shall be yours.  If your letter is ever used on-site, it will be edited and your name changed to protect your identity. 

Want to try out the LIWC program for yourself and analyze your—or your sweetie’s—language style?  Here are some free, fun ways to do it:

—Analyze anyone’s Twitter feed, including your own:http://www.analyzewords.com/

—Test out the LIWC on your own writing samples—about pictures, your thoughts about your own life, or just your description of a bottle— here:http://homepage.psy.utexas.edu/homepage/Faculty/Pennebaker/Home2000/Words.html

The author wishes to thank the following scientists and sources:

James W. Pennebaker, for his expertise, his science, his interview from last week, and his book. 

His book, The Secret Life of Pronouns: What Our Words Say About Us.  The book’s scope incorporates everything from this interview, plus how our words reveal who has the most power in a relationship, how you can tell who’s likely to buy a particular product based on their words, whether a person lives in a (very) particular neighborhood based only on their junk words—and much more.  Highly recommend. 

The LIWC program and its creators, including James W. Pennebaker and Martha E. Francis. 

Related LoveScience articles:

The interview with Dr. Pennebaker regarding his book: http://www.lovesciencemedia.com/love-science-media/junk-words-what-the-above-he-and-i-say-about-your-lovelife.html

Using body language to tell who’s into you: http://www.lovesciencemedia.com/love-science-media/how-to-tell-shes-just-not-that-into-you-or-captain-clueless.html