Every election cycle and presidential debate raises questions about candidates’ honesty and brings forth a bevy of experts claiming to have foolproof methods for unmasking fibbers. In a recent study, U. of I. family communications expert Michael T. Braun explored a widely held theory about the liguistic indicators of deceptive speech. Braun, a research specialist in the School of Social Work’s Children and Family Research Center, spoke recently with News Bureau social work editor Sharita Forrest about his findings.
What does research say about detecting liars and how did that inform your study?
Some researchers claim that there are linguistic cues that signal when someone is lying. For example, when people lie they use fewer first-person pronouns and more third-person pronouns, supposedly because they are psychologically uncomfortable about lying and try to distance themselves from their lies.
I was very skeptical of these claims, and it’s important to note that no one has ever tested whether this distancing effect actually exists, so these claims are merely speculation.
My co-author, Lyn M. Van Swol of the University of Wisconsin at Madison, and I wanted to determine whether there are reliable lexical indicators that someone is lying.
What was your process for analyzing the veracity of politicians’ statements, and what were your findings with regard to speech patterns and deception?
We went on Politifact.com, a website that checks the accuracy of prominent statements made by politicians, to gather data.
We took about 570 statements by politicians, each of which Politifact had classified as a truth, a lie or as “pants on fire,” Politifact’s term for an especially egregious lie. About half of the statements in our sample had been classified as lies.
We captured the text of the statement and a few details about the speaker, such as the context in which the statement was made, the speaker’s sex and their political party.
We considered several variables: Is the person lying or telling the truth? And were they speaking in an interactive context, such as a debate, or a static context, such as writing an op-ed for a newspaper?
We found that there simply are not reliable lexical indicators for what is a lie and what’s a truth.
However, we did find that, overall, context had a far greater impact on the word patterns used than deception did. Moreover, we found that liars used more words than truth tellers, a finding that we call the “Pinocchio Effect.”
Why might politicians be more verbose when they’re lying than when they’re telling the truth?
Politicians are very interested in creating a believable reality in which their policies will lead to prosperity and their opponents’ policies will lead to ruin.
Because they don’t have the truth on their side, they may need to give more details and make more promises in order to make their policy sound appealing. Someone telling the truth may not be thinking about those things because they assume they’ll be believed and that their audience shares the same assumptions, so they don’t need to sell all those details.
Excessive elaboration may be a reliable clue that someone is lying, especially in cases where they need to create this believable reality.
If there are no linguistic cues for identifying dishonesty, what are the takeaways from this study?
Human speech is far too complicated, and the variability in the ways that people choose to talk is far too great, for something simple like pronouns to be reliable indicators of deception or truth.
What this study shows is that we need media organizations that will give scrutiny to the statements politicians make. The more help we can get from the media, the more the public can be informed. There simply are no shortcuts to fact checking, to investigative journalism, to educating yourself on the issues and becoming an expert yourself.
To contact Michael T. Braun, call 217-300-0670; e-mail mtb@illinois.edu
The paper “His Lips Are Moving: Pinocchio Effect and Other Lexical Indicators of Political Deceptions” was published earlier this year in the journal Discourse Processes.
