Is language influencing your beliefs in ways you don't realize?
Updated: Sep 29
Most of us think we’re pretty good at forming accurate beliefs based on the evidence that is presented to us. In fact, in one study we ran we found that, on average, people viewed themselves as better than 70 out of 100 people of their demographic at “evaluating evidence reliability”. Now of course, only 30% of people can be in the top 30% at evaluating evidence, suggesting that most of us are overconfident in this capacity.
We think we can distinguish opinion from fact, and that we can tell when someone might be presenting us with biased information. And yet, we all know that language can have positive or negative connotations without actually claiming anything specific. The way a writer chooses to word a claim can have a large effect on how we perceive that information, even if the facts written are the same. In other words, we're all susceptible to subtle bias that comes from the way things are written.
And we may be prone to biasing others in this way, as well. When we think we’re speaking about something neutrally or objectively, our hunches and beliefs may colour the words that we use, so that we change the impression we leave others with even if the facts we state don't change. The more aware you become of the subtle biases caused by language, the more you start seeing it all over the place. Take a look at some of these situations below. How different would your understanding of the facts be if, in a news article, you only read the first number or the last number?
(1) It is known that revenues will be the same as last month.
(2) Haiden explained that revenues will be the same as last month.
(3) Haiden said revenues will be the same as last month.
(4) According to Haiden, revenues will be the same as last month.
(5) Haiden admitted revenues will be the same as last month.
(6) Haiden could not deny that revenues will be the same as last month.
(7) Haiden confessed that revenues will be the same as last month.
(8) Haiden had no choice but to admit that revenues will be the same as last month.
(1) An insider close to the situation explained that the action was premeditated.
(2) The action was premeditated, as Remy later informed us.
(3) Remy filled us in that the action was premeditated.
(4) Remy said the action was premeditated.
(5) Remy told us the action was premeditated.
(6) Remy claimed the action was premeditated.
(7) Remy was going around saying that the action was premeditated.
(8) If you believe Remy, the action could have been premeditated.
(9) Remy made the claim that the action was premeditated.
(10) Remy tried to claim that the action was premeditated.
(11) Remy wanted people to believe that the action was premeditated.
(1) We learned that the allegations against Pat are entirely false.
(2) Pat explained that the allegations are entirely false.
(3) Pat said the allegations are false.
(4) Pat denied the allegations.
(5) Pat claimed the allegations are false.
(6) Pat would not admit to the allegations.
(7) Pat refused to admit the allegations.
(8) Pat is playing the part of the victim, and telling people the allegations are false.
(9) Pat is still claiming the allegations are false, yet hasn’t given us a shred of hard evidence.
(1) We learned additional useful information about the story by talking to Peyton again.
(2) Peyton later added additional information that helped us better understand the story.
(3) Peyton told us more information about the story.
(4) Peyton’s original story wasn’t complete.
(5) Peyton had omitted information in the original telling of the story.
(6) Peyton had prevented us from fully understanding the story by leaving out information.
(7) Peyton did not tell us the whole story.
(8) Peyton left out important parts of the story.
(9) Because Peyton left out important parts of the story, we were misled about what had really taken place.
(10) Peyton misled us about the story
As you can see, two journalists could write factually accurate descriptions of the same event that send completely opposing messages about that event, purely by means of word choice and implication, while still maintaining the illusion of neutrality. If you are purposefully adding positively or negatively valenced words to the information you are sharing, you have an opinion, and you are potentially altering the reader’s perception of the facts. The facts, as a reader perceives them, are almost never divorced from context and implication.
It’s of course sometimes good to express an opinion, to call something out as bad or good. But when someone is not explicit that they’re stating a belief alongside a fact, their word choice can determine our opinions about the facts without us even realizing it. And when done with sufficient skill and subtly, we may come away falsely assuming that the author had a neutral point of view, that they just plainly told us the facts.
It is important to be able to tell the difference between facts presented in valenced language and facts presented neutrally and objectively, so that we can form accurate beliefs about the world around us. It is just as important that we pay attention to our own word choice and try to determine whether we are presenting more than just facts in order to get the people around us to hold certain opinions.
Can you think of any cases where you thought you were using language neutrally but your opinion came through in the words you used? Or perhaps a situation where you formed an opinion based on the way information was presented to you, when some of that content was merely word choice, rather than factual?
If you found this helpful, you might want to try out our Question of Evidence mini-course or Seeing Other Explanations program. The Question of Evidence teaches you a simple trick, based on Bayes’ Rule, that will help you gauge the strength of evidence more accurately on a day to day basis. Similarly, Seeing Other Explanations walks you through a technique that you can use to figure out the truth in uncertain circumstances.
This article was written by Spencer Greenberg and edited by Holly Muir.