Smiling eyes might by lying eyes.

University of Minnesota researchers have found that anthropomorphic emojis, like language, can get lost in translation, causing significant misunderstandings.

This happens two ways.

There’s the technical translation glitch. What’s sent as a smiling face from a Google Nexus would appear as a frown on the recipient’s Apple iPhone and vice versa.

Then there’s the human component: People interpret identical emojis differently.

Take the “grinning face with smiling eyes” emoji. Across platforms it is varied — sometimes showing teeth, some mouths open, some mouths a straight line. Some have open eyes while others are closed. The corners of the lips are turned up on some while others are turned downward.

That emoji might be sent as a “mildly negative emoji” but will be received as a “relatively positive one,” the study found.

Researchers had recipients rank emojis on a scale of negative 5 to positive 5 in terms of emotional response.

For nine of the 22 emojis tested, the average difference in emotional rating between two platforms was greater than two points.

People also described emojis differently.

When seeing an Apple emoji of a person raising both hands in celebration, people described it as “stop” and “clap.” When describing Google’s version of the same character, people used “praise” and “hand,” wrote Hannah Miller, a third-year Ph.D. student who is part of the university’s GroupLens research lab and posted the findings on its blog.

“We found that in many cases, there is quite a bit of potential for miscommunication,” Miller wrote.

Miller said the finding that “really surprised” researchers was that a “good deal of the potential for miscommunication may come from different interpretations of the exact same emoji rendering.”

In other words: People see things differently.

Miller pointed out that some scholars argue that emoji use represents a seismic shift in language, so understanding its role in human communication is important in developing the next generation of technology.

The idea for the study started germinating when Miller noticed that an emoji she saw on Facebook wasn’t identical to what she saw on her phone.

“That’s kind of an important thing,” she said in an interview. “I think people aren’t aware of it, and now they can realize someone on the other side isn’t seeing what they’re seeing.”

Co-author Isaac Johnson said the students had fun with the study. “It really caught our imagination,” he said.

The study involved 304 people in various pairings to replicate how reactions would differ in two-way conversations.

Miller posted on the blog April 5 and saw interest slowly expand. “I honestly I had no clue it was gonna catch fire like this,” she said. “People are really reacting, and some people are taking it very seriously. We do think that it could cause miscommunication, but we don’t think [emoji interpretations] will cause World War III.”