“I Was Joking” — Why Sarcasm Gets Lost in Text

Company chat, in the middle of a hectic day. Olek gets another urgent request and replies with sarcasm.

— Great. Another urgent task. Exactly what I needed today.
— If the deadline is the issue, let’s discuss it with Karim — Maja replies.

Olek is surprised by her formal tone. Maja is convinced she has just read a complaint.

What really happened

As Olek types the message, he can hear his own voice, timing, and irony in his head. That makes “Great. Another urgent task” an obvious joke to him. Maja does not receive that tone. She receives only the words, which could just as easily be a pointed complaint. The sender therefore has more information about their own intention than the receiver and unconsciously assumes some of that information travels with the text. This is especially easy in chat and email, where voice, facial expression, and immediate correction are missing.

What the research says

Research on email communication found that people significantly overestimate how accurately a recipient will identify their intended tone. The sender already knows whether they are joking, which makes it difficult to mentally remove that knowledge when predicting how someone else will read the message (egocentrism over email; Kruger, Epley, Parker & Ng, 2005). The point is not that text carries no emotion. People do infer emotion from text — they may simply infer a different emotion from the one the sender intended. The more a message depends on irony, subtext, or tone, the more room there is for error.

How to handle it

If a joke only works when the other person hears exactly the tone you hear in your own head, either make the signal clearer or use a richer channel. In a relationship where both people already know each other’s humour, the risk is lower; in a new relationship or a tense situation, it is higher. If a message lands badly, “it was obviously a joke” is rarely useful. The recipient genuinely did not have the same set of cues. The general rule is: do not assume the person reading your message can hear your voice in their head.

How Empatyzer and Em can help

Empatyzer helps in exactly the situations where the sender feels their intention is obvious even though the recipient sees only a few words on a screen. The assessment and past communication patterns may help Em recognise that Olek naturally uses irony and quick, expressive messages, while Maja pays closer attention to tone and may interpret ambiguity more cautiously. In a specific situation, Em can suggest that Olek mark the joke more clearly or rewrite it so the meaning does not depend on a tone that text cannot reliably carry. Maja can be prompted to check the intention with a short question before settling on the worst interpretation. If much of a team’s communication happens in chat, Empatyzer can also reveal broader differences in digital communication style. Micro-lessons build a simple habit: before sending a message that depends on tone, check whether the meaning still works without your voice or facial expression.