Evidence status: conditional support Email experiments find that senders overestimate receivers’ ability to identify sarcasm, sincerity and humour. The intended voice heard by the sender makes it difficult to see the ambiguity in the words alone. Platform features, relationship history, emoji and shared context can alter or improve communication, so early email evidence cannot be mechanically generalised to every digital medium. This is a version 0.1 research draft, with evidence reviewed to August 2026.
How Everyday Psychology Takes Shape · Season Two: How Do We Read Other People? · Article 10
“Fine, of course.”
The sentence can signal sincere agreement, impatience, sarcasm or reluctant compliance. In speech, stress, timing, gaze and the preceding exchange assist interpretation. Once the sentence becomes text, the sender continues to hear an intended voice internally. The receiver obtains only the words.
After a misunderstanding, each party can experience the other as unreasonable. The sender thinks the tone was obvious. The receiver believes they made a sensible interpretation from the wording and relationship. They were not processing identical inputs: one possessed the writing intention and an internal soundtrack, while the other had to generate a voice.
What five email experiments found
Justin Kruger, Nicholas Epley, Jason Parker and Zhi-Wen Ng published five experiments in 2005 on whether people communicate tone by email as successfully as they expect. Participants sent statements intended as sincere, sarcastic or otherwise tonally marked, predicted whether receivers would identify the intention, and were compared with receivers’ actual judgments.
Senders were overconfident on average. Conditions involving spoken delivery or listening also showed how vocal tone could improve some judgments. The authors attributed part of the overconfidence to egocentrism: senders could not readily ignore the intention they knew, and they “heard” prosody that was absent from the receiver’s evidence.
The conclusion is not that writing cannot express emotion. Literature, correspondence and ordinary messages clearly do. The supported claim is narrower. Where a sentence permits several voices and writers assess clarity through their own internal reading, they can underestimate the ambiguity available to a reader.
In 2008, Kristin Byron integrated research on electronic communication and emotion in a theoretical framework. She argued that features of email make receivers likely to interpret ambiguous work messages as more emotionally negative or more neutral than senders intended. Relationship, power, emotional intensity, complexity and choice of medium all matter. As a theoretical review, the article should not be used as an effect-size demonstration that every email becomes more negative.
More than facial expression is missing
Face-to-face interaction provides voice, body, immediate repair and turn-taking rhythm. Text removes some channels but adds other capacities: writers can pause, edit, retain a record and transmit complex detail precisely. A medium is not simply poorer. It changes which evidence is cheap and which is absent.
Asynchrony is important. A receiver may open a message hours later, in another mood and under different task pressure. The scene shared during composition no longer exists. “We need to talk” generates different expectations when sent by a busy manager to an employee and when used between friends arranging a meeting.
Punctuation is not a stable tone dictionary. A full stop is ordinary in formal email but can seem cold in some messaging conventions. Capitals, exclamation marks, ellipses and emoji change meaning across age, group, platform and relationship. A smiling face may clarify friendliness and can also trivialise serious content.
Shared history restores context. Familiar partners know one another’s brief style and private jokes. Familiarity can also generate a new overconfidence: because past messages were understood, this one is assumed to need no checking.
The accessibility of editing creates another paradox. A polished message may feel exceptionally clear to its author because every phrase has been considered. The recipient cannot see the alternatives that were removed or the concern that motivated the final wording.
Why text is especially risky during conflict
When threatened or angry, people are more likely to give ambiguous material a negative interpretation. They may also produce short sentences that feel restrained internally and read as cold externally. The receiver responds to a perceived hostile tone; the sender treats that response as evidence of hostility. Initial ambiguity acquires real consequences through exchange.
Text remains available for rereading. Repetition need not approach the writer’s intention; it can make one interpretation increasingly familiar. A screenshot detached from the earlier thread becomes even more open to a new audience’s meaning.
Power makes clarification costly. An employee receiving “come to my office” may not feel able to ask whether bad news is coming, while the manager thinks only a location was communicated. The sender’s cost of adding context is small. The receiver’s cost of uncertainty can be substantial.
For high-risk, emotional, fragile or complex material, medium choice is therefore part of communicative responsibility. Not everything should become a call: speech may lack a record or be less accessible for some people. A better design may supply written background, offer synchronous discussion and then confirm decisions in writing.
Writing for what a receiver actually receives
After drafting an important message, temporarily remove your intended voice and ask: could these words plausibly come from someone who is angry? If so, do not require the recipient to guess correctly. State the intention directly: “This is not a criticism; I am confirming the time.” “I mean this lightly, but tell me if it lands badly.”
Concrete content is more reliable than decorative tone. Instead of several exclamation marks to signal friendliness, explain the request, reason, deadline and available choice. If negative feedback must be written, separate observable fact, impact and next step. Do not leave one global sentence for the reader to fill with the worst meaning.
As a receiver, preserve tone as a hypothesis: “This reads as quite sharp to me. Is the concern the deadline or the proposal itself?” The sender is not granted infallible final authority—people can revise intentions after the fact—but new evidence enters the exchange.
When an immediate emotional reaction is strong, waiting before replying can prevent the first interpretation from becoming an irreversible action. A pause is not proof that the benign reading is right. It creates room to distinguish the received feeling from the sender’s still-unknown intention.
How digital text distributes the responsibility for understanding
In person, a change in expression can reveal misunderstanding at once. Asynchronous text delays repair and allows a sender to feel that once information has been “sent”, understanding belongs to the receiver. Transmitting characters and completing communication are different events.
Meaning is formed through message, medium, shared history and mutual correction. A sender cannot prevent every interpretation, and a receiver cannot blame every ambiguity on the writer. Responsibility is better allocated by foreseeability. The greater the consequence, the more the sender should reduce predictable ambiguity. The less certain the receiver’s interpretation, the more important clarification becomes before acting on it.
The version 0.1 conclusion is that text is not inherently impoverished, but it encourages writers to count the voice in their heads as part of the message. Emoji and familiarity can help without guaranteeing success. An ordinary joke can tolerate some risk. Work, money, relationship and safety messages should state intention and request more explicitly and leave an entrance for correction. Reliable digital communication is not communication that can never be misread. It is communication in which a misreading can still be discovered and repaired at low cost.
Primary research sources
- Kruger, J., Epley, N., Parker, J., & Ng, Z.-W. (2005), “Egocentrism Over E-Mail”.
- Byron, K. (2008), “Carrying Too Heavy a Load? The Communication and Miscommunication of Emotion by Email”.
- Walther, J. B. (1996), “Computer-Mediated Communication: Impersonal, Interpersonal, and Hyperpersonal Interaction”.
- Riordan, M. A. (2017), “Emojis as Tools for Emotion Work”.
Series navigation: How Everyday Psychology Takes Shape — Season Two overview
Discover more from Geoffrey Chen
Subscribe to get the latest posts sent to your email.