How to Tell If Someone Likes You Over Text: Real Signals
It isn't the heart-eyes emoji. It's the extra four sentences they didn't have to write. In a Cornell study, strangers who had never met chatted for thirty minutes over instant message, and the partner reading along consistently picked up on who was in a good mood and who wasn't — no face, no voice, nothing but text. Assessors rated the upbeat partner as 41.8% happier than the low-mood one, purely from what showed up on the screen — Hancock, Landrigan & Silver (2007).
So if you're staring at a thread trying to reverse-engineer a feeling from a row of gray bubbles, you're asking the right question with the wrong evidence. The signal was never in the wink emoji. It was sitting in how much someone wrote, how they punctuated it, and — the part almost nobody checks — how rarely they pushed back.
Reading someone else's thread is a smaller version of a bigger habit: noticing what a person actually put down, instead of what you wish they'd said.
Write the version of yourself you'd actually want someone to read this closely.None of this is about decoding a written dating profile before you've even matched — that's a different kind of reading, done before anyone's typing to you personally. This is about a live thread, after someone has already started choosing, message by message, how much of themselves to hand over.
In the same study, the upbeat texters simply wrote more — about 29% more words than the low-mood texters, even though both groups sent messages at roughly the same rate per minute. Reply speed, the thing everyone obsesses over, didn't actually separate the two groups. Length did. Someone who's into the conversation keeps adding to it; someone going through the motions answers and stops.
That reframes the familiar ache of watching three dots appear and vanish. The dots aren't the tell. What they type once they commit is.
A five-minute phone call would resolve most of this faster than any of it — tone of voice still beats inference, and it always will. But a lot of early texting happens precisely because neither person wants to spend a call finding out, which is what makes the typed cues worth learning in the first place.
Positive texters in the same transcripts used roughly six times as many exclamation points as the low-mood group, and punctuation turned out to be one of only two cues — alongside how often someone disagreed — that reliably predicted how a partner read their mood, together explaining 34.6% of that judgment Hancock, Landrigan & Silver (2007).
That's a strange thing to hang emotional weight on, and also exactly the point. The cue nobody thinks to fake is the one that keeps working.
The instinct is to listen for praise. The data pointed somewhere blunter: the happy texters disagreed with their partner far less often, and the low-mood group used about five times more negative-feeling words across the same half hour. Compliments are easy to type on command. A pattern of not pushing back is harder to fake for thirty straight minutes.
A separate experiment asked strangers to act out affinity for a partner, either face to face or in synchronous text chat, then had outside raters guess how much each pair actually liked each other — and the raters read the typed version just as accurately as the in-person one Walther, Loh & Granka (2005). Text isn't a degraded channel for liking. It's a different one, with its own tells.
Politeness and interest can look identical for the first few messages, which is exactly why people get stuck rereading the same thread. The split shows up once the small talk runs out: does the reply get shorter, or does it start pulling in a detail from three messages ago?
| Interested | Just being polite | |
|---|---|---|
| Message length | Grows or holds steady past the first few exchanges | Shrinks to the minimum needed to answer |
| Punctuation and tone | Exclamation points, a warmth that reads consistent | Flat, correct, with no real texture |
| Disagreement | Rare — they'd rather ask than contradict | Neutral and non-committal, avoids anything with edge |
| What they bring back | References something you said days ago, unprompted | Sticks to whatever you just wrote |
| Effort over days | Stays roughly the same, or grows | Quietly thins out, message by message |
Strip away the guesswork and five things are worth watching in a real thread, all grounded in how affect leaks into typed language rather than folklore about read receipts — the same reason Anketta turns the guess into an underlined line instead of an inference.
- Length, not speed. Someone invested writes more per message; someone going through the motions answers and stops. Reply time barely moved the needle in the data above — content did.
- Punctuation with texture. Exclamation points, a stray "!!", or a tone that stays warm across a whole thread track real enthusiasm more reliably than any single word choice.
- How rarely they disagree. Not agreement — disagreement. Someone enjoying the exchange pushes back less often than someone merely being civil.
- What they carry forward. A detail from three days ago surfacing unprompted means they were thinking about the conversation between messages, not just answering it.
- Whether the effort holds over days, not one message. A single short reply after a long day means nothing. The same shortness holding for a week, after starting warmer, is the pattern worth noticing — whatever caused it.
Interest rarely disappears all at once — it downshifts, and the downshift itself is the signal. A pace that felt natural at the start starts to feel deliberately slower, and that change reads as retreat even when nobody's said anything is different.
"Once that dance has gotten started, if you slow down to a pace where you're comfortable, that change is going to be interpreted as a lack of interest." — Katherine Hertlein, psychologist, University of Nevada, Las Vegas, Thrive Global (2017)
A slowdown that stays sporadic and thin — a flicker of contact, then nothing, then another flicker — has its own name in the vocabulary that's grown up around this exact pattern, and it's worth knowing the shape even before you're ready to name what's happening to you specifically.
None of this fixes the actual discomfort of not knowing. It just narrows down where to look.
Try writing the thing you'd want someone to read this closely.Does using emojis mean someone likes you?
Not on its own. Emoticons made up only 0.2% of the word count in the Cornell transcripts and didn't differ between happy and low-mood texters — the effort went into words, punctuation, and how often people disagreed, not into icons layered on top of a message.
Is a fast reply proof of interest?
No. Positive and low-mood texters in the same study sent messages at almost identical rates per minute; speed didn't separate the groups. What someone actually wrote, once they replied, carried the signal that speed alone couldn't.
What if they mostly just reply "haha" or "lol"?
That's short by design, which makes it closer to the "just being polite" column than the "interested" one — unless it's paired with real length elsewhere in the thread, a habit like this rarely carries much information by itself.
Can a naturally short texter still be interested?
Yes, and this is where the five signals matter more than any single one. A short-texter's own baseline still shows disagreement rate, follow-through on old details, and whether the pattern holds across days rather than one tired evening.
Does it matter who texts first?
Less than people assume. Initiating once says little; initiating consistently, over days, without being chased for it, says more. A single first text is a coin flip, not a verdict.
What if the energy is high some days and flat on others?
Normal. People have bad days, busy weeks, and phones they forget to charge. The pattern to watch is the trend across a stretch of time, not any single flat reply read in isolation at 11 p.m.
If you want to see what writing instead of guessing looks like, Anketta starts from the same idea — an interested reader underlines a line, and nobody has to reverse-engineer that from punctuation.
Every thread eventually says what it means. The only question is whether you're reading the sentence, or just the silence between two of them.
Unsure about writing? Try reading first.