Sync score
Descriptive, not a prediction. Word lists miss sarcasm; open any moment below to read the messages behind a label.
Your texting, crunched
The story your chat tells about the two of you
Short reads built from your messages. The long version is its own page.
Star signs, strictly for fun
Optional. Pick both signs to see what the horoscope predicts, then what your texts actually did. It also gets a chapter in the story.
Moments
Your best days, the hard ones, and how you come back
Tap a kind of moment, then a day, to read the real messages behind it.
Drift
How the conversation changed
One measure at a time, bucketed by month. The flagged point is the single biggest jump on the measure you are looking at. On I-words: a rise can go with a rough patch or low mood in the research, and also with a busy month. Read it with the notes you left above.
Show the numbers as a table
Who mirrors whom
Who bends toward whom
For each reply, does its word style match the exact message it answers better than a random earlier message from the same person? 50% is chance. Above it means the replier bends toward the other person in the moment.
Check in again
The trend is the useful part
One score is a snapshot. Bring a fresh export in a month and you see which way things are moving. Leave an email and we will nudge you then. Once a month is plenty: a mirror is for a look, not for watching. Nothing from your chat is attached to it.
Style fingerprint
How the sync score is built (the nine word families)
Usage per 100 words. The score is the average of how close the two of you sit on each row. Useful if you want to see which family drives a gap; skip it otherwise.
The words on this page, in plain English
- Sync score
- How alike two people's small words are: our own measure of what researchers call linguistic style matching. 1.00 means the same mix of small words; around 0.50 means very different. Descriptive, not a prediction.
- Function words
- The small words: I, you, the, and, of, not, very. About half of everything we say. We choose them without thinking, which is why they leak attention and mood.
- Bending toward
- When your reply picks up the other person's word style in the moment. Researchers call it accommodation. Above 50% is more than chance.
- Formality
- A 0 to 100 mix of capital letters, full stops, longer words and less slang. Rising formality often reads as distance in texting.
- We-words and I-words
- "We, us, our" versus "I, me, my." More we-talk goes with closer couples in studies. Heavy I-talk goes with stress, low mood and self-focus, but it is a signal, not a diagnosis.
- Tension
- A rough temperature for a day: criticism aimed at each other, absolutes like always and never, capitals, negations. Laughter cools it. It cannot tell sarcasm from anger; use your own memory. Red on this page always means friction between the two of you.
- Support
- One of you vents about something outside the relationship (work, family, a bad day) and the other turns toward it within the hour: a question, an "I'm sorry", a "what happened". Not a repair, because there was no fight between you. Researchers call it responsiveness.
- Banter
- A hot-looking day where a quarter or more of the messages contain laughter. Probably teasing, not fighting.
- Conflict patterns
- Four patterns that observational research on couples (Gottman and Levenson's lab, among others) links to trouble: attacking the person rather than the problem, contempt or mockery, defensiveness, and going silent. Contempt is the most corrosive. Our labels are our own.
- Repair
- Lowering the temperature after friction between the two of you: sorry, you're right, I hear you, can we talk. Counted only when there was a jab aimed at the other person shortly before. Repair attempts matter more than the fight itself.
- 5 : 1
- Stable couples run about five positive moments to each negative one, even during conflict. Here it is counted in emotion words on a single day, so treat it as a rough gauge.
- Bids
- Small reaches for attention. Here, a question. "Turning toward" is answering it soon; "turning away" is letting it sit. Gottman found couples who turn toward most bids stay together.
- Harsh start
- A hard day that opens with criticism in the first few messages. Gottman found how a conflict starts predicts how it ends most of the time.
- Catchphrase
- A word one of you uses far more than the other. Computed by comparing rates, not just counting.
- Themes
- The topic words that keep showing up on the hard days and on the good days. Names and small words are removed first.
Where the ideas come from: the research behind each number
- Sync score. Ireland and Pennebaker (2010) defined language style matching, and Ireland and colleagues (2011) reported that couples with higher matching were more likely to still be together three months later. Later, larger analyses found the link to satisfaction and commitment small or absent, and one of the original authors has said he no longer considers that result robust. We therefore treat sync as a description of attention and rhythm in text, not a forecast. Our word lists and formula are independent implementations of the published idea.
- Bending toward. Giles's Communication Accommodation Theory: people shift their style toward those they want to connect with, and away from those they don't.
- Texting similarity. Ohadi and colleagues (2018) found partners who text alike, in things like how often they initiate and how they use emoji, report higher relationship satisfaction.
- Positive words in messages. Slatcher and Pennebaker (2006) found couples who used more positive emotion words in their instant messages were more likely to still be dating three months later.
- We-words. A 2019 meta-analysis by Karan, Rosenthal and Robbins found we-talk is reliably linked with closer, better-functioning relationships.
- I-words. Rude, Gortner and Pennebaker (2004) linked heavier I-talk to depression; Tackman and colleagues (2019) found across thousands of people that it tracks negative emotionality broadly, with a small-to-moderate effect. A marker, not a test.
- Hot days, repair, 5 : 1, bids, harsh start. Gottman's observational work on couples: the Four Horsemen predict divorce, repair attempts predict who makes it, stable couples hold roughly five positives per negative in conflict, turning toward bids predicts staying together, and a harsh start predicts how a fight ends. Those were coded from video by trained observers; the text-message versions here are rough proxies.
Method and caveats: a mirror, not an oracle
The score is our implementation of linguistic style matching as described in the published research: for each of nine small-word families, 1 minus the normalized difference between the two speakers' rates, averaged. Small words are processed automatically, which is why matching on them tracks attention better than matching on topic.
What differs from the lab version. The word lists here were written from ordinary grammatical categories, not taken from any licensed dictionary, so absolute rates differ from academic tools; the trend shape is reliable, the second decimal is not. The Spanish lists are this app's own. Monthly and per-reply scores are noisier than a whole-conversation number. The bands ("in sync", "out of step") are heuristics, not published cutoffs. Communication compatibility is not relationship compatibility: this page sees how two people text, not their values, their time together, or what happens when the phone is down.
What it can and cannot say. LSM correlates with relationship initiation and stability in published studies, but the effects are modest. A low score is a prompt to look, not a verdict. Drift has boring causes too: a new phone, a new job, texting from a desk instead of a couch. That is what the notes are for. Sarcasm and teasing look like anger to a word list. Laughter cools the heat score and flags banter, but only you know the tone; the excerpt option lets Claude read a few hot days in context if you want a second opinion.
Privacy. Parsing and counting run entirely in this tab. Nothing is sent anywhere for that, and the messages are never stored. If you leave "remember" on, this browser keeps a numbers-only summary so your next check-in can show what changed and so you can compare chats. "Rewrite the cards" sends Claude a summary: the numbers, and with the first box ticked, topic words, repeated phrases and catchphrases. Excerpts are only sent if you tick the second box. Never the whole chat.