September 12, 2026
Small words, big story: what the words you never think about say about how two people text
Function words like I, you, the and not are chosen without thought, which is exactly why they track attention between two people. Here is what a sync score measures, what it does not, and how to read your own.
Ask someone what their partner texts about and they will tell you the topics: dinner, the dog, the thing at work. Ask what words they use and they will shrug. That shrug is the point. The words nobody notices, I, you, we, the, and, not, just, make up more than half of everything we write, and we do not choose them. They come out the way they come out.
Because they are unconscious, they are hard to fake, and because they are hard to fake, they say something. Two people who are paying close attention to each other drift toward the same mix of these small words. The effect was first measured in conversation transcripts and later in text messages, where it turned out to be easy to compute and surprisingly stable.
What a sync score is
Take every message one person sent and count how often each family of small words shows up: personal pronouns, articles, prepositions, negations, and so on. Do the same for the other person. Compare the two profiles family by family and average the result. That is a sync score. Identical profiles score 1.00; wildly different ones sit near 0.50.
What moves the number is not vocabulary or topic. Two people can talk about completely different things and still be in step, and two people can talk about the same weekend in styles that never touch. What moves it is attention: whether each reply is shaped, a little, by the message it answers.
What it is not
A sync score is not a compatibility test and it is not a prediction. The original studies found that couples with higher matching were more likely to still be together months later, but that finding has not held up cleanly in every replication, and the researchers themselves now describe it as descriptive rather than predictive. Treat it as a mirror. It shows you how the two of you text; it does not know how you love.
It also cannot hear tone. Sarcasm reads as sincerity. A fight conducted entirely in polite sentences reads as calm. Anything that happens in person or on a call is invisible. If the score says one thing and your gut says another, your gut has more data.
Why the trend matters more than the number
A single score is a snapshot. The useful thing is the line: the same measure computed month by month across a whole conversation. Feelings are accurate about the present and unreliable about trends, and nobody remembers that their replies slowed in March or that "we" quietly dropped out of their messages in the spring. A slow slide usually tracks attention, which usually tracks the calendar: a new job, a move, a hard season. Seeing when it started is the whole value.
How to read your own
Import a chat export and look at three things before the score. First, the days you were on fire, the ones with the most laughter and warmth, because they tell you what the two of you are like at your best. Second, the moments one of you had a rough day and the other showed up within the hour; researchers call this responsiveness and it is the single best-supported predictor of a relationship feeling good from the inside. Third, whether tense days, if there were any, got closed. Repairs matter more than fights.
Then look at the number, and then at the line. If the line is flat, steady is a result. If it moved, ask what else changed that month. That question is usually more useful than the answer the score gives you.
How We Text does all of this in your browser, without uploading the export, and the other person never knows.
Try it on your own chat. Import a WhatsApp or iMessage export; it runs in your browser and nothing is uploaded.
Open How We Text