Search for the phrases that give a self-centred person away and you get the same answer nearly everywhere: the pronouns. All that “I,” “me,” and “my,” supposedly leaking out of anyone too wrapped up in themselves to notice. It sounds obviously right, which is usually a reason to check. And when researchers finally checked it properly, the obvious answer turned out to be mostly wrong. The real giveaway is quieter, harder to catch, and considerably more interesting.
The pronoun myth
The idea that self-absorbed people betray themselves with first-person pronouns is a durable piece of folk wisdom, and in 2015 it was tested at a scale that could actually settle it. Angela Carey and colleagues, among them James Pennebaker, who more or less founded the psychological study of pronouns, pooled data from 4,811 people across five laboratories, using five different measures of narcissism and two languages. They asked a simple question: do people who score higher in narcissism actually use more first-person singular pronouns?
The answer was essentially no. The correlation was so close to zero that, for practical purposes, counting a person’s “I”s tells you nothing about how self-important they are. Worse for the folk theory, where a rise in first-person singular does carry meaning, it tends to point in the opposite direction. Pennebaker’s broader work links heavier use of “I” not to grandiosity but to distress and low mood, because pain and self-doubt pull attention inward. The person saying “I” a great deal in a conversation is, on average, more likely to be quietly struggling than showing off. The tell everyone watches for is not just weak. It sometimes flags the opposite of what people think.
The real tell is a move, not a word
If not the pronouns, then what? The most useful answer comes from the sociologist Charles Derber, who recorded and transcribed more than a hundred ordinary dinner-table conversations for his book The Pursuit of Attention. What he found was that the signature of self-focus is not a vocabulary but a manoeuvre, one he called the shift response.
The distinction is easy to feel once named. Suppose someone tells you their dog got out and took three days to find. A support response keeps the attention on them: “Oh no, where did you finally track him down?” A shift response swings it quietly back to you: “My dog never leaves the garden, he’s far too nervous.” Both are perfectly normal sentences. But self-centred conversation is a steady accumulation of the second kind, the topic tugged homeward again and again, whoever happened to raise it. The phrases, when you start to hear them, are the small hinges of redirection: “that reminds me of when I,” “the exact same thing happened to me, only worse,” “well, when I did it.” It is not the words that reveal anything. It is the direction they always travel.
Why it is so hard to catch
What makes the shift response a better tell than the pronoun count is also what makes it easy to miss. Derber observed that conversational narcissism is usually subtle rather than blatant, because openly monopolising a conversation gets a person branded an egotist, which most people are keen to avoid. So the practised version mixes in just enough support, a question here, an interested “oh really” there, to seem genuinely engaged, before steering the topic back. No single move is damning. It is the pattern that does the telling, the way a conversation with certain people keeps arriving, as if by gravity, at the same place.
The line that keeps this fair
It is worth being careful here, because this is exactly the kind of idea that curdles into a party trick for judging people. Everyone uses shift responses, and they are not inherently selfish. Sharing a similar experience is one of the main ways human beings show empathy and build connection, and “that happened to me too” can be an offer of solidarity every bit as much as a theft of the spotlight. The line between relating and one-upping is real but fine, and it lies entirely in what comes next: whether the person hands the conversation back or keeps it. That is precisely why a single phrase proves nothing, and why using any of this to diagnose and dismiss the people around you is both unreliable and a little unkind. The pattern is worth noticing. It is most worth noticing in oneself.
What the evidence does and does not settle
A few honest limits belong here. The pronoun study concerns grandiose narcissism in particular and reports a near-zero average across large samples, not a verdict on any individual, and I-talk can still mean plenty of other things. Derber’s work is observational sociology rather than a controlled experiment, an illuminating framework built from listening rather than a law, and both bodies of work are shaped by their settings. Reading character from conversation remains guesswork at the best of times.
Within those limits, the correction stands, and it is more useful than the myth it replaces. The phrases that reveal self-centredness are not the ones the internet keeps circling. They are not pronouns at all, but the quiet redirects that bring every subject back home. And the most valuable place to listen for them is not in other people. It is in the small pause after someone has told you something, when you get to choose between asking them more and making it about you.