Can an apology be too polished to feel sincere?

Here are two things that are both true, according to recent research. People who choose harder, more effortful words when they apologize are perceived as more sincere.

People whose apology sounds too smooth, too rehearsed, are perceived as less sincere.

Set those two findings side by side and they seem to fight each other, since effort and polish usually travel together in everyday speech. They stop fighting each other once it becomes clear they are describing two different kinds of effort.

The case for effort

Shiri Lev-Ari, a psychologist at Royal Holloway, University of London, wanted to know whether the words themselves carry a sincerity signal, separate from tone or body language. In one study, she compared the language people used in apologetic posts on social media against the language the same people used in their other posts, and found that the apologies leaned on longer words. In a second study, participants ranked near-identical apologetic sentences that differed only in word length and how common the words were, and the versions built from longer, less common words were consistently rated as the more apologetic ones.

Lev-Ari argues the mechanism is fairly simple: a longer word costs a little more effort to say and a little more effort to retrieve from memory, and that small, mostly involuntary cost is what a listener seems to be picking up on. As she put it in her write-up of the findings, that extra length “signals effort on the side of the apologizers,” the same way paying for damages or showing up in person signals effort in other kinds of apologies. Choosing the harder word, in other words, works a little like choosing the harder gesture.

Run the same idea through an ordinary text message and it holds up. “Sorry about that” costs almost nothing to type and almost nothing to read. “I feel genuinely awful about how that came across” costs more of both, and the extra length is doing real work: it takes longer to write, it takes a beat longer to say out loud, and something in a reader or listener seems to notice that small delay and count it in the apologizer’s favor. Nobody consciously tallies syllables while deciding whether to forgive someone. The tally appears to happen anyway.

The case against polish

Researchers Florian Bonensteffen, Sven Zebel, and Ellen Giebels ran a different kind of study, showing people a filmed apology and tracking, among other things, how convincing they found it and where their attention went while watching. Buried in the limitations section of their published findings is a detail that cuts against Lev-Ari’s results: several participants said the apology on screen struck them as scripted, and that impression on its own made them trust it less, apart from the actual content of what was said.

The broader findings from that study suggest a reason why. People do not just listen to the words of an apology, they watch for signs that the apologizer is genuinely struggling with what they did: a pause, some visible discomfort, a sentence that trails off before landing. A delivery so smooth it never catches on anything can read as proof that no real struggle is happening at all, whatever the words themselves happen to be. Suffering and responsibility taking, not just phrasing, are what that study’s participants leaned on most heavily when deciding whether an apology was real.

This is a familiar feeling to anyone who has watched a company’s official apology statement after some public failure. The sentences are grammatically flawless, the tone is measured, the structure follows a predictable shape: acknowledge the issue, express regret, describe corrective steps, thank the audience for their patience. It reads like it was drafted by committee and approved by legal, because it almost always was, and audiences tend to sense exactly that. The smoothness is not neutral. It actively signals that the words were chosen for the reader’s benefit rather than pulled, imperfectly, out of the apologizer’s own discomfort.

Effort in the words, not in the performance

What seems to reconcile these two findings is a distinction between two different places effort can show up. Lev-Ari’s effect lives in word choice: a person reaching for a heavier, more specific word instead of a default one is doing something that costs a small, usually unconscious amount of cognitive work, and listeners appear to register that cost as sincerity. The scripted-apology effect lives somewhere else entirely, in pacing and structure and the total absence of hesitation, all of which read less like someone processing regret in real time and more like someone reciting a line they have used before. One kind of effort is invisible and lands as sincerity. The other kind of effort is visible, in the sense that its absence of friction is itself noticeable, and it lands as rehearsal.

Put the two together and the old advice about apologizing, the version that says choose your words with care, starts to look reasonable again, as long as the care shows up as specific, weighty language rather than flawless delivery. An apology that stumbles slightly while still landing on the right word can do more work than one that is stylistically perfect but sounds rehearsed in front of a mirror first. The words can afford to be considered. The performance probably should not be, or at least not so considered that it stops looking like a real-time reaction to having done something wrong.

Where that leaves an actual apology

None of this adds up to a formula for sounding sincere, and trying to reverse-engineer one from these findings is probably its own kind of tell. What it does explain is why an apology that arrives a little rough around the edges, missing its usual polish, sometimes lands better than one that is note-perfect. The words carrying real weight and the delivery carrying visible hesitation turn out to be doing the same job, just from different directions, and a person on the receiving end seems to notice both.