The advice arrives reliably in career guides aimed at women: stop apologising, stop hedging, remove "just" and "sorry" from your emails.

There is a real finding underneath it. There is also a substantial gap between that finding and the advice, and the gap is worth examining before anyone edits another email.

The finding

Diary studies asking participants to record every apology they offered and every offence they committed found that women reported apologising more often than men. They also reported committing more offences.

The number of apologies per offence was similar between groups.

That last sentence is the actual result, and it is almost never quoted. The difference the researchers identified was not in propensity to apologise. It was in the threshold at which an event was classified as warranting an apology at all.

Men in these studies were not suppressing apologies they felt were owed. They had categorised fewer events as owing one.

What that changes

If the difference is in threshold rather than in behaviour, then "apologise less" is instruction to reclassify events — which is a different and much harder request than editing a word out of a sentence.

It also raises the question of which threshold is correct, and there is no obvious reason to assume it is the higher one. A person who does not register that she has inconvenienced someone is not operating at a superior standard. She is less attentive to a real thing.

The relevant problem is not that one group is too apologetic. It is that in professional settings, apology functions as a status signal independently of whether it is warranted, so accurate acknowledgement gets read as weakness.

The two different words

English uses "sorry" for at least two distinct speech acts, and conflating them causes most of the confusion.

The first is an admission of fault: I did this, it was wrong, I accept responsibility. The second is an expression of sympathy or a social lubricant: sorry to hear that, sorry to interrupt, sorry — could you repeat that.

Counts of "sorry" in workplace communication capture both. The second category is doing conversational work that is not self-diminishing at all in most contexts, and removing it produces communication that reads as brusque rather than as authoritative.

Research on communication style suggests that hedging and softening language serves genuine coordination functions — it signals openness to correction, invites input, and reduces the cost of disagreement. It is not merely noise to be stripped out.

The double bind

Here is where the popular advice becomes actively risky.

Evaluation studies have repeatedly found that assertive communication is rated differently depending on who produces it. Directness that reads as confident from one speaker reads as abrasive from another. The same message, attributed to different senders, receives different competence and likeability ratings.

Meaning: a woman who follows the standard advice to strip all softening from her communication may be rated as less warm without being rated as more competent, which is the worst available square of that matrix.

This is not an argument for hedging. It is an argument that the advice is being given as though the constraint were internal, when it is at least partly a property of the audience — and advice that only one party has to follow does not fix an asymmetric evaluation.

Where it starts

Threshold differences are learned early. Observational research on children's play finds that girls' groups more often use conflict-mitigating language and boys' groups more often use direct commands, from primary age onward.

Adults reinforce this differentially — the expectation that a girl will smooth over a social disruption is well documented in classroom observation, and the disapproval directed at girls who do not is greater than that directed at boys who do not.

What is being trained is not politeness. It is responsibility for the emotional state of a room, which is a real and largely uncompensated job.

Advice that survives the evidence

Distinguish the two sorries. Keep the sympathetic one; it is not costing you anything. Examine the fault-admitting one and ask whether the fault is yours. Frequently it is not — "sorry for the delay" when someone else caused the delay is the specific pattern worth catching.

Replace rather than delete. "Sorry this is late" and "thanks for your patience" carry the same social function; only one assigns blame. This is the one piece of the popular advice that holds up well, and it avoids the brusqueness problem entirely.

Notice who is doing the smoothing. In any group, someone is managing the friction. If it is always the same person and it is not in her job description, that is a distribution problem, not a speech habit.

And be sceptical of any advice that locates an entire asymmetry inside the person on the losing end of it. Editing your own vocabulary is available and sometimes worth doing. It is not a substitute for the observation that two people can say the same sentence and be heard differently.