The advice to stop hedging is built on a real finding and the wrong variable
The advice has been going round for about a decade now. Cut just from your emails. Stop apologising. Never end a sentence on a tag question. It resurfaces every January with a browser plugin and a LinkedIn post attached, and with it the suggestion that anyone still typing I just wanted to check is dismantling their own case one word at a time.
Underneath all of it there is a real study. It is older than the advice lets on, and odder.
The courtroom study everything rests on
Researchers in 1978 recorded courtroom testimony, then cut it into versions that differed in style and nothing else. One hedged, hesitated, piled on intensifiers, finished statements with tag questions. The other did none of that. The content never moved. Listeners marked the hedging witnesses down on credibility and on attractiveness, and were less willing to go along with them.
Up to that point the advice describes its source fairly enough. What came after it is the part that goes missing.
The follow up that changed what it meant
Robin Lakoff had offered those features as characteristic of the way women speak, and that is where the list came from. William O'Barr and Bowman Atkins then spent thirty months sitting in a working trial court to see whether it held up.
It did not. The features did cluster, and they clustered reliably, but not by sex. They clustered by standing. Expert witnesses used few of them, whatever their sex; witnesses with less education and lower status jobs used plenty. O'Barr and Atkins concluded that the cluster was neither characteristic of all women nor confined to women, and renamed it powerless language.
The words were not causing the low status. They were tracking it. So advice to delete the words is a proposal to address a symptom, handed to the person with the least power in the room.
That sits a good deal less comfortably than stop saying just, which is presumably why it so seldom survives the trip into a listicle.
How big is the gender gap anyway
Campbell Leaper and Rachael Robnett asked the question head on, pooling 29 studies that covered 3,502 people between them. Women do use somewhat more tentative language. The effect size is 0.23, which is small, and the two groups overlap far more than they differ.
A difference that size will not carry advice aimed at half the workforce. Knowing someone's sex tells you very little about how they will word a request, and a column that opens by telling women they hedge too much has already assumed more than the evidence provides.
And sometimes hedging works
The result that ought to have finished off the simple version arrived in 1990. Linda Carli played 120 listeners one persuasive message, varying nothing but the speaker's sex and whether the delivery was tentative. Tentative women persuaded the men in the audience more. They persuaded the women less.
The same sentences moved opinion in opposite directions, depending on who was sitting in the room. This is a laboratory study from 1990 and it carries those limits, yet nobody has since put back together what it took apart. Whether hedging costs you has no single answer, because the answer depends on who is listening and what they expect.
What we think is actually worth practising
Two kinds of hedge get muddled together, and telling them apart is most of the useful advice going in this area.
One kind carries no information. I just wanted to quickly check whether maybe we could possibly is nine words of throat clearing in front of a request, and cutting them costs the sentence nothing, because they were never saying anything to begin with. The other kind is doing work. The evidence here is contested is a hedge, and striking it out would make the sentence false. This site is full of the second sort on purpose, and we would rather sound less certain than be less accurate.
Day one of our own fourteen day speaking programme drills that distinction and little else, and we should be straight about what a fortnight of forced choices can and cannot do. It will make you better at picking the stronger of two written lines. That is near transfer, the same modest claim we make about every game of this shape in our piece on brain training, and it would be dishonest to claim anything grander two pages later.
What a fortnight can do is make the padding visible. You cannot choose between two versions of a sentence without noticing which one you would have reached for yourself, and that noticing is the part which does not otherwise happen. Whether it survives into a real conversation with real stakes attached, we have not demonstrated, and neither has anyone selling you a plugin.
Hedging, apologising and sounding sure
Does hedging really make you sound less credible?
In the setting it was tested in, yes. A 1978 study played people courtroom testimony that differed only in hedges, hesitations, intensifiers and tag questions, and the witnesses using them were rated less credible. That is a real finding about a jury listening to a stranger. How far it carries into a meeting with people who already know you is a question the study was not asked.
Should women stop saying just and sorry at work?
The advice is aimed at the wrong variable. When researchers tested the features that had been labelled women's language across thirty months of courtroom observation, they sorted by the speaker's social standing rather than by sex, and were found to be neither characteristic of all women nor limited to women. The measured gap between how tentatively men and women speak is also small, about 0.23, with far more overlap than difference.
Is tentative language always a disadvantage?
No, and the cleanest demonstration is from 1990. The same persuasive message, delivered tentatively by a woman, was more influential with men in the audience and less influential with women. Identical sentences, opposite effects, decided by who was listening. Any rule that tells everyone to cut the same words is ignoring that.
What is the difference between a hedge worth cutting and one worth keeping?
Whether it carries information. I think maybe we could possibly look at this adds nothing and costs something. The evidence here is contested is a hedge too, and removing it would make the sentence false. The first is padding. The second is accuracy, and this site runs on the second kind.
Will a game actually make me speak differently?
It will reliably make you better at the game, which is the honest limit on anything of this shape. What a fortnight of forced choices can do is make the padding visible, because you cannot pick between two versions of a sentence without noticing which one you would have said. Whether that carries into a difficult conversation is the part nobody has demonstrated, including us.
Does speaking more slowly and lower actually help?
It is one of the days in our own programme, and it is the one we would defend least confidently. Delivery clearly matters to how people are read. The specific prescriptions attached to it, drop your pitch, slow down, hold the pause, are better supported as stage advice than as findings, and anyone presenting them as settled science is overstating what is known.