On this path · Pragmatics 1. Politeness, Face, and Indirectness
Politeness, Face, and Indirectness
Explain why English softens requests using the idea of face.
Learning outcomes
You can write a grammatically perfect request and still make the reader bristle. “Send me the deploy logs” is flawless English, yet sent to a senior engineer you barely know, it lands like an order. The previous pages gave you the dials of register and tone and the modal toolkit for softening. This page explains the force those tools are working against: a social need called face, and the precise machinery English uses to protect it. Once you see the machinery, the softening stops feeling like random decoration and starts feeling like a calculation you can do.
After studying this page, you can:
- Explain what face is, separate positive face from negative face, and say which one a given sentence threatens.
- Recognize a face-threatening act (a request, a criticism, a refusal, a disagreement) and predict why English will soften it.
- Place the four politeness strategies on a ladder from bald-on-record to off-record and pick the right rung for a situation.
- Use the three variables, social distance, power gap, and size of the imposition, to set how much politeness an act needs.
- Avoid both non-native traps: translating a request that turns blunt in English, and over-softening until the message disappears.
Before we dive in
Imagine you need ten minutes of a senior architect’s time to unblock yourself. In your first language, the natural, perfectly polite way to ask might translate word for word as “Explain to me how the auth flow works.” You send it. The architect reads it as curt, maybe even demanding, and you have no idea why, because in your head you were being polite. Nothing in the grammar is wrong. The sentence simply skipped a layer of social work that English expects and your first language handled differently, or handled with tone of voice that does not survive into text.
That missing layer has a name. In the 1970s and 80s, two researchers, Penelope Brown and Stephen Levinson, asked why speakers across very different languages keep softening the same kinds of sentences in the same kinds of situations. Why do we say “would you mind taking a look” instead of “look at this,” and “I think there might be a small issue” instead of “this is wrong”? Their answer was that every interaction is managing something fragile that they called face. English does not soften requests out of habit or weakness. It softens them because a bare request damages face, and the speaker is paying to repair the damage in advance. This page is about that fragile thing and the four levers English pulls to protect it.
Face: the thing English keeps protecting
Face is the public self-image every person wants to maintain in an interaction. It is not pride or ego; it is the social standing you are claiming right now, in front of this person, and it can be supported or hurt by what others say. The word comes from the everyday sense of “losing face”: being embarrassed, talked down to, or cornered in public.
Brown and Levinson split face into two wants that pull in different directions, and keeping them separate is most of the work.
Positive face is the wish to be liked, approved of, and treated as a member of the group. It is satisfied when someone compliments your work, includes you, agrees with you, or remembers your name. It is threatened when someone criticizes you, ignores you, or treats your contribution as worthless.
Negative face is the wish to act freely, without being imposed on or having your time and choices taken away. (The word “negative” here means freedom from, not bad.) It is satisfied when people leave you room and do not make demands. It is threatened the moment someone asks you for a favor, gives you an order, or commits you to something, because each of those eats into your freedom to do as you please.
A single sentence can threaten one or both. “Your function has a bug” mostly hits positive face: it suggests your work is not good. “Can you rewrite this function?” mostly hits negative face: it imposes a task on you. “You broke the build, fix it now” hits both at once, which is why it stings so much.
Positive face is the wish to be approved of and included. Negative face is the wish to be left unimposed-on and free. Almost every polite move in English is aimed at one of these: warmth and inclusion repair positive face, while indirectness and giving an out repair negative face. Knowing which face an act threatens tells you which kind of softening to reach for.
Face-threatening acts: what puts face at risk
A face-threatening act, often shortened to FTA, is anything you say that puts someone’s face at risk. The category is wide and it covers the everyday moves you cannot avoid at work: asking for help (threatens their negative face, their freedom), criticizing or correcting (threatens their positive face, their competence), refusing a request (threatens their positive face, by rejecting what they wanted), and disagreeing in a meeting (threatens their positive face, by saying they are wrong).
The key insight is that these acts are unavoidable. You cannot do your job without asking for things, pointing out problems, saying no, and disagreeing. So English does not try to skip the act; it tries to perform the act while paying to soften the blow. The softening is not optional politeness frosting. It is the price of doing the threatening thing without damaging the relationship.
How much you pay depends on how big the threat is, and that is what the four strategies and three variables, coming next, let you calculate.
The four strategies, ordered by how much they soften
Brown and Levinson found that the ways speakers handle a face-threatening act line up on a single scale, ordered by how much they soften the blow. Picture a ladder. At the bottom you do nothing to soften; at the top you soften so much you barely say the thing at all. The right rung depends on how threatening the act is: a small, safe act sits low on the ladder, a large, risky one climbs high.
Read the ladder from the bottom up, because each rung adds one more layer of protection.
Bald-on-record is the bottom rung: you state the act directly, with no softening at all. “Send me the logs.” “This is broken.” It seems rude, and often it is, but it is exactly right in two cases. The first is an emergency, where efficiency beats politeness: “Stop the deploy, the database is down” should not be hedged. The second is genuine closeness, where softening would feel cold and fake: between two friends who pair every day, “pass me the cable” needs no cushioning.
Positive politeness is the next rung: you soften by appealing to the other person’s positive face, showing you like them and count them as one of your own. You compliment (“you always spot these things”), you include with “we” (“how should we handle this?”), you make small talk, you use their name and a friendly opener. The request still arrives, but wrapped in solidarity, so it reads as a friend asking a friend.
Negative politeness is the rung above: you soften by protecting the other person’s negative face, their freedom. You make the imposition small and easy to refuse. You are indirect (“would you be able to”), you hedge (“I think,” “maybe,” “a little”), you apologize for intruding (“sorry to bother you”), and you give them an out (“no worries if you are swamped”). The classic shape is “I know you are busy, but would you mind taking a look when you get a chance?” Every clause there is reducing the imposition.
Off-record is the top rung: you do not openly commit the act at all. You hint, and you leave the other person to infer what you want, which lets both of you pretend nothing was asked. “It’s cold in here” can be a request to close the window without ever being one. “I still can’t find those logs” can be a request for the logs without asking. If they choose not to take the hint, no one has lost face, because officially you never asked.
The ladder is not a quality scale where the top rung is the best English. It is a dose scale. Climbing too high for a tiny request (going off-record to ask a teammate for a pen) is as wrong as staying too low for a big one (bald-on-record to ask your CEO for a week of their time). The skill is matching the rung to the threat, which the next section makes precise.
The three variables that set the dose
If the ladder tells you the menu of strategies, three variables tell you how high to climb. Brown and Levinson found that the weight of a face-threatening act, and so the amount of politeness it needs, rises with three things added together.
Social distance is how well you know the person and how close you are. A stranger or a new colleague is high distance; a teammate you have worked beside for years is low distance. More distance means more politeness, because you have no shared history to cushion the act.
Power gap is how much more authority or status the other person has over you. Asking your direct report for something carries a small power gap; asking the CEO or a senior staff engineer carries a large one. More power above you means more politeness, because the act risks more.
Size of the imposition is how big a thing you are asking or how hard the criticism cuts. “Got a second?” is a tiny imposition; “can you spend your weekend on this?” is a huge one. A bigger ask needs more softening, regardless of who it goes to.
The three add up. A small favor from a close peer needs almost nothing, so you stay near bald-on-record. A large favor from a distant senior needs a great deal, so you climb toward negative politeness or even off-record. The table coming up shows this addition working across four real cases.
Four workplace cases, worked
Theory becomes usable when you watch it run on the cases you actually face. The table below takes four common face-threatening acts, shows the too-blunt version that ignores face, then the face-saving version that protects it, and names the strategy doing the work. Read each row as a before and after.
Four face-threatening acts at work: the too-blunt version that ignores face, the face-saving version that protects it, and the strategy used.
| Face-threatening act | Too blunt (ignores face) | Face-saving version | Strategy used |
|---|---|---|---|
| Asking a busy senior for help | Explain how the auth flow works. | Sorry to bother you, I know you're slammed. When you have a moment, could you point me to how the auth flow fits together? No rush at all. | Negative politeness: apology, hedge, give an out |
| Pointing out a bug in a pull request | This is wrong, it'll crash on empty input. | Nice work on this. One thing I noticed: I think this might crash on empty input. Could we add a guard there? | Positive politeness plus hedging: compliment, 'we', 'might' |
| Declining a request from a colleague | No, I can't do that. | I'd really like to help, but I'm fully committed to the release this week. Could we find time next week, or is there someone else who could jump in? | Positive politeness: show goodwill, offer an alternative |
| Disagreeing in a meeting | That won't work. | I see what you mean, and I might be missing something, but I'm a little worried this could slow the API down. Have we considered caching it? | Negative politeness plus hedging: acknowledge, hedge, question |
Notice the pattern across the rows. The blunt versions are not ungrammatical; they are face-blind. The face-saving versions add three kinds of work: a positive-face move (a compliment, a “we”, a statement of goodwill), a negative-face move (an apology, an out, a hedge), and often a question form that hands control back to the other person. Notice too that the strategy shifts with the act: criticism leans on positive politeness because it threatens competence, while a request to a senior leans on negative politeness because it threatens their freedom and time. The hedges here (“I think,” “might,” “a little,” “could”) are the same softening modals from the modal toolkit, now doing face work, and the warmth (“nice work,” “I’d really like to help”) is the warmth dial from register and tone turned up to protect positive face.
You need to tell a teammate, gently, that the API design in their PR has a flaw. Which version best protects their positive face while still making the point?
The two non-native traps
Two opposite mistakes catch advanced learners, and they sit at the two ends of the ladder.
The first trap is under-softening by direct translation. Many languages mark politeness with a verb ending, an honorific, or a tone of voice rather than with the extra clauses English uses. A request that is fully polite in your first language can translate into bare English that lands far too blunt, because the politeness lived in a feature that did not cross over. “Send me the report by Friday” may be the exact, courteous form in your language, but in English it reads as a command. The fix is not to add formality; it is to add face work: a softening modal, an out, a touch of warmth. “Could you get the report to me by Friday? Let me know if that timeline is tight.”
The second trap is over-softening until the message is lost. Learners who have learned that English softens sometimes overcorrect, piling on apologies, hedges, and qualifiers until the actual request drowns. “I’m so sorry to bother you, I know this is probably a silly question and I might be completely wrong, but I was just wondering if it might possibly be okay to maybe ask whether you could perhaps take a quick look sometime, only if you have time of course.” The reader cannot find the request, and worse, the avalanche of apology reads as anxious or even insincere, which damages your own positive face. Real politeness is enough softening to protect face and not one clause more.
More hedging is not more polite past a certain point; it is noise. When every sentence carries an apology and three qualifiers, the reader has to dig for the meaning, and the wall of softening signals that you do not feel entitled to ask at all, which is its own kind of awkward. Aim for one clear positive-face move and one clear negative-face move, then state the request plainly. “Quick favor, no rush: could you send the logs when you get a sec?” protects face in one breath and still says exactly what it wants. If a reader cannot restate your request in one sentence after reading it, you have over-softened.
Mental Model: politeness is a dose, not a constant
The wrong model, and the one that produces both traps, is that politeness is a constant: a setting you turn on and leave on. Under that model you are either “a polite person” or not, and you apply the same level of softening to everything, which is why some learners are blunt everywhere and others are apologetic everywhere. Both are using one fixed dose for situations that call for wildly different amounts.
The better model is that politeness is a dose you calculate fresh for each act. You look at the act (how much does it threaten face, and which face?), you read the three variables (how distant is this person, how much power do they hold, how big is the imposition?), and you climb the ladder to the rung that matches. A tiny ask from a close friend gets a near-zero dose: bald-on-record is correct, and softening it would feel strange. A large ask from a distant senior gets a heavy dose: you climb to negative politeness or hint off-record. Same person can need different doses on different days, because the imposition changed.
This reframes the whole skill. You are not deciding whether to be polite. You are reading a situation and prescribing an amount, the way a doctor doses a medicine to the patient and the illness, not to a fixed rule. Native speakers do this calculation in a flash from long practice. Until it is automatic for you, do it on purpose: name the act, weigh the three variables, pick the rung.
Common mistakes
The first mistake is being too blunt, which reads as rude. This is usually the translation trap: a request that was polite in your first language arrives in English stripped of the face work English expects. The cost is real, the reader feels ordered around or criticized, even though you meant no harm. The fix is to add a softening modal and an out, not to add formality.
The second mistake is being too indirect, where the message gets lost. Climbing to off-record or piling on hedges for an ordinary request leaves the reader unsure what you actually want. Off-record is a real strategy, but it is for high-threat acts and situations where deniability matters, not for asking a teammate to send a file. When in doubt for a normal work request, sit at positive or negative politeness, which are clear and still kind.
The third mistake is apologizing for everything. Starting every message with “sorry” trains the reader to tune it out and slowly lowers your own standing, because constant apology signals you feel you are always in the way. Reserve “sorry to bother you” for a genuine imposition on someone’s time, and let small, ordinary requests stand without it.
The fourth mistake is forgetting that cultures dose differently. The right amount of softening for the same act varies a great deal across cultures and even across companies, so a level that is warm and normal in one place reads as cold or as groveling in another. Calibrating across those differences is its own skill, treated in directness across cultures; for the softening tools themselves, see hedging and stance.
Mastery Questions
Question
What is face, and how do positive face and negative face differ? Give an act that threatens each.
Answer
Face is the public self-image a person wants to maintain in an interaction, the social standing they are claiming right now. Brown and Levinson split it into two wants. Positive face is the wish to be liked, approved of, and included; it is threatened by criticism, for example 'your function has a bug', which questions someone's competence. Negative face is the wish to act freely without being imposed on; it is threatened by a request or order, for example 'can you rewrite this function?', which eats into someone's time and freedom. Warmth and inclusion repair positive face; indirectness and giving an out repair negative face, so naming which face an act threatens tells you which softening to use.
Sources & evidence7 claims · 5 cited
Face, the four strategies as a softening ladder, and the three weighting variables are grounded in Brown and Levinson; Grice and Leech underpin off-record inference. Workplace examples and the dosing mental model are internal reasoning. No quantitative claims.
- Face is a person's public self-image in interaction, and Brown and Levinson divide it into positive face (the want to be approved of and included) and negative face (the want to act unimpeded and unimposed-on).verified
- A face-threatening act is an act (such as a request, criticism, refusal, or disagreement) that runs contrary to the face wants of the addressee or speaker, which is why speakers redress it.verified
- Brown and Levinson order strategies for performing a face-threatening act by increasing redress: bald-on-record, positive politeness, negative politeness, and off-record (hinting so the act is not openly committed).verified
- The weightiness of a face-threatening act, and thus the politeness it needs, increases with three social variables: the social distance between speaker and hearer, the relative power of the hearer over the speaker, and the absolute ranking of the imposition.verified
- Bald-on-record performance of a face-threatening act, without redress, is appropriate in cases such as urgency or efficiency and where the threat to face is very small (including high familiarity), rather than being always impolite.verified
- English commonly redresses requests and criticism with indirect and hedged forms (modal questions like could you, mitigators, apologies, and giving an out), so a request that is polite in another language can land as blunt when translated directly into bare English.stable common knowledge
- Over-redressing an ordinary request by stacking apologies, hedges, and qualifiers obscures the message and can read as awkward or insincere, so effective politeness uses enough redress to protect face and no more.internal reasoning
Cited sources
- Politeness: Some Universals in Language Usage · Penelope Brown and Stephen C. Levinson
- Principles of Pragmatics · Geoffrey Leech
- Logic and Conversation · H. P. Grice
- Cambridge Grammar of English · Carter and McCarthy
- Longman Grammar of Spoken and Written English · Biber, Johansson, Leech, Conrad, Finegan