Politeness Theory (Brown & Levinson)

Face: the public self in interaction

Erving Goffman, writing in Interaction Ritual (1967), proposed that every encounter is staged around face — the positive social value a person claims for themselves. We work, often unconsciously, to maintain our own face and to support the face of others. Politeness, in Goffman's view, is the visible surface of that face-work.

Penelope Brown and Stephen Levinson, in their 1978 essay Universals in Language Usage: Politeness Phenomena (reissued 1987 as Politeness: Some Universals in Language Usage), formalised this insight into a predictive theory. They split face into two complementary wants:

• Positive face: the want that one's self-image be approved of — that others share or at least respect what one values.
• Negative face: the want that one's actions be unimpeded by others — that one's autonomy and territory be respected.

The "negative" in negative face is not pejorative. It marks the want for non-interference, mirroring the use of "negative liberty" in political philosophy.

Face-threatening acts (FTAs)

A face-threatening act is any utterance whose force runs against the face wants of the speaker, the hearer, or both. Brown and Levinson tabulate dozens; the principal classes are:

• Threats to hearer's negative face: requests, orders, advice, warnings, threats. They all impose on H's freedom of action.
• Threats to hearer's positive face: criticisms, contradictions, complaints, insults. They suggest H is not approved of.
• Threats to speaker's positive face: apologies, confessions, accepting compliments awkwardly. They diminish S's standing.
• Threats to speaker's negative face: making promises, accepting offers, expressing thanks. They commit S to future action.

Almost any non-trivial speech act threatens face somewhere. The interesting question is how speakers package them.

The five strategies

Brown and Levinson rank the strategies along a single axis: how much face protection is built into the wording, balanced against how clearly the act is performed.

1. Bald on record. Perform the FTA directly, without redress. "Pass the salt." Used when efficiency outweighs face — emergencies ("Watch out!"), close intimates, or contexts of clear authority.
2. Positive politeness. Perform the FTA but redress it by attending to H's positive face — common ground, in-group markers, compliments, optimism. "Hey buddy, lend me a tenner?" The "we're on the same team" move.
3. Negative politeness. Perform the FTA but redress it by minimising imposition on H's negative face — hedges, apologies, conventional indirectness, deference. "I'm sorry to bother you, but could you possibly pass the salt?" The "I respect your autonomy" move.
4. Off record. Perform the FTA indirectly, leaving plausible deniability. "It's a bit cold in here." (Hint: close the window.) The hearer must work out the implicature; the speaker can disclaim.
5. Don't perform the FTA. The most face-protecting option. The cost is that the goal is not achieved.

Strategy choice is meant to be predictable from a weighting:

W_x = D(S, H) + P(H, S) + R_x

where W_x is the seriousness of FTA x, D is social distance between speaker and hearer, P is power difference (H's power over S), and R is the ranking of imposition for the act in the culture. The higher W_x, the higher-numbered the strategy a rational actor selects.

Comparing the five strategies

Strategy	Face protection	Clarity of FTA	Typical context	English example
1. Bald on record	None	Maximum	Urgent, intimate, or hierarchical	"Pass the salt."
2. Positive politeness	Aimed at hearer's positive face	High	In-group, friendly equals	"Hey, you're great with this — pass the salt?"
3. Negative politeness	Aimed at hearer's negative face	Moderate	Strangers, asymmetric power	"Could you possibly pass me the salt, if it's no trouble?"
4. Off record	Maximum (deniable)	Low	High-risk requests, taboo topics	"This soup needs something."
5. Don't perform	Total	Zero (no act)	Cost outweighs benefit	(Reach across the table yourself.)
Compare diagonally: as you go down, face protection rises and clarity falls.

A worked example: borrowing money

Suppose a speaker wants to borrow $1,000 — a high-R imposition. The strategy choice is sensitive to D and P:

• Sibling (low D, low P), low-stakes loan: bald on record. "Lend me a thousand bucks?"
• Old friend (low D, low P), serious loan: positive politeness with hedges. "Look, I hate to ask, but could you spot me a grand? I'll pay you back next month, swear."
• Boss (high D, high P), serious loan: negative politeness or off-record. "I'm in a tight spot at the bank this month — anything you might be able to do would be a huge help."
• Stranger on a bus (high D, no P, very high R): don't perform.

The same propositional content (request: borrow $1,000) is wrapped in radically different linguistic forms because D, P, and R weight differently in each scenario.

Cross-linguistic data

Brown and Levinson claimed universality on the basis of comparing English, Tzeltal (a Mayan language of Chiapas), and Tamil. Subsequent typological work has both supported and complicated their claims.

• The T-V distinction. Many European languages (French tu / vous, German du / Sie, Spanish tú / usted, Russian ты / вы) grammaticalise the negative-politeness choice into the second-person pronoun. The marked form (V) raises the address by one notch on the social hierarchy. English lost its V form (thou originally being the T form, you the V) and now relies on lexical hedges.
• Honorific systems. Japanese has a layered honorific system — sonkeigo (respectful: actions of the addressee), kenjōgo (humble: actions of the speaker), and teineigo (polite: desu/masu forms) — that grammaticalises power and distance into verb morphology. Korean similarly distinguishes seven speech levels.
• Javanese and Balinese have speech levels — entire registers (Javanese ngoko / madya / krama) chosen based on relative status. The lexical replacement is so extensive that calling it "politeness" understates how thoroughly social structure is encoded.
• Tzeltal, in Brown's own fieldwork, uses heavy positive-politeness markers (in-group address, repetition, point-of-view operations) where English would use negative politeness. This was a key piece of evidence in the original argument.

The cross-linguistic comparison is what made Brown and Levinson's framework controversial. Matsumoto (1988) argued that Japanese politeness rests not on protecting individual autonomy but on placing oneself correctly within a group hierarchy — making "negative face" a poor fit. Gu (1990) made parallel arguments for Chinese, where politeness norms are rooted in moral and social obligation (limao) rather than face-saving strategy.

Inside positive and negative politeness

Brown and Levinson identify fifteen positive-politeness sub-strategies and ten negative ones. A subset:

• Pos 1: Notice/attend to H. "You must be tired — let me get you a drink."
• Pos 4: In-group identity markers. "Lend us a fiver, mate?"
• Pos 7: Presuppose common ground. "We're all in the same boat here, right?"
• Neg 1: Be conventionally indirect. "Could you possibly…?"
• Neg 2: Question, hedge. "I was wondering if you might…"
• Neg 5: Give deference. "I know you're busy, but…"
• Neg 6: Apologise. "Sorry to bother you…"
• Neg 7: Impersonalise. "It would be appreciated if…"

Strategies stack. A high-W FTA may carry several markers at once: "I hate to bother you, but I was just wondering if you might possibly be able to look at this when you have a moment." Five negative-politeness moves in one sentence.

Critiques and refinements

• Cultural specificity. Discernment-based models (Hill et al. 1986; Ide 1989) argue East Asian languages encode politeness as wakimae — knowing one's place — rather than strategic face-work.
• Non-strategic politeness. Watts (2003) and Eelen (2001) argue much polite behaviour is normative and habitual, not rational FTA-weighting.
• Im/politeness. Culpeper's work on impoliteness (1996, 2011) extends the framework to deliberately face-attacking acts — insults, mockery, sarcasm.

Why politeness theory matters

• Cross-cultural communication. Mismatched D / P / R weighting produces miscommunication that feels like rudeness — Anglo-Americans often read Japanese deference as evasiveness; Japanese speakers read American directness as crudeness.
• Workplace interaction. Power-asymmetric requests (boss to subordinate, doctor to patient) are calibrated through politeness markers; mismatches are heard as either tyrannical or insubordinate.
• HCI and conversational AI. Voice assistants and chatbots calibrate politeness markers — the explosion of negative politeness in Apple's and Google's prompts is a direct application.
• Forensic linguistics. Threats, harassment, and coercion are sometimes diagnosed partly by their politeness profile: a "request" using bald on record from a stranger is read as a threat.
• Translation. Honorific systems make politeness one of the hardest things to translate from Japanese / Korean into English without flattening.

Common pitfalls

• "Negative politeness" sounds rude. It isn't — it's the cluster English uses with strangers and superiors ("Could you possibly…?"). The label refers to the kind of face being protected, not the speaker's demeanour.
• Ranking strategies as good vs bad. They are not a hierarchy of decency; bald on record between intimates is more polite than negative politeness, which would feel cold.
• Forgetting D, P, and R interact. A small request from a high-power speaker still needs framing for the relationship; a large request from a friend might still go bald-on-record. The W formula is additive, not lexical.
• Treating politeness as universal in form. Honorifics, T-V pronouns, and discernment registers are language-specific implementations; the framework predicts variability there, not uniformity.
• Ignoring impoliteness. Brown and Levinson modeled face protection. Real interaction includes deliberate face attack; modern theories handle both.