Honorifics: Grammar That Encodes Hierarchy
What an honorific actually is
An honorific is not vocabulary about politeness. It is politeness as morphosyntax โ the social variable fused into the verb stem, the pronoun paradigm, the noun, the suffix. You cannot construct a grammatical sentence without first taking a position on the hierarchy. The grammar refuses to let you stay neutral.
Linguistic typology, following Comrie (1976) and the framework formalized in Brown & Levinson (1987), splits honorific reference along four axes:
| Axis | Marks the relationship between speaker and... | Example languages |
|---|---|---|
| Addressee | the person being spoken to | T/V in French, German, Russian; Japanese ใงใ/ใพใ |
| Referent (subject) | the person being talked about | Japanese ๅฐๆฌ่ช (sonkeigo), Korean -์- |
| Referent (object) | the person affected by the action | Japanese ่ฌ่ญฒ่ช (kenjลgo) |
| Bystander | a third party physically present | Dyirbal "mother-in-law language" (Jalnguy); Pohnpeian |
A language can mark any subset of these. English marks essentially none grammatically โ it leans on lexical choice ("request" vs "ask"). French marks one (tu/vous). Japanese marks three. Dyirbal famously marked all four, with an entire parallel avoidance lexicon used in the presence of taboo affines, documented in R. M. W. Dixon's 1972 grammar.
The World Atlas of Language Structures surveyed 207 languages on pronoun politeness alone:
| System | # of languages | Share | Example |
|---|---|---|---|
| No politeness distinction | 136 | 65.7% | English, Mandarin, Finnish |
| Binary (familiar/polite) | 49 | 23.7% | French (tu/vous), Russian (ัั/ะฒั) |
| Multiple politeness levels | 15 | 7.2% | Hindi (เคคเฅ/เคคเฅเคฎ/เคเคช) |
| Pronouns avoided in polite contexts | 7 | 3.4% | Japanese, Korean, Burmese |
Japanese sits in the rarest cell. Pronouns are not graded โ they are evaded. You don't conjugate "you"; you replace it with a name, a title, or nothing at all, and route the politeness load through the verb. That is the structural fact most introductions miss.
Why Japanese is the limit case
Japanese keigo (ๆฌ่ช, "respect language") is, by consensus among typologists, the most morphologically elaborated honorific system in any living language with a major literature. Masayoshi Shibatani (1990) calls it "the most fully developed honorific system among the world's major languages." Ho-min Sohn (1999) argues Korean rivals it in grammatical scope but lacks Japanese's lexical depth. The closest competitor for sheer stratification is Javanese โ more on that later.
Three properties make Japanese the limit case:
- Triadic structure. Three independent honorific registers โ exalting, humbling, polite โ combine multiplicatively, not additively.
- Suppletive verbs. Common verbs have entirely different lexical roots in honorific registers. Not affixation. Suppletion.
- Mandatory marking. Unlike T/V languages where you can stay "neutral" in casual contexts, every Japanese sentence forces a register choice. No off switch.
The three standard business bows โ eshaku (ไผ้, ~15ยฐ), keirei (ๆฌ็คผ, ~30ยฐ), saikeirei (ๆๆฌ็คผ, ~45ยฐ) โ gradate respect by angle the same way keigo gradates it by morphology. Both systems are mandatory: there is no neutral, zero-angle posture in formal Japanese interaction. Source: Wikimedia Commons.
The three pillars
| Register | ๆผขๅญ | Honors | Mechanism | Example: "to eat" (taberu) |
|---|---|---|---|---|
| Sonkeigo (ๅฐๆฌ่ช) | ๅฐๆฌ่ช | the subject (raises them) | suppletion, prefix ใ/ใ, passive form, ใใใซใชใ | ๅฌใไธใใ (meshiagaru) |
| Kenjลgo (่ฌ่ญฒ่ช) | ่ฌ่ญฒ่ช | the object (lowers speaker, indirectly elevating addressee/object) | suppletion, ใใใใ pattern | ใใใ ใ (itadaku) |
| Teineigo (ไธๅฏง่ช) | ไธๅฏง่ช | the addressee (general politeness) | ใงใ/ใพใ endings, prefix ใ | ้ฃในใพใ (tabemasu) |
The 2007 Bunkachล (Agency for Cultural Affairs) reform split kenjลgo into two โ ่ฌ่ญฒ่ชI (toward a specific honored person) and ่ฌ่ญฒ่ชII (general humility, formerly called ไธ้่ช) โ and broke off ็พๅ่ช (bikago, "beautification language": ใ้ , ใ่ถ) from teineigo. The official scheme is now five-way. Most learners still operate with the classical three because the practical contrasts haven't moved.
The combinatorics matter. One utterance can stack: ใ่ชญใฟใซใชใฃใฆใใใฃใใใใพใ (o-yomi ni natte irasshaimasu, "[the honored person] is reading") layers sonkeigo (ใใใซใชใ) + sonkeigo aspect (ใใใฃใใใ as the honorific copula for ใใ) + teineigo (ใพใ). Three independent honorific operations on one verb.
Suppletive verbs: the cliff
For about a dozen high-frequency verbs, keigo is not formed by affixation but by replacing the lexical root entirely. This single feature is what gives learners the steepest cliff.
| Plain | Sonkeigo (raise subj.) | Kenjลgo (lower self) | Teineigo (polite) |
|---|---|---|---|
| ใใ (do) | ใชใใ | ใใใ | ใใพใ |
| ่กใ (go) | ใใใฃใใใ | ๅใ / ไผบใ | ่กใใพใ |
| ๆฅใ (come) | ใใใฃใใใ / ใใใงใซใชใ | ๅใ | ๆฅใพใ |
| ใใ (be/exist) | ใใใฃใใใ | ใใ | ใใพใ |
| ้ฃในใ (eat) | ๅฌใไธใใ | ใใใ ใ | ้ฃในใพใ |
| ้ฃฒใ (drink) | ๅฌใไธใใ | ใใใ ใ | ้ฃฒใฟใพใ |
| ่ฆใ (see) | ใ่ฆงใซใชใ | ๆ่ฆใใ | ่ฆใพใ |
| ่จใ (say) | ใใฃใใใ | ็ณใ / ็ณใไธใใ | ่จใใพใ |
| ่ใ (hear/ask) | ใ่ใใซใชใ | ไผบใ / ๆฟใ | ่ใใพใ |
| ็ฅใ (know) | ใๅญใใ | ๅญใใ / ๅญใไธใใ | ็ฅใฃใฆใใพใ |
| ใใใ (give) | ใใ ใใ | ๅทฎใไธใใ | ใใใพใ |
| ใใใ (receive) | ใๅใๅใใซใชใ | ใใใ ใ / ่ณใ | ใใใใพใ |
ใใใฃใใใ collapses three plain verbs (่กใ / ๆฅใ / ใใ) into one honorific form. Context disambiguates. This is genuine suppletion โ ๅฌใไธใใ shares no morphological relationship with ้ฃในใ; you cannot derive one from the other. The English analog is go/went, where the past tense is a fossilized fragment of an entirely different Old English verb (wendan).
Which verbs go suppletive? Eating, drinking, going, coming, seeing, saying, doing, being. The most-rehearsed verbs in any language. This is the Zipfian pattern of irregularity Lieberman et al. (2007) documented for English: regularization rate is inversely proportional to frequency to the 1/2 power. High-frequency verbs preserve archaic morphology. Keigo's suppletive class follows the universal โ these are not random verbs, they are the cognitively most-rehearsed.
The productive patterns
For verbs without dedicated suppletive forms, two productive patterns generate honorifics:
| Pattern | Register | Formation | Example: ๆธใ (write) |
|---|---|---|---|
| ใใใซใชใ | sonkeigo | ใ + verb-stem + ใซใชใ | ใๆธใใซใชใ |
| ใ(ใ)ใใ | sonkeigo (lighter) | passive form repurposed | ๆธใใใ |
| ใใใใ / ใใใใใ | kenjลgo | ใ + verb-stem + ใใ/ใใใ | ใๆธใใใ |
| ใใพใ | teineigo | stem + ใพใ | ๆธใใพใ |
The passive-as-honorific (ใใใ/ใใใใ) is the structurally fascinating one. Japanese uses the same morphology for passive voice, potential mood ("can write"), spontaneous occurrence ("comes to mind"), and honorific. Four readings, one form. Susumu Kuno (1973) argued these are not four homophones but reflections of a deep "agent-defocusing" function โ when you make the agent grammatically backgrounded, you can read it as defocused (passive), uncontrolled (spontaneous), capable (potential), or socially exalted (too high to point at directly). One mechanism, four social interpretations.
The honorific-as-passive trick is widespread cross-linguistically. Korean -์- is the canonical subject-honorific suffix, but it has clear historical links to passive/causative morphology. The pattern is robust enough that Brown & Levinson listed "point-of-view distancing" as one of their universal positive-politeness strategies.
Honorific prefixes: ใ and ใ
| Prefix | Origin | Used with | Examples |
|---|---|---|---|
| ใ | native Japanese (ๅ่ช, wago) | mostly native nouns | ใ่ถ, ใ้, ใๆฏใใ, ใๅผๅฝ |
| ใ | Sino-Japanese (ๆผข่ช, kango) | mostly Sino-Japanese compounds | ใ้ฃฏ, ใๅฎถๆ, ใๆณจๆ, ใ้ฃ็ตก |
The split is etymological: ใ for native vocabulary, ใ for Chinese borrowings. Exceptions exist (ใ้ป่ฉฑ is a native prefix on a Sino-Japanese noun) and they are linguistic fossils worth noting โ they tend to mark words that entered general usage so early they "naturalized."
Bunkachล reclassified bare ใ/ใ uses (ใ้ , ใ่ถ) as ็พๅ่ช (bikago, "beautification language") rather than true keigo. The distinction: when ใ marks respect toward someone (ใๆฏใใใฎใ่ถ = "your mother's tea"), it's keigo; when it just marks formal register (ๅฅณๆงใฏใ่ถใ้ฃฒใ, "women drink tea"), it's bikago. The re-analysis was driven by corpus evidence that bare ใ/ใ is now used regardless of social context, particularly in advertising and women's speech.
The suffix system: san, sama, kun, chan, sensei, senpai, dono
Japanese's referent honorific suffixes attach to names. Choosing the right one is a sociolinguistic decision per relationship.
| Suffix | ๆผขๅญ | Register | Used for | Notes |
|---|---|---|---|---|
| ใใ | โ | neutral polite | most adults, default | Etymologically a contraction of ๆง (sama) |
| ๆง/ใใพ | ๆง | high formal | customers, deities, letters | ็ฅๆง (kami-sama, "the god") |
| ๅ/ใใ | ๅ | familiar (typically male juniors) | younger men, schoolboys, subordinates | Used for women in some workplaces โ controversial |
| ใกใใ | โ | affectionate | children, close female friends, pets | Phonetic softening of ใใ |
| ๅ ็ | ๅ ็ | professional respect | teachers, doctors, lawyers, novelists, politicians | Replaces other suffixes; standalone term of address |
| ๅ ่ผฉ | ๅ ่ผฉ | seniority | senior at school/work | Hierarchical, not neutral; pairs with ๅพ่ผฉ (kลhai) |
| ๆฎฟ/ใฉใฎ | ๆฎฟ | archaic-formal | written documents, official letters | Survived in legal/administrative usage |
| ๆฐ | ๆฐ | journalistic-formal | newspapers, public figures | Distancing, used in third-person writing |
| ้ฃไธ | ้ฃไธ | "Your Excellency" | ambassadors, generals | Diplomatic only |
| ้ไธ | ้ไธ | "His/Her Majesty" | the Emperor/Empress only | One referent in modern Japanese |
The suffix is not optional. Naked names โ yobisute (ๅผใณๆจใฆ) โ are reserved for intimates, subordinates, or contempt. Calling your boss "Tanaka" instead of "Tanaka-san" registers as a serious breach in any but the most informal startup. The empty slot is itself a marker.
Takao Suzuki (้ดๆจๅญๅคซ) in his classic Words in Context (1973/1978) showed that Japanese second-person reference operates on a positional logic: you don't address someone by their identity, you address them by their role-relation to you. A child calls his older brother onii-san (literally "older brother"), but the older brother calls the younger by name. The same person is otลsan to one speaker and jichan (uncle) to another, simultaneously. Pronouns are evaded because there is no fixed perspective from which a single pronoun would be valid.
Scene from the Genji Monogatari Emaki ("Yadorigi" chapter), c. 1130, Tokugawa Art Museum. The hierarchical Heian court society depicted here โ segmented by curtains, screens, and seating position rather than direct address โ is the social world in which the keigo system crystallized. Murasaki Shikibu's prose in this period already deploys a fully suppletive honorific verb system. Source: Wikimedia Commons.
How much keigo is in the wild
Patrick Heinrich (2012) and the Bunkachล annual surveys on Japanese language attitudes give us actual usage data:
| Statistic | Value | Source |
|---|---|---|
| Adults who report difficulty using keigo correctly | 64% | Bunkachล Kokugo Yoron Chลsa, 2018 |
| Workers who received explicit keigo training in onboarding | 81% | NHK Workplace Language Survey, 2019 |
| Verbs in core suppletive keigo paradigm | ~13 | Natsuko Tsujimura (2013) |
| Keigo-related entries in Kลjien (ๅบ่พ่) | 1,200+ | Iwanami Shoten, 7th ed. |
| Words that take ใ/ใ prefix in modern usage | ~1,800 | NINJAL BCCWJ corpus |
The 64% self-reported difficulty figure is the striking one. This is not a learner's problem โ it's a native speaker's anxiety. Keigo is taught explicitly in junior high (ไธญๅญฆ), reviewed in high school, and most large companies run mandatory manner kenshลซ (ใใใผ็ ไฟฎ) for new hires with 30+ hours on keigo and bijinesu manฤ. Books like Iitai Koto ga Kichinto Tsutawaru "Keigo" no Hon sell tens of thousands of copies a year; the genre is its own publishing vertical.
The folk anxiety tracks a real complexity. Niyekawa's 1991 Minimum Essential Politeness โ written for diplomats โ runs 300+ pages and concludes there is no shortcut.
The honorific belt
Honorific systems comparable to Japanese cluster geographically across East and Southeast Asia:
| Language | Family | Mechanism | Distinct lexical strata |
|---|---|---|---|
| Japanese | Japonic | suffix + suppletion + prefix | 3โ5 |
| Korean | Koreanic | -์- subject suffix + 6 sentence-final levels | 6 (ํด๋ผ/ํด/ํ๊ฒ/ํ์ค/ํด์/ํฉ๋๋ค) |
| Javanese | Austronesian | parallel lexicons (ngoko/madya/krama) | 3 |
| Balinese | Austronesian | 4 strata + caste-restricted lexicons | 4 |
| Tibetan (Lhasa) | Sino-Tibetan | suppletive verbs + nominal honorifics | 3 |
| Thai | Tai-Kadai | royal vocabulary (rachasap) + pronoun cascade | 5+ |
| Vietnamese | Austroasiatic | kinship-based pronoun system | gradient |
Joseph Errington's Shifting Languages (1998) on Javanese is the canonical study: the krama (high) lexicon and ngoko (low) lexicon are largely non-overlapping โ over 800 distinct word pairs. A high-status interlocutor demands a near-complete lexical substitution, not just a register shift. By that measure Javanese is more honorific-elaborated than Japanese. But it has been losing speakers and ground to Indonesian since 1945, while Japanese keigo remains stable in adult usage.
Korean is closest to Japanese in grammatical mechanism. The subject-honorific suffix -์- (-si-) attaches to the verb stem just like a Japanese sonkeigo passive: ๊ฐ๋ค (kada, "go") โ ๊ฐ์๋ค (kasida, "[honored person] goes"). Then the addressee-honorific level (ํด์ฒด, ํด์์ฒด, ํฉ์ผ์ฒด, etc.) layers on top. Two grammatical dimensions, multiplicative โ exactly the Japanese sonkeigo ร teineigo split. Sohn (1999) is the definitive treatment.
English is the outlier even among Indo-European languages โ it lost the T/V distinction in the 17th century when thou fell out of use, leaving you as the universal pronoun. Brown & Gilman's 1960 paper "The Pronouns of Power and Solidarity" traced the European T/V system and is required reading for anyone interested in honorifics โ it formalized the power and solidarity axes that have driven sociolinguistic theory ever since.
Is keigo eroding?
Not really. The function is shifting.
Heinrich (2012) and Patricia Wetzel (2004) document a 50-year shift from vertical keigo (status-based, asymmetric) to horizontal keigo (distance-based, symmetric). In pre-war Japan, keigo encoded a fixed status hierarchy: a wife used keigo to her husband; an employee used keigo to a boss who responded in plain form. By 2000, that asymmetry had collapsed in domestic and many workplace contexts. Today's keigo is more often bilateral โ both parties use polite forms because they don't know each other well, not because one outranks the other.
One marker says it cleanly: Bunkachล surveys show "I use keigo with my own family" dropped from 38% (1995) to 9% (2018). "I use keigo with strangers regardless of age" rose from 51% to 78%. Function shifted from rank to relational distance.
The much-discussed baito-keigo (ใใคใๆฌ่ช, "part-timer keigo") โ phrases like yoroshikatta deshล ka ("would that have been alright?", with redundant past-tense politeness) heard in convenience stores โ is technically ungrammatical by classical rules but represents a productive innovation: workers under-trained in formal keigo invent new hyper-polite forms by stacking redundant markers. Miyako Inoue (2006) read this as evidence that keigo is alive and re-grammaticalizing, not dying.
The cognitive argument
Niyekawa (1991) and others have argued that keigo enforces perspective-taking: you cannot frame an utterance without first positioning yourself, your addressee, and any third-party referent on a hierarchy. Non-trivial cognitive load. Also a forced empathy drill.
Recent psycholinguistic work has tested it. Hasegawa et al. (2014) used fMRI to compare brain activity when Japanese subjects processed plain vs. keigo sentences. Keigo activated the right temporo-parietal junction (rTPJ) โ the area most consistently associated with theory of mind and perspective-taking (Saxe & Kanwisher, 2003). Plain forms didn't. Reading the appropriate keigo register isn't just lexical retrieval; it is a social-cognitive task with measurable neural correlates.
Whether keigo speakers exhibit measurably different perspective-taking outside language is unsettled โ the Sapir-Whorf-shaped version of the claim is hard to test. The fMRI signal is real, though. Keigo is not decoration on top of propositional content. It is a parallel social-modeling computation, running on different cortex.
Recommended reading
In roughly increasing technicality:
- Niyekawa, A.M. (1991). Minimum Essential Politeness: A Guide to the Japanese Honorific Language. Kodansha International. โ Practical, written for diplomats. The most-referenced learner book.
- Wetzel, P.J. (2004). Keigo in Modern Japan: Polite Language from Meiji to the Present. University of Hawai'i Press. โ Sociohistorical, traces the 1900โ2000 shift.
- Suzuki, T. (1978). Words in Context: A Japanese Perspective on Language and Culture. Kodansha. โ On pronoun avoidance and role-based reference. A classic.
- Shibatani, M. (1990). The Languages of Japan. Cambridge UP. โ Chapter 12 is the typological treatment.
- Brown, P. & Levinson, S. (1987). Politeness: Some Universals in Language Usage. Cambridge UP. โ The universal framework. Required reading.
- Sohn, H.-M. (1999). The Korean Language. Cambridge UP. โ For the closest typological comparison.
- Errington, J.J. (1998). Shifting Languages: Interaction and Identity in Javanese Indonesia. Cambridge UP. โ For Javanese stratification.
- Bunkachล (2007). Keigo no Shishin (ๆฌ่ชใฎๆ้). โ The official 5-part reform. Free PDF.
- Brown, R. & Gilman, A. (1960). The Pronouns of Power and Solidarity. โ The foundational T/V paper.
- Helmbrecht, J. (2013). Politeness Distinctions in Pronouns. WALS Online, ch. 45. โ Cross-linguistic typology, free.
The map
| # | Feature | Key insight |
|---|---|---|
| 1 | Honorific axis | Speaker can encode relation to addressee, subject, object, or bystander |
| 2 | Japanese rarity | Pronouns are evaded, not graded โ top 3.4% of WALS sample |
| 3 | Three pillars | Sonkeigo + Kenjลgo + Teineigo combine multiplicatively |
| 4 | Suppletion | Top-frequency verbs use entirely different lexical roots |
| 5 | Productive forms | ใใใซใชใ / ใใใใ / ใใพใ scale to all verbs |
| 6 | ใ vs ใ | Etymological split: native vs Sino-Japanese |
| 7 | Suffix system | san/sama/kun/chan/sensei/senpai mark referent on each name |
| 8 | Pronoun avoidance | Role-relations replace fixed-perspective pronouns |
| 9 | Cross-linguistic | Korean -์- mirrors structure; Javanese rivals depth |
| 10 | Drift | Vertical (status) โ horizontal (distance) over 1945โ2020 |
| 11 | Cognitive load | rTPJ activation suggests parallel social-modeling computation |
Every grammatical Japanese sentence is a hierarchy claim. There is no neutral form. That is the fact worth understanding before any vocabulary list.