This article is an non-exhaustive list of rules that govern the Tokyo pitch accent of non-common nouns. For other rules, see Japanese pitch accent rules.
Proper nouns Given names All bimoraic given names are initial accented, and some have different accent from the common nouns or verbs from which they derive:
Most trimoraic names derived from common nouns or other types of names retain the original accent:
But: :
All trimoraic names derived from verbs are unaccented regardless of the verbs' original accent:
All trimoraic names derived from adjectives and adjectival nouns are initial-accented:
Some feminine names often spelt only with kana are unaccented: , , , etc.
Foreign trimoraic names tend to be initial-accented: , , , , , , etc.
Given names with specific suffixes Certain suffixes make names with predictable accent. An accent shift away from a voiceless mora is also possible. Unaccented:
: , , , , etc. : , , , , etc. , etc. Unaccented or penultimate-accented: : Initial-accented:
Unaccented or initial-accented: : (Pre)antepenultimate-accented: : : , , etc. , , etc. But: : , , , etc. , , , , etc. But: , etc.
A name's accent may also correlate with the mora count of the component preceding its suffix. , : Bimoraic: Initial-accented: Trimoraic: (Ante)penultimate-accented: , , , : Unimoraic: Unaccented: Bimoraic: Unaccented, final- or penultimate-accented: : Unimoraic: Initial-accented: Bimoraic: Unaccented, final- or penultimate-accented: : Unimoraic: Initial-accented: Bimoraic: Unaccented or final-accented: : Unimoraic: Unaccented: Bimoraic or longer: Antepenultimate-accented: : Bimoraic or shorter: Unaccented: Trimoraic or longer: Antepenultimate-accented: : Unimoraic: Initial-accented: Bimoraic: Unaccented: , : Bimoraic or shorter: Unaccented: , , etc. , , , , etc. Trimoraic or longer: Antepenultimate-accented: , etc. , , , etc. , : Unimoraic: Unaccented: , , etc. , etc. Bimoraic with an autonomous second beat: Unaccented or accented on the suffix's first mora: , , etc. Bimoraic with a special second beat: Initial-accented or accented on the suffix's first mora: , , , etc. , , , etc. Trimoraic or longer: Accented on the suffix's first mora: , , etc. : Unimoraic, or bimoraic with an autonomous second beat: Unaccented: , , , , etc. , etc. Bimoraic with a special second beat: Initial-accented or accented on the suffix's first mora: , etc. Trimoraic or longer: Accented on the suffix's first mora: , etc. , etc.
Extended ‑rō given names A name with the suffix , as the second component of a compound, can be further expanded regularly. If the second component is trimoraic, the name's accent correlates with both components. If the second component is quadrimoraic, it alone determines the accent. Unimoraic Bimoraic, 2nd mora being Bimoraic, 2nd mora being Bimoraic, 2nd mora being Bimoraic, 2nd mora being anything else
Some kabuki performers' stage names are exceptional: , , , etc.
Historical given names with no specific suffixes The native names of historical figures tend to follow the antepenultimate rule. Some more accentually established names are unaccented, but some of them have become regularized as antepenultimate-accented: , , etc. , , , etc. , , , , , , , , , , etc. , , , , etc. , , , , etc. But: , , , , etc.
Fully or partly Sino-Japanese names also tend to follow the antepenultimate rule, and are simultaneously initial-accented in many cases. Some more accentually established names are unaccented, final- or penultimate-accented, but some of them have become regularized as antepenultimate-accented: , , , etc. , , etc. , , etc. , etc. , , etc. , , , , , , , , , , , , etc. , , , etc. , , etc. , , etc. , etc. , , , etc. , , , , , etc. But:
, , , etc. , , , , etc.
Family names Most trimoraic or shorter family names are initial-accented, and some may have different accent from the common nouns they derive from:
But:
Western names generally follow the antepenultimate rule: , , , , , , , , , , , , , , , etc. Although a few retain the accent in the source languages: English: Edison → German: Marx → Schubert → French: Sartre →
Family names with specific suffixes Family names with specific suffixes are generally predictable. Unaccented:
But:
But: (Pre)antepenultimate-accented: : , , etc: : But:
A name's accent may also correlate with the mora count of the component preceding its suffix. , : Unimoraic: Unaccented: Bimoraic: Initial-accented: But: , : Unimoraic: Initial-accented: Bimoraic: Unaccented: But: , , : Unimoraic: Initial-accented: But: Bimoraic: Unaccented: But:
Long compound family names Long compound family names tend to follow the rules of long modifier–head compound nouns.
Full names and dvandvas Japanese names Contemporary Japanese full names are not meant to be compoundified, but pronounced as separable minor phrases: , , , etc. , , , etc. , , , etc. , , , , , etc. , , , , etc. , , , , etc.
Phonetically, given names even if downstepped still have slightly higher pitch contours than common nouns, for example occupational titles (see ). In other words, Haruka in has a slightly higher contour than hakase in .
Historical full names, on the other hand, are more likely compoundified: , , etc. , , , etc.
Stage family names and titles work similarly to family names: , , , etc. , , etc.
Dvandvas work similarly, though it is common to pronounced them as one phrase: , , , , , etc.
Sino-Japanese Chinese names Two-kanji Chinese names are accented on the first kanji's first mora: , , , , etc.
Three-kanji names are accented on the second kanji's first mora: , , , , etc. , etc.
Note that these rules only account for Sino-Japanese names, not Modern-Mandarin-derived transcriptions of names such as Mao Tsōton ().
Western names Western names are regular recent-loanword compounds with the given names as modifiers and the family names as heads (see ): , , , , , etc.
Placenames Many placenames in Tokyo and nearby areas are unaccented, which conforms to the aforementioned principle of deaccentuating high-frequency or familiar words: , , , , , , , , , etc.
A select few western placenames, some borrowed earlier on, are also unaccented: , , , , , , , , , etc.
Most other placenames conform to the antepenultimate rule; if they are trimoraic or shorter, they tend to be initial-accented: , , , , , , , , , , , , , , , , , , etc. , , , , , , , , etc. , , , , , , , , , , , , , , , , , etc.
Some placenames are alternatively penultimate-accented or unaccented if the antepenultimate mora is special or devoiced (the dagger marks obsolete or obsolescent accent): , etc. , , , etc. , , , , etc.
Long compound placenames behave like long compound common nouns, whether compoundified or non-compoundified see ():
Dvandvas tend to be non-compoundified:
Clipped proper nouns Bimoraic clippings Bimoraic clippings tend to be initial-accented:
Tri- and quadrimoraic clippings Tri- and quadrimoraic clippings tend to be unaccented:
Some are accented like short Sino-Japanese dvandvas (see ):
Some elliptical clippings may retain the original noun's accent:
But: :
Quinquemoraic or longer clippings Quinquemoraic or longer clippings tend to behave like long compounds:
Pronouns, attributives, adverbs and conjunctions Pronouns, , adverbs and conjunctions generally behave like nouns before particles and auxiliaries:
Some unaccented words can become final-accented:
Old-Japanese-derived ‑re personal pronouns, demonstratives and indefinites Most of the Old Japanese-derived ‑re personal pronouns and their genitive forms are initial-accented: Singular : : : → dare: : : Plural : : : : : Ga genitive : : :
The ko‑, so‑ and a‑ demonstratives are mostly unaccented, while the ka‑ demonstratives and do‑ indefinites are mostly initial-accented: : → : : : : : :
Other indefinites are usually initial-accented: , , , , , , , , , etc. But: , , , , etc.
Verbal attributives Attributives derived from archaic verbal attributives (see ) simply keep their regular accent:
Noun- and demonstrative-derived personal pronouns Personal pronouns that are etymologically nouns or demonstratives with pronominal uses generally do not have different accent (they, them and their are singular and gender-neutral here): Primarily first-person:
→ watashi: → watai: → wate: → wasshi ~ watchi: → : → wai: → atakushi: → atashi: → atai: → ate: → asshi: → ashi:
: (compare )
Primarily second-person:
→ nishi:
→ temē: : → → omē: → omai: : → → anta: Primarily third-person:
→ yakko: → yatsu:
Nouns used as adverbs In principle, when a noun is used as an adverb (see Japanese counter words for the grammar), its accent does not change:
But: , , , , etc.
However, final- and medially penultimate-accented nouns with special final morae, that express quantity or time, can be deaccented as adverbs:
Some younger speakers may ignore this rule:
To ideophonic adverbs Adverbs that are ideophones and end in the particle to follow certain patterns as follows.
Trimoraic or shorter to adverbs whose penultimate mora is special tend to be initial-accented: , , , , , , , , , , etc. But: , , , etc.
Trimorae whose penultimate mora is not special can be either initial- or penultimate-accented: , etc.
Quadrimorae and sexamoraic reduplicatives whose penultimate mora is tend to be accented on the antepenultimate mora, unless that mora is : , , , , , , , etc. , , , , etc. , , , etc. A case where is accentable:
Other to adverbs' stems with specific endings: ‑ri tems tend to be equally likely penultimate- or final-accented if trimoraic, and mostly penultimate-accented if quadrimoraic: , , , , , , , , , , etc. , , , , , , , , , , , , etc. But: ‑n stems tend to be penultimate-accented: , , , etc. , , , , , , , , , , , , etc. But: ,
Reduplicatives whose stem can combine with morphemes other than the particle to behave differently depending on what follows the stem: The stem on its own, and when it is followed by to or directly by a verb, tends to be initial-accented, with a possible rightward shift away from the devoiced first mora: , , , etc. , , , , , , , , , , , , , , , , , , , , , , , , etc. , , , , , , , etc. , , , , , , etc. However, verb- and adjective-derived stems, tend to be penultimate-accented if they can be followed by the particle to or the verb suru ( ), but not by the particle *: , , , etc. , , , , etc. Verb- and adjective-derived stems can also be antepenultimate-accented if they can be followed by to, but not by *suru nor *: , etc. , , etc. , etc. , etc. , etc. The same stems above, as well as other types of stems, tend to be unaccented if they are uniquely followed by the copular finite verbs da/na and the copular infinitive particle ni (see ). Compare:
Other examples: , , , , , , , , etc. , , , , etc. , , etc. , etc. Some can still be accented: , , , etc.
Adjectival and auxiliary nouns Adjectival and auxiliary nouns are often treated as conjugational (hence the terms and ). Their so-called "conjugation," however, relies solely on a following copula (see ), typically exemplified by de aru → da ().
Adjectival nouns ending in ‑ka are mostly antepenultimate-accented: , , , , , etc.
, , , , , , , etc.
, , , etc. , etc.
Adjectival nouns ending in ‑ra, ‑re and ‑ro are mostly unaccented: , , , , , , , etc.
Interjections Interjections’ accent is often influenced by intonation. Interrogative interjections may sound final-accented because of the rising intonation. Non-derivative interjections tend to be initial-accented, especially those used for calling or responding: , , , , , , , , etc. , , , , etc. , , , etc. , , , , , etc. But: , , , etc.
Derivatives tend to retain the original accent of the source words: , , , etc. , , , , etc. , , , etc. , , , etc. But:
The demontrative-derived interjections are initial-accented: , , Some greetings are final-accented, which can only be realized in the form of a quotation before a particle:
→
→