This article is an non-exhaustive list of rules that govern the Tokyo pitch accent of affixes. For other rules, see Japanese pitch accent rules.
Prefixes Native prefixes The prefixes , , and can form regular long compounds:
But: :
Short compounds with su‑ and fu‑/bu‑ tend to be initial-accented:
Short compounds with ma‑, on the other hand, tend to be unaccented:
But:
Some prefixes may not affect the original accent of the prefixed components, while others add accent: : But: :
Emphatic prefixes, especially those that express contempt or antagonism, such as , , , , etc tend to result in initial-accented compounds, although younger speakers may regularize them (the dagger marks obsolete or obsolescent accent):
Ko‑ without such a negative connotation can make initial-accented or unaccented short compounds:
Ō‑ otherwise still often results in initial-accented short compounds:
Ai‑ and can form non-compoundified compounds:
Sino-Japanese prefixes A number of Sino-Japanese prefixes are prone to make initial-accented or non-compoundified compounds: : , , , etc. : , , , etc. : , , , , etc. : , , etc. : , , etc. : , , , , , , etc. : , etc. : , etc. : , , , , , , , etc. : , , , , etc. : , , etc. : , , , , etc. : , , , etc. : , etc. : , , , etc. : , etc. : , etc. : , , , etc. : , , , etc. : , , , , etc. : , , etc.
Some compounds with may have different accent patterns depending on the different meanings, namely and : , etc. , etc. , , , , , , , etc. , etc.
The honorific prefixes spelt with O‑ O‑ + noun Bimoraic compounds prefixed by the native o‑ tend to be unaccented: , , , , , , etc.
Trimorae correlate with their prefixed component and tend to be: Unaccented if the component is originally unaccented: , , , , , , , , etc. Unaccented or sometimes accented on the component's first mora if the component is originally final-accented: , , , , , , , , , , etc. , , , , , , etc. Accented on the component's first mora or sometimes unaccented if the component is originally initial-accented: , , , , , , , etc. , , , , , , etc. , , etc.
Long-compound quadrimorae tend to be: Regularly accented on the component's first mora if it is originally unaccented or initial-accented: , , , , , , , , , , etc. , , , , , , , etc. But some old ones are unaccented: , , , , etc. , etc. Irregularly unaccented if the component is originally final- or medial-accented: , , , , , , , , , , , etc. , , , etc. But: , , etc.
Quinquemorae tend to be: Irregularly unaccented, or sometimes regularly accented on the component's first mora, if the component is originally unaccented: , , , , , , , , etc. , , , etc. Regularly accented if the component is originally initial- or medial-accented: , , , etc. , , , , etc.
Contractions with o‑ tend to be accented on the prefixed component's first mora: , , , , , , , , , , etc.
Feminine names and expressions of quantity and time tend to keep their accent even when prefixed by o‑: : (compare : ) : (compare : ) : (compare : )
O‑ + verbal infinitive Verbal-infinitive-derived compounds tend to be unaccented:
But:
Exaltative and polite o‑ verbal-infintive-based constructions are unaccented, regardless of the infinitives' original accent, when combining with the verbs , , etc, and only the "doing" verb at the end determines the accentual outcome:
O‑ + adjective/adjectival noun Compounds formed with adjectives and adjectival nouns tend to be unaccented:
On‑ On‑ () tends to form initial-accented or non-compoundified compounds.
But: :
Mi‑ Mi‑ tends to form unaccented compounds. , , , , , , , , etc. , , , etc. But: , , etc.
Go‑ Go‑ tends to have no effect on the nouns that it prefixes: , , etc. , , , , , , , etc. , etc. , , etc.
Gyo‑ Gyo‑ can only form short compounds with Sino-Japanese morphemes. Bimorae with it tend to be initial-accented, and trimorae tend to be unaccented: , , , , , , etc.
Suffixes ‑chū, ‑jū, ‑go, ‑zen, ‑mae and ‑zumi , , , and are productively and regularly dominant after nouns (but not necessarily as short compounds' components, for example in , etc; and some idiomatic expressions such as ), as follows: ‑chū, ‑jū and ‑zumi are deaccenting: , , , etc. , , , etc. , , etc. ‑go is initial-accented or deaccenting: , , etc. ‑zen and ‑mae are initial-accented: , , etc.
‑sama, ‑chama, ‑san, ‑chan, ‑don, ‑ya and ‑ki and its derivatives ‑chama, ‑san and ‑chan, as well as ‑don (derived from ‑dono below), ‑ya and , are recessive and unaccented, always so after proper nouns and mostly so after common nouns. Adding the prefix in these cases does not make a difference: , , , , , , , etc. , , , , , , , etc. , , , , , , , , , , etc. , , , , etc. , , , , , etc.
‑sama sometimes exert its original final accent, or becomes a regular long-compound head, in some common-noun compounds:
‑samasama is final-accented and can form its own minor phrase:
‑dono is recessive, and unaccented or initial-accented: , , etc. , , etc. , , etc.
‑shi and ‑uji is recessive and preaccenting, and alternatively forms it own downstepped minor phrase after an accented name: , , , etc. , , etc. , , etc.
‑shi () is dominant and preaccenting: , , etc. , , etc.
‑uji ( ) is recessive or dominant, and preaccenting: , etc. , etc.
‑ke is typically dominant, and deaccenting or preaccenting, but is additionally recessive and not deaccenting after accented bimoraic names: , , etc. , etc. , , etc. But: , etc.
‑kun, ‑gimi and ‑ue is recessive and unaccented, or preaccenting only after unaccented family names: , etc. But: , , etc. , , etc. , , etc.
Short Sino-Japanese compounds with ‑kun tend to be initial-accented: , , etc.
and tend to be dominant and preaccenting: , , etc. , , , , , , , , etc.
‑go tends to be dominant, and deaccenting in trimoraic compounds but preaccenting in longer nouns: , , , , etc. , , , , , etc.
‑kō is recessive and unaccented, or dominant and preaccenting: , etc. , , etc.
Short Sino-Japanese compounds with ‑kō tend to be initial-accented: , , etc. , , etc.
-kō () is dominant and initial-accented: , etc.
‑bō and ‑nbo(o) is dominant and preaccenting: , , etc.
‑bō () is dominant and initial-accented: , , , etc.
‑nbo(o) tends to be recessive and unaccented: , , etc. , etc. But (words with no equivalent nouns are parenthesized): ,
‑ō is dominant and initial-accented: , , , , etc.
‑ya tends to be dominant and preaccenting in trimoraic shop names: , , etc.
‑ya as used in some stage family names and derogatory terms is dominant and preaccenting as well: , , , , etc. , , etc.
In quadrimoraic or longer shop and stage family names and derogatory terms, ‑ya tends to be dominant and deaccenting: , , , , etc. , , etc.
‑ya () works similarly, with some exceptions: It is preaccenting in most trimoraic compounds: , , , , , etc. , , , , , , etc. But: , , etc. , , , etc. , , etc. It is deaccenting in most quadrimoraic or longer compounds, especially if the modifier is originally final-accented: , , etc. , , , , etc. , , , , etc. , , , , etc. But: , , , etc.
Long Sino-Japanese occupational titles and establishment types Long (two kanji long or longer) Sino-Japanese occupational titles ("Doctor Ono," "Master Yamada," "Emperor Reiwa," etc) and establishment types ("Satō's Shop," "Yoshida's Barbershop," etc) are generally recessive or even non-compoundifying (see ), but accented titles have been made dominant by some younger speakers. The following is a non-exhaustive list of examples: : , etc. , etc. , etc. : , , etc. , etc. , , , etc. : , etc. : , etc. : , etc. , etc. : , etc. , etc. , etc.
Pluralizing suffixes ‑ra and ‑tachi and tend to be dominant and preaccenting, except after some initial-accented pronouns: , , etc. , , etc. But: , , etc.
‑domo tend to be dominant and initial-accented, except after some initial-accented nouns and except in some quadrimoraic compounds: , , , etc. , etc. But: , etc. , , , etc.
‑gata and ‑bara and tend to be dominant, and either deaccenting or preaccenting: , etc. , , etc.
Nominalizing suffixes ‑sa ‑sa tends to make unaccented nouns out of unaccented adjectival attributives and nouns:
Out of accented adjectival attributives, it tends to create minimally antepenultimate-accented nouns if the mora count permits. Optional rightward shifts may occur:
Quadrimoraic adjective-based nouns are additionally penultimate-accented:
Out of accented adjectival nouns, ‑sa tends to make penultimate-accented nouns:
‑mi tends to make unaccented nouns out of unaccented adjectival attributives and nouns:
Nouns based on accented adjectival attributives and nouns tend to be final-accented. Quadrimoraic or longer ones are additionally penultimate-accented or unaccented:
‑ke/‑ge tends to make unaccented nouns out of originally unaccented source words:
Trimoraic nouns based on accented source words tend to be final-accented (verbal infinitives are parenthesized):
While quadrimoraic or longer nouns tend to be penultimate-accented:
‑me tends to make unaccented nouns out of originally unaccented adjectival attributives:
And final-accented or unaccented nouns out of originally accented attributives:
Verbalizing and adjectivalizing suffixes Verbalizers and adjectivalizers generally make accented verbs and adjectives: , , , etc. , , etc. , , etc. , , etc. , , etc. , etc. , etc. , , , , etc. , , , etc. , , etc. , , , , etc. , , etc. , , etc. , , etc. , , , , , etc. , , , , , , , etc. But if the suffixed word is originally unaccented: , , , , etc.
Notes
References
Bibliography
External links Online Japanese Accent Dictionary