From Japanese Pitch
About this page
This page was written in an effort to make information on Japanese pitch accent, as it is spoken in the Tokyo area, easily accessible to learners of the language. The ultimate goal is to make pitch accent easier to learn and use, and perhaps easier to teach; consequently, some impractical exceptions or counter-examples may have been left out. Pitch accent is almost never taught to second language learners of Japanese and incorrect pitch is a strong characteristic of a foreign accent in Japanese. Anyone wishing to improve their accent or aiming for near-native accent is encouraged to study the information found in this page and the suggested links.
In our opinion, learning pitch consists of three important steps:
a) understanding the pitch system: students need to understand how Japanese pitch works, so that when presented with words and their pitch, they can independently and correctly predict and pronounce an entire sentence;
b) understanding accent changes: students need to study the patterns that govern accent displacement, such as how accent placement is determined in compound nouns or how verb endings (gobi/語尾) carry or assign pitch information;
c) acquiring independence: students need to learn the pitch of every word through the use of the proper tools, namely paper, electronic or online dictionaries; students might want to establish a methodology or structured study plan.
For the purpose of clarity, low morae are written in lowercase, high morae in uppercase. Accented morae are in bold. For instance, in taBETAi, ta is low, BE is high, TA is high and contains the word's pitch, which is then followed by i, a low mora.
Overview of Standard Japanese Pitch Accent
Most Japanese dialects, including Standard Japanese, exhibit pitch accent (高低アクセント kōtei akusento). In this system, every mora (1) is pronounced with a high (H) or a low (L) pitch (2). This is quite different from languages that exhibit stress, such as English.
Whether a word contains an accented mora (ie. marked for pitch) or not, its morae will follow a predictable pattern whereby the first mora is L and all subsequent morae are H until the end of the word or phrase is reached (3), or an accented mora occurs. An accented mora is followed by a downfall and all subsequent morae until the end of the word or phrase will be L. Not all words are accented. Consider these examples:
LHHH no accent
LHHL accent on third mora
HLL accent on first mora
Note that there can only be one downfall per word and the first 2 morae must have different height, ie. LH or HL.
For instance, the word nihongo (日本語) is not accented and is therefore realized as niHONGO. If a particle is added, it will also be high: niHONGO HA, niHONGO NO, etc.
When a word contains a downfall, the H mora will be followed by L morae: JIsho ga.
Consider these three words distinguished only by pitch, followed by the particle ga, and which can be heard here:
HAshi ga = chopstick (HLL, accent on first mora)
haSHI ga = bridge (LHL, accent on second mora)
haSHI GA = edge (LHH, no accent)
In general, pitch is fixed and doesn't change in nouns, except in compound nouns. However, the pitch of adjectives and verbs will vary depending on the ending.
When two words form a compound, pitch often changes. Although where the pitch of the new compound will fall is unpredictable, it often falls on the first mora of the second word (4). Ex.: Umi + hiRAki = uMIBIraki but shiMA + kuni = shiMAguni
Here is some information on nouns' pitch found in Gramatyka japońska (Romuald Huszcza, Maho Ikushima ISBN 83-86483-69-5)[notation conventions differ]:
N = 名詞 めいし NOUNS to be learnt by heart N composed of n morae – n + 1 possible pitch distributions
N + particles
1. N + は が に を へ で と も や から しか だけ
2. N + の
2a. N with the last mora accented change to unaccented: 男otokò -> 男のotokono
2b. the rest – no change
3. The majority of two- and three-mora particles (except for から、しか、だけ) より, さえ, でも, のみ, まで, かしら, すら, とも, だって, など, なり, こそ, では
3a. unaccented N – the first mora of the particle is accented
3b. accented N – no change
4a. either according to 1. or
4b. all N become unaccented (both are correct)
Surnames (last names) + titles
1. + さんsan, 様 さまsama
2. + 君 くん
2a. accented surnames – no change
2b. unaccented surnames – no change or the last mora of the surname is accented, both are correct
3. + 先生 せんせい sensèi 博士 はくしhàkusi 大使 たいしtàisi
3a. unaccented surnames – the title is accented
3b. accented surnames – no change, but there is a tendency to accent the title (particularly by the younger generation)
N = 名詞 めいし, AN = 形容動詞 けいようどうし + the copula
the copula であるde aru （だ、です、だった, etc）
forms of the copula with -màsu, -masèn, etc, are accented
1. unaccented N, AN – the first mora of the copula is accented (except for だ)
2. accented N, AN – no change
A table indicating pitch variations in adjectives can be found here. The pitch will vary depending on whether the adjective is accented or not, and the accent of the ending.
Unlike nouns, verbs' pitch is a lot more volatile. The proportion of verbs that carry pitch is also a lot higher -- I'd venture that most do. Where the pitch will fall will be determined by the verb suffix (such as -te, -tara, -nai, etc.), which usually carries 2 possible patterns: one for accented verb, one for unaccented verbs.
A pdf chart of the most common verb suffixes can be downloaded here. A perhaps unorthodox attempt was made to categorize affixes into 3 groups to simplify acquisition: weak morphemes that only affect the pitch of unaccented verbs, strong morphemes that affect both accented and unaccented verbs but in different ways, and fixed morphemes that have their own intrinsic accent, regardless of the verb they accompany. More traditional tables can also be found here, here and here.
Let's take a few verbs as examples. unaccented -- suru, shiru (5) accented -- shiRABEru, waKAru, taBEru (6)
The -te suffix is a 0/-3 suffix, meaning that on a pitch verb, the downfall will occur on the 3rd mora from the end, and that on a pitchless verb, there will be no downfall. As a result, we get:
shiTE IRU, shiTTE IRU shiRAbete iru, waKAtte iru, TAbete iru
(1) In Japanese, ん and っ are considered morae.
(2) Height is relative; through terracing, high and low morae tend to decrease in pitch as a sentence progresses, so that a high mora could potentially be the same pitch as another low mora earlier in the sentence.
(3) The first two morae must have different pitch, and will surface as [LH or [HL. This is not the case in other dialects.
(4) I haven't been able to determine whether the fact that either or both of the words have pitch or not affects where the compound's pitch falls, but quick analysis seems to indicate that there is no consistency.
(5) By default, shiru will be realised as shiRU, but for notation purposes, *shiRU would indicate a word with pitch on the last mora.
(6) It's very common for 3-mora verbs to be LHL.
Consult the User's Guide for information on using the wiki software.