Help:IPA/日本語

以下の表は、ウィキペディアの記事中の日本語と沖縄語の発音を、国際音声記号 (IPA) で表したものです。ウィキペディアの記事にIPA文字を追加する方法については、{{IPA-ja}}、{{IPAc-ja}}、こちらの英語版での説明をご覧ください。

表の例は、ヘボン式ローマ字で翻字した日本語の単語です。

日本語の音声のより詳細な説明は日本語の音韻を参照してください。

子音
IPA	例		英語の近似音
IPA	仮名	ローマ字	英語の近似音
b	ばしょ, かびん, ヴァージョン	basho, kabin, vājon	bug
bʲ	びょうき	byōki	beauty
ç	ひと, ひょう	hito, hyō	hue
ɕ	した, いっしょう	shita, isshō	sheep
d	どうも, どうどう	dōmo, dōdō	doctor
dz^[1]	ざぜん, あんぜん, キッズ	zazen, anzen, kizzu^[2]	cards
dʑ^[1]	じょじょ, かんじゃ, エッジ	jojo, kanja, ejji^[2]	jeep
ɸ	ふじ	fuji	phew!（にほぼ近い）
ɡ^[3]	がっこう, りんご, ぎんこう	gakkō, ringo, ginkō	goat
ɡʲ	きぎょう	kigyō	argue
h	ほん, はは	hon, haha	hat
j	やくしゃ, ゆず	yakusha, yuzu	yacht
k	くる, はっき	kuru, hakki	skate
kʲ	きょうかい, けっきょく	kyōkai, kekkyoku	skew
m	みかん, せんぱい, もんもん	mikan, senpai, monmon	much
mʲ	みゃく	myaku	mute
n	なっとう, かんたん	nattō, kantan	not
ɲ	にわ, こんにゃく, きんちょう	niwa, konnyaku, kinchō	canyon
ŋ^[3]	りんご, なんきょく	ringo, nankyoku	pink
ɴ^[4]	にほん	nihon	long（にほぼ近い）
p	パン, たんぽぽ	pan, tanpopo	span
pʲ	はっぴょう	happyō	spew
ɾ	ろく, そら	roku, sora	atom（アメリカ英語）
ɾʲ	りょうり	ryōri	party（アメリカ英語）
s	する, さっそう	suru, sassō	soup
t	たべる, とって	taberu, totte	stop
tɕ	ちかい, けっちゃく	chikai, ketchaku^[2]	itchy
ts	つなみ, いっつい	tsunami, ittsui^[2]	cats
w^[5]	わさび	wasabi	was（にほぼ近い）
ɰ̃^[6]	ふんいき, でんわ, あんしん	fun'iki, denwa, anshin	sin
z^[1]	ざぜん, つづく	zazen, tsuzuku	zoo
ʑ^[1]	みじかい, じょじょ	mijikai, jojo	vision
ʔ	あつっ！	atsu'!	uh-oh

母音
IPA	例		英語の近似音
IPA	仮名	ローマ字	英語の近似音
a	ある	aru	father
e	えき	eki	bet
i	いる	iru	meet
i̥^[7]	した	shita	meet（小声で）
o	おに	oni	story
ɯ^[8]	うなぎ	unagi	shoot
ɯ̥^[7]	すきやき	sukiyaki	shoot（小声で）

超分節音素
IPA	説明	例	英語の近似音
ː	長母音	hyōmei, ojiisan	re-equalize
ꜜ	ダウンステップ^[9]	[kaꜜki] (牡蠣, 'oyster'), [kakiꜜ] (垣, 'fence')	[ˈmæri] (marry), [məˈriː] (Marie)
.	分節法	nin'i [ɲiɰ̃.i]	higher [ˈhaɪ.ər]

註釈[編集]

^ ^a ^b ^c ^d Voiced fricatives [z, ʑ] are generally pronounced as affricates [dz, dʑ] in word-initial positions and after the moraic nasal /N/ ([n] before [dz] and [ɲ] before [dʑ]) or the sokuon /Q/ (only in loanwords). Actual realizations of these sounds vary (see Yotsugana).
^ ^a ^b ^c ^d When an affricate consonant is geminated, only the closure component of it is repeated: [kiddzɯ, eddʑi, ittsɯi, kettɕakɯ].
^ ^a ^b A declining number of speakers pronounce word-medial /ɡ/ as [ŋ] (Vance 2008:214), but /ɡ/ is always represented as [ɡ] in this system.
^ The utterance-final nasal is traditionally described as uvular [ɴ], but instrumental studies have found that this is inaccurate and the actual realization varies (Maekawa 2021). However, an alternative transcription has yet to be established, so ⟨ɴ⟩ is used.
^ [w] is phonetically a bilabial approximant [β̞], but it is traditionally described as a velar [ɰ] or labialized velar approximant [w] and transcribed with ⟨ɰ⟩ or ⟨w⟩ (Maekawa 2020).
^ The syllable-final n (moraic nasal) is pronounced as some kind of nasalized vowel before a vowel, semivowel ([j, ɰ]) or fricative ([ɸ, s, ɕ, ç, h]). [ɰ̃] is the conventional notation undefined for the exact place of articulation (Vance 2008:97).
^ ^a ^b Close vowels [i, ɯ] become voiceless [i̥, ɯ̥] when short and surrounded by voiceless consonants within a word. When the second consonant is [ɸ], [ç] or [h], or when both consonants are fricatives (including the second component of an affricate), devoicing is much less likely to occur (Fujimoto 2015), so vowels in such environments are not transcribed as voiceless (nor are word-final or non-close vowels, whose devoicing is also less consistent). Where close vowels that would be devoiced according to the above rules occur in succession, however, usually whichever vowel is accented or the second vowel, if neither is accented, remains voiced (Fujimoto 2015:189), so transcribe them accordingly: [kɯꜜɕi̥kɯmo, tsɯ̥kɯɕi]. These rules may be overridden by citing a reliable source that marks devoicing, such as NHK (2016) or Kindaichi & Akinaga (2014), if the word being transcribed appears in it.
^ [ɯ], romanized u, exhibits varying degrees of rounding depending on dialect. In Tokyo dialect, it is either unrounded or compressed [ɯᵝ], meaning the sides of the lips are held together without horizontal protrusion, unlike protruded [u].
^ A pitch drop may occur only once per word and does not occur in all words. The mora before a pitch drop has a high pitch. When it occurs at the end of a word, the following grammatical particle has a low pitch.

参考文献[編集]

Fujimoto, Masako (2015). “Vowel devoicing”. In Kubozono, Haruo. Handbook of Japanese Phonetics and Phonology. Berlin: De Gruyter. pp. 167–214. doi:10.1515/9781614511984.167. ISBN 978-1-61451-252-3
Kindaichi, Haruhiko; Akinaga, Kazue, eds (2014) (Japanese) (2nd ed.). Tokyo: Sanseido. ISBN 978-4-385-13672-1
Maekawa, Kikuo (2020). “Remarks on Japanese /w/”. ICU Working Papers in Linguistics 10: 45–52. doi:10.34577/00004625.
Maekawa, Kikuo (2021). “Production of the utterance-final moraic nasal in Japanese: A real-time MRI study”. Journal of the International Phonetic Association: 1–24. doi:10.1017/S0025100321000050.
NHK Broadcasting Culture Research Institute, ed (2016) (Japanese). Tokyo: NHK Publishing. ISBN 978-4-14-011345-5
Vance, Timothy J. (2008). The Sounds of Japanese. Cambridge University Press. ISBN 978-0-5216-1754-3

[yotsugana-1] Voiced fricatives [z, ʑ] are generally pronounced as affricates [dz, dʑ] in word-initial positions and after the moraic nasal /N/ ([n] before [dz] and [ɲ] before [dʑ]) or the sokuon /Q/ (only in loanwords). Actual realizations of these sounds vary (see Yotsugana).

[gemination-2] When an affricate consonant is geminated, only the closure component of it is repeated: [kiddzɯ, eddʑi, ittsɯi, kettɕakɯ].

[g-3] A declining number of speakers pronounce word-medial /ɡ/ as [ŋ] (Vance 2008:214), but /ɡ/ is always represented as [ɡ] in this system.

[4] The utterance-final nasal is traditionally described as uvular [ɴ], but instrumental studies have found that this is inaccurate and the actual realization varies (Maekawa 2021). However, an alternative transcription has yet to be established, so ⟨ɴ⟩ is used.

[5] [w] is phonetically a bilabial approximant [β̞], but it is traditionally described as a velar [ɰ] or labialized velar approximant [w] and transcribed with ⟨ɰ⟩ or ⟨w⟩ (Maekawa 2020).

[6] The syllable-final n (moraic nasal) is pronounced as some kind of nasalized vowel before a vowel, semivowel ([j, ɰ]) or fricative ([ɸ, s, ɕ, ç, h]). [ɰ̃] is the conventional notation undefined for the exact place of articulation (Vance 2008:97).

[devoicing-7] Close vowels [i, ɯ] become voiceless [i̥, ɯ̥] when short and surrounded by voiceless consonants within a word. When the second consonant is [ɸ], [ç] or [h], or when both consonants are fricatives (including the second component of an affricate), devoicing is much less likely to occur (Fujimoto 2015), so vowels in such environments are not transcribed as voiceless (nor are word-final or non-close vowels, whose devoicing is also less consistent). Where close vowels that would be devoiced according to the above rules occur in succession, however, usually whichever vowel is accented or the second vowel, if neither is accented, remains voiced (Fujimoto 2015:189), so transcribe them accordingly: [kɯꜜɕi̥kɯmo, tsɯ̥kɯɕi]. These rules may be overridden by citing a reliable source that marks devoicing, such as NHK (2016) or Kindaichi & Akinaga (2014), if the word being transcribed appears in it.

[8] [ɯ], romanized u, exhibits varying degrees of rounding depending on dialect. In Tokyo dialect, it is either unrounded or compressed [ɯᵝ], meaning the sides of the lips are held together without horizontal protrusion, unlike protruded [u].

[9] A pitch drop may occur only once per word and does not occur in all words. The mora before a pitch drop has a high pitch. When it occurs at the end of a word, the following grammatical particle has a low pitch.

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]