Japanese phonology

This article deals with the phonology (i.e. the sound system) of the Japanese language.

Consonants

	Bilabial	Dental	Alveolar	Post- alveolar	Palatal	Velar	Glottal	Place- less
Stop
Flap
Fricative
Affricate
Nasal
Approximant

Voiceless stops are slightly aspirated: less aspirated than English stops, but more so than Spanish. Voiced stops do not always achieve full occlusion, being sometimes realized as fricatives or approximants. is realized as in many dialects, especially in eastern Japan.
are apical and denti-alveolar (i.e. the tongue apex contacts the back of the upper teeth and the front part of the alveolar ridge). Before , these sounds are alveolo-palatal; before they are alveolar.
are laminal alveolar. Before , these sounds are alveolo-palatal.
(transcribed above) is a lateral apical postalveolar flap. It is similar to the Korean r. To an English speaker's ears, its pronunciation lies somewhere between a flapped r (as in American English better and ladder), an l, and a d, sounding most like d before , and most like l before .
The compressed velar is essentially a non-moraic version of the vowel . It is not equivalent to IPA since it is pronounced with lip compression rather than rounding. (English is rounded. The IPA symbol does not quite capture this pronunciation.)
N is a moraic nasal, fully a stop before another stop, where it becomes homorganic with that consonant, but not achieving full occlusion before fricatives or between vowels, where it is realized as a nasal vowel. Word finally before a pause, it may be realized as a uvular nasal stop, a bilabial nasal stop, or as a nasal vowel. Not all analyses include this abstract archiphoneme; some treat the coda nasal as /n/.
assimilates to before , and to before .
Q is realized as the first half of a geminate obstruent. Other, less abstract analyses reject Q in favor of simple geminate consonant clusters, e.g., /pp/, /tt/, /ss/, etc.

Note that this table does not cover all the consonantal variation in the Japanese language. Please refer below for the details of pronunciation.

Vowels

Japanese has 5 vowels:

Japanese vowels are pronounced as monophthongs, unlike in English; they are similar to their Spanish or Italian counterparts. However, the high back vowel is somewhat centralized as well as "compression rounded", rather than protrusion rounded as , or unrounded as . More precisely, is pronounced with the lips compressed toward each other but not spread to the sides. The IPA transcriptions on the right side of the diagram at right are suggested by Okada (1999). Note, however, that there is no IPA symbol for lip compression, so no transcription will be complete. is transliterated as u.

Japanese a is a low, non-palatal, non-retracted low vowel, IPA , though it is also often represented as . It is between the English a in "father" and the English a in "dad." The Japanese o is a "flat" o, unlike the English one, which is a diphthong. Try to keep your tongue lowered while pronouncing the Japanese o. The i is like English ee in "feet." The e sounds to English speakers like a mix between short e in as in "bed," and long e as in "lay," though it is closer to the former than the latter.

Vowels have a phonemic length distinction (i.e., short vs. long). Cf. contrasting pairs of words like ojisan "uncle" vs. ojiisan "grandfather", or tsuki "moon" vs. tsūki "airflow".

In most phonological analyses, all vowels are treated as occurring with the time frame of one mora. Phonetically long vowels, then, are treated as a sequence of two identical vowels, i.e. ojiisan is not .

Although the phonotactics of Japanese lead some to believe that the language lacks diphthongs, this is not correct. A diphthong is defined as "two vowels pronounced in one syllable," so Japanese, like so many other languages, has them. However, unlike English, in which diphthongs are perceived as single vowels, Japanese diphthongs are perceived as sequences of two different vowels. These vowel sequences are phonetically different from the diphthongs that occur in languages like English. In English, a diphthong such as the one in eye is pronounced as a vowel with a following off-glide: or ; while in Japanese the sequence in ai 愛 'love' is pronounced as (as in naïve) where each vowel segment is of equal length. A glide plus a vowel is analyzed as a sequence of consonant and vowel. Furthermore, unlike English, Japanese distinguishes between 'to meet' and 'blue,' as well as between 'love' and 'dressing,' and also between , e.g., koi 'carp' and , e.g., koe 'voice.' Furthermore, because Japanese has long vowels, it also distinguishes between "regular" and "long" diphthongs, e.g., between 'hey' and 'to be numerous.'

Within words and phrase, Japanese allows long sequences of phonetic vowels without intervening consonants, although the pitch accent and slight rhythm breaks help track the timing when the vowels are identical.

{| cellpadding="3" | hōō o (鳳凰を) 'phoenix (direct object)' tōō o ōu (東欧を覆う) 'to cover Eastern Europe'
(this artificial example is not something that would normally be said)

Phonological processes

Japanese contains a number of phonological processes which greatly alter the phonetic realization of consonants and vowels. A few are listed below.

Consonant processes

Weakening

Non-coronal voiced stops between vowels may be weakened to fricatives, especially in fast and/or casual speech:

		bilabial fricative :	abareru 暴れる 'to behave violently'
		velar fricative :	hage はげ 'baldness'

However, is further complicated by its variant realization as a velar nasal . Standard Japanese speakers can be categorized into 3 groups (A, B, C), which will be explained below. If a speaker pronounces a given word consistently with the allophone (i.e. a B-speaker), that speaker will never have as an allophone in that same word. If a speaker varies between and (i.e. an A-speaker) or is generally consistent in using , then the velar fricative is always another possible allophone in fast speech.

may be weakened to nasal when it occurs within words — this includes not only between vowels but also between a vowel and a consonant. There is a fair amount of variation between speakers, however. Some, such as Vance (1987), have suggested that the variation follows social class; others, such as Akamatsu (1997), suggest that the variation follows age and geographic location. The generalized situation is as follows.

At the beginning of words:

all present-day standard Japanese speakers generally use the stop at the beginning of words: → gaiyū 外遊 'overseas trip' (but not )

In the middle of simple words (i.e. non-compounds):

A. majority of speakers uses either or in free variation: → or kagu 家具 'furniture'
B. minority of speakers consistently uses : → (but not )
C. smaller minority of speakers consistently uses ^{^*}: → (but not )

In the middle of compound words morpheme-initially:

B-speakers mentioned directly above consistently use :

So, for some speakers the following two words are a minimal pair while for others they are homophonous:

sengo 1,005 (せんご) 'one thousand five' = for B-speakers
sengo 戦後 (せんこ゜) 'postwar' = for B-speakers^{^*}

To summarize using the example of hage はげ 'baldness':

A-speakers:
B-speakers:
C-speakers:

Palatalization and affrication

The palatals and palatalize the consonants they follow:

		→ palatalized :	umi 海 'sea'
		→ palatalized :	gyōza ぎょうざ 'fried dumpling'
		etc.

The coronals and glottal are affected as follows:

		→ alveolopalatal fricative :	shio 塩 'salt'
		→ alveolopalatal or :	jishin 地震 'earthquake'; gojuu 50 'fifty'
		→ alveolopalatal :	niwa 庭 'garden'
		→ alveolopalatal affricate :	chijin 知人 'acquaintance'
		→ palatal fricative :	hito 人 'person'

Of the allophones of , the affricate is most common, especially at the beginning of utterances and after (or , depending on the analysis), while fricative may occur between vowels. Both sounds, however, are in free variation. The (laminodorso-)alveolopalatal allophone differs from a palatalized apico-dental , a palatalized apico-alveolar nasal, or a palatal nasal . Similarly, while the symbols and may be encountered, they are not strictly correct, as they represent palatal stops, whereas the Japanese sounds are articulated more forward as alveolopalatal. Since there are no IPA symbols for alveolopalatal stops, and are reasonable compromises, if properly explained.

In the case of the , , and , when followed by , historically, the consonants were palatalized with merging into a single pronunciation. In modern Japanese, these have become separate phonemes:

		(Romanized as sh):	shabon シャボン 'soap'
		(Romanized as j):	じゃがいも 'potato'
		(Romanized as ch):	cha 茶 'tea'

The vowel also affects consonants that it follows:

		→ bilabial fricative :	futa ふた 'lid'
		→ dental affricate :	tsugi 次 'next'

Moraic nasal

Some analyses of Japanese treat the moraic nasal as the archiphoneme . However, other, less abstract approaches treat a syllable-final nasal as a regular coronal . In either case, it always follows vowels (never consonants) and undergoes a variety of assimilatory processes. Within words, it is variously:

uvular at the end of utterances and in isolation.
bilabial before and ; this pronunciation is also sometimes found at the end of utterances and in isolation. Singers are taught to pronounce all instances of this sound as .
dental before coronals and ; never found utterance-finally.
velar before and .
(a nasalized vowel) before vowels, approximants ( and ), and fricatives (, , and ). Also found utterance-finally.

Some speakers produce before , while others produce a nasalized vowel before (see Akamatsu 1997).

Moraic obstruent

In some analyses of Japanese, the archiphoneme is posited. However, not all scholars agree that this is the best analysis. In those approaches that incorporate the moraic obstruent, it is said to completely assimilate to the following obstruent, resulting in an geminate (that is, double) consonant. The assimilated remains unreleased and thus the geminates are phonetically long consonants. does not occur before vowels or nasal consonants. This archiphoneme has a wide variety of phonetic realizations, for example:

		before :	→ nippon 日本 'Japan'
		before :	→ happyaku '800'
		before :	→ kassen 合戦 'battle'
		before :	→ satchi 察知 'inference'
		etc.

Another analysis of Japanese dispenses with /Q/ and other archiphonemes entirely. In this approach, the words above are phonemicized as shown below:

		before :	→ nippon 日本 'Japan'
		before :	→ happyaku '800'
		before :	→ kassen 合戦 'battle'
		before :	→ satchi 察知 'inference'
		etc.

neutralization

The contrast between and is neutralized before and . By convention, it is often assumed to be , though some analyze it as , the voiced counterpart to . Among younger speakers, the contrast between and has been reintroduced through loan words.

Vowel processes

Devoicing

Japanese vowels, especially and , tend to be devoiced when between unvoiced consonants except when they are in accented moras. Additionally, and are optionally devoiced following a voiceless consonant and at the end of an utterance.

			kutsu 靴 'shoe'
			suhada すはだ 'bare skin' (^* is not devoiced since it's accented)
			hikan 悲観 'pessimism'
			himo 紐 'string' (^* is not devoiced since it accented)
		or	hikaku 比較 'comparison'

To a lesser extent (and even rarer ) may be devoiced with the further requirement that there be two or more adjacent moras containing .

kokoro 心 'heart'

Devoicing is common in even normal slow speech and is not restricted to only fast speech.

The common sentence-ending copula desu is pronounced .

Gender roles also play a part: it is regarded as effeminate to pronounce devoiced vowels, particularly the terminal "u" as in "arimasu". Basilectic varieties of Japanese can sometimes be recognized by their hyper-devoicing, while in some Western dialects and some registers of formal speech, every vowel is pronounced.

Nasalization

Japanese vowels are slightly nasalized when adjacent to nasals . Before the moraic nasal , vowels are heavily nasalized:

seisan 生産 'production'

Glottal stop insertion

At the beginning and end of utterances, Japanese vowels may be preceded and followed by a glottal stop , respectively. This is demonstrated below with the following words (as pronounced in isolation):

		:	en 円 'yen'
		:	kishi 岸 'shore'
		:	u 鵜 'cormorant'

When an utterance-final word is uttered with emphasis, this glottal stop is plainly audible, and is often indicated in the writing system with a small letter tsu っ called a sokuon.

Moras and phonotactics

If considered as a system of morae (or moras) instead of syllables, (as the katakana and hiragana phonetic writing systems explicitly do) the sound structure is very simple: The language is made of morae, each with the same approximate time value and stress (stress, here, being correlated with loudness, not pitch). The Japanese mora may consist of either a vowel or one of the two moraic consonants, and (the less abstract analysis that dispenses with archiphonemes defines possible moraic consonants as any voiceless obstruent, or a nasal, in the syllable coda position. Scholars disagree over whether the coda nasal is limited to or can also include ). A vowel may be preceded by an optional (non-moraic) consonant, with or without a palatal glide .

Mora Type	Example	Japanese	Morae per word
V		i 胃 'stomach'	1-mora word
CV		te 手 'hand'	1-mora word
CjV		kya きゃ '(surprised or scared scream)'	1-mora word
N	in	yon 四 'four'	2-mora word
Q	in	mittsu 三つ 'three'	3-mora word

In this table, the period represents a division between moras, rather than the more common usage of a division between syllables.

Consonantal morae are restricted from occurring word initially, though utterances starting with are possible. Vowels may be long, and consonants may be geminate (doubled). Geminate consonants are limited to a sequence of plus a voiceless obstruent, though some words are written with geminate voiced obstruents. In the analysis without archiphonemes, geminate clusters are simply two identical consonants, one after the other.

In the writing system, each kana corresponds to a mora. The moraic (i.e., the first half of a geminate cluster) is indicated by a small "tsu" symbol called a sokuon (subscript ッ in katakana, or っ in hiragana). Long vowels are usually indicated in katakana by a long dash following the first vowel, as in sābisu サービス 'service'. The direction of this dash follows the direction of writing.

In English, stressed syllables in a word are pronounced louder, longer, and with higher pitch, while unstressed syllables are relatively shorter in duration. In Japanese, all morae are pronounced with equal length and loudness. Japanese is therefore said to be a mora-timed language.

On the other hand, since all syllables have equal stress in Japanese, some unstressed syllables in European languages tend to be inaudible to the Japanese ear, leading to confusion.

(Compare to the syllable system of Finnish and Italian.)

Prosody

In Japanese, an accented mora is pronounced with higher pitch than the following mora. This is part of the Japanese intonation pattern.

Notes

Akamatsu (1997) speculates that only 10% of population are consistent users.
Note that the symbol ゜is used by Japanese academia to distinguish between and .

Bibliography

Akamatsu, Tsutomu. (1997). Japanese phonetics: Theory and practice. München: LINCOM EUROPA. ISBN 3-8958-6095-6.
Akamatsu, Tsutomu. (2000). Japanese phonology: A functional approach. München: LINCOM EUROPA. ISBN 3-89586-544-3.
Bloch, Bernard. (1950). Studies in colloquial Japanese IV: Phonemics. Language, 26, 86–125.
Haraguchi, Shosuke. (1977). The tone pattern of Japanese: An autosegmental theory of tonology. Tokyo: Kaitakusha. ISBN 0-8704-0371-0.
Haraguchi, Shosuke. (1999). Accent. In N. Tsujimura (Ed.), The handbook of Japanese linguistics (Chap. 1, p. 1–30). Malden, MA: Blackwell Publishers. ISBN 0-631-20504-7. ISBN 0-6312-0504-7.
Kubozono, Haruo. (1999). Mora and syllable. In N. Tsujimura (Ed.), The handbook of Japanese linguistics (Chap. 2, pp. 31–61). Malden, MA: Blackwell Publishers. ISBN 0-6312-0504-7.
Ladefoged, Peter. (2001). A course in phonetics (4th ed.). Boston: Heinle & Heinle, Thomson Learning.
Martin, Samuel E. (1975). A reference grammar of Japanese. New Haven: Yale University Press. ISBN 0-3000-1813-4.
McCawley, James D. (1968). The phonological component of a grammar of Japanese. The Hague: Mouton.
Okada, Hideo. (1999). Japanese. In Handbook of the International Phonetic Association: A guide to the usage of the International Phonetic Alphabet (pp. 117–119). Cambridge, England: Cambridge University Press. ISBN 0-5216-5236-7; ISBN 0-5216-3751-1 (pbk).
Pierrehumbert, Janet; & Beckman, Mary. (1988). Japanese tone structure. Lingustic inquiry monographs (No. 15). Cambridge, MA: The MIT Press. ISBN 0-262-16109-5; ISBN 0-262-66063-6.
Sawashima, Masayuki; & Miyazaki, S. (1973). Glottal opening for Japanese voiceless consonants. Annual Bulletin of the Research Institute of Logopedics and Phoniatrics, University of Tokyo, Faculty of Medicine, 7, 1-10.
Shibatani, Masayoshi. (1990). Japanese. In B. Comrie (Ed.), The major languages of east and south-east Asia. London: Routledge. ISBN 0-4150-4739-0.
Shibatani, Masayoshi. (1990). The languages of Japan. Cambridge: Cambridge University Press. ISBN 0-5213-6070-6 (hbk); ISBN 0-5213-6918-5 (pbk).
Vance, Timothy J. (1987). An introduction to Japanese phonology. Albany: State University of New York Press. ISBN 0-8870-6360-8; ISBN 0-8870-6361-6 (pbk.).