What are the differences between kana writing and pronunciation?

While Japanese kana writing mostly represents the pronunciation, there are some cases where it differs. For example, the object particle is written を ("wo") but pronounced "o". Tokyo is written とうきょう in kana, but pronounced とおきょお, unlike the verb omou, "to think", which is written おもう in kana and also pronounced with a う, not as おもお.


The particles pronounced as on the left are written in kana as on the right.
Pronunciation Spelling Grammatical function
o (お) wo (を) Object marker
e (え) he (へ) Direction marker
wa (わ) ha (は) Topic marker
This is a hangover from the historical kana usage. See also Particles (助詞) for more about the uses of these particles.

Long vowels (おう, えい)

In the case of the combinations ou and ei, there are two possible pronunciations of the same kana depending on whether or not they are chōon (長音), long vowels.

In the long vowel ou in おう, こう, そう, etc. of Chinese-derived words such as koushin (更新) (renewal), こしん, the う (u) is pronounced as a continuation of the こ (ko), as if こしん. However, the う (u) of omou (思う), おも, "think", is pronounced as う (u) not お (o).

The combination of kana ei (えい) is usually pronounced ee (ええ). This is sometimes reflected in spelling. For example, keitai, mobile phone, is sometimes written as ケータイ in katakana with a chōon symbol (see What is the long line symbol used in katakana?). However, in some words, such as tameiki (ためいき) "sigh", meiru (滅入る) "be depressed", or ei (エイ) "ray" (the fish) the ei combination is pronounced as ei, not ee.

Syllabic n (ん)

There are four different pronunciations of the kana ん (the syllabic n, see What is syllabic n?), depending on the sound that comes after it.

  1. ん is pronounced as an M as in "Mark" if one of the sounds [p] [b] or [m] follows it. For example, kimpatsu (金髪) (blond), kembutsu (見物) (sightseeing), amma (按摩) (massage). In the original Hepburn romanization, this is reflected in the romanization of the word. See What is Hepburn romanization? and What are the systems of romanization of Japanese?
  2. ん is pronounced with an N as in "snow" if one of the sounds [t], [d], [r], [z], [dʒ] (the "j" sound) or [n] follows it. For example, hantai (反対) (opposite), shindai (寝台) (bed), shinrai (信頼) (trust), kinniku (筋肉) (muscle)
  3. ん is pronounced as in "think", as [ŋ], if the sounds [k] or [g] follow it. For example, tenki (天気) (weather), bungaku (文学) (literature).
  4. ん is pronounced slightly differently when it occurs in the medial position succeeded by a vowel [a,i,u,e,o] or consonants [s], [w], [j], [h], [S] (the "sh" sound) and [F] (voiceless bilabial fricative). With the other n sounds, the air flow is completely through the nose. What sets this one off is that air flow is largely (and this varies from speaker to speaker a bit) through the mouth, although nasal air flow is also important.

When ん occurs with nothing immediately following, there is nothing for it to assimilate to, and it may become any of the above sounds. It is usually a nasalized continuation of the preceding vowel, perhaps shifted in the direction of a high central position [ɰ].

Consonant g (がぎぐけご)

There are two different pronunciations of the consonant "g" in がぎぐげご (ga/gi/gu/ge/go):

This is hyōjungo (標準語) (standard pronunciation), and many native speakers will differ from this.

づ and ぢ

The kana つ and ち with a dakuten are pronounced zu and ji, exactly the same as ず and じ. Most of the old spellings of the historical kana usage which used づ and ぢ turned into ず and じ. For instance itadura (いたづら), 'mischief', became itazura (いたずら), which is pronounced the same way. What did not change were the words which clearly originated from a compound where the tsu or chi of one of the compound words had been nigoried in the process of compounding. For instance, kanadukai. meaning 'kana usage' from kana and tsukai, did not change its spelling.

Hepburn and Kunrei romanization (see What are the systems of romanization of Japanese?) map both the sounds づ and ず and ぢ and じ onto the same romanized letters, zu and ji, because they are the same sound, but it makes converting back to kana difficult. Nippon-shiki romanization (see What is romanization?) maps づ to du and ぢ to di.

Other differences

Other differences include:

  1. Kana does not show the pitch accent of the words. See What is Japanese pitch accent?
  2. The verb "iu" is often pronounced "yuu".
  3. There are at least two ways each to read the kana き (ki), く (ku), し (shi), す (su), ち (chi), つ (tsu), ひ (hi), and ふ (fu), because the u and i vowels may or may not be devoiced (see What is a devoiced vowel?).

Historical kana usage

Japanese spelling was drastically reformed in the period following the second world war. Prior to this, Japanese spelling (in kana) was quite irregular. There were several ways of spelling things which were in fact all pronounced the same way. (see also What is historical kana usage?)

For example, ikimashou (いきましょう), was written as ikimaseu (いきませう), the sokuon character (the small tsu () indicating an extended consonant, see What is the small kana "tsu" used for?) was written as a large tsu, the verb iru (いる) was written as wiru (ゐる), and so forth. People didn't speak like that: they spoke as they do now, but they wrote like that.

In the period of spelling reform, most of this old stuff was swept away and replaced with the modern system, which is more phonetic. For example the kana for we (ゑ) and wi (ゐ) are no longer used in modern Japanese because these sounds have disappeared. This meant for instance that several old kana writings were mapped onto one new one. For example both kuwa and ka were mapped onto ka. So, it's relatively easy to make a set of rules for making new spellings from the old ones, but one cannot work out the old spelling from the current one. You can find a complete list of the different types of old spelling, and which new sound they are mapped onto, in one of the appendices of the Kōjien dictionary. You'll also find the old-style spelling in this book, and in almost any other Japanese dictionary.


Thanks to Bart Mathias for several corrections. Thanks to Mike Lyford for some suggestions for additions.

If you have questions, corrections, or comments, please contact Ben Bullock or use the discussion forum / Privacy policy

Book reviews Convert<br>Japanese<br>numbers Handwritten<br>kanji<br>recognition Stroke order<br>diagrams Convert<br>Japanese<br>units