Hiragana and the Japanese Sound System, Part 2

In this section you'll learn the conventions used for writing sounds beyond the basic 46. Add to that a couple spelling rules and you'll be able to read and write any Japanese word in Hiragana. You'll find a complete chart including everything on this page under the Hiragana and Katakana resources.

From there we'll talk about romanization schemes, including the popular Hepburn romanization system. The Beginning Lessons include both Hiragana and Romaji, so you'll be able to get started on grammar and such while you're still mastering Hiragana. In this case, you might want to read A Guide to Japanese Pronuncation first.

Recommended background:

Hiragana and the Japanese Sound System, Part 1

Voiced Sounds (Daku-on)

Daku-on ("muddy sounds"), are the voiced equivalents of Kana starting with a voiceless consonant.

If you aren't familiar with the concept of voiced and voiceless consonants, put your fingers over your vocal cords and make a 'k', 's', or 't' sound. Now do the same for 'g', 'z', or 'd'. The sounds in the second set are the voiced equivalents of those in the first. This gives us the 'ga', 'za', and 'da' gyou. Many other consonants exist only in voiced or unvoiced form, but not both.

Japanese also has one set of handaku-on ("half-muddy sounds"), the 'pa' gyou. Although phonetically 'b' is the voiced equivalent of 'p', for historical reasons both are related to 'h' (which doesn't really have a voiced equivalent), so 'p' becomes "half-voiced" with 'b' being voiced. Voiced sounds are marked with the dakuten ( ゛), also known informally as tenten. Half-voiced sounds are marked with the handakuten ( ゜), also called maru (circle).

We now have five new columns:

p b d z g
ぱ pa ば ba だ da ざ za が ga
ぴ pi び bi (ぢ *ji) じ *ji ぎ gi
ぷ pu ぶ bu (づ *zu) ず zu ぐ gu
ぺ pe べ be で de ぜ ze げ ge
ぽ po ぼ bo ど do ぞ zo ご go

Here too, there are several inconsistencies. Phonetically, 'ji' should be the voiced version of 'chi', but in fact じ in the 'za' gyou has this pronunciation as well. (Not many languages have the real voiced version of 'sh', but English and French do – it's the 'z' in "azure".) In the middle of words the pronunciation of the 'j' sound can become closer to 'z' in some dialects.

ぢ and づ in the 'da' gyou are pronounced identically to their neighbors じ and ず in modern standard Japanese, and are now seldom used, having been replaced by their 'za' gyou counterparts in most words. Where づ is used, its pronunciation is sometimes closer to 'dzu', which makes sense since it's technically the voiced version of 'tsu'.

For all practical purposes, there is one place where you are likely to see ぢ and づ: the start of the seond part of a compound word.

ex. はな (hana) "nose" + ち (chi) "blood" -> はなぢ (hanaji) "nosebleed"

This phenonemon where the first consonant of a compound word becomes voiced is called randaku, and in the case of 'chi' -> 'ji' and 'tsu' -> 'du', the same base kana is kept.

Combination Syllables (You-on)

As you learned in Part 1, the combination [consonant + 'y' + vowel] also makes a valid mora in Japanese. To write these in Hiragana, take any character from the 'i' dan (including the daku-on and handaku-on) and add a small 'ya', 'yu', or 'yo' to make one of the you-on ("contracted sounds").

ry my py by hy ny (*j) *ch *j *sh gy ky
りゃ rya みゃ mya ぴゃ pya びゃ bya ひゃ hya にゃ nya (ぢゃ *ja) ちゃ *cha じゃ *ja しゃ *sha ぎゃ gya きゃ kya
りゅ ryu みゅ myu ぴゅ pyu びゅ byu ひゅ hyu にゅ nyu (ぢゅ *ju) ちゅ *chu じゅ *ju しゅ *shu ぎゅ gyu きゅ kyu
りょ ryo みょ myo ぴょ pyo びょ byo ひょ hyo にょ nyo (ぢょ *jo) ちょ *cho じょ *jo しょ *sho ぎょ gyo きょ kyo

The semi-vowel 'y' is very close to the vowel 'i', so it seems reasonable that one should be deleted.

As you might expect by this point, し, じ, ち, and ぢ yield slightly different results. This makes sense when you consider that the semi-vowel 'y' we're adding is a palatal glide, but these four Kana are already palatalized, so the 'y' is basically redundant. In other words, whenever you want to say "shya", you drop the 'y'.

ぢゃ, ぢゅ, and ぢょ are not used in modern Japanese, since their pronunciation would be identical to じゃ, じゅ, and じょ, but I've included them for completeness.

Doubled Vowels (Chou-on)

Chou-on, literally "long sounds", are made by adding a second vowel to the Kana to be doubled. This is the first place you'll need to remember a couple of true inconsistencies of spelling.

Hiragana Example Romaji Pron.
[あ dan] + あ かあ kaa kaa
[い dan] + い しい shii shii
[う dan ] + う つう tsuu tsuu
[え dan] + い / え ねい、ねえ nei, nee nee
[お dan] + う / お おう、おお ou, oo oo

The only words using the spelling "ee" and "oo" you'll encounter any time soon are "oneesan" (older sister) and "ookii" (big). Otherwise the long 'e' and long 'o' will be spelled as 'ei' and 'ou' respectively.

However, not every sequence of '[C]ei’ and ‘[C]ou' you'll encounter is actually a you-on. Whenever a compound word is made with one component ending in an 'e' dan Kana is directly followed by an 'i', the two vowels are pronounced distinctly. The same rule applies for 'o' and 'u'. Fortunately, in practice these two possibilities turn out to be easy to distinguish, for reasons that you'll understand as you start to learn Kanji.

Anyway, the same rules apply for the you-on.

Hiragana Example Romaji Pron.
[い dan] + ゃ + あ じゃあ jaa jaa
[い dan] + ゅ + う りゅう ryuu ryuu
[い dan] + ょ + う ぎょう gyou gyoo

Curiously, while just about any combination "[C]you” is used in an actual word, I can’t think of any words with the spelling “[C]yoo".

Doubled Consonants (Soku-On)

Finally, we have the soku-on ("assimilated sounds"), which is marked by a small 'tsu', っ, and represents what is often called a "doubled consonant", though this is not quite accurate. In the case of a stop consonant (k, t, ch, ts, p), the mouth takes the shape of the following consonant, but waits for one mora before pronouncing it. For other consonants (s, sh), the consonant sound is extended into the added mora.

っぱ ppa った tta っさ ssa っか kka
っぴ ppi *っち tchi っし sshi っき kki
っぷ ppu っつ ttsu っす ssu っく  kku
っぺ ppe って tte っせ sse っけ kke
っぽ ppo っと tto っそ sso っこ kko

*In the case of the doubled 'chi', 'tchi' is used in the Hepburn romanization system as discussed below, while 'cchi' or 'tti' (but annoyingly, not 'tchi') are used for computer input, which will be discussed in another article.

Voiced sounds are never preceded by a soku-on in native Japanese words, but the consonants of unvoiced combination syllables can be (ex. kasshu "singer").

It's also not unusual to find both soku-on and long vowels in the same word, for example gakkou "school, pronounced ga-k-ko-o. An extended 'n' sound is written with the moraic nasal (ん) rather than a soku-on (っ).

Spelling Rules

So far we've come across a few places where the high level pattern breaks down, but few real spelling inconsistencies (i.e., し is always 'shi', not 'si', こう is always 'koo', not 'kou'). In the case of the two possible spelling for 'ji' (じ・ぢ) and 'zu' (ず・づ), I'll always alert you to when the less common spelling is used, and as you'll see, even this is predictable.

The only other notable spelling rules for Hiragana are each for a single particle, or grammar word. Here they are:

Particle Hiragana Function
wa は (ha) marks the topic of a sentence
e へ (he) marks the direction of movement
o を (wo) marks the object of a verb

These spellings are a remnant of classical Japanese that was retained despite the post-war spelling reform. The use of the Hiragana 'wo', though, is useful because it distinguishes the particle 'o' (を) from the polite prefix 'o' (お). There are several dozen particles in Japanese, and you'll start learning the most important ones soon enough.


Languages that don't use the Latin alphabet often have multiple romanization schemes, each of which will have various advantages and disadvantages. The most common Japanese romanization system in the English speaking world is the modified Hepburn romanization system, which allows English speakers to pronounce most words more accurately than with the Kunrei-shiki system, which more closely approximates Kana and is used more often by Japanese people in Japan.

The consonant spellings I've used here basically agree with most variants of the Hepburn system. The main difference is that while the Hepburn system uses macrons (lines) over long vowels other than 'ii' and 'ei', I have no way to produce these, so instead I've been spelling all long vowels as written in Hiragana, a style known as wapuro romaji (wapuro meaning "word processor").

Remember how I've been writing "gyou" instead of "gyoo"? Some books always use 'ee' and 'oo' for long 'e' and 'o', but this makes it impossible to know what the Hiragana spelling is. With words like Tokyo (とうきょう / Toukyou) that have been assimilated into English, the long vowels may be dropped altogether in non-language-learning contexts.

One place where romanization can be ambiguous is with the moraic 'n' (which in modified Hepburn is never written as 'm'). To understand this, consider the sequence "kinen". The first 'n' might go with the following 'e', giving きねん (ki-ne-n) "commemoration", or it might be the moraic nasal, giving きんえん (ki-n-e-n) "no smoking".

To differentiate the two, when the moraic nasal is followed by a vowel or a 'y', we write an apostrophe in between. So, きねん is "kinen" and きんえん is "kin'en". This also allows us to differentiate chou-on like にゃ (nya) from sequences like んや (n'ya).

What Next?

By learning Hiragana you've completed you first major step towards speaking and writing Japanese. A few remaining issues in pronunciation will be covered in the next lesson:

Pitch Accent and Vowel Devoicing

Having trouble understanding something? Anything I've missed? Please send your feedback using the contact form and help me improve this site. All questions, comments, and corrections are welcome.