Japanese pronunciation: every sound, beat and common mistake, with audio

The 5 Japanese vowels

Japanese has five vowels, always in the same order: a, i, u, e, o. They are short and pure — none of them glides into a y or w the way English "day" and "go" do — and each one is pronounced the same in every word. Four of them are close to Spanish or Italian vowels; the fifth, う, is the one English speakers need to change.

Rhythm: every beat (mora) takes the same time

Japanese is timed in morae — beats — not syllables. Each kana is one beat, and every beat gets roughly the same length. Three things count as a beat of their own even though they are not a full syllable in English: the second half of a long vowel, the small っ, and ん.

So 日本 (にほん, "Japan") is three beats: に-ほ-ん. 東京 (とうきょう, "Tokyo") is four: と-う-きょ-う — not the two English syllables of "TO-kyo". Clapping or tapping the beats is the fastest way to hear the rhythm.

Japanese consonants, row by row

The kana chart is organized in rows by consonant: ka-ki-ku-ke-ko, sa-shi-su-se-so and so on. Most rows keep one consonant throughout, and that consonant is close to English. The surprises are the kana where the consonant changes before i or u (し, ち, つ, ひ, ふ, じ) and the r-row.

Long vowels, small っ, ん and small ゃゅょ

These four spellings are where Japanese rhythm lives. Three of them add a beat (a long vowel, small っ, ん); the fourth, small ゃ ゅ ょ, takes a beat away by merging two kana into one. Get them wrong and you say a different word.

Vowel devoicing: why です sounds like "dess"

In Tokyo Japanese, the vowels i and u often become whispered — devoiced — when they sit between two voiceless consonants (k, s, t, h, p, ch, sh, ts, f) or after a voiceless consonant at the end of a phrase. The vowel is not deleted: the mouth still makes it, but the voice is off. That is why です sounds like "dess", ます like "mass", and した like "shta".

Devoicing is natural but optional, and it varies by region and speed. If you say the vowel fully, you will be understood; if you devoice it, you will sound more like a Tokyo speaker.

Pitch accent: high and low, not loud and soft

English marks words with stress: one syllable is louder and longer. Tokyo Japanese marks them with pitch instead: each mora is high or low, and words differ in where the pitch drops. Loudness and length stay even. 箸 (はし, chopsticks) is high-low; 橋 (はし, bridge) is low-high. The kana are identical — only the melody differs.

Pitch accent matters for sounding natural, but context almost always makes the word clear, and the patterns vary between regions. Most learners get the sounds and rhythm right first, then work on pitch.

The recorder on this page scores sounds and rhythm, mora by mora — not pitch accent. No score you see reflects pitch. To check pitch, use a pitch-accent dictionary or a teacher.

The 8 mistakes English speakers make in Japanese

1. 1

   Stressing one syllable the English way

   English makes one syllable loud and long and squeezes the rest: "ka-ra-OH-kee". Japanese gives every mora the same length and strength: カラオケ is ka-ra-o-ke, four even beats.

   カラオケkaraokeひろしまHiroshima
2. 2

   Rounding the う

   English "oo" pushes the lips forward. Japanese う is short, with the lips relaxed and nearly flat.

   うみすし
3. 3

   Cutting long vowels short (or stretching short ones)

   Length changes the word: おばさん is aunt, おばあさん is grandmother; ここ is here, こうこう is high school. Count the beats and hold long vowels for two.

   おばさんおばあさん
4. 4

   Skipping the small っ

   The small っ is a full silent beat. Without it, きって (stamp) becomes きて (come), and かっこ (brackets) becomes かこ (the past).

   きってちょっと
5. 5

   Using the English r

   The r-row is a single light tap of the tongue tip, like the tt in "butter". The English r — tongue pulled back, lips rounded — is not used.

   りんごありがとう
6. 6

   Gliding え and お

   English "say" and "go" end in a y or w glide. Japanese え and お are pure, and えい/おう are long vowels, not glides: せんせい is "sen-seh" with a held e, ありがとう ends in a long, steady o.

   せんせいそうso; right · sō
7. 7

   Shortening or splitting ん

   ん is a full beat, and it does not link to the next vowel. 日本 has three beats (ni-ho-n), and きんえん (no smoking) must not turn into きねん (commemoration).

   日本きんえん
8. 8

   Splitting きょ, しゃ and other small-kana sounds

   Small ゃ ゅ ょ merge into one beat. 東京 is to-u-kyo-u — four beats — not "to-ki-o". びょういん (hospital) and びよういん (beauty salon) differ exactly here.

   東京びょういん

Minimal pairs: words that differ by one beat

Most Japanese minimal pairs are about timing, not sound: a long vowel, a small っ or a small ゃゅょ. Listen to both words, tap out the beats, then say them.

A note on the score: the recorder scores each word mora by mora against the word you were asked to say, so a missing beat shows up as a weak mora — but it cannot tell you what you said instead, and it does not score pitch accent.