Japanese Long Vowels and Double Consonants Explained

Why one beat changes the word

English vowel length is a by-product of stress and the consonant that follows, so English speakers rarely listen for it. Japanese uses length as a building block of the word itself. The unit of timing is the mora, a beat that is roughly equal whether it holds a full syllable like ka, a single n, the second half of a long vowel, or the silent hold of a double consonant. Obasan has four beats and means aunt. Obāsan has five and means grandmother. Nothing else changes, and nobody will guess from context if you ask a shopkeeper about your grandmother's discount.

The pairs below are the ones every teacher uses because they come up in ordinary talk. Read each column aloud and make the second word noticeably longer, not louder. Most learners make their long versions too short and their short versions slightly too long, so the two blur together. Exaggerate the gap in practice until the natural gap feels normal.

How long vowels are spelt in hiragana

Hiragana has no length mark, so a long vowel is written by adding a second kana. For a, i and u the added kana matches the vowel: おかあさん, おじいさん, くうき. For e and o the spelling is historical. Most long e sounds are written with い, as in えいご and せんせい, and most long o sounds with う, as in おとうさん and ありがとう. In everyday speech えい is usually pronounced as a held e rather than a glide, although careful speakers and singers sometimes keep the i audible.

A short list of native words spells long o with お instead of う: おおきい, とおい, とおり, こおり, とお. Learners meet these early, so it is worth memorising them as a set. Romaji marks long a, o and u with a macron. Long i is usually written ii, and adjectives ending in ii such as oishii are genuinely two i beats, so this post writes ojīsan and chīzu with a macron only where the length is the whole point of the contrast.

Katakana and the long-vowel bar

Katakana solves the same problem with one symbol: the bar ー, which simply repeats the previous vowel for one more beat. It runs horizontally in horizontal text and vertically in vertical text, which is why it is called the long-vowel mark and not a dash. Loanwords are full of it because English stressed vowels tend to be borrowed as long ones: coffee becomes kōhī with two long beats, and cake becomes kēki. English speakers often shorten these back to English rhythm and say something closer to kohi, which sounds clipped.

The small tsu is a silent beat

The small っ, called sokuon, doubles the consonant that follows it. In speech it is not a sound but a beat of preparation: your tongue or lips move into position for the next consonant and hold there for one mora before releasing. In gakkō the back of the tongue closes for k, waits one beat in silence, then releases into kō. In kitte the tongue tip presses behind the teeth and waits. Before s and sh the hold is a lengthened hiss instead of silence, so zasshi has a drawn-out sh before the final i.

In native and Sino-Japanese vocabulary the small tsu appears before k, s, t and p, including their partners sh, ch and ts. Romaji doubles the consonant letter, with one exception: before ch, Hepburn writes tch, so マッチ is matchi and 一致 is itchi. The common mistake is to skip the hold and say kite for kitte, or to fill the silence with a vowel and produce something like kitsute. Neither is understood as a stamp.

The small tsu at the end of exclamations

Written at the end of a word with nothing after it, the small tsu marks a sharp cut-off: the breath stops abruptly in the throat, like the catch in the middle of the English uh-oh. You see it in あっ and えっ, in a sudden 痛っ when something hurts, and in casual writing of any word that is bitten off. There is no released consonant, only the stop. Romaji has no standard letter for this, so the rows below use an apostrophe to mark it.

Casual speech also inserts a small tsu into adverbs for emphasis, turning とても into とっても and すごく into すっごく. Both are informal and carry feeling, so they belong with friends rather than in a job interview. Practise them anyway, because the added beat is exactly the same mechanism as in kitte, and hearing it in emotional speech helps you notice it everywhere else.

Why you shorten long sounds under pressure

Learners who hold long vowels perfectly in a drill often lose them the moment a real person is waiting for an answer. The reason is rhythm, not knowledge. Under stress, speech speeds up and falls back on the rhythm of your first language, and in English that rhythm compresses everything that is not stressed. A long o in the middle of a word is exactly the kind of sound that gets squeezed. You also tend to mark importance with loudness, so the long vowel gets louder instead of longer, which a Japanese listener does not register as length at all.

The remedy is to make the beat physical. Tapping a finger once per mora gives the long vowel two taps and the small tsu one silent tap, and the hand keeps time even when the mouth is nervous. Slow the whole sentence rather than only the long sound, because a long vowel inside a rushed sentence still ends up short. Once the sentence is steady, bring the speed back up in stages while the taps stay even.

• Tap once per beat on your leg or the table, and keep tapping through the silent hold of a small tsu.
• Say the sentence at half speed first, then at three-quarter speed, then at normal speed, checking that the taps stay even.
• Make the long vowel longer, never louder; loudness is stress, and stress is not length.
• When you feel yourself rushing in a conversation, slow the whole sentence down instead of trying to rescue one vowel.
• Record the same sentence calm and then after a short burst of exercise to hear how much pressure shortens your beats.

Contrast drills inside real sentences

Isolated pairs are useful for hearing the difference, but the skill you need is keeping the difference while thinking about grammar and meaning. These sentences put both members of a pair, or a long sound next to a short one, inside a natural line. Read each one three times: slowly with taps, at normal speed with taps, and at normal speed with your hands still. If the third reading collapses the contrast, go back to the second.

A typical reply to koko wa kōkō desu ka is hai, sō desu or iie, chigaimasu, and the person answering will unconsciously copy your length. If they answer about koko, here, rather than kōkō, you have your feedback without anyone correcting you. Listen to the reply as closely as you listened to your own line.

Speed ladders for long sentences

Once single contrasts are steady, stack several long and doubled sounds in one sentence and climb a speed ladder: half speed, then normal, then slightly faster than comfortable, then back to normal. The overshoot step matters, because normal speed feels slow afterwards and the beats stay intact. These sentences are deliberately crowded. If one word keeps collapsing, isolate it, tap it, and put it back.

• Read the sentence at half speed with taps and confirm every long vowel gets two taps.
• Read it at normal speed with taps, then without taps.
• Read it once slightly faster than feels comfortable, accepting a small loss of clarity.
• Return to normal speed and record it; this take is the one to compare against a native model.

How to check yourself

Length is hard to judge from inside your own head, because you know which word you meant. Record yourself, listen back the next day with the meanings covered, and try to tell which member of each pair you said. If you cannot decide, a listener certainly cannot. Playing a native recording of the same sentence straight after yours shows whether your long vowels are too short or your short ones are drifting long. The pronunciation coaching from your AI Sensei in Unihongo's immersive 3D classroom gives the same kind of check in conversation, flagging the words where your timing blurred while you were busy thinking about what to say.

• Record the minimal-pair table once a week and listen back blind, guessing which word each recording was.
• Shadow a native sentence and record yourself over it, then listen for places where your version finishes early.
• Ask a speaking partner or your AI Sensei to write down what they heard rather than what you meant, and compare.
• Keep a short list of the words you personally shorten, such as kōhī or gakkō, and open every session with them.