How to Pronounce the Japanese R Sound

What the Japanese r actually is

Put the tip of your tongue on the bumpy ridge just behind your upper front teeth, where the tongue sits at the start of an English d. Now flick it down and away in one quick movement while your voice keeps going. That single brush is the Japanese r, known to phoneticians as an alveolar tap [ɾ]. There is no curling of the tongue backwards, no bunching in the middle of the mouth, and no lip rounding, all of which are part of the English r and all of which sound wrong in Japanese.

If you speak American English, you already make this tap every time you say butter, water or city quickly: the tt or t between the vowels is not a real t but a flap of the tongue tip. Say butter, notice the middle sound, then say bara with the same middle. If you speak Spanish, the single r in pero is essentially identical. British English speakers who pronounce a full t in butter can build the movement from d instead, as shown below.

Why it sounds like l to some ears and d to others

The tap is made at the same place as both l and d, so it gets pulled towards whichever the listener expects. For l, the tongue tip stays on the ridge while air flows around the sides, so the sound can be held. For d, the tongue closes fully, pressure builds, and the release is a small burst. The tap does neither: contact is complete but so brief that no pressure builds, and it cannot be held. An English ear files any brief tongue-tip contact between vowels as a t or d, so it hears something d-like, while a learner listening for a foreign r hears the l-like lightness.

Japanese speakers themselves are not perfectly uniform. Before i and e the tongue is a little further forward and flatter, so ri and re often sound closer to l. Before o, u and a, especially at the start of a word, the tap is firmer and can sound closer to d. Neither is wrong, and native listeners hear both as the same sound. What they will not accept as the same sound is an English r, which they hear as a w-like glide, or a trill, which carries attitude.

Finding the position from a d sound

The most reliable route to the tap is through d, because d already has the tongue in the right place. Say da-da-da slowly, feeling the full closure and the small burst. Speed up until the closure has no time to build pressure and the burst disappears. At that point you are making ra-ra-ra. The words dare and dore are ideal first targets because they contain a real d and a tap side by side: the d is firm and the r is light, and you can feel the two in the same word.

• Say da-da-da-da, then speed up until it blurs into a light ra-ra-ra-ra; that blur is the target.
• Say the English phrase a lot of at conversational speed and feel the t between lot and of; that flap is the Japanese r.
• Hold your jaw still with a finger on your chin so the movement comes from the tongue tip alone, not from the whole jaw.
• Keep the lips completely relaxed; if they round even slightly, an English r is creeping in.
• Check in a mirror that the tongue tip is what moves and that it does not stay pressed against the ridge.

Kore, sore, are and dore are the best early drill words in the language because they appear in almost every beginner conversation and put the tap between two vowels, which is its easiest position. A shop assistant asked dore desu ka will reply with kore desu or kochira desu, so you hear the native tap straight back at you.

Ra, ri, ru, re, ro one at a time

The five kana in the r row are the same tap followed by five vowels, but each vowel pulls the mouth into a different shape and creates its own mistake. Ra is the easiest. Ri needs the tongue tip slightly further forward, and it is where learners most often produce a held l instead of a tap. Ru is the most common failure point because English speakers round their lips for anything spelt ru; keep the lips flat and the u short and unrounded. Re should stay short, without the English glide of ray. Ro needs relaxed lips, not the tight circle of an English row.

Ryo, ryu and rya in a single beat

The combinations りょ, りゅ and りゃ are one mora each, exactly as long as ro, ru or ra. The y is not a separate sound but a narrowing of the tongue that happens during the tap, so ryo is a ro made with the middle of the tongue raised towards the roof of the mouth. English speakers stumble here in two ways. The first is to say ri-yo, adding a beat and turning ryokō, travel, into something with four beats instead of three. The second is to give up on the r and say yo, so ryōri becomes yōri.

The fix is to start from the vowel and add the tap last. Say yo, then yo-yo-yo, then flick the tongue tip at the very start of each yo without lengthening it: ryo-ryo-ryo. Do the same for yu and ya. When you can say ryo in the same time as ro, put it into words. Ryōri and ryokō come up constantly, ryūgaku matters for students, and ryōshin, parents, will appear in almost every self-introduction.

Words that stack r sounds

One tap is manageable. Two or three taps in the same word are where the sound falls apart at speed, because the tongue has to reset between them and English wants to turn the sequence into a glide. The table climbs from single taps between vowels to word-initial taps, then to the ry combinations, and finally to words with repeated r. Work down it in order and do not move a level until the previous one is clean at conversational speed.

Renraku and rirekisho are the two words learners most often ask to be corrected on, because each has two taps close together and one of them sits at the start of the word, where there is no vowel to launch from. Say the first syllable as a light da if you must, then lighten it again; a slightly d-like first tap is far better than an English r.

Why you should not roll it

A trilled r, where the tongue tip vibrates several times, does exist in Japanese, and learners with Spanish, Italian or Russian backgrounds sometimes reach for it because it feels close to the tap. In Japanese it carries a strong social meaning. It is the rolled r of tough characters in films, of shouted threats, and of some very rough casual speech, which is why it is sometimes called makijita, rolled tongue, and treated as a marker of aggression. Using it in ordinary conversation makes an innocent sentence such as kore o kudasai sound like a demand.

If your first language has a trill, the adjustment is to stop after a single contact. Say your rolled r deliberately slowly and notice that the tongue bounces several times; now cut it to one bounce and release. The word arigatō is a good monitor, because the r comes after a vowel and any second bounce is easy to hear when you record it. Aim for a sound so light that a listener might mistake it for a d.

A two-week drill

Five minutes twice a day is enough, provided you record on the days marked and compare against a native model. The first week builds the movement and the five kana; the second week takes it into the hard combinations and then into unscripted speech, where the sound has to survive while you think about something else. Do not skip the final day, because the whole point is a tap that holds up in conversation rather than in a mirror.

Phrases to shadow

These lines are chosen so that the tap appears in every position: between vowels, at the start of a word, before y, and repeated within a word. Play a native recording or your AI Sensei's model, speak over it with a half-second delay, and keep your hands still. The replies show what you will hear back in a real exchange, and they contain their own r sounds to copy.

How to check yourself

The tap is hard to monitor from the inside, because the movement is fast and your ear fills in the sound you intended. Record the shadow phrases on day seven and again on day fourteen, then play each line directly after the native model with your eyes closed. Listen for an r that sounds like a w, an r held long enough to be an l, and any word with more than one bounce. Pronunciation coaching from your AI Sensei in Unihongo's immersive 3D classroom adds the check you cannot do alone, picking out the r sounds that slid back to English while you were mid-sentence.

• Record ra ri ru re ro and listen back at half speed; a tap sounds like a soft d when slowed, while an l stays an l.
• Say kore to a speaker and listen to how they say kore back; copy the lightness rather than the letters.
• Keep a personal list of the words where your r slips, and start every session with those five words.
• Read the two-week table again after the fortnight and repeat only the days whose words still feel wrong.