English Pronunciation for Japanese Speakers

Japanese and English build syllables in opposite ways. Japanese has five vowels, one sound that covers both r and l, and syllables that almost always end in a vowel. English has many more vowels, two separate liquids, and stacks consonants at both ends of a syllable: street starts with three of them.

The good news is that the errors this produces are predictable. Ian Thompson's chapter on Japanese speakers in Learner English (Swan and Smith, 2001) describes the same core problems this guide covers. Each one comes from a habit that is correct in Japanese, so each one has a specific fix. If you are going the other way, as an English speaker learning Japanese, our guide to Japanese IPA, pitch accent and the R tap is the mirror image of this one.

Every transcription in the tables comes from a pronunciation lexicon IPAtics maintains, checked by hand and printed exactly as it appears there. The US column puts the stress mark before the stressed syllable, so computer is /kəmˈpjutɚ/. It writes the English r as /ɹ/ and marks no vowel length. The UK column marks length with ː, writes /tʃ/ as /t͡ʃ/, and often leaves the stress mark out.

L and R: Two English Sounds for One Japanese Sound

Japanese has a single liquid consonant. It is usually an alveolar tap [ɾ], where the tongue tip flicks the ridge behind the upper teeth once, and it is sometimes closer to [l] (see the Wikipedia article on Japanese phonology). English splits that space into two phonemes, /l/ and /ɹ/, and relies on the difference to keep many words apart. Learner English lists the confusion of /l/ and /r/ as a typical problem for Japanese speakers.

The fix starts with dropping the tap for both sounds. The tap touches for an instant. English /l/ touches and stays: press the tongue tip against the ridge behind your upper teeth, keep it there, and let the voice flow out around the sides of the tongue. Hold it for a full second before the vowel, llll-ight, then shorten it step by step.

English /ɹ/ touches nothing. Pull the tongue tip back so it hangs in the air behind the ridge, or bunch the whole tongue up and back, round the lips slightly, and voice. A good way in is the vowel of her, which in American English already carries the r quality: start on that vowel and slide into the word, er-right, then drop the start. Our page on the English r sound /ɹ/ has more words to practise with.

The tap itself still has a job in English. American English uses it for a t between vowels, as in butter and water. It never stands for an English r or l.

The clusters in the lower half of the table are harder than a single sound at the start of a word, because the tongue has to go from /ɡ/ or /b/ straight into the liquid. Take them slowly and watch in a mirror: the lips round for /ɹ/ and stay neutral for /l/.

Hearing the difference can lag behind making it. Sheldon and Strange (1982) found Japanese learners whose production of /r/ and /l/ was more accurate than their perception, so do not wait until you can hear it before you practise saying it. For the ear, a 1991 training study by Logan, Lively and Pisoni improved Japanese listeners' identification of /r/ and /l/ with minimal pairs recorded by several different speakers. The /l/ and /r/ pairs in our list of English minimal pairs are a place to start.

Extra Vowels: Why Street Grows to Five Beats

Japanese syllables are built on a consonant-plus-vowel pattern, and the only sounds that can close one are the moraic nasal ん and the first half of a doubled consonant (っ). English allows three consonants before a vowel and several after it. Learner English notes that Japanese speakers tend to break up consonant clusters with vowels and to add a vowel after a final consonant. English loanwords in Japanese show the same process:

Street is one syllable in English. The loanword sutorīto adds a vowel after the s, another after the t, and a third at the end. The added vowel is usually u, and o after t and d, as in beddo and dorinku. Miruku shows the liquid problem too: the l of milk becomes a Japanese r.

Japanese already gives you a tool for this. In desu, the final u is usually whispered, and between voiceless consonants it can almost vanish. Use that whispered vowel as a stepping stone: say desuku with both u sounds whispered, then take them out until only the hiss and the stops are left. That is desk.

For clusters at the start of a word, build from the end. Say eat, then treat, then street, adding one consonant at a time to a word you can already say without a vowel break. For consonants at the end of a word, stop and do not open again. In bed, the tongue closes on the ridge for /d/ and the word ends there, in silence, instead of opening into an o.

TH Is Not S or Z

Japanese has no dental fricatives, and the closest sounds it has are s and z. Learner English lists /s/ for /θ/ and /z/ for /ð/ as typical substitutions for Japanese speakers. Minimal pairs show what that costs:

To make /θ/, put the tip of the tongue lightly against the back of the upper front teeth, or just between the teeth, and blow. The air hisses over a flat tongue tip, not through the narrow groove behind the teeth that makes /s/. /ð/ is the same position with the voice on: put a hand on your throat and feel it buzz. Check in a mirror. If you cannot see the tip of your tongue at all, you are probably making s. There are more example words on our page for the /θ/ sound.

V Is Not B, and F Is Not ふ

Japanese has no /v/ of its own, and a borrowed [v] is normally not distinguished from /b/. Learner English lists /b/ for /v/ among the typical problems, which makes very and berry the same word.

/b/ closes both lips and releases them. /v/ never closes anything: the upper teeth rest on the lower lip and the voice buzzes through the gap. A simple test is to hold the sound. A long vvvv is possible; a long bbbb is not.

/f/ has the same lip-and-teeth position with no voice. The Japanese sound written ふ is different. It is a bilabial [ɸ], air blown between the two lips with no teeth involved, and Learner English notes that Japanese speakers often use it for English /f/. Next to a real /f/ it sounds soft and breathy. Japanese uses it only before u, while English puts /f/ in front of every vowel, so practise it before vowels other than u too.

Who and hood are there for the opposite reason. In Japanese, h before u is also pronounced [ɸ], so ふ covers ground that English splits between /f/ and /h/. Say who with the lips apart and nothing but breath before the vowel. If the lips come together, it starts to sound like ふ.

See, She and Tea: S and T Before I

In Japanese, s before i is pronounced [ɕ], as in し, and t before i is [tɕ], as in ち. There is no native "si" or "ti". Learner English describes the result: Japanese speakers tend to pronounce English /s/ as /ʃ/ and /t/ as /tʃ/ before an ee or i vowel. See turns into she, sip into ship, and tip into chip. The same shift puts a swear word in place of sit.

Older loanwords follow the Japanese rule: team is チーム chīmu and ticket is チケット chiketto. Newer ones use the katakana spelling ティ to keep the t, as in パーティー pātī for party.

The fix is to keep the tongue where it is for sa or ta and change only the vowel. Say sa, hold the hiss, ssss, then move to the vowel of see without letting the middle of the tongue rise toward the roof of the mouth. Keep the lips spread, as in a smile; /ʃ/ rounds and pushes them forward. For /t/, touch the ridge with the tongue tip and let go cleanly into the vowel. The release should be a short puff of air, not the hiss that turns it into ち. Tea should start with the same /t/ as top.

Vowels: More of Them, and Length Is Not Everything

Japanese has five vowels, and each one can be short or long. English has far more. British RP, as Peter Roach describes it, has twelve pure vowels and eight diphthongs. Learner English describes Japanese speakers merging English vowels that share the same nearest Japanese vowel: /iː/ and /ɪ/ around i, /uː/ and /ʊ/ around u, and /æ/, /ʌ/ and /ɑː/ around a.

Length first. Japanese speakers already hear vowel length, because it changes meaning in Japanese, and that is a real advantage: an い held for two beats gets you close to /iː/. But the short partner is not a short い. English /ɪ/ is lower and more relaxed, and /ʊ/ differs from /uː/ in the same way. English length also shifts with the next sound. Roach calls it pre-fortis clipping: a vowel is shortened before a voiceless consonant, so the vowel in heat is shorter than the one in heed, yet it is still /iː/. If you listen only for length, heat can sound like hit. The US column makes the same point: it marks no length at all, and hit and heat still differ, because the vowels are different symbols.

The Japanese う is made with the lips barely rounded, so fool and Luke need more lip movement than feels natural.

The a sounds and the vowel of bird are the second group. Katakana writes the bird vowel and the car vowel the same way, as a long ā: card is カード kādo and girl is ガール gāru. So heart and hurt look identical in katakana, while English keeps them apart:

The last pair is the pattern behind work and walk, another common mix-up: work has /ɜː/ with neutral lips, and walk has /ɔː/ with rounded ones. In the US column the /ɜː/ vowel appears as /ɝ/, the r-coloured version.

Rhythm: Stress Timing Against Mora Timing

Japanese is mora-timed: each mora takes roughly the same amount of time. English is usually described as stress-timed. Stressed syllables are long, loud and clear, and the syllables between them are squeezed, with their vowels often reduced to schwa /ə/. Learner English notes that Japanese speakers tend to give every syllable similar weight and a full vowel. The loanwords show how far apart the two systems are:

Banana is the telling one. It has three beats in both languages, but English stresses only the middle syllable and reduces the other two to /ə/. Say it with three equal, full a sounds and the word sounds foreign even though every consonant is right.

Stress also moves within a word family, and the vowels change with it:

The first vowel is a full /oʊ/ when it carries the stress and /ə/ when it does not. The second vowel does the opposite. The UK entries for these words carry no stress mark, so read the stress from the US column, where ˌ marks a secondary stress.

The fix is to find the stressed syllable first and make it longer, louder and a little higher, then let the rest go short. Tap the table once per stressed syllable, not once per syllable. The same applies across a sentence: small words such as to, for, and and can are normally unstressed and reduced. Rhythm is hard to train one word at a time, which is why the shadowing technique, speaking along with native audio, suits it well.

Where to Start

You do not have to work on everything at once. One sensible order is to start with extra vowels and rhythm, because they show up in almost every sentence, then move to L and R and the vowel pairs, and finish with TH, V, F and the see/she shift.

A weekly routine that covers one section at a time:

1. Listen first. Play the pairs from one table and try to tell them apart before you say anything.
2. Say them slowly. Use the mouth descriptions above and a mirror. Exaggerate at first.
3. Record and compare. Your own recording is harder to fool than your ear in the moment.
4. Move to sentences. Put the words in short sentences and shadow a native recording.
5. Check new words before you say them. Paste them into the English to IPA converter to see which vowel they use and where the stress falls.

Trying It Yourself

Most of these problems hide in words you already know. When a word catches you in an article or an email, check it on the spot instead of guessing. IPAtics shows the IPA for any text you select, in any app, so the stress mark and the vowel are in front of you before you say the word.

Sources

• Thompson, I. (2001). "Japanese speakers." In M. Swan and B. Smith (eds.), Learner English: A Teacher's Guide to Interference and Other Problems (2nd ed.). Cambridge University Press. The source for every statement above about errors Japanese speakers typically make.
• Roach, P. (2009). English Phonetics and Phonology: A Practical Course (4th ed.). Cambridge University Press. The RP vowel inventory, pre-fortis clipping, stress timing and weak forms.
• Wikipedia, Japanese phonology. The tap [ɾ], [ɸ] before u, [ɕ] and [tɕ] before i, vowel length, mora timing, and /v/ merging with /b/.
• Sheldon, A. and Strange, W. (1982). "The acquisition of /r/ and /l/ by Japanese learners of English: Evidence that speech production can precede speech perception." Applied Psycholinguistics 3(3).
• Logan, J. S., Lively, S. E. and Pisoni, D. B. (1991). "Training Japanese listeners to identify English /r/ and /l/: A first report." Journal of the Acoustical Society of America 89(2).