When English words enter Japanese, they are not simply spelled out in a new alphabet — they are rebuilt from the ground up to obey Japanese sound rules. The result often looks and sounds so different from the original that English speakers fail to recognize their own words: milk comes out as three beats, ミルク (miruku), and McDonald's stretches to six, マクドナルド (makudonarudo). The good news is that this rebuilding is not random. It follows a small set of consistent rules, and once you know them you can often predict the katakana form of a word you have never seen written — and, just as usefully, work backwards from a katakana word to the English source.
Rule 1: Every consonant gets a vowel
English lets consonants pile up (str-, -lk, -nd) and end a syllable bare (cat, bus). Japanese does not. A word like strike, one syllable in English, has to be pulled apart into a chain of consonant-plus-vowel beats, and a vowel is glued onto every stranded consonant.
今のは完全にストライクだろ、なんでボールなんだよ。
ima no wa kanzen ni sutoraiku daro, nande bōru nan da yo
That was a clear strike — how was that a ball?
Strike → ス・ト・ラ・イ・ク (su-to-ra-i-ku): the s gets u, the t gets o, the r gets a, and the final k gets u. One English syllable becomes five Japanese beats (morae). This is why loanwords feel so much longer in Japanese — see Mora vs Syllable for why the beat, not the syllable, is the unit that matters.
Which vowel gets inserted? The default is /u/ (ウ), the shortest, most neutral Japanese vowel. But there are two systematic exceptions:
- After t and d, the inserted vowel is /o/, because tu and du would come out as tsu and zu and distort the consonant: bed → ベッド (beddo), hint → ヒント (hinto).
- After ch and j, the inserted vowel is /i/: match → マッチ (macchi), bench → ベンチ (benchi).
明日、数学のテストがあるのに全然勉強してない。
ashita, sūgaku no tesuto ga aru noni zenzen benkyō shite nai
I've got a math test tomorrow and I haven't studied at all.
Test → テ・ス・ト (tesuto): the s takes u, the final t takes o.
Rule 2: L and R both become the Japanese tap
English distinguishes light from right; Japanese has a single liquid, the tap ラ行 (ra, ri, ru, re, ro). Both English l and English r collapse onto it. This is why light and right become the identical loanword ライト (raito) — context alone tells them apart.
部屋のライトが切れたから、電球を買ってこないと。
heya no raito ga kireta kara, denkyū o katte konai to
The light in my room burned out, so I need to go buy a bulb.
コーヒーにミルクと砂糖を入れますか。
kōhī ni miruku to satō o iremasu ka
Do you take milk and sugar in your coffee?
Milk → ミ・ル・ク (miruku): the l becomes the tap (ル), and the final k gets a u. Notice too that the -lk cluster is impossible in Japanese, so a vowel splits it.
Rule 3: TH → s/z, V → b (or ヴ), F → フ
Japanese lacks the English th, v, and (native) f sounds, so each is mapped to the nearest available consonant:
- th (voiceless, as in think) → s: thank you → サンキュー (sankyū).
- th (voiced, as in smooth) → z: smooth → スムーズ (sumūzu).
- v → b in the traditional system: violin → バイオリン (baiorin), television → テレビ (terebi).
- f → the syllable フ plus a small vowel (ファ・フィ・フェ・フォ): fork → フォーク (fōku).
夜はだいたいテレビを見ながらご飯を食べる。
yoru wa daitai terebi o minagara gohan o taberu
In the evening I usually eat dinner while watching TV.
Television → テレビ (terebi): the v becomes b, and the word is clipped short (Japanese loves to shorten long loans). To write foreign sounds more faithfully, modern Japanese has an extended katakana set — ヴ for v (ヴァイオリン, vaiorin), ティ for ti, ウィ for wi — covered on Extended Katakana for Foreign Sounds. In everyday speech, though, most people still pronounce ヴ as b.
手続きが思ったよりスムーズに進んで助かった。
tetsuzuki ga omotta yori sumūzu ni susunde tasukatta
The paperwork went more smoothly than I expected — what a relief.
Rule 4: Final consonants — vowel-add or gemination
A bare final consonant is repaired in one of two ways. Usually it simply gets a vowel (Rule 1). But when the English word has a short, stressed vowel right before the final consonant, Japanese doubles the consonant with a small っ (the sokuon) before adding the vowel — capturing the "clipped" feel of the English syllable.
| English | Katakana | Romanization | Repair |
|---|---|---|---|
| bag | バッグ | baggu | gemination + u |
| cut | カット | katto | gemination + o |
| bed | ベッド | beddo | gemination + o |
| kick | キック | kikku | gemination + u |
| bus | バス | basu | plain vowel + u |
| pen | ペン | pen | -n → moraic ん |
美容院で髪をばっさりカットしてもらった。
biyōin de kami o bassari katto shite moratta
I got my hair cut nice and short at the salon.
Only -n and -m escape needing a vowel: English word-final n/m map onto the moraic ん (see The Moraic N (ん)), as in pen → ペン (pen).
Rule 5: Stress becomes length — or vanishes
English is a stress-timed language; Japanese is not. So English stress has nowhere to go. Two things happen:
- Long or stressed English vowels, and the vowels of -r-colored syllables (car, beer, printer), turn into long vowels written with the ー bar (see The Chōonpu ー).
- Everything else flattens into an even sequence of equal beats with pitch accent replacing stress entirely.
仕事終わりのビールが、一日で一番うまいんだよなあ。
shigoto owari no bīru ga, ichinichi de ichiban umai n da yo nā
The after-work beer is the best part of the whole day.
Beer → ビール (bīru), with a long ī. This length is not decorative — it is contrastive. Drop it and you get a different word:
| Katakana | Romanization | Meaning |
|---|---|---|
| ビル | biru | building (from "building") |
| ビール | bīru | beer |
An English speaker who says both with the same short vowel will be misheard, because in Japanese the extra beat is the difference between a skyscraper and a drink.
The predictability payoff
Because the rules are consistent, you can run them forward on a brand-new word. Take printer: p-r-i-n-t-e-r → break the pr cluster (pu-ri), keep n as ん, add o after t, turn the final -er into a long ー → プリンター (purintā). Or elevator → エレベーター (erebētā), v → b, final -or → ー.
Two closing points worth internalizing. First, this same machinery explains the famous case of one English word borrowed twice: ストライク (sutoraiku, a baseball strike) and ストライキ (sutoraiki, a labor strike). Both come from strike, borrowed at different times for different senses, and the final vowel froze differently.
電車がストライキで止まっていて、会社に遅れそうだ。
densha ga sutoraiki de tomatte ite, kaisha ni okuresō da
The trains are stopped because of a strike, and I'm probably going to be late for work.
Second, adaptation sometimes drags the meaning along with the sound. マンション (manshon, from "mansion") means an ordinary concrete apartment, not a stately home; サービス (sābisu, "service") often means "free of charge." Sound-adapted words that have drifted in meaning are false friends — treated in full on False Friends & 和製英語.
Common mistakes
Mistake 1: Pronouncing katakana loans with English phonology. English speakers instinctively re-import the original stress, clusters, and coda consonants — and are not understood.
❌ milk(英語の1音節のまま)
'milk' as one clipped English syllable
Incorrect — Japanese ears expect three separate beats.
✅ ミルク
mi-ru-ku
Milk — three equal morae, each consonant with its own vowel.
Mistake 2: Dropping the epenthetic vowels. Saying sutrike or strike instead of the full five-beat su-to-ra-i-ku. The inserted vowels are not optional padding; they are the word.
✅ ストライク
su-to-ra-i-ku
Strike (baseball) — five morae, no clusters.
Mistake 3: Skipping vowel length. Because English does not use length contrastively, learners flatten ビール to ビル and コーヒー to コヒー. Length is meaning here.
❌ コヒー
kohī (or kohi)
Incorrect — the first vowel must be long.
✅ コーヒー
kōhī
Coffee — a long ō and a long ī, four morae total.
Mistake 4: Assuming the katakana word means what the English word means. マンション is not a mansion; ハイテンション (hai-tenshon) means "excited/hyper," not "tense." Learn the Japanese sense, not the English one — details on the false friends page.
Mistake 5: Mishearing a spoken loanword by reverting to English structure. When a native says マクドナルド at speed, learners "hear" the English syllables and get lost. The fix is to expect the Japanese mora structure: ma-ku-do-na-ru-do, six even beats, no English stress peak.
駅前にマクドナルドができたから、今度みんなで行ってみようよ。
ekimae ni makudonarudo ga dekita kara, kondo minna de itte miyō yo
A McDonald's opened by the station — let's all go check it out sometime.
Key takeaways
- Japanese bans clusters and (most) final consonants, so loanwords are rebuilt by inserting vowels — default /u/, but /o/ after t/d and /i/ after ch/j.
- l and r merge onto the single tap; th → s/z, v → b (or ヴ), f → フ-series.
- English stress becomes vowel length (ー) or disappears; short stressed finals trigger gemination (っ).
- The rules are regular enough to predict a word's katakana — and mishearing a loanword usually means you slipped back into English syllable structure.
Now practice Japanese
Reading grammar gets you part of the way. The exercises are where it sticks — free, no signup needed.
Start learning Japanese→Related Topics
- Extended Katakana for Foreign SoundsN4 — How modern katakana stretches to spell fa, ti, wi, che, v and other sounds Japanese did not originally have.
- When Katakana Is UsedN4 — The full set of jobs katakana does — loanwords, mimetics, scientific names, branding, and native-word emphasis — and why it is not just an 'English marker'.