Romanization Systems: Hepburn, Kunrei, Nihon-shiki

Sooner or later you will see the same Japanese word spelled two different ways in Latin letters — Fuji on one map and Huzi in a linguistics paper, shinbun in your textbook and sinbun on a Japanese schoolchild's worksheet — and wonder whether you misheard something. You didn't. Japanese has three competing romanization systems (ローマ字, rōmaji), and they disagree on exactly the sounds English speakers find hardest: し, ち, つ, ふ, じ. This page lays them side by side so you always know which one you are looking at. This guide, like almost all material aimed at learners, uses Hepburn — but knowing the other two will save you real confusion.

The three systems at a glance

Read down the columns and the personalities emerge. Hepburn writes what an English speaker's eye expects. Kunrei-shiki keeps the consonant constant across a whole row. Nihon-shiki does the same but also preserves distinctions that have since merged in speech.

Hepburn (ヘボン式) — the one you'll see everywhere

Hepburn (Hebon-shiki) was popularized by the American missionary James Curtis Hepburn in his 1867 Japanese–English dictionary. Its guiding idea is pronunciation for a foreigner: spell each kana so that a reader of English will land close to the Japanese sound. That is why し is shi (not si), つ is tsu, and ふ is fu — an English si, tu, hu would mislead the eye.

新聞 is shinbun in Hepburn (Kunrei: sinbun), 写真 is shashin (Kunrei: syasin). Hepburn is the system on Japan's road signs, on station names, in passports (a "modified Hepburn" variant), and in essentially every textbook for non-Japanese learners — including this one.

Kunrei-shiki (訓令式) — the schoolroom system

Kunrei-shiki ("cabinet-directive system") was set by Japanese government order in 1937 and reaffirmed in 1954; it is the ISO 3602 standard and the romanization taught in Japanese elementary schools. Its guiding idea is the opposite of Hepburn's: regularity of the kana grid, not English intuition. Since た, ち, つ, て, と form one t-row in Japanese phonology, Kunrei spells them all with t: ta, ti, tu, te, to.

This is why a Japanese-authored worksheet writes ち as ti and し as si. It is not an error — it is a different, internally consistent system, and to a Japanese child it is the system.

In Hepburn that mountain is Fuji; in Kunrei it is Huzi. Same に-ほ-ん word, different letters.

Why linguists prefer Kunrei

Kunrei's grid-regularity is not just tidiness — it makes Japanese morphology visible in a way Hepburn hides. Take the verb 立つ (tatsu, "to stand"). Its conjugation is perfectly regular in the kana: a single stem た-t plus each row-vowel.

In the Kunrei column the stem is obviously constant — tat- — and the verb looks as regular as it actually is. In the Hepburn column the same verb seems to lurch between tat-, ta-chi, and ta-tsu, disguising a perfectly clean pattern. That is the "phonemic honesty" linguists value: Kunrei describes the system, while Hepburn describes the surface sound. Both are right about different things.

Nihon-shiki (日本式) — the strict ancestor

Nihon-shiki (1885) is the oldest and most rigidly one-to-one of the three. It matches Kunrei on the t- and s-rows (ti, tu, si, hu, zi) but goes further: it keeps apart two pairs of kana that merged in modern pronunciation. Today じ and ぢ sound identical (both ji-ish), and ず and づ sound identical (both zu-ish) — a merger called yotsugana. Hepburn and Kunrei write the merged sound (ji, zu); Nihon-shiki insists on the kana's identity and writes ぢ as di and づ as du.

続く is tsuzuku in Hepburn and tuzuku in Kunrei, but つづく in Nihon-shiki is tuduku — the づ shown as du, faithful to the kana even though no one pronounces a d there anymore. Nihon-shiki is essentially confined to historical linguistics and certain traditionalists today, but it explains why you occasionally meet spellings like Ohno → Oho or place names romanized with wo and du.

Long vowels: macron, circumflex, or doubled letter

The systems also disagree on how to show a long vowel, and this is where casual romanization gets messiest.

Hepburn marks length with a macron (ō, ū); Kunrei and Nihon-shiki use a circumflex (ô, û). Passports and everyday signage often drop the mark entirely (Tokyo, Osaka) — which is why so many English speakers never realize those names have long vowels at all. The spelling you type at a keyboard is a fourth thing again: it mirrors the underlying kana, so 東京 is entered toukyou (お + う). This guide uses Hepburn with macrons throughout the romanization line, so you always see where a vowel is held. For where the length lives in the kana itself, see spelling long vowels.

The same word, three ways

Every row is the same Japanese word, unchanged. Only the Latin dress differs.

Common mistakes

The core error is treating a spelling difference as a sound difference. Shi and si, Fuji and Huzi, chawan and tyawan are the same words; the systems merely draw them differently. Pick one system (this guide picks Hepburn), read consistently within it, and translate on sight when you meet another.

Key takeaways

• Hepburn (shi, chi, tsu, fu, ji) targets English intuition; it rules signs, passports, and learner materials — and this guide.
• Kunrei-shiki (si, ti, tu, hu, zi) targets grid-regularity; it is taught in Japanese schools and exposes verb stems (tat-i, tat-u).
• Nihon-shiki additionally keeps merged pairs apart: ぢ = di, づ = du, を = wo.
• Long vowels: Hepburn macron (Tōkyō), Kunrei/Nihon-shiki circumflex (Tôkyô), signage often omits it (Tokyo), keyboards spell the kana (toukyou).
• A different spelling is not a different word — learn to convert between systems on sight.