Why romanization differs between textbooks

Two systems

There are two main ways to write Japanese in Roman letters: Hepburn (ヘボン式) and Kunrei (訓令式).

Hepburn is tuned so that an English speaker reading it aloud lands near the right sound. Passports, road signs, station names and English-language textbooks all use it.

Kunrei mirrors the logic of the gojūon chart instead. Every か-row kana takes k, every さ-row kana takes s, consistently. It is taught in Japanese schools, and Japanese keyboard input is closer to it.

Where they differ

Everything else is identical in both. The differences are confined to parts of the さ, た and は rows, plus the yōon built from them.

Which to use

As a learner, default to Hepburn. Nearly all romaji you will see in Japan is Hepburn, as are English dictionaries and course books.

Kunrei is worth a look because its regularity exposes the chart’s structure. Seeing sa si su se so in a row makes it obvious that し belongs to the same family.

Either way, romaji is a training wheel. Leaning on it too long stalls your kana reading speed. Drop it the moment you can read without it.

Long vowels are their own mess

Even within Hepburn, long vowels are not standardized. とうきょう appears as Tōkyō, Tokyo, Toukyou and Tohkyoh.

Passports omit length entirely, so いとう is rendered ITO. Road signs use the macron, Tōkyō.

The upshot is that the English spelling of a Japanese place name often hides its real length. Osaka is おおさか — four beats.