Translate Document from English to Japanese

Going into Japanese, the text shrinks and the characters grow

English arriving in Japanese usually contracts. A great deal of what English spells out is either carried by a single kanji or dropped altogether, since Japanese has no articles, marks plural only when it matters, and routinely leaves out a subject that the reader can infer. Count the characters and the Japanese version is commonly a good deal shorter than the English it came from.

The page does not shrink to match, because Japanese characters are full width. Each one occupies a square roughly as wide as it is tall, where Latin letters vary and most are narrower. Fewer characters at greater width means the line length lands somewhere unpredictable, and it lands differently on every line.

The places that suffer are the ones with no room to give. A button label or a menu item sized to the English word. A table column set to the width of its longest English entry. A slide bullet that fit on one line. A header or footer inside a fixed box. Running text absorbs the change without complaint, so it is worth looking at everything except the paragraphs when you check the result.

Going the other way the problem inverts: Japanese into English expands, and the layout has to be re-fitted so nothing overflows. If your file is a PDF and the typesetting side is what concerns you most, the PDF to Japanese page covers font embedding, vertical text and line breaking in detail.

Names, dates and figures follow local conventions

Personal names

Japanese puts the family name first, the reverse of the order an English document uses, so a name copied across unchanged can be read as the wrong person. Non-Japanese names are written in katakana, and most admit more than one defensible spelling, which means the same person can end up rendered two ways in one file. Settle each name once and then reuse that spelling everywhere. A signatory, an author or an executive whose name appears differently on two pages is a problem in a contract, not a detail.

Dates

The order is year, month, day, written with the characters 年月日 rather than slashes, so 28 August 2026 becomes 2026年8月28日. Official and administrative paperwork also uses era years, currently 令和, counted from the start of the reign rather than from the Western epoch. Converting between the two systems is arithmetic that has to be done deliberately and checked, never guessed at, because an era year silently treated as a Western one is off by two millennia.

Full width and half width digits

Digits and Latin letters exist in two widths in Japanese text, and they are distinct characters rather than a font setting. Technical and horizontally set material tends to use the narrow half width forms, while formal and vertically set text prefers the full width ones. Either is correct. Mixing them within a single table column or a single figure is what looks wrong, and it is easy to do without noticing.

Large numbers

Japanese groups large figures in units of ten thousand, 万, and a hundred million, 億, where English groups in thousands and millions. A number that reads as 1.2 million in English is naturally expressed as 120万, and the mental conversion is not a matter of moving a comma. In financial and commercial documents this is where errors hide, so figures are worth checking against the source rather than skimmed.

Punctuation

Japanese ends sentences with 。 and separates clauses with 、, both full width, with no following space. Quotation uses 「 」 and parentheses have their own full width forms. A document that keeps Western periods and commas in Japanese sentences reads as unfinished even when every word in it is correct.

Decide once what your product is called in Japanese

Borrowed vocabulary is written in katakana, which is why so much modern Japanese business language is recognisably English underneath. That flexibility is also a trap: a single English term can usually be handled three ways, and all three appear in real Japanese documents. It can be transcribed into katakana, kept in Latin letters, or replaced with an established Japanese word. Every one of those is defensible. What is not defensible is using a different one on each page.

Product names, feature names, job titles, legal terms and the names of internal systems are where this shows. Fix them in a glossary before the first file goes through, and the choice is made once for every document that follows instead of being re-made silently each time. Custom glossaries and shared team access come with every DocTranslator plan, so the same list applies across everything your team uploads.

Where a name is trademarked or already established in the Japanese market, keeping the Latin form is usually the safer decision. Where an English term has a settled Japanese equivalent that the audience already uses, the Japanese word reads better than a katakana transcription of the English one.

What to check, and when to bring in a person

Automatic translation handles this pair well for the bulk of the work, and the review that follows is quick if you know where to look. Read the sentence endings for a consistent politeness level. Check names, dates and figures against the source. Look at fixed width elements, tables and headings for text that no longer fits. Confirm your product and system names match the glossary. That is a short list, and it catches most of what goes wrong.

Bring in a native reviewer when the document carries obligation or risk: contract wording, safety instructions where a misread step causes harm, regulatory submissions, and anything a customer will read as the voice of your company. The cost of a review on those is small next to the cost of getting the register wrong in front of a client.

Anything filed with an authority is a separate category again. Official submissions generally require a certified translation signed by a human translator, which is a procedural requirement rather than a judgement about quality. See certified translation for what that involves.