The Official Basis for Katakana Spelling Names Company Names as Out of Scope
Somewhere in the work of entering Japan, a question about katakana stops being a question about spelling and becomes a question about who decides. At that point the brand wants a ruling. Two katakana spellings are on the table, both defensible, and somebody asks whether there is an official answer.
There is a document. It is a Cabinet notification, issued in 1991 over the Prime Minister’s name, and it is still the reference in force. It is worth reading, and what it says is not what a brand in that position is hoping for.
What it says it is establishing
The proclamation is one sentence. Here it is, with the word that matters left in the original (our translations throughout):
The よりどころ for “the writing of loanwords” — for setting down the modern national language in ordinary social life — is established as follows.
A よりどころ is something you rely on, a thing to go by. It is not the word Japanese uses for a rule. The distinction is doing real work in that sentence: what the Cabinet establishes is not a set of requirements but a reference that exists to be consulted.
The opening notes use the word again in the same position — the document is “a thing that indicates the よりどころ” for writing loanwords across law, official documents, newspapers, magazines and broadcasting.
And it appears a third time, in the part of the package that is an instruction rather than a notice. A Cabinet directive issued the same day tells the organs of government what to do with what had just been proclaimed:
From now on, each administrative organ shall take this as the よりどころ for “the writing of loanwords” in setting down the modern national language.
A directive of that kind is an order to the organs of state, and the verb here is accordingly “shall”. The noun it reaches for is still the one that means a thing to lean on. Across the proclamation, the directive and the opening notes, the word 規則 does not appear. Three times, in the three places where this package says what it is for, it declines the word that would have made it a rule.
That reading matters because everything below it is consistent with it, and would look like a series of failures if you read the document as a rulebook instead.
The first exclusion, by name
The opening notes are five numbered items. The second and third narrow the scope, and the third does it by listing examples:
This “Writing of Loanwords” does not extend to proper nouns and the like — for example, personal names, company names, product names and so on — where following it is difficult.
The item before it has already set aside specialist fields and individual practice:
This “Writing of Loanwords” is not something that seeks to extend to the writing of the various specialist fields — science, technology, the arts and so on — or of individuals.
So a brand looking for the official answer finds its own category named in the exclusions, in the document’s third sentence about itself. Not buried in a schedule. In the opening notes, with two of the three it names being exactly the thing the brand is trying to spell.
It is worth counting the five opening items by what each one does. The first says where the basis applies: laws, official documents, newspapers, magazines, broadcasting and the like, in ordinary social life. The fifth describes the document’s own structure. The other three all mark an edge — specialist fields and individual practice, proper nouns of the kind a brand owns, and writings used in the past.
Three of the five opening items are about the reach of the document rather than its content, and that is before the main text begins.
A search that stops here produces the wrong conclusion — that the document would have settled the question if only it applied. It would not have.
Inside the scope, and still not settled
Read on into the part that does apply, and the same posture appears in a different form. The first set of notes, the ones the document calls general, includes this:
Where a word has variant forms — ハンカチ and ハンケチ, グローブ and グラブ — no attempt is made here to settle which form it should be.
That is a declaration about ordinary loanwords, well inside the scope. The document is not saying it lacks the authority. It is saying it is not trying.
The next item explains what it does instead:
As for word forms and how they are written, where usage has settled, that is what to follow. Where different fields have settled on different usages, there is no objection to each going by its own.
And the detailed notes open by limiting themselves before the reader reaches a single example:
The word examples shown in the items below are given as one example of the use of each kana; they do not mean that the word must always be written that way.
The appendix of worked examples carries the same statement, and adds a count of its own:
Loanwords and foreign place and personal names are in many cases such that usage for the word form and its writing has not settled on one, and varies. In this list too, the word forms and writings shown here are one example, and are not a denial of other ways of writing.
Then, having said that, the appendix explains that it will print some entries twice — and names the shape it uses:
Further, for the main ones among the words for which the detailed notes give two ways of writing, both are listed together, in the form バイオリン／ヴァイオリン.
So the list does it: インタビュー／インタヴュー, エルサレム／イェルサレム, ギリシャ／ギリシア. Two spellings on one line, neither marked as preferred.
Two more declarations sit in the same document, and both run the same way. The fourth opening note:
This “Writing of Loanwords” is not something that seeks to deny the various writings used in the past.
And the third item of the main text, on sounds the two kana tables cannot represent:
As for how to write special sounds that cannot be written with the kana shown in Table 1 and Table 2, no arrangement is made here; it is left free.
A document that wanted to standardize would treat both of those as unfinished business. This one records them as decisions.
Outside it, and inside it too
Put them together and the shape is unusual.
A document that had drawn one boundary would leave a brand somewhere to go: outside the scope, yes, but with a settled practice visible just inside it, available to imitate. That is roughly what a style standard offers.
This one does not leave that. A brand name sits outside the scope by explicit naming. And the words that sit inside the scope are, on the document’s own account, not settled either — not because the work is unfinished, but because settling them was not the goal. There is no inner circle to copy.
That is also why the reading of よりどころ matters. If the Cabinet had established a 規則, the failure to settle ハンカチ against ハンケチ would be a gap. Because it established a basis to go by, the same fact reads as a design decision. A reference does not have to resolve everything, and this one says so in six separate places: once in the opening notes, once in the main text, three times in the notes on use, and once in the appendix.
The limits of a document about itself
The document is evidence about itself and nothing else.
Everything above is the notification describing its own reach. That is a narrow thing to have established, and the edges of it deserve stating plainly. Whether a given company’s katakana is a good one, whether readers find it easy, whether one form outperforms another in a market — the notification is silent, and so is this article. How the names actually come out is a separate exercise, and we have set out what that looks like in practice. Whether comparable notifications exist in other countries is a question we have not opened. And what happens legally if the document is simply ignored is a different subject from what the document says it covers.
One more limit worth naming. The Agency for Cultural Affairs publishes the body of the notification split across thirteen pages, with that body available separately as a single download. The proclamation and the directive quoted above are not in that download; they sit on their own page, ahead of the material the download begins with. The six statements counted above sit on five different pages of that set: the front matter, the main text, the notes on use twice over, and the explanatory preface to the appendix. No one page carries the pattern. It is visible only if you read the whole document.
What is actually available to a brand
The absence of an answer is not the same as the absence of information, and this document supplies a fair amount of the second kind.
The question was considered by the body that sets the national basis for this spelling, and then set aside deliberately. The appendix shows which forms are common enough to have been printed in a national list, and where that list prints two rather than one. And the note on usage points somewhere specific. Its first sentence is not permission but instruction: where usage has settled, that is what to follow. The precedent that governs is sectoral, not national. That is a narrower question, and unlike the first one it has an address — the people already writing in that field.
None of that decides the name. What it does is move the decision back to where the notification leaves it: with the people who will have to keep writing it.
Sources: 外来語の表記 / The Writing of Loanwords, Cabinet Notification No. 2 of 1991 (平成3年内閣告示第2号). Agency for Cultural Affairs, full text: https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/gairai/ — downloadable in one file at https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/gairai/pdf/94380801_03.pdf. Retrieved 14 September 2026. One word, よりどころ, is left in the original in the body, because picking an English equivalent for it would decide the question this article is asking. Everything else quoted here was put into English by us.