forum

[Rule] Romanization of Japanese

posted
Total Posts
29
Topic Starter
Melophobia
There are some methods of romanizing Japanese, the most common ones are Hepburn and Nihonshiki. In short, Hepburn is based on how we pronounce Japanese. Nihonshiki is based on how we type Japanese into an IME system.

The method used in osu! needs to be consistent, it just leads to inconsistency in metadeta of beatmaps and stupid unranks otherwise. I think Hepburn romanization is the best one since it's closest to what we do in osu! now.

However, long vowels such as "おう" "うう" should be romanized into "ou" "uu" instead of "o" "u", since we cannot use a macron in a non-unicode field. It has already taken root in osu! so you probably find it more natural. And using "o" "u" also makes us unable to tell the difference between 2 different words which have the same romaji, for example "Yuka" and "Yuuka". And "ou" "uu" show their actual pronunciation so they help you know how to pronounce them correctly.

For example:
O Touhou
X Toho

O Shippuuden
X Shippuden

Also there are many variants of Hepburn romanization. The two most common styles are Traditional Hepburn and Modified Hepburn. They're basically the same but different in some respects. See Wikipedia for their detailed features. As you can see in osu, we romanize particles "を" "へ" "は" into "o" "e" "wa" so modified Hepburn seems closer to what we have in osu!

tl;dr - Refer to Hepburn romanization when you romanize Japanese. Long vowels such as "おう" "うう" are romanized into "ou" "uu" because we cannot use a macron in a non-unicode field.

See also:
[Rule] Romanisation of Chinese, Japanese and Korean titles

Any suggestion?
Kytoxid
Also words derived from English should be written in their English forms, correct?

Since this is somewhat complicated, I recommend we open a new page on the Wiki for Romanization Standards and add a rule to the Ranking Criteria a rule that says romanization must follow the Standards.

No issues with the proposed system though.
Topic Starter
Melophobia

Kytoxid wrote:

Also words derived from English should be written in their English forms, correct?
Yes. But it's already stated in: p/2418829 I'm not sure if this one needs to include it too.

Kytoxid wrote:

Since this is somewhat complicated, I recommend we open a new page on the Wiki for Romanization Standards and add a rule to the Ranking Criteria a rule that says romanization must follow the Standards.
Sounds good to me. I'm willing to be charged with making it.
Garven
The problem with converting to the foreign equivalent is when things like puns enter the picture (https://osu.ppy.sh/p/beatmaplist?q=mushikui). It might be better to just have a blanket stay-true-to-the-original when it comes to the romanization since that doesn't give any room for interpretation.
Sure
so when we romanize Japanese title,

Loanword -> do not romanize Japanese
Original Japanese word -> use (modified or traditional) hepburn style

right?
Topic Starter
Melophobia
Slightly fixed the OP and title..

ykcarrot wrote:

so when we romanize Japanese title,

Loanword -> do not romanize Japanese
Original Japanese word -> use (modified or traditional) hepburn style

right?
Exactly. Regarding loanwords please refer to SG's thread. It'd been discussed there for a long time.
TKS
I agree with Melophobia.
Having a consistent is beneficial for all players and modders. At present, it has giving the confusion to everyone. In fact, Hepburn style is a bit unnatural for Japanese. As Melophobia said, Hepburn romanization is the best one. I bet it will work.
alacat
finally! this rule :)
But it seems we need to check this problem because most of the mappers seems no attention at the moment. It seems most of foreigners aren't interested or don't even care, I hope they get to know about Hepburn.
KSHR
Apparently most of mappers have been confused with romanization for a long time, because the rule regarding the method hasn't been stated in the Ranking Criteria yet. And even native Japanese speakers sometimes make a mistake with that. Therefore this problem deserves to be discussed to say the least. Of course, I will gladly help as a native Japanese / a Japanese BAT.
Kodora
I hope this rule will prevent lots of unranking over wrong titles, have my support.

Like Kytoxid said, we need a different page on wiki about correct romanization rules - hopefully people will pay attention to that.
Topic Starter
Melophobia
I think this matter has been discussed adequately within the team and in public, and it seems all people are fine with it.
Kytoxid
Something else I just thought up and wanted to clarify, romanization should be based on ruby if it differs from the normal pronunciation, right?

Like, Toaru Kagaku no Railgun instead of Toaru Kagaku no Choudenjihou.
Topic Starter
Melophobia
Exactly, it's based on its actual pronunciation, and that's one of reasons why I push Hepburn.
SapphireGhost
Attempt to condense this into a rule:
Songs with Japanese titles must use the Modified Hepburn method of romanisation in the "Romanised Title" field. As a non-unicode field, long vowels such as "おう" and "うう" should be romanised into "ou" and "uu" to avoid macrons. Loan words should be expressed using romanisation from the original language.
Another potential issue is this: if the creator of the song has provided a Nihonshiki or other alternate romanisation, should this be used instead? People may say yes because it follows the intentions of the creator, but people may say no because it creates inconsistencies and still does not count as the "original" title that is being romanised.

Like with the custom difficulty name thread, I would like to suggest to Melophobia to create a tentative deadline for finalisation and update the rule with suggestions from others in the thread.
Lanturn
A bit of a random spot to ask this, but it seems to be the thread that mostly fits it.

Does beatmap ranking requirements already force us to use 'o' instead of 'wo' or 'e' instead of 'he'?

I'm not fond at all of using 'o' or 'e' because to me, its pretty much taking out the k in knife because it's silent. I'm just wondering if this is a forced rule or not yet.
Kytoxid

Lanturn wrote:

A bit of a random spot to ask this, but it seems to be the thread that mostly fits it.

Does beatmap ranking requirements already force us to use 'o' instead of 'wo' or 'e' instead of 'he'?

I'm not fond at all of using 'o' or 'e' because to me, its pretty much taking out the k in knife because it's silent. I'm just wondering if this is a forced rule or not yet.
It does not currently, but this new rule will enforce it, so I recommend you use 'o', 'e', and 'wa' for consistency with future maps.
Ephemeral
Hepburn (not modified) seems to be the gold-standard here.
Ephemeral
Actually, we had an entire discussion on the staff side about this, and came to the decision that Nihonsiki is what we will be using for this.
Sure
つづく(= to be continued)
Hepburn : tsu zu ku
Nihonshiki : tu du ku

teach me how to pronounce when using nihonshiki ... ( ´д`)
TheVileOne
Wait so Touhou == Toho? I'm confused. Did the naming convention change?
Topic Starter
Melophobia
When did we come to the conclusion that we'll make use of Nihonshiki? Few people actually supported it but the majority of staff members agreed to Hepburn if I remember correctly. Nihonshiki is indeed closer to how we type Japanese so it's natural for native speakers but other than that there's no point in using it imo.

P.S: I'm sorry i've abandoned this thread badly. :(
Charles445
Isn't romanization supposed to help people who can't read the original text?
Hepburn would be better because then we can see how to pronounce things properly.
B1rd
Yeah I think Hepburn, as far as I know it's the standard for romanisation.

The modern one though with n instead of the m before consonants such as p.
Love

SapphireGhost wrote:

Attempt to condense this into a rule:
Songs with Japanese titles must use the Modified Hepburn method of romanisation in the "Romanised Title" field. As a non-unicode field, long vowels such as "おう" and "うう" should be romanised into "ou" and "uu" to avoid macrons. Loan words should be expressed using romanisation from the original language.
Could we start getting this finalized now? Suggestions/changes would be nice to hear.
SapphireGhost

Love wrote:

SapphireGhost wrote:

Attempt to condense this into a rule:
Songs with Japanese titles must use the Modified Hepburn method of romanisation in the "Romanised Title" field. As a non-unicode field, long vowels such as "おう" and "うう" should be romanised into "ou" and "uu" to avoid macrons. Loan words should be expressed using romanisation from the original language.
Could we start getting this finalized now? Suggestions/changes would be nice to hear.
I still stand by getting this added as is. I think it's thoughtfully worded and works in the best interest of the situation. With no rule, we're left with ambiguous cases like this that it would be best to avoid.
Love
Hmm, instead of avoiding that, I think it would be better to let the mapper pick if they want the correct translation or the romanised title.
peppy
How is this not added yet? We've all come to an agreement ages ago in trello.
Topic Starter
Melophobia
I'm fine with Sapphire's condensation, but there's a few things to be addressed before getting this finalized;

SapphireGhost wrote:

Another potential issue is this: if the creator of the song has provided a Nihonshiki or other alternate romanisation, should this be used instead? People may say yes because it follows the intentions of the creator, but people may say no because it creates inconsistencies and still does not count as the "original" title that is being romanised.
I would say yes, considering romanization is superior to translations in terms of respecting the creator's intention, which shall not be contravened.

Ephemeral wrote:

Hepburn (not modified) seems to be the gold-standard here.
I'm all for Modified Hepburn, because it uses "o" "e" "wa" so is closer to their actual pronunciation.
Loctav
I have no idea why no one amended this yet. Consider this as finalized.
Please sign in to reply.

New reply