forum

[Proposal/Discussion] Change the requirement of Romanised Title and Romanised Artist

posted
Total Posts
18

Do you agree with the proposal?

Yes
44
84.62%
No
8
15.38%
Total votes: 52
Topic Starter
[Crz]Alleyne
Some current metadata things about title and artist in Ranking Criteria:

Guidelines:

A:
If multiple metadata options are available, priority should be given to the option which is most easily recognisable and traceable back to the original song or source. Official romanisations and translations are preferred for romanised fields so long as they are easily found and commonly recognised.
B.
If the same song exists in the Ranked or Loved sections already, the metadata should be followed unless it breaks other rules in the ranking criteria or the official sources state something completely different.
Allowance:

C.
If a Unicode song title or artist has an official translation or romanisation provided by the artist, it may be used in the respective romanised field. If both a translation and romanisation are available, either may be used.
Here we can see, when a beatmapset gets ranked or loved, the title and artist fields of the next beatmapsets should follow this ranked or loved one, according to B. But if the artist or title in the ranked beatmapset is already a fault? Then A can do nothing to the B even if you are allowed to use other optional titles or artists. ...unless it breaks other rules in the ranking criteria or the official sources state something completely different. The point is whether wrongly using translated or romanisation title or artist can be considered as "the official sources state something completely different".

In my view, we should respect the artists for the music they are making. So at least we should follow their given translation or romanisation. You can only do translation or romanisation when they are not given. For C, here is my proposal:
If a Unicode song title or artist has an official translation or romanisation provided by the artist, it must be used in the respective romanised field. Self-romanisation can be only available if there's no official translation or romanisation. If both a translation and romanisation are available, either may be used.
Since this contains "must", it should be moved under Rules section, Standardisation sub

Then A should be like this (Removed the last sentence):
If multiple metadata options are available, priority should be given to the option which is most easily recognisable and traceable back to the original song or source.
I believe the limitation (or maybe a conflict) that B gives to A should be gone. Personally, I hope the official can be always final. We cannot modify that without the artist's permission, even if there's ranked one that disobeys the artist's intention. We cannot change it for our selfishness.

---------------------

For the finalization (please make sure you have read this.):

This is following Akasha's ideas.

"A" reverted to the original (No change)

If multiple metadata options are available, priority should be given to the option which is most easily recognisable and traceable back to the original song or source. Official romanisations and translations are preferred for romanised fields so long as they are easily found and commonly recognised.
"B" changed to:

If the same song exists in the Ranked or Loved sections already, the metadata should be followed unless it breaks other rules in the ranking criteria or the official state it's allowed to be used. For the artists who have multiple romanisations or translations, mappers can choose one to use.
- This is to avoid being forced to follow the already ranked or loved mapsets with wrong metadata for consistency.

"C" changed to

If a Unicode song title or artist has an official translation or romanisation provided by the artist, it may be used in the respective romanised field. Only if none are available may the fields be self-romanised. If both a translation and romanisation are available, either may be used.
- This is to lead into a deeper defination of when self-romanisation should be used and avoid it when official romanisation / translation exists.

Unresolved questions: Quenlla's post, Serizawa Haruki's post
Noch Einen
Based from Akasha's map discussion (HAWATARI NIOKU CENTI)
I'd say rules are meant to test, but not to break. However, if the official translation (explicitly stated by the artist/group band/composer) exist, using that metadata should be the main priority, at least what we could do to respect the song creator.

I'd agree wholeheartedly, regardless how ridiculous the artist named their song (since it's problem of nominator, not a mapper).


Selfish / preference shouldn't be a word / used as reason there, since "we are their listener or just a user from this platform (osu), not the creator of the song"
Akasha-
By any means, yes.

It rubbed me the wrong way why do we have to choose what's best for us when the artist's intention is as clear as day so why don't we just go with that instead?

Also to get it right, it should be prioritized rather than a 100% must to be able to handle edged/complicated cases

Should an official be used or shouldn't be with a proper reason.
Drum-Hitnormal
follow official is
1) respect artist intention, its their song not ours.
2) easy to follow since you dont need to come up with ur own translation/romanisation
3) enforce consistent since everyone should be looking at same source unless official is released after you ranked personal romanisation, in that case future maps should still follow official, and old ranked map should have new official added to tags

However, i disagree with special case of aritist name like Inori Minase

Time 1: artist no official name so use romanised Minase Inori
Time 2: artist has official english name of Inori Minase which is simply to confirm to western naming convention

essentially it is not completly different name, but would cause inconsistency in ranked maps. and make it weird when a song has 3 artist, 1 with western order and 2 with jp order.

so would suggest add 1 clause to handle this case

if official artist name is just western order, apply original romanisation to keep consistency with past maps and other artist without western name
Nokashi
I just want to say that the same exact rule was enforced for about... 2 weeks (sabaku de proposal sep 2021 ish) in the form of forcing provided romanization (it was just for artist name at the time) and it ended up creating a deadlock due to label and artist inconsistencies. It was then bandaid-fixed to what we have now offering flexibility to avoid such deadlocks. regressing back to that feels like history repeating itself and its a matter of time until a deadlock appears again.

Im not gonna pop up in this thread anymore i just want to post these references for all of you to give a read and understand the possible deadlocks forcing official romanizations/translations could result in. Trying to reach a compromise for song titles could be made imo. but names and especially japanese names, is a huge headache just cause of how Japanese names are formatted (Last Name First Name)

+ beatmapsets/1557051/discussion/-/generalAll#/2594751
+ community/forums/topics/1415070?n=1
+ community/forums/topics/1317453?n=1
SuzumeAyase
Yes I'm agree

If official already give their translation then why we need to use another translation?
Its like you've been given a good sword but you chose a broken sword
Except if official doesn't give their translation we can use another translation/unofficial translation

Using official translation is means we respect and we support their intention
Follow Ranked Maps its not always good or better
Because we really don't know whether the previously ranked map has perfect metadata or not

Forcing your opinion and your will won't make you look cool ~ Me 2k22
Quenlla
There's definitely a need to break the ambiguity and contradictions of RC regarding official romanisation vs official translation vs correct romanisation, but I don't fully agree with the proposed solution.

Here's my thoughts
  1. In terms of supporting of official translation vs official/correct romanisation, I think either of them is fine, but there should be a rule clearly indicating which is prioritary (here I agree with the original proposal)
  2. The support/enforcing of "official romanisations" shouldn't be applied to cases where they break our own romanisation rules AND have no stylistic/artist gimmickery going on (e.g. official but incorrect "Zankyo Sanka" vs the correct "Zankyou Sanka"). Otherwise, the game's attempts of providing a reliable romanisation system to standardise titles and names is rendered completely useless.
  3. The spirit of preferring "official" romanisations should be exclusively for either:
    1. Reflecting a clear stylistic gimmick or choice in the artist/title name, like "yanaginagi".
    2. Avoiding ambiguities in things like kanji reading (e.g. should we read 佐坂 as "Sasaka" or "Sazaka")
  4. In the particular case of japanese names, standardisation is urgently needed, we can't have people constantly flipping names only because jp artists sometimes need to resort to using FN LN so foreigners don't get confused (and to appeal to the global market). The most respectful standardisation imo is to generally enforce LN FN; the japanese government has even released multiple press notes urging other governments and foreign companies to use LN FN to respect their traditions.
Serizawa Haruki
I tried explaining here and here that the guideline is worded incorrectly but nothing has been done about it.
It's important to distinguish romanization and translation because they're not the same. Preferring official romanizations to "self-made" romanizations makes sense (to respect the artist's intention) but prioritizing translations does not because it's a whole different name/title (it's called "romanized field" after all).
YuEast 2018
agree.
official = title/artist/source
commonly used/following already ranked/the case official made a mistake = add to tags
(example like "Rish" with "Risshuu/Rissyuu")

and there might be the third case that there's contradict official sources, in that case an order of priority should be listed. Firstly ordering by how well it can represent the artist their selves, then how well it can be easily understood or commonly used.
Topic Starter
[Crz]Alleyne

Quenlla wrote:

There's definitely a need to break the ambiguity and contradictions of RC regarding official romanisation vs official translation vs correct romanisation, but I don't fully agree with the proposed solution.

Here's my thoughts
  1. In terms of supporting of official translation vs official/correct romanisation, I think either of them is fine, but there should be a rule clearly indicating which is prioritary (here I agree with the original proposal)
  2. The support/enforcing of "official romanisations" shouldn't be applied to cases where they break our own romanisation rules AND have no stylistic/artist gimmickery going on (e.g. official but incorrect "Zankyo Sanka" vs the correct "Zankyou Sanka"). Otherwise, the game's attempts of providing a reliable romanisation system to standardise titles and names is rendered completely useless.
  3. The spirit of preferring "official" romanisations should be exclusively for either:
    1. Reflecting a clear stylistic gimmick or choice in the artist/title name, like "yanaginagi".
    2. Avoiding ambiguities in things like kanji reading (e.g. should we read 佐坂 as "Sasaka" or "Sazaka")
  4. In the particular case of japanese names, standardisation is urgently needed, we can't have people constantly flipping names only because jp artists sometimes need to resort to using FN LN so foreigners don't get confused (and to appeal to the global market). The most respectful standardisation imo is to generally enforce LN FN; the japanese government has even released multiple press notes urging other governments and foreign companies to use LN FN to respect their traditions.
I'm thinking about whether a special style in romanisation can be kept for the artist's intention. No one can ensure whether Zankyo is the artist's intention compared to Zankyou. If this is still in the argument, we can ask the artist for the final result.

There're also some cases the official changed their romanisation or translation, like sakuzyo -> Sakuzyo. In this case, we just need to check under which romanisation or translation is this song released. It could be better if the artist changes all the meta in his old works.

When there is no official for these controversial romanisations, I think there should be some rules or guidelines to finalise these problems. Probably need GMT from Japan since I'm non-local and just a beginner for Japanese.

About the Japanese names (and not only for Japanese names), there's already a rule:
When romanising an artist's name yourself and no official romanisation is available, it must be romanised in the order it is printed in the unicode field.
If still needed, we can discuss this more since I can see there're still conflicts now. (I haven't involved this too much.) I understand what the order of LN and FN means.
SaltyLucario
i agree with proposed rc change that translations/romanisations coming from artists should always take priority, that's literally how artist wants the title to be seen for english speakers so we definitely should respect that
it also makes the most sense as if there's a translation available, the song is usually more well known by that name rather than by the original one
Protastic101
Merged

Psych, as Akasha- mentioned below, there seems to be a misunderstanding. I currently have a PR (https://github.com/ppy/osu-wiki/pull/8380) in the making to undo the change, so fear not.
Serizawa Haruki
How exactly was the consensus reached that official translations are forced now? It doesn't make sense to prohibit romanisation in the romanised fields just because a translation exists. There should be a choice as to what to use in the romanised field depending on what is more recognizable (which is also stated in the same rule), but this change removed that option which is most likely an oversight. Also, this rule only mentions artist name and song title so it's unclear how the source field is handled because the RC only says:

Songs with Japanese metadata must use the Modified Hepburn romanisation method in romanised fields. Further reference can be found on in-depth romanization tables. The same applies to the Source field if a romanised Source is preferred by the mapper.
But I think this is more of a general issue because the rules regarding the source field shouldn't be written as part of the romanisation methods of various languages but rather as its own paragraph.
Akasha-
The proposal wasn't implemented probably and it wasn't adjusted based on the community's feedback. This was a mistake from lack of communication and the proposer thoughts it was finalized, hence the change.

There's no one at fault, we're all having a miscommunication regarding this thread. I'd like everyone to have a proper discussion now that the thread is revived, should the wording change from the main post or be different, let your voice be heard. A small reminder for participants to highlight the important words and suggest a proper change to the proposed wording alongside with an explanation.

For the time being, the thread will be moved back to pending, revert the change until further notice.

METADATA > RULES > ROMANISATION (REMOVE) (1)
If a Unicode song title or artist has an official translation or romanisation provided by the artist, it must be used in the respective romanised field. Only if none are available may the fields be self-romanised. If both a translation and romanisation are available, either may be used.

METADATA > GUIDELINES (REVERT/MODIFY) (2)
If multiple metadata options are available, priority should be given to the option which is most easily recognisable and traceable back to the original song or source.
(Revert back into the Original)
If multiple metadata options are available, priority should be given to the option which is most easily recognisable and traceable back to the original song or source. Official romanisations and translations are preferred for romanised fields so long as they are easily found and commonly recognised.

(1), (2) will be changed shortly

The guideline in which should have been modified to give complexity to the title choices, excluding artist
METADATA > GUIDELINES > STANDARDISATION (A)
If the same song exists in the Ranked or Loved sections already, the metadata should be followed unless it breaks other rules in the ranking criteria or the official sources state something completely different.

(should be changed to something else that the artist remains consistent -given that it's correct- and the title is flexible and not being force-to-change when the options given are valid)
Topic Starter
[Crz]Alleyne

Serizawa Haruki wrote:

How exactly was the consensus reached that official translations are forced now? It doesn't make sense to prohibit romanisation in the romanised fields just because a translation exists. There should be a choice as to what to use in the romanised field depending on what is more recognizable (which is also stated in the same rule), but this change removed that option which is most likely an oversight. Also, this rule only mentions artist name and song title so it's unclear how the source field is handled because the RC only says:

Songs with Japanese metadata must use the Modified Hepburn romanisation method in romanised fields. Further reference can be found on in-depth romanization tables. The same applies to the Source field if a romanised Source is preferred by the mapper.
But I think this is more of a general issue because the rules regarding the source field shouldn't be written as part of the romanisation methods of various languages but rather as its own paragraph.
Sorry this is my fault that I didn't take good communication. Here I'll explain some points in your post

Even the romanisation exists, but we still need to check if it's official. It's just applied to the official. What I proposed is to stop to use any third-party romanisation / translation when official exists. If we are worrying about the consistency between ranked mapsets, after applying this rule, there's probably no more inconsistent title or artist. Follow the official is to respect the artists and the original artists' meanings. We cannot involve in any changes unless someone has already contacted the artist.

Those guidelines, standardisations are only available when there's no official romanisation / translation, which means when we meet the Romanised Artist field but no outside support, we need to handmade ourselves following the guidelines and standardisations. So because of this, I'd like to make it in the rules to make sure it's the most major. For more about relative RC things, I'll soon check one by one to see if there're more conflicts that exist.

---

To all participants, considering that this is also changing rules, keep eyes sharp. This also caused another problem under miscommunication so I wish this time we could have more discussion about it and make it clear.

---

For Akasha's changes at the end

I'm actually changing these pieces to make sure the artist's mind is major. The conflict point is between "prefer" and "must". But maybe I missed some points, like beatmapsets/1469980/discussion/-/generalAll#/2614698 there I actually want to follow the very official one but still changed for the standardisation, which is much more like "prefer". I'm gonna take your proposal to the end of the topic with a few slight changes.

And sorry for the trouble that miscommunication takes. And hope you can have further opinions about the official source about the consistency in the future. Like when there're two options for the artist and one of them has been used and ranked, how about the other one? Are we allowed to use that when there's a ranked set with different meta?
Quenlla
It's crucial to clearly rule this stuff but as I warned some weeks ago we really have to be careful about coherency and cohesiveness with the rest of the RC, and with dangerous assumptions about what artists and labels "prefer" or mean. It's hard to propose a rule directly just yet, so I will just share some thoughts:

1). Existance of an official translation but not an official romanisation can't be assumed as an artist preference: It's dangerous to assume "this is what the artist preferred" because a lot of times the translation is just a concession made by the artist so foreigners have something to understand the song title. A good example was 驟雨の狭間 (Rainshower) / Silentroom × Shun. Only the official translation existed, but upon contacting the artist we learnt that he rather preferred to use a "unofficial" romanisation (which we romanised as "Shuu no Hazama").

2). An official but "incorrect" romanisation is typically harmful for the integrity of the ranked section: Multiple romanisation systems exist, and osu is currently settled on ALA-LC: we chose that, but other labels/artists/localizator choose others. If an artist/label/localizator provides a romanisation that adds no artistic/stylistic value in a different romanisation system, it doesn't make sense to use it without adapting it to ALA-LC. Otherwise, why have romanisation rules in the first place? e.g. 太田あすか ("Oota Asuka") was romanised by the MKWii localization system as "Asuka Ohta", because they used a different system; it's official, but it's incorrect per our romanisation rules, so using it destroys our standardisation efforts.

3). Japanese name romanisation needs urgent standardisation: This is a bit more tangential but we can't just keep flipping and rewriting names constantly because we keep running into the problems described as 1) and 2): a lot of japanese artist use deliberately incorrect romanisations (even acknlowedging it) only so some foreigners dont confuse their family names with given names or we e.g. pronounce "おお" (oo) as "u". These break ALA-LC, and we should avoid using them unless they hold clear, meaningful artistic intention (e.g. "yanaginagi", "Meychan").

4). The "official romanisation vs official translation" dilemma: This is really the big important part that sparked this thread, yet it was not changed. If an artist provides both things, we need some criteria to choose which one to choose, leaving "the first mapper chooses which one becomes the ranked consistency" as the last resort.

I hope these insights are useful, love y'all
Serizawa Haruki

[Crz]Alleyne wrote:

Even the romanisation exists, but we still need to check if it's official. It's just applied to the official. What I proposed is to stop to use any third-party romanisation / translation when official exists. If we are worrying about the consistency between ranked mapsets, after applying this rule, there's probably no more inconsistent title or artist. Follow the official is to respect the artists and the original artists' meanings. We cannot involve in any changes unless someone has already contacted the artist.

Those guidelines, standardisations are only available when there's no official romanisation / translation, which means when we meet the Romanised Artist field but no outside support, we need to handmade ourselves following the guidelines and standardisations. So because of this, I'd like to make it in the rules to make sure it's the most major. For more about relative RC things, I'll soon check one by one to see if there're more conflicts that exist.
It's important to keep in mind that most songs with a title written in characters different from the latin alphabet don't provide an official romanisation or translation, which means that most song titles are actually self-romanised. This might be a bit different when it comes to artist names, but either way it's weird to treat self-romanisations as secondary whenever an official alternative is available. After all, the purpose of the romanised fields is to give people who can't read the original name/title the possibility to read/write it. It's not quite the same as the unicode fields which always consist of the original name/title provided by the artist. So I think that for the romanised fields officiality is less important because that is not their main purpose.

Quenlla wrote:

1). Existance of an official translation but not an official romanisation can't be assumed as an artist preference: It's dangerous to assume "this is what the artist preferred" because a lot of times the translation is just a concession made by the artist so foreigners have something to understand the song title. A good example was 驟雨の狭間 (Rainshower) / Silentroom × Shun. Only the official translation existed, but upon contacting the artist we learnt that he rather preferred to use a "unofficial" romanisation (which we romanised as "Shuu no Hazama").

2). An official but "incorrect" romanisation is typically harmful for the integrity of the ranked section: Multiple romanisation systems exist, and osu is currently settled on ALA-LC: we chose that, but other labels/artists/localizator choose others. If an artist/label/localizator provides a romanisation that adds no artistic/stylistic value in a different romanisation system, it doesn't make sense to use it without adapting it to ALA-LC. Otherwise, why have romanisation rules in the first place? e.g. 太田あすか ("Oota Asuka") was romanised by the MKWii localization system as "Asuka Ohta", because they used a different system; it's official, but it's incorrect per our romanisation rules, so using it destroys our standardisation efforts.

3). Japanese name romanisation needs urgent standardisation: This is a bit more tangential but we can't just keep flipping and rewriting names constantly because we keep running into the problems described as 1) and 2): a lot of japanese artist use deliberately incorrect romanisations (even acknlowedging it) only so some foreigners dont confuse their family names with given names or we e.g. pronounce "おお" (oo) as "u". These break ALA-LC, and we should avoid using them unless they hold clear, meaningful artistic intention (e.g. "yanaginagi", "Meychan").

4). The "official romanisation vs official translation" dilemma: This is really the big important part that sparked this thread, yet it was not changed. If an artist provides both things, we need some criteria to choose which one to choose, leaving "the first mapper chooses which one becomes the ranked consistency" as the last resort.
I fully agree with point 1, this is what I've been saying the whole time. Translations should be optional, not preferred.

Regarding point 2 I also concur because it's a similar logic again, but I'm unsure as to how incorrect/inconsistent romanisations can be defined. For instance, the artist KISIDA KYODAN & THE AKEBOSI ROCKETS uses a different romanisation method than the RC ("si" instead of "shi") and every recently ranked map uses this official romanisation. Is this considered a stylistic choice, and if yes, how can it be distinguished from a simple variation in romanisation system?

The solution to point 4 is already part of the guideline "If multiple metadata options are available, priority should be given to the option which is most easily recognisable and traceable back to the original song or source."



As mentioned previously, the usage of the source field and its respective rules/guidelines/allowances need to be discussed too. It's different from the other fields due to the lack of a separate romanised field. However, official romanisations are allowed and official translations are de facto allowed, despite there being no explicit statement about it in the RC. The question is whether metadata priorities should be identical in this case too or not. Personally I'm in favor of always keeping the source in its original language and adding the other options to the tags because this would be in line with the concept of officiality, avoid ambiguity as to which option should be used, and increase consistency.
Topic Starter
[Crz]Alleyne

Serizawa Haruki wrote:

It's important to keep in mind that most songs with a title written in characters different from the latin alphabet don't provide an official romanisation or translation, which means that most song titles are actually self-romanised. This might be a bit different when it comes to artist names, but either way it's weird to treat self-romanisations as secondary whenever an official alternative is available. After all, the purpose of the romanised fields is to give people who can't read the original name/title the possibility to read/write it. It's not quite the same as the unicode fields which always consist of the original name/title provided by the artist. So I think that for the romanised fields officiality is less important because that is not their main purpose.
Yeah, I understand it now. I found my intentions lacking in some points so it's probably not good. Now I agree with no force to follow the official, especially after Akasha's and Quenlla's posts

But there's still one point that I want to change in the origin, which is to decrease the limitation of following the ranked / loved mapsets with metadata. This is bad when the ranked / loved one is just one situation while there's another one that exists. That means when there're multiple available options for the metadata, only keeping using the same as the ranked / loved version is kinda stubborn. It should give mappers choices for the metadata, like a mapper is trying to respect the artist while the ranked one uses different.

--- 11/26 edited

oh I misunderstand point 4. This is what we are gonna change. Only keeping using the ranked versions gives no more choices for those mappers who have different available options. Like Mada Aoi and Stay blue, they're both official. but in different regions, people have different preferences. Some thinks Mada Aoi is more official, the other thinks Stay Blue is more official. So this needs it more open while choosing multiple available official options of artist's names.
Please sign in to reply.

New reply