Uploaded image for project: 'MusicBrainz Server'
  1. MusicBrainz Server
  2. MBS-5192

Missing alias attribute SCRIPT (as tracklists' one)

    • Icon: New Feature New Feature
    • Resolution: Unresolved
    • Icon: Normal Normal
    • None
    • None
    • Edit system, Schema Change
    • None

      This ticket stands for all aliases (works' and artists').

      As for pseudo release tracklists we need the SCRIPT attribute on aliases.

      An example : 陰陽座's one LATIN script alias would be Onmyo-za (japanese name in this case).
      Currently, without the LATIN script attribute on alias, we would have to either

      • 1. create one french Onmyo-za alias (wrong as other latin languages won't see any aliases and wrong to say that Onmyo-za is a french word) or
      • 2. create as many Onmyo-za aliases as there are latin languages (stupidly long alias list and wrong to say that Onmyo-za is both french, english, german, dutch, etc. word) or
      • 3. create one Onmyo-za alias without language attribute set (wrong for all non latin languages and some random editor will eventually come and auto-edit set whatever language they fancy and we're back to 1.)

      Solution : This Onmyo-za alias must only be set as LATIN script alias, period, end of problems.

      AFAIK the language attribute is less important, if at all.

      Related stuff

      PS. The ultimate use of aliases I guess is internationalisation if you want this or that release tracklist in latin for instance, it will get works aliases latin script aliases and artists latin script aliases to make up the tracklist instead of using the current but to be dropped pseudo-release system.
      And for browsing internationalisation, if you say you're in one of the latin based languages those aliases will be used (instead of problem 2. stupidly long alias list).

          [MBS-5192] Missing alias attribute SCRIPT (as tracklists' one)

          jesus2099 added a comment - - edited

          Now that artist credit is possible on pseudo releases (on both official or pseudo releases).

          So what to we need artist aliases now?

          • Search hints,
          • information, and maybe the
          • localised display in MB website if one chooses an option to not display original artist names?

          Now that artist credit exist for releases, I think this ticket should be reviewed.

          jesus2099 added a comment - - edited Now that artist credit is possible on pseudo releases (on both official or pseudo releases). So what to we need artist aliases now? Search hints, information, and maybe the localised display in MB website if one chooses an option to not display original artist names? Now that artist credit exist for releases, I think this ticket should be reviewed.

          jesus2099 added a comment - - edited

          jesus2099 added a comment - - edited Another related topic: https://community.metabrainz.org/t/alias-type-for-transliterated-legal-names/479054/11?u=jesus2099

          MonkeyPython added a comment -

          I very much agree with the addition of script in the locale list.
          But I also think we still need languages. They are not super useful for artists (except in the rare few situations like Yaz/Yazoo)

          But for things like Instruments, they are so necessary: https://beta.musicbrainz.org/instrument/b3eac5f9-7859-4416-ac39-7154e2e8d348/aliases
          (note how there are several aliases in "Cyrillic" script that are not identical, they correspond to the language)
          of course I am not against having the ability to put "Japanese in Latin script" ie transliteration, so exactly what this ticket asks for. This will be useful also for instruments.

          MonkeyPython added a comment - I very much agree with the addition of script in the locale list. But I also think we still need languages. They are not super useful for artists (except in the rare few situations like Yaz/Yazoo) But for things like Instruments, they are so necessary: https://beta.musicbrainz.org/instrument/b3eac5f9-7859-4416-ac39-7154e2e8d348/aliases (note how there are several aliases in "Cyrillic" script that are not identical, they correspond to the language) of course I am not against having the ability to put "Japanese in Latin script" ie transliteration, so exactly what this ticket asks for. This will be useful also for instruments.

          jesus2099 added a comment -

          The need for this is no longer really for pseudo releases as we now have the artist credit that is even more convenient.

          But the Latin script alias could be useful for displaying alias instead of artist name (internationalisation):

          jesus2099 added a comment - The need for this is no longer really for pseudo releases as we now have the artist credit that is even more convenient. But the Latin script alias could be useful for displaying alias instead of artist name (internationalisation): MBS-7830 https://community.metabrainz.org/t/use-of-disambiguation-for-translation-of-non-latin-script-languages/340132/7?u=jesus2099

          jesus2099 added a comment -

          BTW just to let you know that I now wonder what we need in aliases exactly.
          I'm now completely clueless. http://musicbrainz.org/edit/36811479

          The only things I remember :

          • align system with tracklist language/script... but why now that we use AC on tracks
          • avoid having to create the same latin alias for all languages using the latin script... maybe the last valid issue with current Locale system.

          jesus2099 added a comment - BTW just to let you know that I now wonder what we need in aliases exactly. I'm now completely clueless. http://musicbrainz.org/edit/36811479 The only things I remember : align system with tracklist language/script... but why now that we use AC on tracks avoid having to create the same latin alias for all languages using the latin script... maybe the last valid issue with current Locale system.

          jesus2099 added a comment -

          It's great idea (root) and seems very keep consistent !

          jesus2099 added a comment - It's great idea (root) and seems very keep consistent !

          Ulrich Klauer added a comment -

          To make this actionable, I'd like to propose the following approach:

          • Add new locales, with one locale per script. The code should be root_Latn, root_Cyrl, etc., where root is the "root locale" from which all other locales inherit. This makes it a valid Unicode Locale Identifier, just as zh_Hant or ar_EG that we currently use. (The corresponding IETF/BCP 47 tag, by the way, would be und-Latn etc., where und is the ISO 639-2 registered code for "undetermined language". See BCP 47 Language Tag Conversion for the correspondance.)
          • Define the meaning of multiple aliases, whose locales overlap, as follows: A script-only locale applies to all locales that use this script, but with lower precedence than an alias with a more specific locale. (This is not something to be implemented in code, it's just a definition "how we understand it" that should be documented somewhere.)

          E.g., Michael Jackson could get these aliases:

          • "Michael Jackson", primary for locale root_Latn
          • "Maykl Cekson", primary for locale az
          • "マイケル・ジャクソン", primary for locale ja
          • (some more)

          Then someone interested in the most appropriate name for en_GB could automatically derive that it's the root_Latn alias (because there is no en_GB nor en alias, but a script-only locale with the same script); someone else looking for the most appropriate name for az_AZ would find (no az_AZ, but) an az alias and stop there.

          Ulrich Klauer added a comment - To make this actionable, I'd like to propose the following approach: Add new locales, with one locale per script. The code should be root_Latn , root_Cyrl , etc., where root is the "root locale" from which all other locales inherit. This makes it a valid Unicode Locale Identifier , just as zh_Hant or ar_EG that we currently use. (The corresponding IETF/BCP 47 tag, by the way, would be und-Latn etc., where und is the ISO 639-2 registered code for "undetermined language". See BCP 47 Language Tag Conversion for the correspondance.) Define the meaning of multiple aliases, whose locales overlap, as follows: A script-only locale applies to all locales that use this script, but with lower precedence than an alias with a more specific locale. (This is not something to be implemented in code, it's just a definition "how we understand it" that should be documented somewhere.) E.g., Michael Jackson could get these aliases: "Michael Jackson", primary for locale root_Latn "Maykl Cekson", primary for locale az "マイケル・ジャクソン", primary for locale ja (some more) Then someone interested in the most appropriate name for en_GB could automatically derive that it's the root_Latn alias (because there is no en_GB nor en alias, but a script-only locale with the same script); someone else looking for the most appropriate name for az_AZ would find (no az_AZ , but) an az alias and stop there.

          Another example which does not fit to the script-only concept are Slavic female form surnames.

          For example in Czech Republic and Slovakia surnames of foreign women (often) get a female suffix -ová:
          Michelle Obama -> Michelle Obamová
          Sinéad O'Connor -> Sinéad O'Connorová
          Audrey Hepburn -> Audrey Hepburnová
          Ella Fitzgerald -> Ella Fitzgeraldová

          Though for artists adding of the suffix seems not to be common especially when the artist's name is a pseudonym. E.g I do not remember I have ever seen Édith Piafová or Alanis Morissettová.

          See http://en.wikipedia.org/wiki/Czech_name http://en.wikipedia.org/wiki/Family_name#Slavic_surname_formation

          I think it is a good idea to have the script attribute especially to be able to create a default transcription alias for a certain script but this attribute cannot replace the language attribute. There will be special transcriptions / aliases for some languages.

          • there are different sets of transcription rules (even for transcribing from the same script as Ulrich Klauer showed)
          • there are other language specialities like gender name suffixes

          Václav Brožík added a comment - Another example which does not fit to the script-only concept are Slavic female form surnames. For example in Czech Republic and Slovakia surnames of foreign women (often) get a female suffix -ová: Michelle Obama -> Michelle Obamová Sinéad O'Connor -> Sinéad O'Connorová Audrey Hepburn -> Audrey Hepburnová Ella Fitzgerald -> Ella Fitzgeraldová Though for artists adding of the suffix seems not to be common especially when the artist's name is a pseudonym. E.g I do not remember I have ever seen Édith Piafová or Alanis Morissettová. See http://en.wikipedia.org/wiki/Czech_name http://en.wikipedia.org/wiki/Family_name#Slavic_surname_formation — I think it is a good idea to have the script attribute especially to be able to create a default transcription alias for a certain script but this attribute cannot replace the language attribute. There will be special transcriptions / aliases for some languages. there are different sets of transcription rules (even for transcribing from the same script as Ulrich Klauer showed) there are other language specialities like gender name suffixes

          Ulrich Klauer added a comment -

          Another example that just sprang to mind: Hungarian names are family-first, e.g. Bartók Béla; other languages with Latin script invert the name (Béla Bartók).

          Ulrich Klauer added a comment - Another example that just sprang to mind: Hungarian names are family-first, e.g. Bartók Béla; other languages with Latin script invert the name (Béla Bartók).

          Ulrich Klauer added a comment -

          Another case to keep in mind: Most Latin-script languages retain the original spelling of names from other Latin-script languages even though it does not match their pronunciation rules; e.g., it is "Michael Jackson" in French, German, Swedish, Turkish, Vietnamese, etc., so we probably don't want to add hundreds of identical aliases for each of these languages. However, one notable exception is Azeri: It uses a Latin script, but transcribes names in the same way as is usually only done for foreign scripts. Michael Jackson is written "Maykl Cekson" there.

          So we'd want to have a "script default" that can still be overridden by a more specific, language-based alias.

          Ulrich Klauer added a comment - Another case to keep in mind: Most Latin-script languages retain the original spelling of names from other Latin-script languages even though it does not match their pronunciation rules; e.g., it is "Michael Jackson" in French, German, Swedish, Turkish, Vietnamese, etc., so we probably don't want to add hundreds of identical aliases for each of these languages. However, one notable exception is Azeri: It uses a Latin script, but transcribes names in the same way as is usually only done for foreign scripts. Michael Jackson is written "Maykl Cekson" there. So we'd want to have a "script default" that can still be overridden by a more specific, language-based alias.

            Unassigned Unassigned
            jesus2099 jesus2099
            Votes:
            19 Vote for this issue
            Watchers:
            12 Start watching this issue

              Created:
              Updated:

                Version Package