Release Title sort name style?

It is possible to enter a sort name for a Release title, by adding an Alias to that Release, and filling out the Sort Name field of the Alias. This seems like a nice thing to have, especially if alphabetising a discography list of Releases by a particular Release Artist. However, I do not see many Releases which contain such aliases. One of the few I have found is Instruments of the Orchestra, The: The Oboe.

The (Style / Titles](Style / Titles - MusicBrainz) guidelines do not appear to mention sort names for Release titles.

So, is it a good or a bad thing to enter an Alias for a Release which applies to the home area of that Release, but includes a sort name? If I go ahead and enter a few hundred such aliases, am I causing a Style problem or muddying the database?

I have an application where I could extract and use such sort names. I am tempted to make such entries for all Releases for which it helps my application.

There aren’t any, and as far as I know, no attempt has ever been made at writing any.

Since the purpose of the field is to make things machine-sortable, it would have to involve things like,

  • dropping quotation marks and/or adding missing punctuation
  • translating roman numerals to Arabic numerals
  • left padding Arabic numerals with zero(s)
  • otherwise not literally copying the name of the release as printed

  • Normalise Unicode forms (this is variation like é as one character vs e+´ as two characters combined.
  • Maybe defining a sort order that the sortnames are supposed to work with: does Å sort with A or after Z?

No need to define this, as there is no general rule. This must be considered at the time of sorting and the order depends on the user’s display locale. E.g. in German ä sorts just like a, in Czech ä comes after a and in Swedish ä comes after z. Alphabetical order - Wikipedia has a nice overview over sorting in various languages.

1 Like

You are correct, and I was clumsy with my wording. I meant, we would maybe want to define which language’s sort order rules apply to aliases which have a particular locale setting.

However, I just looked. The “locale” field on Aliases has values which are a language perhaps modified by a country, not an Area (i.e. geography). This is a good design.

Thus the obvious rule is, the value of an alias’s sort name field is collated according to the rules of the language given by that Alias’s locale field. And we would probably want to define how to collate sort names for Aliases which have their locale field empty.

No, that’s not quite correct. The rule must be that the application doing the sorting should choose an alias best matching the user’s locale. But the alias itself does then always sort the same. E.g. in a German user interface I expect a release starting with “ä” to sort the same as “a”, it doesn’t matter what language the name is.

Empty sort name can always fall back on the release name.

Yes, I think Release Group Aliases are slightly more useful than having to repeat an Alias on each Release.

And even Release Group Aliases are not that useful.
I am not surprised we don’t see many.

There’s no specific style, but I’d just sort as anything else: same as the title but with articles and whatnot at the end. There’s nothing bad about adding aliases matching the title or the title language (same as with other entities).


Once again, thank you for the correction, and my language was sloppy. I should have said, editors who want a Release collated well for a locale can enable that by defining an Alias for that locale with a proper sort name.

I think we might be talking past each other here.

I was talking about the case where the Alias has an empty locale field. In that case, I’m not sure how software collating Releases should consider that Alias. It does seem like an empty locale field in an Alias of type Release Name should be a deficiency to correct (exception: it is OK for search hint aliases).

I think you are talking about the case where the Alias has an empty sort name field, regardless of the locale field. In that case, we have a choice. We could interpret the empty sort name field as meaning, “same as the Alias’s alias field”, or as “sort name not specified”. I prefer the latter interpretation. It lets us treat an empty sort name field as a deficiency to correct. While the deficiency exists, then it is reasonable for collating software to use the alias field of the Alias to substitute for the missing sort name value.

Good point. Maybe the rule for software collating Releases should be, when collating Releases by Title, check the Release Group for Aliases also. (There probably should be a rule for the case where the Release has a different Title value from the Release Group’s Title.)

It seems that Release (Group) title sort names are useful anytime Releases get collated by title. That probably happens in the MusicBrainz web app. It certainly happens in the directory structure where I store my music files.

One place where this happens is an Artist’s Discography section. See for example the “Various Artists” Discography/Album listing. It is sorted by year, then by Release title. In the 1960 listing, shouldn’t “The Blues Roll On” be sorted by “Blues”, before “Hawaiian Slack Key Guitar”. Shouldn’t “The Newport Folk Festival” be sorted by “Newport”, before “Paras n°2”? Instead, they are both sorted by “The”, after “Roots of the Blues”. Where else does it happen in the web app?

My own directory tree of music files is divided into a layer of directories named after Release Artist, then a layer of subdirectories named after Release Title. Out of 477 Release-level directory names, 57 begin with “The”, “A”, or “An”, and 10 German-language names begin with “Die” or “Das”. That is about 14% of my releases. They could get alphabetised better, if I were to make 67 fairly simple edits.

I am hearing an absence of objections, and a consensus that adding aliases with sort names is reasonable. I think my next step is to add the sort name alias to the next Release I enter or edit. I will leave that edit open for voting, and post a link to the edit here. Looking at a few practical cases might help clarify things.

Thank you for the good discussion! Let’s keep it up.


it could possibly be beneficial when there’s multiple variant release titles that aren’t tied to locales. for example, on this version of Elvis’ Christmas Album/It’s Christmas Time, note that the German release has an English title. could also apply when there’s no linguistic content in the title, such as 4×4=12 by deadmau5.

that’s not to say they’re mutually exclusive, as we could do both~

1 Like

This is why I don’t usually fill in the language when the same title is used in multiple countries.

Common Example: If you have a German release of an English band, with an English album title - how would it be labelled?

And what happens when a French language request is made for a sort order of the above?

So I leave it blank unless it is a specific translation from the default title.

The locale as I understand it describes the alias, so an English title will be set to English. It doesn’t matter where the release was released, or where the band originates from.

But maybe an alias with empty locale is useful for specifying the default sort name. For clients the logic for the title usually would be to select the alias with locale matching the user preference, if it exists. If not fall back to default release title. But since there is otherwise no default release sort name it could be beneficial to fallback to alias without locale and use that sort name. But also alias matching the language set on release would be a sane choice.


[quote=“outsidecontext, post:12, topic:592743”]

Agreed. I understand the “locale” field of an Alias entry to describe the language of the text in the “alias” and “sort name” fields. It makes no claim about the languages of whatever Release or Release Group that Alias is associated with.

Maybe so, but I think this is the point where we are discovering a shortcoming in the data mode, and bandaging it with another mechanism. The data model for Release and Release Group includes a “title”. That title ought to be described by a language identifier, and supplemented by a variant title for sorting. Sadly, the data model does not have a place for language identifier and sorting variant of title. So we try to use Aliases as a bandage. We could do it. Or we could try to improve the data model.

It seems like this is not a hard problem. The sorting mechanism looks for an Alias of that Release or Release Group which is labelled with a French locale. It does not find such an Alias. Then it has to fall back, probably to the Release’s own title. If the user does not like that result, they can add a French-language Alias to that Release or Release Group.

That example does not seem to require an empty locale field. The Release Group has a single title, in a single language. Each of the seven Releases in that Group has a single title, each in a single language. If I want to add a Ukranian-language Alias to the Group or to a Release, I translate the single title, and set the locale field to Ukrainian.

That particular title does not seem to require an Alias. But if it did, the locale field of the Alias would describe the language of the Alias’s title and sort name fields, not the alias of the Release title. And if the Release data model added a field for locale of Release Title… I say we have to accept that creative people like to mess with boundaries, and sooner or later a Release will violate whatever labelling system we cataloguers try to set up. It does not eliminate the value of the labelling system for the vast majority of Releases, however.