Naming guidelines of classical music works

Following the discussion on J.S. Bach naming of works, I have started to think about general work naming style guidelines. I think that it would be helpful to improve naming consistency to provide such guidelines in MB documentation.

Here is a first attempt draft reflecting my thoughts on the subject, feel free to comment …

Canonical names of classical music works

Naming of classical music works faces many challenges, as many works have not been published, or are available only from manuscript copies of the composer original work.
The names of classical music works should be generally structured as follows:

[Original name of work][Key][(additional work information)],[catalog work reference number]


[Original name of work]: name of the work as printed on the first edition of the work published by the composer. If the work was first published posthumously, the following sources should be used, in decreasing order of priority:

  1. Autograph manuscript of the work
  2. Name of work in composer catalog of works
  3. Name of work in IMSLP
  4. Other name sources [examples ?]

Please note that for work published posthumously, the first published edition of the work may not be the most accurate source to name the work, as more recent musicology research may have provided better information or sources for the unpublished original work.
The name should be quoted in the original language of the publication, or of the manuscript.
When a work has been attributed a common name posthumously, or after the initial publication of the work, this name should be listed as alias. [examples]

[Key]: when relevant, the key of the work should be quoted, in the language of the work.

[(additional work information)]: when relevant, additional work information used to name the work. [examples]

[catalog work reference number]: mandatory whenever a catalog of the composer work exists.

These are general guidelines, please refer to the annotations of the composer MusicBrainz entry for specific guidelines for the composer.

Given the diversity of situations in terms of availability of information on works, I believe that the best way to ensure consistency would be to add composer specific guidelines in the composer’s MB entry annotations. This would enable to provide information on available sources to use to determine the proper work names. Given the diversity of situations in classical music, generic guidelines cannot properly cover all the specific cases.

We also would need a way to capitalize on the forum discussions which describe how specific composers works are to be named.

1 Like

If a work has a well-known name everybody use, I would prefer it to be a part of the MB name, even if it was not used by the composer himself. A good example is Beethoven’s piano sonata no. 14, known as “Moonlight”. A lot of people never heard of a “Sonata quasi una fantasia”, but know the “Moonlight sonata”. And no, having it as an alias is not enough.

In my opinion, MB will be better off not as a purely academic project created by musicologists for musicologists, striving for absolute correctness even if the result is less useful to an average person, but as a user-oriented project where pragmatic solution and user-friendliness may justify a compromise.

For example, a possible compromise may be to add a new optional property “canonical name”, “name of the first publication” or something similar.

My second point is about catalogue references. For some composers there is no a single authoritative catalogue, but several catalogues which partially intersects. Take Tchaikovsky: we have works by opus number, The Tchaikovsky Handbook and Thematic and Bibliographical Catalogue. Some works are present only in one of these catalogues, some in all 3. In such cases it probably makes sense to add to MB composer entry an annotation about priorities, to prevent multiple catalogue numbers in some names.


I have proposed this to be consistent with other MB naming rules. For example for releases or tracks, the name should match what is written on the media, even though it may have other names.
That being said, I agree that very often the work title on the original music score is not enough, which is why I included the [(additional work information)] part in the name.

So to take your example, the name could be:

Sonata quasi una fantasia in cis-Moll (No. 14, “Mondscheinsonate”), op.27 no. 2

“Moonlight” would in any case be an alias, as English translation of original German title.

I fully agree with you. This is why I proposed to have author specific guidelines in the author’s annotations. When there are several catalogs, we could specify which catalog reference should be used as the main work number, and the priority of others.

1 Like

If I understand correctly, you consider important the following points:

  • The primary work name shall be the name used in the first publication.
  • All parts of the primary name shall be in the native language of the composer (or maybe in the language of the country of the first publication).

Could you please explain why you consider both points important? I agree that they are important for a musicologist, but from my point of view they are secondary to the convenience of a mass user. I think there is a large difference between Bach Digital and MB: the former targets “researchers and practical musicians”, as they write on the first page, the latter is primarily for a non-professional users cataloguing their collections of CDs.

Current name of our example is “Sonata for Piano no. 14 in C-sharp minor, op. 27 no. 2 “Moonlight”” violates both your points. The name is not as in the first print (not “Sonata quasi una fantasia”), the key is in English notation instead of German (“C-sharp minor” instead of “cis-Moll”), and the additional work information is “Moonlight”, not “Mondscheinsonate”. Why do you consider this name to be inferior to “Sonata quasi una fantasia in cis-Moll (No. 14, “Mondscheinsonate”), op.27 no. 2” for an non-professional user?

Some of name parts may cause controversion, but may be discussed. For example, what are advantages of writing the key in a native language of the composer (or shall it better be the language of the country of the first publication)? Why is it better for the user if works of Bach and Mozart contains a key “Cis-dur”, works of Tartini “Do diesis maggiore”, of Boccherini (published during his employment in Spain) “Do sostenido mayor”, of Sibelius “Cis-duuriasteikko”, of Tchaikovsky “до-диез мажор”, and let’s not forget famous Armenian composer Արամ Խաչատրյան for whom I can’t find how to write C-sharp major in his native language and script. English is not my native language, I fluently speak German and Russian, I understand Italian and French music terminology, but considering a possible mess I would prefer English key names used uniformly.

1 Like

Indeed I consider these points important, as guiding principles, but there should be flexibility, as we face very diverse situations from one composer to another. What is key, is to have consistency in naming the work of a given composer. This is why I believe that guidelines at composer level will be very useful to address the diversity of situations.

On the topic of languge, MusicBrainz is an international database, used by people from different countries and languages, I believe that we need to acknowledge that this means that ideally we would need the works names in several languages.

What is important from my point of view, is that for a given composer, all works primary names are in the same language.

If you look at works entries, it is a mess, with names in various languages. For example:

Unless I am mistaken, English is not defined as the primary or default language of MB.
I do not feel that using English as default language for works names or key indications would be right.

I am not a native English speaker either, and if I had to name the Moonshine sonata, I would name it

Sonate pour piano No. 14 en sol majeur, op. 27 no. 2, “Clair de lune”

So my proposal to use the language in which the work was initially published (or native language of the composer) was meant to provide a consistent rule, which do not rely on selecting arbitrarily a language as “primary” or “default”.

As you pointed out, it is a rule which is already used to name composers, but MB features enable to find easily “Aram Khachaturian”, where most non Armenian speaking users would struggle recognising " Արամ Խաչատրյան"

I would use the same language for key indications as I believe that the key indication in a work name should be consistent with the name language. I would clearly dislike using in work names key notation which is not linked to work name language: it would be arbitrary, and it would negatively impact the users for which this standard key system is not native. Aliases of the name in different language would have the key indication translated. Also MB has a key attribute which enables the translation of the key in the language used by the MB user.

Regarding the primary work name, I proposed the rule “name used in the first publication (when available” to be consistent with the way MB names other entities such as recording and tracks.

Many classical music works have been named differently in different times, and MB is able to capture these names as aliases, which do not impair at all the convenience of users to find works. For example, the “Moonshein” sonata was first named “Arbor sonata”.

This rule is not necessarily the rule that musicologist would use in a work catalog, the purpose is to be faithful to the composer intent, not to be “musicology correct”. If you take the above example of the works of Jean-Marie Leclair, the names used in the catalog of its works (“uniform titles”) are not the names used in the original published music score. In this example, I would not recommend using these musicology correct “uniform titles”, but use the original music scores titles.

In practice, there may be few composers which most original published works are readily available, and accessible to the public. I believe that in many cases, we would revert to another source to name the works.

I hope that this clarifies how I came to these draft style guidelines. This was intended as a first proposal, to get the discussion started.


I’m not sure where this discussion is headed. On the one hand, I can see the benefit of greater consistency in the naming of works in MB. On the other hand, to move away from a style that has usability as well as consistency as an objective might be counterproductive. I don’t think this is necessarily a language issue. After all, no-one would want to rename “Eine kleine Nachtmusik” to be “A little night music” (or, more accurately, “a little serenade”) would they(?), but equally “Moonlight Sonata” is more universally recognized than “Mondscheinsonate.”
Incidentally, the MB entry for the former is Serenade no. 13 for Strings in G major, K. 525 “Eine kleine Nachtmusik” - a mixture of English (generic) and German (specific). Are we seriously going to re-name all the generic components of work names depending on what language they were first published in? In any case, these generic components of the work name (which are important in achieving consistency and usability) are frequently omitted from or post-date the published manuscripts. (Ironically, Mozart himself only wrote the words “a little serenade” as a purely generic description of the work.)

I fear that an over-enthusiasm for academic purity in work-naming will only:

  1. reduce usabilty
  2. create a huge amount of re-naming effort by what is a fairly small band of classical MB editors
  3. increase inconsistency as a result of an incomplete effort (resulting from (2))

So I think a compromise is needed. Where works of a composer are consistently named in a widely-understood manner and in accordance with current CSG, I do not think we should start wholesale renaming. Where a composer’s works are poorly and inconsistently named, I do agree with the idea of having an agreed and documented style for that composer. Some time ago, there was an effort at selecting individual composers for a “tidy-up” by classical MB editors. This worked pretty well, I thought - perhaps it is time to revive it?
A word about aliases. I fear that this is not a solution - i.e. putting an “academic” work name as the main name and the commonly-used name as the (what language?) alias. Aliases are not subject to any guidelines and can be changed at will by editors without any voting; they should be seen principally as additional information that is useful for search purposes.
In conclusion, I think that it is right to raise the question of better guidelines for work naming, as the existing CSG guidelines focus on track titles not work names, but that the way forward is to be consistent with existing guidelines and to focus on achievable improvements that enhance MB’s usability.


The only guideline that I have found related to naming works is:

Name is, unsurprisingly, the name for the work. Usually this will be the same as the recording name… but not always! For example, remixes don’t usually get their own work, so a recording like “Cool Song (Someguy remix)” will also be under the work for “Cool Song”. Also, classical music is more complicated, as it usually is - please make sure you read the guidelines for classical works before doing anything with classical music. This is the only required field .

Especially, the guidelines for classical works do not include any information on naming works.

Have I missed something? This guideline do not look very relevant for classical music.
It seems drafted for current music, where the recording is the way new music is published, with music score published afterward.

I’m sorry if my proposal looks too academic. I’m totally open to improve it, especially if it makes it more practical.

If I raise this question, it is because I am working on retagging my music tracks, to add works tags, and would like to do so through MB. However, at the moment I am a little at a loss as to how I can correct work names. Track names are easy: the guideline are clear, and they should match the CD booklet.
But works so far are mostly messily named. The question is how could I contribute to correct this.

Indeed, I totally agree with you. Unfortunately, for many composers the naming is not consistent, at least that’s my experience so far, but I’m quite new to MB. I may have been unlucky …

You are right, we need a better approach to move forward. An idea could be to start by listing how are named works for composers where names have been cleaned and are now consistent. This will provide what should not be changed, and gives us good examples to draft guidelines.

For example works of Claude Debussy are named very well and very consistently.
Looking at these names, they have the following structure:

[Name of work][key indication], [catalog references][, "incipit"]
Disambiguation: [type of ensemble][additional note]

[Name of work]: the name of the work in French (original language), very consistent with the catalog of works. Main difference: version indication specified in Disambiguation.
[key indication]: when applicable, in the title language (French)
[catalog references]: the two catalog references are listed " L. 27, CD 40" as there are two versions of the catalog.
[, "incipit"]: for works with lyrics, the incipit of the lyrics, as indicated in the catalog of works

Disambiguation is written in English, with the following information
[type of ensemble]: type of ensemble the work was written for (“for voice and piano”, “for orchestra”, …)
[additional note]: version information, “unfinished”
Disambiguation is missing for a few works.

@MetaTunes, would you have other examples of composers which works names have already been cleaned?


It may be for you, but that is not a universally held belief—or even an officially stated purpose of MusicBrainz—and indeed many things that MusicBrainz can (and does) capture are entirely unrelated to cataloguing CD collections.

You might be interested in reading through this discussion from a couple of years ago:


Debussy was the first subject of the classical “community clean-up”. Other specific composers were Mahler and Dvorak. That leaves quite a few. Interestingly, the clean-up did not specifically mention workname corrections in its objectives. Nevertheless, Debussy does look fairly tidy. There are some inconsistencies still (e.g. ‘Preludes Book’ vs ‘Preludes Livre’), not to mention some Japanese pictograms at the end of the list of works. Most of his works are commonly known by their French titles (an interesting exception is “Children’s Corner”) and so don’t present too many difficult choices - unlike other works discussed here where the commonly-used title is not the ‘original’ name.

That may be true, and it certainly is true that MB is far more than just a tagging source. Nevertheless, it is quite apparent that the focus of MB is on recorded music. For example, it is irritating when trying to review a composer’s works, to have all the recordings of those works listed, cluttering the page and often making huge gaps between works. Similarly, in the clean-ups mentioned above, the focus was on recordings, not works. And, as already discussed, the style guidelines for works (as opposed to track titles) are virtually non-existent (hence this thread).
I am very supportive of MB becoming a more authoritative encyclopedia. In particular, it would be nice to improve its reputation as a source for classical music data. I think this is quite a long-term project with a number of different strands. Work-naming is one of these strands, but separate and related strands must be the database structure and the UI. For example, when listing a composer’s works, aside from the ‘clutter’ problem, it is not possible to see a hierarchical view, or just to list ‘top’ works. As regards Picard, my experience with the Classical Extras plugin highlighted that accessing the MB database for works which are parts of other works is not really catered for in the webservice - each requires an additional lookup with a 1 second rate-limiting.
Apologies for broadening the topic out somewhat, but I think the broader context is relevant. Agreeing a style for work names and then implementing it is important but will take quite a lot of effort. I think it is doable: although I haven’t counted how many classical works there are in MB, implementation can be structured by composer and popularity. But to make it worth doing, there needs to be a belief that MB aims to be a better (and preferably pre-eminent) resource for classical music.


Reading through the discussion above, it is clear that there are different opinions. Maybe a partial solution may be easier agreed upon, for example:

  • Very broad general rules which are more recommendations than strict requirements. For example, if there is an authoritative catalogue for the composer, use primary name from the catalogue; otherwise prefer the name of the first publication, or maybe the name under which this work is best known.

  • More detailed rules for particular composers, for example for J. S. Bach “the work name must be from the BWV catalogue, followed by BWV reference, the key must be in German style”. Such rules may be agreed upon when starting clean-up of a particular composer, as suggested by @MetaTunes. For some composers pragmatical considerations may require different rules for different works. For example, at the moment for Tchaikovsky most vocal works are entered in original language (mostly Russian, some French), whereas instrumental works are entered mostly in English (that is, “String Quartet no. 2 in F major, op. 22” instead of “Струнный квартет №2 в фа мажоре, соч. 22”). Shall Tchaikovsky’s instrumental works be renamed? I don’t think so, since probably it would not benefit most MB users.


I think this is a useful suggestion to build on. Also, to include the guidelines for each composer in the annotations.

I have not read the whole thread - too much to read. All I want to add is it is always impossible to be a single answer for every case. Whatever is decided will upset a large group.

Classical is complex. And the complexity needs to be there for the true fans.

But please leave a way in for idiots like me. I have very little classical in my collection, and I need those simple and common names. I need a way of being able to see what the music is as shown on the CDs in my hand. Yes, it will be great to dig down and find the true details as you discuss, but having these as the first results coming back from Picard will not draw me into the Classical World. It will turn me away with the complexity.

As someone noted above, in the non-classical parts of MB the rule is generally to put into the database what is written on the CD covers. Clearly we need both.

There is a classical addon for Picard, and it would help if there is a simple way to set a mode for Beginners. Don’t leave us out with the technical language. :slight_smile:


The track titles should always show this (with a few modifications for Classical Style Guidelines). This thread is about the work names, which the general user may not care about. Picard only picks up the lowest level work name and puts it in the “work” tag, which most users don’t use.

I wrote the plugin because Picard does not pick up the work structure and therefore does not replicate what is shown on a typical classical CD. Hopefully the plugin achieves that (see pic below). The intent was to allow users to customise the tagging to show what they want - this could be based on the track titles (with added structure) or ‘canonical’ work names or some combination. I agree that this results in a large number of options which could be confusing at first. The idea of a “simple way to set a mode for Beginners” is intriguing and I’ll look into it, but I might need a bit of help in deciding what is both simple and effective :wink:


Freso and I would like you to make it intuitive as well. :rofl:

I arrived here at MB thinking, “How difficult can cataloguing well be?”.

And now think that simple and effective cannot both be achieved.
Let alone intuitive.


From what I have found, Debussy’s works with “book” in name refer to orchestrations of Debussy’s original works. For example: Préludes, Book 2: Brouillards (orch. Matthews)

The title of the orchestrated work may have been published with the work name translated in English.

These seem to be translation of Debussy original works (or Japanese orchestration on Debussy’s lyrics)
So far, Debussy works seem very clean, great cleaning work.

Good to know too, I’ll move to these composers after tagging my Debussy albums.

Actually in case of Debussy, I have found quite a number of track names to be corrected …

Fully agree with you … There are a few database changes which would be useful, such as “tracks groups” which are available in other Music DB such as Discogs, and would enable matching how tracks are grouped. To keep this thread on the initial topic, I’ll open another topic …

A few change in UI may greatly simplify correction of works names, so it’s good that you raised this topic.

Actually I am not sure that there is such a difference in opinions.
My initial proposal for general rules was probably too normative, and I’m totally fine with more general rules which are more recommendations and guiding principles. We just need to find the right balance to have overall consistency across composers, but flexibility at composer level to allow for the huge diversity of classical works.

As @MetaTunes, I find your suggestion very useful to build on, and create a consensus on general works naming guidelines.

Thank you. Understand better now and will go away and be quiet. :slight_smile:

1 Like

This topic has been quiet for some time, which hopefully means that all who had an opinion have been able to express it.

I’ll try to redraft the classical works naming guidelines based on the comments received so far: less prescriptive, more principle based, more emphasis on the need to specify composer specific guidelines.


I’m also very interested in this topic, agreeing to a guideline should increase user base.

Assuming that there is an agreement how the main work title should look like, I would like to see more guideline details regarding aliases.

Aliases should be used to provide translations of work names in other languages.

Especially within a locale. What I’m after are further suggestions how to handle aliases. For example

  • “main” alias for locale follows composer work guidelines in a given locale

  • “common” alias for locale lists a full common name of the work as it is widely used, when relevant

The purpose of the common alias would be to provide an alternative to work name as specified in the guideline which might be too academic for some. For example in “en” locale for Beethoven String Quartet No. 4


Quartet for 2 Violins, Viola, Cello no. 4 in C minor, op. 18 no. 4


String Quartet no. 4 in C minor, op. 18 No. 4

Picard ( Classical Extras ) and others could provide an option to use only main work names within a locale or provide common if present . Perhaps this will satisfy the needs for a musicological naming guideline and every day use one.


Thank you for raising this topic, @Algwyn. I am not in a position to give a detailed contribution now, but I have some high-level thoughts.

  • This is an important but complex subject. Don’t expect good solutions to be easy.
  • I believe classical works do not generally have single names. They have multiple names. Each name is preferred by some users, for valid reasons. Any solution that proposes a single name will be unsatisfactory.
  • Satisfactory classical work names are based on conventions of the reader’s language. Thus any good solution will need to include some form of language variants for each work name.
  • Classical work names are based mostly on descriptions. They combine factual phrases using culture-specific templates. I imagine a solution which involves a dozen or so fields storing facts about the Work (e.g. type is Symphony, number is 9, key is D-minor), templates based on conventions from the reader’s culture, and a way of rendering facts into the reader’s language (mapping key to “D-minor” or “d-moll” etc.).
  • Usability will require some guided way to enter all these fields into the Work entry reasonably.

Thus Work Name becomes not a text field in a database, but a function which generates text strings from data fields based in the reader’s preferences.

1 Like