I appreciate the response - quite obviously thought went into it. However, I admit, I find it rather confusing, and somewhat contradictory, and I really just have to disagree here.
>> In my opinion, any concept of a release language, for classical, is >> rather bogus. > >The MBReleaseLanguage cannot encompass all which might be interesting to >note about languages for a great many classical releases. This does not >make it bogus. Per the current definition of what data the "Release Language and Script" fields hold, I would disagree: "Releases have two linguistic attributes, language and script. The language attribute (e.g. French) records the language of the release title and track titles, (not the lyrics and not the extra information on the disc sleeve), and the script attribute (e.g. Latin) records the general type of characters in which the release and track titles are written." - http://wiki.musicbrainz.org/ReleaseLanguage I emphasize the "(not the lyrics and not the extra information on the disc sleeve)", as essentially, what you describe in most of your email is exactly one or the other of these - the lyrics, or the extra information on the sleeve. >(Right now I admit I can only think of 1) Languages used in the original >titles for the music and 2) Languages used in/on the packaging.) #1 - This assumes that the music originally had a title. With the exception of opera, very little classical has titles in the original, other than perhaps the generic "Sinfonia" "Sinfonie", "Symphony", etc. - Most canons are untitled (Mozart: 6 out of 64) - Most songs end up using the first line of the libretto for a title, as they too are untitled (Mozart: 24 out of 48) - Most symphonies are untitled (Mozart: less than 20%, and (almost?) all are names others gave them years later) - Most concertos are untitled (Mozart: ignoring the untitled, unnumbered ones, 3 out of 38 - again, all after-the-fact titles) I could keep going, but you get my point. For the large majority, not just Mozart, but for most classical composers, they didn't title the works. Most names any named works do have came after the fact, as people began to call them by this or that title. So there is no "original titles" we could even use, much more often than not. #2 - See "the extra information on the sleeve" above. Plus, with regards to both suggestions, the unofficial ClassicalReleaseLanguage page goes further than ReleaseLanguage in addressing these suggestions. Under "How to determine the language of a classical release", we have 5 differing ideas on this, each with various problems: 1. "the release title will usually give hints like the words like "symphony" or "sonata" which change from language to language can be used" A subset of what I think you mean by "2) Languages used in/on the packaging." Looking again at the 20 examples I gave before, using this idea, 9 would become English (including the 2 CD release which has 7 out of 13 tracks titled in English on CD1, and none on CD2), 1 would become French, the Handel release with work/track titles all in English would now be Latin ("Musica Sacra"), 1 could be any of a bunch of languages ("Requiem"), 2 would be "no" language (group name as the title), and 7 would still be in three different languages within the title. Problem is, these titles have essentially nothing to do with the contents - "Beethoven: Symphony No. 9" with the work title listed in 3 languages and all tracks titled only as tempos would now become an English release based solely on the single word "Symphony" being the English spelling. 2. "other words like "concerto FOR piano/concerto POUR piano" determine release language" Another subset of what I think you meant... Yet, again, this does us no help when we're dealing with the majority of releases either having only tempos for titles, and/or more than 1 language used to list works. It's better than just "Symphony", but we're now basing the release language on "symphony" and "for"... 3. "you should not look at the language of a work's name to determine the release language (Der Ring des Nibelungen is not commonly translated for English releases, neither are the song titles)" This indication, while unofficial, is the exact opposite of your first suggestion... 4. "character names (in operas, for example) may be translated, but not always" I'd suggest it's more like "may be translated, but almost never" - not a single opera CD I have (nor that I can recall ever seeing) has translated names for roles. 5. "the way the key signature is written out ("in D minor" is English and this would change if it were a French or German release)" This is again a subset of using the liner - and has all the same problems. 2, 3, 4, 5+ languages on the liner to pick from... > Any decent non-budget release (Naxos, Philips, etc) >> tends to list either multiple languages or single listings in a mix of >> languages. > >"Tends to" would be correct: Naxos has offices in both Norway and Sweden >and some releases have catalogue numbers ending in "N" or "S" to >indicate that the packaging is in Norwegian or Swedish. (Some of these >are available without the N/S, some are not.) I think we can all find exceptions, but in at least my glance through 500 or classical CDs at the public library the other day, I can say that less than 50 had only one language on the liner. >We've typically selected one, rather than enter the same >> classical release multiple times. The non-decent releases tend to be >> the ones with a single language listing - but so minimal ("Adagio", >> "Allegro", etc) as to be rather not worth consideration. > >Milestones in the history of recording they are not. My suggestion: >waste little extra effort on releases you don't care about. We don't >need to have everything perfect, but we do need to have some areas >pretty well-covered, so it is easy for new editors to add good data. I just can't help but disagree with this notion. This is essentially the same concept as "ignore new (add) edits unless you care about that artist and that release". We strive for clean listings in non-classical; why should any other standard hold within the listings for classical composers? >> Point is, "the language of the release" makes sense for most things. >> Classical, though, it just doesn't. > >How about this for an algorithm to determine the MBRelLang for classical: > >For each track on the release, consider: >Is it 1) a vocal work, or 2) a non-vocal work. > >If 1), then a fitting language for this track is quite probably the >language sung/spoken on this track. (Or the main language in case there >is more than one. Conceivably, even here you might want to say >MultipleLangs). > >Much more often than not, this language would be in the track title, so >we're following MBRelLang thinking here. Yes, you do find "The Magic >Flute", in stead of "Die Zauberflöte", and you do find "Act I", which is >not German, but my suggestion here is to stick more closely to the >lyrical content of the track, not the packaging. (Unless you're adding a >translation, of course.) > >(Even if a track title is in a different language than the lyrics, and >possibly not even in the lyrics at all, I think that what's sung is more >important than the small set of words (a foreign phrase or a name, e.g.) >used in the title.) Here's where your suggestion gets confusing for me. Using your example of "Die Zauberflöte", yes, if we use the language of the recording, at least we do avoid the problem of the (common) CD listing "The Magic Flute / Die Zauberflöte / La flûte enchantée". However, it solves the problem in what I would consider a suboptimal way. First, it's the exact opposite of the "not the lyrics" principle used for non-classical. (See the quote above.) If this could solve the classical language problem, I'd be all for it. But this really only solves it for opera, songs, songspiel, etc. As you point out, it does nothing for non-vocal classical. It also has the problem of assuming that people know what language works are in. German, Italian, English, or French might be easy, but in my own collection, I know I couldn't identify Norwegian vs Dutch vs Swedish by just listening. Or we get into the problems of "Old French", "Old English", etc - do we even have some of the archaic languages in the system to be able to support this concept? >If 2), you have a non-vocal work, then the title of the work/movement >should be your guiding beacon. This may be translated by the label from >which the recording originates, and if so, this should be used, much >like "The Magic Flute" should be used if the opera was sung in English. <snip> This just brings us back to where we already were, but now we would have carved out one part of classical and made it non-standard to the common ReleaseLanguage principle, while making yet other classical still adhere to the common ReleaseLanguage principles (in some classical-specific fashion). Actually, the problem here would be even greater - now we'd have taken out the works which do at least sometimes get titles, only to then try to then set the release language based on the title for the group of works with, for the large part, don't have them. You could say "Symphony" is at least a "title". However, now we'd be looking at the problem of people having to research (if the info even exists such that enough research could find if the original was "Symfonie", "Symphony", "Sinfonia", "Symphonie", etc.), and we're dealing with a rather limited subset of possible spellings for the worktype words. <snip> >Coming to an end for this section: Which MBRelLang? If a substantial >proportion (which in my head usually translates to one third or more of >the track titles) differ from the rest, it is MultipleLanguages. If more >than two thirds of the titles are Czech, then it should be Czech, even >if it was released by DG under the title "Love Songs", and full >translations are provided for several languages. On this, under the current way we do it, at least I agree. But I think this is the easy case - obviously, if you have 10 titled vocal tracks from 6 composers, in different languages, "multiple languages" is the obvious choice under the current way we handle it. (But if we ever turn on the TrackLanguage field, this "easy out" goes away...) <big snip> >Again: A group of well understood and well-maintained classical >MB-releases, will enable us to point the interested editors in the right >direction. We shouldn't worry to much about languages where we cannot >help out. Rather, we should be glad if people in Portugal added >Portuguese stuff. Quite honestly? We don't have it. My first few classical edits, way back when, were to add a copy of Mozart's Requiem and to try and fix up the listing for the soundtrack to "Amadeus". I read through all of the CSG stuff in the wiki (which is confusing enough, given that we don't actually have *any* official classical guideline in the wiki). I searched out each release with the works, trying to find ones which had been done right. I even made sure to find ones the editors whose names were in the wiki as having essentially written the CSG guidelines had *entered* and copy/pasted them over. The result? Each track had me doing 5 to 8 series of edits, fixing things other CSG editors found wrong with the track titles. We did not have a single correct entry in the entire database for Mozart's Requiem - every single listing, of even the best CSG style listings, had at least 5 different things wrong that some CSG editor or another pointed out when I copied those titles exactly as they had been listed elsewhere. <snip> and this, then, is where we get into the "language used for the listing", which was my original point, not the "release language for classical" problem, which is related, but very different. >We shouldn't lament this, but rather let user demand guide us. The >common practice of not interfering with the submitter's choice of which >language to use is currently working out quite fine, isn't it? If >someone _really_ wants to add a French version of something that already >exists in German (or Polish, for that matter), why shouldn't we let them? What common practice? Where in any official guideline for classical do we say "pick a language for classical releases and use that, then noone will ever again change that language"? Essentially, to my eyes, this is an argument for Balkanization of classical. It ignores that most releases come in various packaging sizes, much more often than non-classical. It ignores that most releases come with either multiple languages or essentially no language on the packaging. Instead of bringing as many people together on the same releases, to get the most number of eyes finding typos and adding ARs, this encourages people to just pick the language they want it in, and add a new release (or multiple). Now instead of 2 people looking at and working on the same release, we have two releases to manage. Now, instead of 1 release for CSG editors to AR and fix up, we have 2. My point is, instead of our having it just be "pick the language you want, pick the box set / single CD you want, then add it", leading to numerous essentially duplicate entries for works, let's concentrate the efforts to a single listing for that same release, then do it right. ARs for classical take forever - there's a reason so few people do them. For other things which tend to be entered as pseudo-releases, the works at least have "real" titles. The same isn't true for essentially anything to which CSG applies. (The only part which would, as you say, we *don't* translate!) So, given that we have a group of releases with works which essentially have no one "correct language" title; given that we have a group of releases which (across the vast majority of classical releases) either list "no" language or multiple languages; given that we have a group of group of releases that take perhaps the most effort and research to enter even the track titles properly, and given that we have a group of releases that are both the most-time consuming (other than perhaps musical theater) and the most-rarely done for ARs... ...Why would we not standardize? It solves the release language problem easily. We create a "classical" language, with the understanding that it's essentially saying "this release is classical and has all the issues surrounding classical releases and there not being any one "correct" language for most releases, etc". It (partially) solves the Balkanization problem, as we now have only one listing per any release. (We still, though, have the box-set problem). It massively simplfies the work for editors, both CSG-skilled and not. It allows a single master works list to be referenced and simply copy/pasted, just as I had done that first time for the Requiem - but this time, copying from a list which is correct per CSG. (This assumes more works lists are put together, but I think it can be expected that, where there is interest or demand for one, one of us will take the project on, as is already happening.) Thus we avoid the problems of people not knowing CSG, doing CSG incorrectly, mis-translating, typo's in translation, etc. The master list also has the extra benefit of now taking the editors who might not be CSG editors, but who do have an interest in at least a few classical CDs, and focuses them not on a "release in the language I picked when I entered the CD", and not even on just "the one common release listing for this CD in a standardized language", but rather, it focuses all that effort and all those eyes on one single listing for that work, a listing which now truely has a chance to be improved. I've seen lots of edits to the wiki (or pms to me pointing one out) to fix typos and tempo issues with the Mozart list. We may not have GenericWorks, but the wiki lists are the next best thing; if we're going to try to clean up classical, let's get the effort focused as best as possible, rather than continuing to approach it in a manner which spreads a very few people across far too many things in need of cleanup. A popular pop CD gets hundreds of eyes on it each time someone pulls it up or tags from it. A totally obscure non-classical release may not get the same numbers of eyes, but when people do look at it, problems are rather apparent to anyone. (Notice all the hundreds of edits all the time to fix typos and such on non-classical.) A classical release, on the other hand, may have many errors, but between the very few numbers of people who even look at any particular listing, listings being in the complex CSG style, typos being much less obvious to anyone with only a passing interest in classical (say, "Andantiano" instead of "Andantino"), and the basic apathy problem ("all of classical is a mess - it's not worth my time to fix that"), if I can say it, as you say, "We don't need to have everything perfect" - no, we can't get perfect, but shouldn't we at least try for perfection and fail, instead of accepting that it's essential a HugeMess and NotWorthTryingToCleanUp? One further benefit of standization in track titles. If we standardize track titles to a master list, the apathy and the HugeMess issues go away. You see a typo in a listing for Chopin? No problem - you edit to fix it, you edit to fix it in the wiki, people subscribed to that wiki page see the fix and either revert it (as being correct the way it was) or then can easily do a quick search, find all other instances of that same work, and fix that typo on the others too. Now, on the flip side, what do we lose? Well, we've now standardized classical to whatever CSG language is used in the wiki master list for that composer. So, where I have that non-common CD Chopin that's only in English, or only in Dutch, or only in Swedish, or Russian, or whatever, it's now being entered in French CSG (as the Chopin list uses French CSG). Is this really a loss, though? Is it any different from what just about every site/book/magazine listing/newspaper review/etc selling/listing/reviewing classical does anyhow? I go to Amazon.de, I see classical works on a CD listed in German. I go to Amazon.fr, I see that same CD now in French. .com? English. Etc. (All of these looking at the useful classical-only works list that Amazon shows under the track list, not the questionable-quality-anyhow tracklist they show on classical and non-classical) As I see it, we can keep going as we are, and classical becomes just more and more of a HugeMess as people continue to add classical. We can continue to pretend that the language used in the track titles, and the release language setting for classical releases, has some inherent meaning. We can keep going, doing each release one by one, rebuilding the CSG track titles over and over, correcting typos and missing/bad data in those titles, etc. We can get create overly complex methods for determining the language of a release, etc. Or we can standardize, make the master lists, clean the listings up, then focus on the harder parts - tracking down the correct work ID for the releases which don't ID the works, finding and fixing the mis-identified work listings, removing the bogus listings, merging the dupes, etc. The former option - do it like we do it now - to me, is a recipie for classical essentially being even more of a untouchable dumping grounds than it already generally is. The latter, on the other hand, seems to me a way to get the entire mess cleaned up in some orderly and reasonable way, such that we can say something no other music database can, that our classical listings are clean, accurate, correct, and detailed. Sorry for the long reply, but I just think we're seriously in trouble in the classical area if we either ignore the problem, or we make it even more complex. A formula for building up opera and classical titles that can cover just about any case, a standardized work list, and a little effort, we could make something really good. Brian
_______________________________________________ Musicbrainz-style mailing list Musicbrainz-style@lists.musicbrainz.org http://lists.musicbrainz.org/mailman/listinfo/musicbrainz-style