Cataloging Systems

[BKhon]

I recently wrote this:

http://imslp.org/wiki/IMSLP:Cataloging_Methods

More will be employed later, but I stopped due to the fact that I think it's an utter failure. Might it be better if I go by composer, rather than system? Eventually I plan to go more in depth, and use this as a "quick guide" reference, which might come in handy. Maybe not, but at least it's a fun read

[pml]

Not bad, though the transatlantic spelling wrangling will start with the first word of the page, “cataloguing”.

Given that IMSLP does not follow the rule of a space following after a period, I would express my preference for the abbreviation KV for the Köchel-Verzeichniss (Mozart’s works) rather than the resulting ugliness of K.

It's also not uncommon for multiple catalogues to annex the same letter, if the abbreviation consists of a single letter, so there are some creative reassignments, e.g. Anthony van Hoboken (actually Hob. is preferred, rather than H., for works of Joseph Haydn); David Kern Holoman (H without full stop, for Hector Berlioz); Hugh Wiley Hitchcock (H., for Marc-Antoine Charpentier).

Fun fun fun.

Regards, PML

[BKhon]

Given that IMSLP does not follow the rule of a space following after a period, I would express my preference for the abbreviation KV for the Köchel-Verzeichniss (Mozart’s works) rather than the resulting ugliness of K. PML

On Mozart's category page, everything is currently listed as K. Why do you think it's ugly?

[Davydov]

I think it would be hard to justify switching to "KV" for Mozart's 490 works on IMSLP, particularly as the English-speaking library system uses "K." or "K.Anh." in its catalogues (courtesy of the Library of Congress's catalog [sic]). That shouldn't be the only consideration of course, and no doubt Philip will point out that the libraries always use spaces after "Op." But without any automatic way of carrying out a global search and replace on IMSLP, we need to target our volunteers' energies where the problems are most serious...

[Boccaccio]

For Carl Philipp Emannuel Bach: Shouldn't it read Wq. for Wotquenne?

[vinteuil]

For Carl Philipp Emannuel Bach: Shouldn't it read Wq. for Wotquenne?

And Gluck.....
Problem: Fux+Mozart have Köchel...etc....

[pml]

<curmudgeon>
Davydov rightly points out that I view the lack of a space after a full stop as an irredeemable infraction of correct style and typography, besides noting that libraries inevitably put the space in after an abbreviation. Therefore, to keep within IMSLP’s style guide, I would consider KV 339 as being infinitely preferable to K.339; however, I do not share either the view that changing the titles of 500 Mozart works would present any particular problem other than the size of the task, or the contention that the Library of Congress must be correct in using K. for the abbreviation and should be followed without question, when both the International Mozart Foundation and the Neue Mozart-Ausgabe both use KV throughout when referring to works, and almost never use the alternative abbreviation.
</curmudgeon>

It’s also worth noting that the Mozart page titles are corrupt in another way, because they usually cite only the first version of Köchel’s catalogue, and this is arguably not best practice for the works where the old number has been replaced in later editions by a new catalogue number. Also, as Davydov points out, there’s far more to be concerned about than the lack of a space after a full stop, or that the Mozart titles are generally a mess. It’s just a pity the titles weren’t planned with better foresight. Perhaps when we incorporate some of the NMA scores it might be a good time to address these problems.

Regards, PML

[imslp]

Someone's angry

Not to address the substantive issues, but I do want to mention this. It is understandable that as musicians we are also perfectionists. However, when dealing with a database such as IMSLP we must also remember that the cost of perfection may be too high. The reason IMSLP is able to grow so large was in part because we emphasized efficiency, sometimes at the cost of imperfections. I would especially mention that systematic and robotic changes such as adding spaces after "Op." or "K." should really not be done by contributors (unless the scope of the change is small, which does not seem to be the case here), as it is an inefficient use of their time. If it becomes enough of a problem, I could, with relatively minimal effort, do a global search and replace and add the spaces.

Also a slight substantive defence. The original reasoning behind using K1 was that all numbers that exist in K1 exist in K6 (and points to the same piece), whereas the reverse is not true. In other words, K1 is the lowest common denominator of all Kochel versions, and hence, even if musicologically not entirely accurate, may be practically more useful (K6 numbers are theoretically supposed be in brackets in the title anyway). However, I am not a musicologist, hence I will defer the decision to people who actually know something about the Kochel catalogue.

[Melodia]

Personally I prefer it without the period, IE K 550.
I've seen it all four ways -- K.550, K550, K. 550 and K 550. The KV isn't as common, I've found...

[Davydov]

Apologies for this long response, but please bear with me!

Libraries have extremely lengthy and precise rules for coming up with work titles, so they should (in theory) be meticulous in matters of spacing, capitalization and punctuation. Their rulebooks run to hundreds of pages (just Google "AACR2" or "RDA" to see why), and our easiest option would be just to follow their title schemes, and use "Concertos, flute, harp, orchestra, K. 299, C major (Mozart, Wolfgang Amadeus, 1756-1791)", or "Nozze di Figaro (Mozart, Wolfgang Amadeus, 1756-1791)". Lots of arguments settled instantly (except for works not in the library catalogues), and with a lot of effort the job could be finished in a few of months. But...

Would that be an improvement over what we have now? Well, not necessarily, as we would lose a lot of opus or catalogue numbers from our existing titles, and end up with the unfamiliar "Shchelkunchik. Suite (Tchaikovsky, Peter Ilich, 1840-1893)", instead of "The Nutcracker, suite, Op.71a (Tchaikovsky, Pyotr Ilyich)", for example.

The library system isn't the only one out there, of course. Follow any of the Amazon links on IMSLP, and you'll see they're building up their own database of works, e.g. "Nutcracker, suite from the ballet, Op. 71a", "Alinde ("Die Sonne sinkt in's tiefe Meer"), song for voice & piano, D. 904 (Op. 81/1)", "Piano Sonata [No. 2] in B flat minor". Probably much too long for our page titles, but IMSLP's direct links to Amazon scores are influenced by the titles we choose, so it quite literally wouldn't pay us to ignore them completely.

So... can we come up with our own rule for page titles, using the best aspects of these other systems? Probably, but it would have to be easy to understand (for the page-namers and the users alike), and produce titles that don't become too long and unwieldy. I'm the first to admit that the recently-updated IMSLP:Score submission guide/Manual of Style, has its deficiencies, and among the questions we need to address are:

• Which language should be used for titles? — the composer's original / English always preferred / user has the option to choose their own language?
• Should opus numbers and catalogue numbers always be included? — if there is more than one numbering system, which one should take preference?
• Should key signatures always be included?
• Should the instrumentation, where not obvious, be included?
• Should terms like"Concerto for Piano and Orchestra" and "Sonata for Cello and Piano" be shortened to "Piano Concerto" and "Cello Sonata"? If so, what about "Concerto for Horn and Strings", or "Concerto for 2 Pianos".
• Should there be a maximum length (number of characters) for titles? If so, should we try to keep composers' names as short as possible?
• Should composer names include their dates, e.g. "Mozart, Wolfgang Amadeus, 1756-1791"?
• Should there be spaces after "Op.", "No." and "K."?

If we're serious about this, then we need to come up with something that's not only aesthetically pleasing, but which functions on a technical level, and won't need thousands of hours of manual input to introduce. But if it was that easy, it would have happened long ago

PS: @Feldmahler — if a global search and replace is possible, can you do something about the 700 pages containing "monocrome"?

[Boccaccio]

I don't know if it is technically feasible, but after reading about the different possibilities to abbreviate catalogues, set periods, put spaces etc., which furthermore seems to lead to personal preferences that differ among different people, I came up with some alternative idea: Can one generate the titles shown on the composer pages somehow dynamically such that each user can set his own preferences? The basis of this automatically generated titles would then of course have to come from the information on the work pages (as I write this, I note that this might help with problems of which language to choose and whether to include catalogue/opus numbers while it probably will not help to settle the K. vs. KV problem unless each user can specify for each catalogue how to display it).

You see this idea is not very developed at the moment and there are certainly many problems to be solved before it can be realized, but nevertheless I wanted to inform you about my idea.

[pml]

My disorganised thoughts...

Someone's angry ;)

Not really!

Not to address the substantive issues, but I do want to mention this. It is understandable that as musicians we are also perfectionists. However, when dealing with a database such as IMSLP we must also remember that the cost of perfection may be too high. The reason IMSLP is able to grow so large was in part because we emphasized efficiency, sometimes at the cost of imperfections. I would especially mention that systematic and robotic changes such as adding spaces after "Op." or "K." should really not be done by contributors (unless the scope of the change is small, which does not seem to be the case here), as it is an inefficient use of their time. If it becomes enough of a problem, I could, with relatively minimal effort, do a global search and replace and add the spaces.

I think the open model (with the capability to make mistakes) is the better one for IMSLP to follow, especially if one compares say Wikipedia to similar projects that have insisted on too great a degree of scrutiny and professional oversight before action can be taken.

One of the tasks we carry out here is (sometimes arbitrarily) moving page titles to deal with style issues, and I find it ironic that in the past we’ve had complaints to various contributors on account of them breaking the style guide, by following the natural inclination to include the space after a full stop (which is in line with typography) – I remember seeing an acrimonious discussion on a contributor’s page from one of the “movers” to one of the people who was steadfastly following the space after full stop rule.

Also a slight substantive defence. The original reasoning behind using K1 was that all numbers that exist in K1 exist in K6 (and points to the same piece), whereas the reverse is not true. In other words, K1 is the lowest common denominator of all Kochel versions, and hence, even if musicologically not entirely accurate, may be practically more useful (K6 numbers are theoretically supposed be in brackets in the title anyway). However, I am not a musicologist, hence I will defer the decision to people who actually know something about the Kochel catalogue.

The intention was good, since the initial Köchel remains good for many of the works (Mozart’s own written catalogue of his works, started in Vienna, meant that many of the later work numbers are well-substantiated and have never required amendment). Conversely though, not all Mozart works have an assignment in KV^1 – a small number of works only entered the catalogue as of the 2nd or 3rd edition; yet these are tacitly included in the page titles even though they do not derive from KV^1. Admittedly, if Neal Zaslaw ever gets around to publishing the New Köchel (ninth edition!), this would involve minimal change to IMSLP’s page naming, since in his basic plan the works which are no longer chronologically sorted gain an asterisk, and the confusing scheme of extra catalogue numbers is done away with (or so I’m told). The main issue is that because the later catalogue numbers supplant the earlier ones, libraries and publishers that are aware of the more recent numbers often use the 3rd or 6th edition numbers only, which confuses searching for these works.

On Davydov’s talk page I noticed a case of a work in Chopin’s œuvre where the omission of a key signature in a page title had in part delayed the recognition of a mistaken assignment of the work number. This has often been a fault of IMSLP titles, especially for large collections of works with similar titles (Mozart and Haydn are cases in point). I think the argument that the (often arbitrary) numbers assigned by a publisher, or the catalogue number are the best identifiers for these sorts of works, are weak, when numbers are far from being the most memorable feature of a musical work.

I’ve just noticed two concert overtures by Berlioz have had their titles changed to remove the fact that they are independent overtures, which seems particularly odd: the works were explicitly titled as being concert overtures by the composer without reference to having also composed a theatrical work, say incidental music to the play King Lear, or an operatic version of the novel Waverley by Walter Scott, say. Now, perhaps under the scheme cited above a work such as the King Lear Overture might be referred to as “Roi Lear, overture, orchestra, C major (Berlioz, Hector, 1803–1869)”. In actual fact, the work is now referred to in Holoman’s catalogue and the New Berlioz Edition as “Grande ouverture du roi Lear”, so shouldn’t the library be keeping up with the times and amending its titles to follow the most recent scholarship? But they don’t, presumably because the task is so enormous to update their catalogues to keep up with the latest research, which means that titles inherited from libraries are prone to following the path of least resistance. The short titles are also confusing since the assumption might be that when looking at a work with the title like “King Lear”, that the overture is merely an introduction to a more substantial work – when in fact it consists only of the overture, and there is no evidence to suggest the composer intended to provide anything more.

Finally, I did google the AACR2 (Anglo-American Cataloguing Rules, Second Edition) and was amused to notice that according to Wikipedia, “Despite the claim to be 'Anglo-American', the first edition of AACR was published in 1967 in somewhat distinct North American and British texts. The second edition of 1978 unified the two sets of rules (adopting the British spelling 'cataloguing') and brought them in line with the International Standard Bibliographic Description.”

So, even they agreed that it’s “cataloguing”, in the end. :-P

EDITED TO ADD:

I was unfair above to criticise libraries for “following the path of least resistance”. However, I would reiterate there is something drastically wrong with a Uniform title which takes a composer’s work title such as “Grande ouverture de Waverley” and reduces it to “Waverley” by completely discarding half of the information inherent in the title – which describes not merely the work’s content (i.e. program music based on a literary source) but also the form (specifically, a concert overture). PML

[Davydov]

Thanks for your comments, Philip, to which I've responded fully on my talk page.

In the meantime I've been conducting a little experiment by comparing the titles of works by Mozarts as they appear on IMSLP, the Library of Congress, and the website http://www.allmusic.com, and you can see the results on this sandbox page. Sorting on each of the first three columns highlights the differences between the sources, and a lack of consistency of approach in page naming has
which in some cases has led to the same works appearing on two different IMSLP pages (indicated in red). This is why I believe that we need an easy-to-follow scheme that will avoid these kinds of problems, which are wider just than Mozart.

Taking on board some of the comments that have been received, I've added two columns to the right of the table suggesting new methods of dealing with the titles. These include extra spaces after "K." and "No.", inclusion of the keys for works that don't have distinctive titles, and (specific to Mozart) the numbers from the 6th Köchel catalogue given in brackets, where these are different from the first edition. The dual numbering is a practice used by Allmusic, although I corrected several wrong numbers, and there may be others that have slipped through.

The second suggested title type differs from the first only in that some terms are more fully spelled out, "Piano Concerto" (type 1) is shown as "Concerto for Piano and Orchestra" (type 2), while "Violin Sonata" is "Sonata for Violin and Piano". This has some effect on the way that the works will be sorted on the category page.

I didn't include "(Mozart, Wolfgang Amadeus)" after the suggested title types just to save space in the table, but obviously this will be a factor in the overall title length. The titles generated by Allmusic are impractically long for us to adopt as they stand, but our categorization system allows our users to sort works by instrumentation and type anyway. Some of the Library of Congress's choices of titles are a little surprising, and only a minority of works appear in their listings anyway.

The only other point I should probably make is that our page titles serve an index to the works on IMSLP, and they shouldn't have to correspond to the exact form used by the composer or on the printed editions. There are spaces on the work pages themselves to include fuller versions of the titles, with nicknames and alternatives in other languages. Instead, consideration could be given to how they appear in the alphabetical lists of works generated by the category walker and in the composer categories.

Anyway, there are probably many better naming schemes than the ones I've suggested, but hopefully this will help set the ball rolling. Comments and constructive criticism are welcome....

[pml]

Thanks Davydov.

The Library of Congess titles are interesting in their inconsistencies – the majority of secular vocal works and some sacred works lose any reference to their Köchel numbers, as do variations works; some of the capitalisation, work categories and details of instrumentation are haphazard; and unlike allmusic.com, no identification of which version of the Köchel catalogue is employed in a title (any that use a letter of the alphabet cannot derive from KV^1). I like that allmusic’s listing includes the vocal incipit for small vocal works, though this would unduly lengthen titles (LoC by contrast, dispenses with titles for recitatives and sorts by the main aria text, for the combination recitative and aria works!)

Looking at the allmusic.com list, many of the works have useful identifying features that are implicit in our works tagging system; these therefore probably don’t need to be copied over to page titling. I'm still inclined to think the major improvement would be the inclusion of keys for a number of the work types: although these are not as useful as the catalogue numbers to uniquely identify specific works, they are far less arbitrary than the work numbers assigned for individual categories such as the 41 symphonies (2 of which are not by Mozart; and various left-overs range up to “No. 55” or fail to get a number) or the 4 horn concertos (where the first numbered concerto is in reality the last). The Library of Congress notably drops the ball when it comes to Horn concerto no. 1.

I’m also inclined to think some more work types demand the instrumentation as part of the title than we presently permit. Having been doing some work compiling a sortable list of works for Mendelssohn based on the new MWV catalogue, I’ve noticed that we have a variety of works which appear in the Mendelssohn category simply as:
# Lieder, Op. #
In the MWV catalogue, these would fall into 4 broad categories, (K) lieder for voice and piano; (J) lieder for two voices and piano; (F) lieder for mixed-voice choir a cappella; and (G) lieder for male-voice choir a cappella. There is no way of discerning which ones are which without going to the category walker, and although this is one of the tasks for which the category walker was designed for, I am also of the feeling that the page titles are themselves under-descriptive.

To return to the Mozart list, likewise I see a number of works with unusual instrumentation (e.g. mechanical organ or glass harmonica anyone?) which again would turn up in the category walker, but would otherwise go completely unnoticed when looking at the IMSLP titles: is this a good thing, if visitors to the site may not immediately think of using the category walker? (In fact, I imagine novice visitors are almost certain to look for a search field, rather than using the category walker.)

Regards, PML

[KGill]

Maybe an alternate solution would be to make the Category Walker more linked-to and more intuitive to use. Most people probably wouldn't first look to it because it's hidden in the links at the top of the composer page, which can be rather far to get to if you're browsing a category with hundreds of members. (Also, the names of the commands on the CW itself should really be changed to something more self-explanatory.) Could there be some sort of JavaScript thing that would make it so that if you hover over a link to a work page, you could see how it's tagged?
Or, pursuing the search idea, perhaps the search function could automatically look in the tagging system to see if anything matches, and if so provide relevant links to the CW. That could get complicated, though...