lazykoblog   occasional musings about knowledge organization

Ralph Papakhian, head of technical services in the music library at Indiana University, passed away last winter. Ralph was an inspiration to many people over the years. Back in the early days (ah, youth), when I was the head of music cataloging at Illinois, he and I had something of a music cataloging cabal going on, which is a fancy way of saying we helped each other out. We conducted a huge study of music in the then-new OCLC, which was a good survey of the conditions nascent automated music cataloging operations had to deal with in those days.* Here is a photo (I knew I had this someplace!) of my graduate assistant Constance Wernersbach (Connie) and Gretchen my beloved cat who graced my life from 1973 until 1985. Connie was sorting OCLC printouts and calculating some variable, and Gretchen was helping. You can tell, just look at her ears all perked up there! That’s the floor of our living room in Urbana, which as you can see, was then under renovation … I finished that woodwork the night Lech Walensa took over the government of Poland. Oh my, how the world changes. Well, notice Connie is using an old-fashioned calculator; this was before home computers folks! (That’s Brad’s gateleg table, for those of you who’ve been to visit us, its still sitting in our front hall; he got it at a flea market in New Orleans.) Sorry … brief moment there ….

I remember the study chiefly for being the impetus for my doctorate. It was such a difficult problem to sort through all of the data and it was very frustrating to discover one had gathered data that wouldn’t be useful in the end. I said that famously to Arlene Taylor, and she said “come to Chicago and learn how to do it right,” by which she meant learn research methods for real, and I did, and the rest is history, I suppose.

But we won an award for that study and we dined out on it for awhile. When Ralph passed away, I was asked to contribute a paper to a Festschrift, and the first thing I thought of was his study of personal name frequency in music catalogs. I will eventually write an essay about the theoretical value of that study, which reaches far beyond the obvious influence of the stated findings.

Be that as it may, I was reminded also of the OCLC study, and a little bird suggested I should perhaps replicate that study. I thought that was a good idea, and as I was about to teach music cataloging at UWM, I also thought using it as the backdrop for the course might be a good thing for those students.** Plus, it would bring them within Ralph’s orbit, however briefly. So, instead of having those students write term papers I’ve divided them into three different sets of small groups over the course of the semester and they’ve all been working on a replication of sorts. Eventually we will all be coauthors of an article that will appear in the Papakhian Festschrift.

Of course, we can’t exactly replicate the original study, nor would we want to. The original study included analysis of the music holdings by comparing OCLC to the Basic Music Library, to see whether one could (at that time) expect to find standard repertory. There’s clearly no point of replicating that–even in 1981 OCLC had copy for 91.5% of the essential scores and books. But we did replicate the analysis of timeliness for new publications. We searched lists of recently published books on music, music scores, and musical sound recordings from December 2009 and March 2010. Interestingly, while the books and recordings were almost all present, only 70% of the scores were found. So that’s just a small preliminary indication about rapidity of coverage. And I guess it also tells us what sorts of things an original music cataloger is likely to need to be working on these days.

We also received random samples of bibliographic records for scores and recordings from OCLC (thanks to Ed O’Neill and the OCLC Research Division). We searched all of those records and took some basic bibliographic “demographics.” Here are some tidbits from the preliminary analysis:

Scores: 44% AACR2 descriptions, 10% have pre-AACR2 ISBD descriptions, 24% are full level, 50% are M level (less than full, from batch-loading), 94% have Source: d (not LC), dates entered are consistently even from 1972 until 2002, and, the rate of record replacement is constant over time.

Sound recordings: 19% are full level, 48% are M level, 87% have Source: d, dates entered fall into two large clusters early and late, but otherwise are consistent.

(Results, such as they are, are drawn with 95% confidence +/- 5%.) Interestingly, only about half contain cataloging matched to current standards. That is another indicator of the kind of work music cataloging divisions can look forward to.

In the original study, because both Ralph and I ran cataloging departments, we included a sample of workforms from our two divisions, so we could comment on the sorts of changes we were making. For the replication we located the bibliographic records that had been changed the most (in all cases, more than 7 times, and as much as 22 times). Those are being analyzed independently by the students. When they’ve turned in their results I’ll post a summary here.

Cat-lovers among you might be wondering about Gretchen, so here she is, basking in the glory of the award she won for this research (okay, it was the afternoon sun in my study on High Street in Urbana). Still, she was always majestic.

(*Oops, sorry–the original study was Smiraglia, Richard P. and Papakhian, Arsen R. 1981. Music in the OCLC Online Union Catalog: a review. Notes 38: 267-74.)

(**We’ll have to ask the students what they think, but pedagogically, I’ve used the study to drive blogging exercises all through the course. I opened with the tale of how I dropped an entire drawer of music shelflist, and had to refile it, and in reading something like 1100 cards learned a lot. So each of these students has analyzed several hundred bibliographic records, in addition to the 7 they’ve created. I think it’s a good learning opportunity.)

Forgot to post a summary. The research is ongoing, of course; here are some basic results:

Most of the recently published works were found, which means coverage in WorldCat is excellent. That’s quite a change from the early days. The lowest rate of coverage, less than 70%, was for scores. Searchers’ notes raise some interesting questions, including the noise created by apparent duplicates (these are caused by multiple batch-loadings from different sources), the advantage of having unique item numbers to search with (instead of name-title combinations). Of 306 records for scores: 44% AACR2 descriptions, 10% have pre-AACR2 ISBD descriptions, 24% are full level, 50% are M level (less than full, from batch-loading), 94% have Source: d (not LC), dates entered are consistently even from 1972 until 2002, and, the rate of record replacement is constant over time. Of 309 records for sound recordings: 19% are full level, 48% are M level, 87% have Source: d, dates entered fall into two large clusters early and late, but otherwise are consistent.