Content area
Online Computer Library Center's (OCLC) decision to load Library of Congress (LC) preliminary cataloging for monographs into the online union catalog resulted in the addition of a significant number of these records to the database over the 15-month period before the project was suspended in March 1991. A description is given of the problems experienced in dealing with these items by the cataloging staff of a medium-sized research library. The data are from a project at Morris Library at Southern Illinois University. The problems illustrate why the project came to an abrupt end. Twenty percent of initial searches yielded Level 5 records that duplicated member input full (or minimal-level) cataloging. Catalogers discovered that items with Level 5 records required almost as much time to catalog as those without copy. In addition, the heads of cataloging at 57 Association of Research Libraries institutions in the US were asked what they thought of Level 5 records. Their responses revealed that the records created more work, especially for higher level staff.
OCLC users were notified in February 1990 (via a logon message) that the previous month the Library of Congress (LC) had begun to include Levels monographic records as part of its tape-distribution service. Since it had little information as to their nature, OCLC initially processed these as "normal records" until members complained that some of them had replaced member copy. More than 900 records were affected out of the first 6,800 loaded. OCLC quickly revised its tape-load procedures so that a Level 5 record was added only if matching member copy was not in the database; if a match occurred the Library of Congress' DLC" holding symbol was added to the existing record.
In a network newsletter issued shortly thereafter, OCLC acknowledged member complaints and warned that "no assumptions can be made about whether the Level 5! record is correct...no authority work has been done, nor subject analysis provided."(1) These brief records had no call numbers, no added entries, and no notes; they occasionally lacked series statements, and some contained typographic errors (see figure 1) (Figure 1 omitted) The descriptive cataloging elements for Level 5 monographic records are even briefer than those required by National Bibliographic Record Books for standard minimal-level cataloging. They include only the following data:
1xx: first personal name on the title page 245: title and statement of responsibility in full 250: full edition statement 260: first place and publisher, one date except for multipart items 300: simplified description 4xx: series in full as it appears on the piece, whether traced or not 5xx: acquisitions date 020: first or most appropriate ISBN 010: LCCN supplied 050: IN PROCESS note supplied Fixed fields: language, country of publication, priority(2)
Responding to those who might be concerned about the prospect of using such bare-bones records for copy cataloging, OCLC promised to monitor the situation and to consider "additional changes to record processing." Meanwhile, members had the option of upgrading Level 5 copy to K-or I-level standards.
LC's plan to share information about its recent acquisitions via its tape-distribution service dated back a number of years. The Technical Services Directors of Large Research Libraries (TSDLRL), which met at the ALA Annual Conference in Dallas in 1984, reacted favorably to LC's proposal to distribute in-process data.(3) This proposal came at a time when some research libraries also were beginning to consider standardized minimal-level record creation as an effective means of attacking cataloging backlogs and promoting resource sharing.(4) A 1985 report prepared for the Association of Research Libraries (ARL) Committee on Bibliographic Control explicitly recommended inclusion of LC in-process data in the national utilities. The expectation was that this information would decrease duplicate cataloging efforts by indicating the relative priority given to a work by LC. The report's authors also believed that the appearance of in-process data would allow libraries to "predict when LC copy may be forthcoming.(5)
In the late 1980s LC reevaluated its methods for assigning cataloging priority levels to incoming materials (including monographs, serials, and microforms). The revised guidelines were published in Cataloging Service Bulletin (CSB) in the 1991 winter issue. LC reemphasized an item's intrinsic research value, i.e., its content, over any former preference for language or source of material. LC also provided detailed information on the terms used to evaluate incoming materials and to place them in one of four cataloging priority levels. The decision in each case was to be based upon a careful balance of factors specific to the item, as well as LC's internal needs. CSB 51 also featured a brief description of LC's preliminary, or in-process, records. LC cautioned, as OCLC had, that "No assumptions can be made as to whether headings in the record reflect established forms, nor whether the record meets the National Level Bibliographic Record minimal-level cataloging specifications."(6) Neither article specified when more complete cataloging would be supplied. LC expected that "once the revised priority assignment procedures have been operational for a reasonable time, the Library will periodically provide figures on how long items in each priority category take to go through the series of required processing steps."(7) Its plan was to catalog completely items in priority levels 1-3; those assigned to Priority 4 and titles remaining in its cataloging arrearage for three years or more would be candidates for minimal-level cataloging.
STUDY
Because previously loaded preliminary records will remain in the OCLC database for the foreseeable future, it may be useful to describe the problems experienced in dealing with these items by the cataloging staff of a medium-sized research library. The data presented here are from a project at Morris Library at Southern Illinois University, Carbondale, designed by the authors.
From July through December 1990, we reviewed all Level 5 records identified during OCLC searches of monographic titles entering the cataloging department. We recorded the OCLC record number and the item's priority level, general subject area, language, place of publication, and number of holdings. Following the test period, we searched these records again to see whether they had been upgraded, merged with a duplicate record, or replaced by LC complete cataloging. We also noted any holding changes.
Of ninety-two records encountered during the six months, thirty-nine (42 percent) were eventually upgraded by Morris Library catalogers.(8) Other OCLC members upgraded twenty-eight records (30 percent). Fourteen records (15 percent) had been replaced by full DLC copy, and duplicate record mergers had occurred four times. Seven items (8 percent) had not been upgraded when the records were researched in February 1991. The six Priority 4 records replaced by LC had received complete cataloging. Rather surprisingly, LC had replaced twice as many Priority 4 records as it had Priority 3s.
Using the before-and-after versions of the records, we compiled a profile of the group (see figure 2 and table 1). (Figure 2 omitted) (Table 1 omitted) More than one-third of the records had an original priority assignment of Level 4; 10 percent had been assigned Priority 2. The "age" of the records at the time we first encountered them can be summarized as follows: Priority 2, two weeks to nine months; Priority 3, two weeks to twenty-eight months; Priority 4, two weeks to eight months. Since "date of entry" in the fixed field was sometimes as early as 1988, that element obviously referred to the date LC had created the record rather than when it entered the OCLC database.(9)
We assigned each item to one of three subject areas: social studies (sixty-eight items), science (twelve items), or humanities (twelve items). These were subsequently matched with the priority levels assigned by LC. We found that most humanities and social studies items were placed in Priorities 3 and 4, while science items were in Priorities 2 and 3 exclusively (see table 2). (Table 2 omitted)
Items in foreign languages (chiefly German) accounted for 38 percent of the entire sample. No Priority 2 items were in a foreign language. Out of the thirty-five foreign-language items, 86 percent were Priority 4, 14 percent were Priority 3. While only a third of the Priority 2 records were for items with a non-US. imprint, foreign imprints accounted for nearly half of Priority 3 items and for almost all of those in Priority 4. Foreign imprints made up 65 percent of the entire sample (see figure 3). (Figure 3 omitted)
As noted earlier, the authors compared the number of holdings on a record at the beginning and end of the project. The average number of OCLC holdings at the point of receipt was 3.5; following the test period the average had risen to 20.6.
DISCUSSION
The problems our catalogers had with these records may help to illustrate why this experiment came to an abrupt end. For instance, 20 percent of our initial searches yielded Level 5 records that duplicated member input full (or minimal-level) cataloging. Previously, when the searching staff found duplicate records they were instructed to select DLC copy. Now it was necessary to have all records printed if one of them displayed encoding Level 5. This additional searching was the first delay in cataloging of these items,
The library received fifteen to twenty items with preliminary copy each month. It did not seem practical to defer cataloging until LC supplied full records, nor did we wish to impede user access for an indefinite period. Since the cataloging department had already developed procedures for upgrading minimal-level records, Levels records were integrated into this work flow.(10) This meant that after searching, Level 5 items were immediately routed to the original cataloging unit.
Catalogers discovered that items with Level 5 records required almost as much time to catalog as those without copy. In addition to authority work performed for name, subject, or series entries added to the record, catalogers had to verify any existing main entries, because these headings frequently were not in AACR2 form. They also assigned a Dewey call number, added notes, and revised the fixed field as appropriate. The LC priority designation and the message "in process (online)" were deleted. All this took considerable time for an item with ostensible cataloging copy. Additional effort was required if duplicate records existed. Catalogers had the option of either approving member input copy and returning the item to the copy cataloging unit or upgrading the DLC record if member input copy lacked access points or other significant information. OCLC was notified of the existence of duplicate records in all cases.
ARL SURVEY
Since Morris Library's catalogers generally regarded Level 5 records as a nuisance, we were curious to know what catalogers in other research libraries thought of them. As part of a survey on minimal-level cataloging policies distributed during the fall of 1990, the authors asked heads of cataloging at ninety-nine ARL institutions in the United States to estimate how many LC in-process records were encountered monthly, how they were edited, and whether they were viewed as a problem (see appendix A). (Appendix A omitted) Since this study focuses on Level 5 records used for cataloging in OCLC, we excluded twenty-two institutions known to be Research Libraries Group (RLG) members. This left a potential pool of seventy-seven ARL/OCLC libraries. Fifty-seven (74 percent) of these responded to our survey. One of the fifty-seven responses was incomplete and was excluded.
The responses showed that 60 percent of the OCLC libraries had formal policies on the creation and use of minimal-level cataloging; 50 percent used some form of minimal-level cataloging themselves. Eighty-four percent indicated that their libraries upgraded minimal or preliminary records in OCLC; of this subgroup 80 percent reviewed the entire record beforehand, rather than selectively checking and modifying fields. Level 5s were upgraded by 79 percent of the respondents. Notes, subject headings, and added entries were the fields added most often to LC preliminary records. The in process (online)" message was usually not deleted by the editing library. Sixteen libraries (29 percent) reported seeing more than fifty Level 5 records a month, and several said they encountered hundreds per month.
Various methods of editing and/or upgrading preliminary or minimal-level records were cited. A few libraries edited the record without upgrading; twenty-five (45 percent) upgraded in compliance with OCLC Bibliographic Input Standards, while fourteen (25 percent) combined or varied these methods, depending on the nature of the material being cataloged. Professional catalogers processed these materials 50 percent of the time. Two percent of the libraries used paraprofessionals exclusively, while 48 percent used a combination of both professionals and paraprofessionals, depending on the type of material.
Respondents identified many of the same drawbacks in these records that our catalogers had discovered: they created more work, especially for higher-level staff; they resulted in the creation of backlogs and disruption of the work flow; DLC records could no longer be considered authoritative as a matter of course. Some libraries reported that they did not have the time to upgrade all the Level 5s they encountered. Others held items for as long as a year in anticipation of a complete record. Opinions on Level 5 records were very negative overall. Only three respondents commented favorably on them, expressing the viewpoint that "something is better than nothing."
CONCLUSIONS
Whatever their contribution to collection development and resource sharing might have been, the value of LC in-process records as sources of cataloging copy appears to have been negligible, chiefly because of their duplicative nature and the special routines needed to handle them. Their inclusion in a national database subverted the long-held notion that DLC records could be used with little editing.
Several observations may be made concerning the sample records that we studied. We found a preponderance of foreign language and foreign imprints had been assigned to the lowest cataloging priority, despite LC guidelines that no longer discriminate in favor of English language or domestic publications. The substantial growth in holdings per record during the test period indicates that research libraries are likely to acquire and require cataloging copy for many items considered by LC to have low research value and need. If a particular library cannot afford the time or staff to upgrade or fully edit Level 5 records, the ultimate result will be less access in an online catalog.
In April 1991 OCLC announced that it had stopped loading LC in-process monographic records because of widespread complaints about their quality and because of continuing technical difficulties in matching them with existing member copy. At that point approximately 171,000 Level 5 monographic records remained in the database.(11) Since many had multiple holdings attached, they could not be deleted. OCLC expected their number to diminish gradually as they were upgraded by member libraries, merged, or replaced with more complete LC copy. An experiment designed to foster resource sharing had foundered because of its incompatibility with the needs of network members for a standardized and reliable cataloging source.
REFERENCES AND NOTES
1. "Encoding Level 5 Records in the OLUC," Bits and Pieces, no.116:2-3 (Mar. 1990). 2. "Simplified In-process Record Specifications," addendum to Sept. 13, 1990, memorandum from Duane E. Webster, Association of Research Libraries, to directors of ARL Libraries, 3. Colleen Bednar, Library Exchange, RTSD Newsletter 9, no.6:75-76 (1984). 4. Michael Gorman, "Report on the Technical Services Directors of Large Research Libraries Survey on Minimal Level Cataloging," Information Technology and Libraries 3, no.4:382-84 (Dec. 1984); Nancy E. Douglas and Shirley Leung, "Use of the Full MARC Record: "Myth and Reality," in Academic Libraries: Myths and Realities, ed. Suzanne C. Dodson and Gary L. Menges (Chicago: Association of College and Research Libraries, 1984), p.177-82. 5. Carol A. Mandel and Susan F. Rhee, "Shared Cataloging: Some Remaining Issues," Cataloging & Classification Quarterly 7, no.2:33,37 (Winter 1986). 6. Cataloging Service Bulletin 51:50 (Winter 1991). 7. Ibid., p.4. 8. This number includes two "reupgraded" records in which it was apparent that the in-process record had bumped member copy. It was necessary, among other things, to reverify and correct the main entries for these items. 9. This is with the exception of the items referred to in note 8. 10. Mary Anne Fox and Barbara G. Preece, "Upgrading Minimal Level Monographic Records: A Study and Conclusions," Technical Services Quarterly 8, no.4:35 (1991). 11. Personal communication with OCLC Office of Quality Control; ILLINET/OCLC Service, Information Bulletin, no, 230:2 (Apr. 15, 1991).
Barbara C. Preece is Head, Online Catalog Maintenance, Cataloging Department, Morris Library, Southern Illinois University, Carbondale, and Mary Anne Fox is Head, Original and Special Materials Cataloging, Cataloging Department, Morris Library, Southern Illinois University, Carbondale.
Copyright American Library Association Mar 1992