by Denise Soufi (Middle Eastern Cataloger, The University of North Carolina at Chapel Hill)
and Nanako Thomas (Catalog Librarian for Japanese Language Resources, Duke University)
and Natalie Sommerville (Team Lead, Monographic Original Cataloging, Duke University)
The cataloging of non-Roman script materials poses special challenges that usually require a cataloger to possess the requisite language expertise. When an institution possesses many of these materials but is unable to hire the needed cataloger, a common solution is to outsource the cataloging to a vendor. In the case of Duke University and the University of North Carolina at Chapel Hill (UNC), however, the question was asked: would it be more cost effective and efficient to leverage the expertise of our own non-Roman script catalogers through an exchange of materials? As two of four member institutions of the Triangle Research Libraries Network (TRLN), this type of project falls well within the scope of TRLN’s mission “to marshal the financial, human, and information resources of their research libraries through cooperative efforts in order to create a rich and unparalleled knowledge environment that furthers the universities’ teaching, research, and service missions.” (“About.” TRLN, Triangle Research Libraries Network, https://www.trln.org/about/).
In October 2017, the TRLN Collections Council approved the proposal for a cooperative cataloging pilot between Duke and UNC as well as the formation of a task group to implement the pilot. The Cooperative Cataloging Pilot Task Group was charged with examining the logistics, workflows, efficiency, cost and benefits of TRLN cooperative cataloging compared to outsourcing, and Nanako Thomas and Denise Soufi were named the project managers at Duke University and UNC, respectively.
With input from our supervisors and department heads, we first prepared a set of guidelines outlining cataloging procedures and workflows in a project charter and a cooperative cataloging agreement. In the charter, we defined the project’s objective as assessing the efficiency and cost-effectiveness of TRLN cooperative cataloging as compared to outsourcing, while factoring in the quality of the resultant bibliographic records. The project scope was limited to 100 titles sent by each institution for cataloging. UNC agreed to send 100 Japanese-language titles to Duke while Duke agreed to send 100 Arabic-language titles to UNC. We set deadlines for deliverables and milestones, the most important being that the books would be sent out by the beginning of April 2018, the cataloging would be completed by June 22, and all records would be imported by the end of June. We also outlined our team members, their roles in the project and the estimated time they would spend.
In the agreement, we set down the details of how the project would be carried out. We decided to ship our books using the library truck that circulates among the four TRLN libraries. We barcoded the books prior to shipment and tracked their location using local procedures. We agreed to catalog according to our own institutional standards, using the Worldcat authorization provided by the owning library’s institution and following the BIBCO Standard Record. UNC staff set up Google spreadsheets for recording all cataloging statistics. For each record, Nanako and Denise recorded the type of cataloging along with the Worldcat number and barcode. For type of cataloging our options were New, a record created from scratch; Enhanced Copy, a record created by another institution that we edited in Worldcat; or Copy, a record created by another institution that required no edits. For Duke, Denise recorded the barcode in the record and saved it to Duke’s online save file in accordance with Duke’s needs for end-processing. The catalogers also tracked what we called peripheral cataloging, which we defined as any cataloging tasks that would not normally be provided by an outsourcing vendor, such as creating authority records and Program for Cooperative Cataloging (PCC) level bibliographic records. In addition to recording the type of cataloging, Nanako and Denise kept track of the time spent cataloging, recording the total amount of time in hours spent creating, enhancing and copy cataloging bibliographic records. They recorded separately the amount of time in hours spent on peripheral cataloging work.
Although we assumed a higher level of cataloging quality that we could not expect from a vendor, we thought it was important to perform some quality control in order to compare our results with vendor records. We agreed to take a random sample of one-third of the records to check for the accuracy and completeness of the subjects, classification and MARC coding elements.
In the agreement we also gave a brief outline of the local procedures we would follow for importing the records. Natalie Sommerville was able to leverage an existing service in Duke’s integrated library system (ILS) to load the records cataloged by Denise; while at UNC Denise and various team members developed a workflow to batch import the records cataloged by Nanako into UNC’s ILS. Thanks to our supervisors’ prioritization of this project, Nanako and Denise were able to complete the cataloging by the end of May, and Natalie and Denise were able to complete record loading by the end of the June deadline.
The following table includes the raw data listing the number of titles cataloged by Nanako and Denise and the number of hours spent on cataloging. Of significance is the fact that Nanako spent nearly twice as much time cataloging as Denise did. This is due to two factors. First, three-quarters of the Japanese-language titles required original records, whereas more than half of the Arabic-language titles were copy or enhanced copy. Second, about 20% of the Japanese-language titles were in an older, difficult-to-read script requiring more cataloging time. Additionally, in the process of cataloging Denise recorded four extra titles. One was simply an extra book. There was also a group of four books that had presumably been sent as a four-volume monographic set; however, in Worldcat each volume had been cataloged separately as part of a series, and, after consulting with Nanako, Denise used those records instead of creating a new one.
Number of Titles Cataloged and
Number of Hours Spent Cataloging
Next is a table which summarizes the time spent on non-cataloging tasks, some of which were performed by other colleagues; these tasks include book selection, tracking and packing/unpacking, quality control and importing records into the local ILS. Again, there are a few areas of difference. First, UNC spent less time on book selection because the Chinese language cataloger had sorted the Japanese-language books requiring original cataloging in the previous year. So, it was straightforward for her to select books to send to Duke. Second, UNC’s time spent on importing records is much higher than Duke’s because it includes the development and documentation of a workflow to batch import records; in the future our numbers should be similar.
Time, in Hours, Spent on Non-Cataloging Tasks
*Includes time spent on development and documentation
In order to compare with vendor outsourcing costs, Natalie and Denise calculated the cataloging cost per title based on the hours logged by the pilot project catalogers and the average cataloger’s salary, including benefits, at their respective institutions. They also included time spent on other tasks that would not be performed for vendor cataloging, namely tracking, packing and unpacking books. Time spent on importing records was not included since similar work is involved with importing vendor-created records.
For comparison, Natalie supplied data for the average cost per title paid to Duke’s non-Roman language outsourcing vendors based on costs during the years leading up to the pilot project, as well as Duke’s internal costs for scanning and quality control. The cataloging cost is for original cataloging, which, as defined by Duke, consists of both new records and enhanced copy records. The scanning cost is based on the time spent scanning a certain set of pages from the book (generally the cover, spine, title page, title page verso, colophon, table of contents and the first few pages of the introduction), which are then sent to the vendor for use as the basis for cataloging.
The following chart illustrates a comparison of costs for vendor cataloging, both with and without quality control, and cataloging by Duke and UNC, both with and without added peripheral cataloging work and quality control.
Vendor “cost per title plus” includes quality control.
Duke/UNC “cost per title plus” includes peripheral
cataloging and quality control.
Several caveats must be noted in order to correctly interpret the data. First, the fact that Duke’s cost is dramatically higher than UNC’s is due not only to the slightly higher average salary, but also to the fact that Nanako spent more time cataloging due to the special needs of the Japanese books mentioned previously, particularly the fact that most required new records. While Duke’s cost appears to be on par with the vendor cost, the latter is based on a more even mix of new and enhanced copy titles. Second, it must be noted that the cost per title spent on quality control for Duke– and UNC-created records was minimal (approximately $.50/title vs. $2.38/title for vendor records) due to the high quality of records created at both institutions.
Duke aims to perform systematic quality checks on vendor records for approximately two months. Therefore, for long-term projections of cost, we thought it fruitful to compare the cost of vendor records without the cost of quality control with the average cost of Duke/UNC records with peripheral cataloging work but without quality control.
Not only is the Duke-UNC average cost significantly less than the vendor cost, it includes the bonus of peripheral cataloging work such as NACO and PCC records.
Based on the cost comparison alone, we recommend pursuing consortial cooperative cataloging projects when there is language expertise available in the network. In addition to lower costs, there are added benefits that cannot be provided by a vendor. First, the quality of cataloging is higher and more reliable, due to both the expertise of the catalogers as well as the availability of the entire book rather than a few scanned pages. Second, the catalogers were able to provide the bonus of NACO and PCC records, and both were willing to submit classification and subject proposals had they been necessary. This extra work is credited to the owning institution and enriches records that will likely be used by other libraries.
However, several factors should be considered before initiating a cooperative cataloging project. First, the fact that the books are shipped out means there is potential for damage or loss. We minimized this risk by using the truck that routinely delivers library materials among the TRLN institutions. Second, urgent local needs may take priority over the cooperative cataloging work, causing unwanted delays. Strong support from supervisors to prioritize the project helped us to meet our deadlines and ensure timely completion.
Third, a project cataloger may decide to leave his/her position, resulting in the loss of a language expert. There is, unfortunately, little that can be done to allay this risk. Lastly, the amount of time spent cataloging may be unbalanced between the two institutions; while the number of new records to be created can be ascertained, it is more difficult to determine the difficulty of cataloging the materials. Although we did not try to address the imbalance we experienced, in the future it may be possible for the institution that is spending less time cataloging to take on extra materials to even out the time commitment by each institution.
We highly recommend that any institutions undertaking a cooperative cataloging project create a project charter as well as an agreement that outlines detailed cataloging procedures, including any local needs, and a workflow for shipping and tracking materials and tracking data. We also recommend framing cooperative cataloging work as a project that can be prioritized with sensible deadlines. The framing should include an in-depth discussion of scope and sustainability within the local context. For example, depending on the size of the backlog and the available staff, some institutions may choose to limit the scope and duration of the project to address cataloging of high-priority materials. Conversely, other institutions may be able to use this model on an on-going basis to address a backlog or to prevent its growth. Duke and UNC look for opportunities to use this cooperative cataloging model for size- and time-limited projects, but currently do not have a project in the pipeline. In either case, the model used in the pilot is flexible enough to be adapted accordingly.