College and Research Libraries Automation in U.S. East Asian Libraries in the United States: A Review and Assessment Sarah Su-erh Elman The RUN CJK system has been in use for more than six years and the OCLC CJK350 system for more than three years. Have they brought the holdings of East Asian collections into the bibliographic mainstream, as they were expected to do? This paper briefly examines the history and special features of these two CJK systems. A national survey also was conducted among academic and research member libraries in the United States to learn how these systems have been incorporated into their local automated library systems. Findings of the survey indicate that the advantages of the two systems are not fully delivered to the general user because local automated library systems still are incapable of processing and displaying non-Roman languages. More efforts are needed to develop this capability in order to integrate fully non-Roman collections into the general collection. rom the experimental main- frame systems in the 1960s to the various mini- and micro- computer-based, integrated, and stand-alone systems that are avail- able today, the development of auto- mated library systems has come a long way and has accomplished a great deal.1 However, in the first two decades of their development, most attention was focused on the automatic processing of English- and other Roman-language materials. The ideo a h·c ature of East Asian language , esp_eciall C mese, made East A-sian--li-braries- on yo servers of library automation until the develop- ment of the Research Library Informa- tion Network-Chinese, Japanese, and Korean (RLIN CJK) system in 1983 and the Online Computer Library cemer- Chinese, Japanese, and Korean (OCLC CJK350) system in ~The implementa- tion of these two systems has implications for East Asian libraries, in particular, and scholarly communities, in general. They not only assist member libraries in the automation of .CJK materials, but foster national and international information exchanges and resource sharing of these materials. As Research Libraries Group (RLG) vice president John W. Haegar says, ''The implementation of CJK placed a marker at the end of the old world and the beginning of a new one for East Asian libraries, and broug-ht their holdingsl nto--the bibliographic main- streaJ.ll." 2 After several years of implementation, it is useful to examine the current use of these two CJK systems in the biblio- Sarah Su-erh Elman is a 1990 graduate of the Graduate School of Library and Information Science, University of California, Los Angeles, California. Currently, she is traveling abroad and working for both the Modern History Institute Library, Academia Sinica, and the National Central Library in Taiwan. 559 560 College & Research Libraries graphic community as a whole to see whether they are fully utilized by the institutions that have purchased them. The major purpose of this research proj- ect is to learn how academic and re- search libraries in the United States incorporate the RLIN CJK and the OCLC CJK350 systems into their local systems. I have collected for analysis information re- lated to functions, merits, and weaknesses of the two systems, as well as to their relationship with local bibliographic systems. Three similar surveys have been per- formed, one each by Karen T. Wei (1986), Wen-kai Kung (1986), and Hee-Jung Lee (1985).3 However, these surveys were conducted before the OCLC CJK350 sys- tem was implemented. This project up- dates their findings. RLIN CJK AND OCLC CJK350 Much has been written about the his- tory of the development and the charac- teristics of these two systems by authors such as John W. Haegar, Jay Lee, Alan Tucker, Andrew Wang, and Karen T. Wei.4 I will not repeat their efforts, but only mention some of the important fea- tures of these systems. RLINCJK The hardware of the first-generation RLIN CJK system is a CJK cluster com- posed of a cluster controller, one to four CJK terminals, and an optional printer. The RUN CJK uses a specially designed keyboard with a total of 179 keys, of which ten are control keys, 133 are char- acter-composing keys, and 36 are func- tion keys. 5 In September 1988, RLG introduced a second-generation RLIN CJK terminal, the MultiScript Worksta- tion (MSW), which is an IBM PC/ AT- class machine configured to work as a standard RUN terminal, but that also supports the processing of bibliographic information in Chinese, Japanese, Ko- rean, Hebrew, and Cyrillic scripts.6 The new MSW not only has improved hard- ware and software, but costs much less than its predecessor (approximately $6,000, compared with $27,000 for the previous CJK cluster). November 1991 RUN CJK's input method is based on a character-composing system for both Chinese and Korean characters. Users have to key in the character components in the correct keystroke sequence to re- trieve desired characters. Familiarity with the writing sequence of characters is important to operate the RUN CJK efficiently. RUN CJK provides various ac- cess points, such as commonly used stan- dard numbers and codes, author, title, corporate body, and subject, and several display formats. Truncation, Boolean logic operators, and qualifiers can also be used in searching. Truncation, Boolean logic operators, and qualifiers can also be used in searching. Since the Library of Congress entered the first CJK record into RUN on Sep- tember 12, 1983, the total number of CJK records in RUN has increased rapidly and has exceeded 400,000 as of spring of 1990.7 Currently, the Library of Congress and 23 academic and research libraries in the United States use the RUN CJK system. OCLCCJK350 Unlike RUN, which designed a com- pletely new system for CJK materials, OCLC modified its existing IBM PC/XT configurated OCLC M300 workstation to accommodate the capability of pro- cessing and displaying CJK characters. When linked with the OCLC online sys- tem, the workstation is capable of process- ing information in Chinese, Japanese, Korean, English, French, German, Malay, Spanish, Vietnamese, and other Roman- alphabet languages. The workstation also can be used as a stand-alone micro- computer.8 OCLC CJK350 provides both charac- ter-based and pronunciation-based input methods. However, its character- based input method, Tsang-chieh, is dif- ferent from that of RLIN. Instead of pressing the character component keys directly to retrieve desired characters, users key in, in the correct sequence, the Roman alphabets that represent the char- acter components. OCLC CJK350 pro- vides pronunciation-based input methods for four Romanization schemes: Wade- Giles and Pinyin for Chinese characters; modified Hepburn for Japanese kanji, katakana, and hiragana; and McCune- Reischauer for Korean hancha and hangul.9 The OCLC CJK350 provides fewer ac- cess points than the RLIN CJK system. Only commonly used standard numbers and codes, titles, authors (personal or corporate), and combinations of authors and titles are searchable. (On January 6, 1990, OCLC released its online reference system, the EPIC service, which provides keyword or phrase and subject searches, Boolean operators, truncations, and many other useful features. 10 However, it can search and display Roman or Romanized records only. OCLC CJK350 users do not benefit directly from it.) Qualifiers such as format and year(s) of publication can be used in searching. Although group display, collective display, and truncated record display are used when more than one record is retrieved, the only display format for a single · CJK record is the MARC format. OCLC CJK350 also al- lows users to print out locally catalog cards, with both Romanized and vernac- ular information. Officially introduced in January 1987, the OCLC CJK350 system had 70 user libraries worldwide as of October 1989, including 20 academic libraries in the United States. 11 The total number of unique CJK records in the OCLC Online Union Catalog had exceeded 320,000 as of January 1990. Comparisons The most important merits of the RLIN CJK system are the size of its database and its powerful search capa- bilities. RLIN CJK user libraries are major East Asian academic and research libraries, holding approximately 60% of the total East Asian collections in the United States. RLIN CJK also provides Boolean logic operators, truncation, and keyword and subject search capabilities. Automation in East Asian Libraries 561 These capabilities make it possible to use RLIN CJK as an online public access cat- alog, provided users can overcome the difficulty of its input methods. The most important merits of the OCLC CJK350 system are its diversified input methods and its card production capability. The different input methods are convenient for users of various back- grounds. The card production capability is a great help for most East Asian librar- ies because most of them still rely on the card catalog, with vernacular characters, for public access. The most important merits of the OCLC CJK350 system are its diversi- fied input methods and its card production capability. In addition to incorporating member libraries' contributions to original cata- loging and record upgrading, both OCLC and RLG are seeking records from other sources, such as institutions in China, Japan, and Taiwan. If this up- grading is accomplished, the two sys- tems will eventually provide access to CJK materials worldwide. Meanwhile, RLIN and OCLC have agreed to ex- change their records and to cooperate in the implementation of standards for computer linkages. 12 If further commu- nication and cooperation between these two major bibliographic utilities occur, both East Asian libraries and scholarly communities in the United States can greatly benefit. RESEARCH QUESTIONS Understanding the status of RLIN and OCLC CJK systems in the greater biblio- graphic community requires knowledge of how these systems are used in individ- ual user institutions. This research ad- dresses the following questions: 1. What functions do the RLIN CJK and OCLC CJK350 systems serve for member libraries? 2. What are the merits and weaknesses of these CJK systems, as seen by member libraries? 562 College & Research Libraries 3. What are the relationships be- tween these two CJK systems and member libraries' local automated library systems? 4. What direction is automation in East Asian libraries heading from this point? Is it possible, in the near future, to integrate these two CJK systems with member libraries' local automated library systems to perform total online library ser- vices? What are the major barriers? RESEARCH METHODS To investigate these questions, a na- tional survey among academic and re- search member libraries regarding both the RLIN CJK and the OCLC CJK350 systems was conducted. A questionnaire was designed to gather information on the following areas: A. General information on the East Asian collection: the location of the collection, the size of the collec- tion and the staff, and the public access mechanism to the collection; B. Information on the use of the RLIN and OCLC CJK systems: the date the CJK system was installed, the number of terminals used, the functions the CJK system serves for the collection, and the benefits and problems with regard to using the CJK system; C. Information on the local auto- mated library system: the name of the local system, when it was im- plemented, how it was developed, what functions it serves, and its capability of handling non-Roman scripts (If the local system cannot handle non-Roman scripts, what are the problems? Is the institution going to develop the capability? If yes, when and how? If no, why not?); and D. Relations between the local auto- mated library system and the CJK system: is the CJK system parallel to the local system? Or is the CJK system used mainly to download records to the local system? If so, how does it work out? How are the vernacular characters treated? November 1991 RLIN and OCLC provided the mem- ber lists of these two systems. In mid- January 1990, questionnaires were sent to 43 academic and research member li- braries of these two systems in the United States-20 for OCLC CJK350, 23 for RLIN CJK. However, because two libraries at Columbia University use the RLIN CJK system, the number of institutions in- cluded in this research is actually 42. Follow-up questionnaires were sent out in late February. The data-gathering pro- cess was completed in mid-March, with the return rate close to 90% (38 out of 42 institutions returned the questionnaire). SURVEY RESULTS The survey results are described in the following text and tables. A. General Information on the East Asian Collections Table 1 shows the location and lan- guage coverage of the responding librar- ies; table 2 shows the size of their collections (monographs only) versus the size of their staffs. Physical Location. Among 38 respon- dents, 30 (79%) reported having either a separate East Asian library or a separat.e collection housed in the general library (see table 1 ). Language Coverage. Thirty-seven li- braries answered the question on lan- guage coverage. All of them have Chinese and Japanese materials; 32 (86%) have Korean materials; and 10 (26%) have some materials in languages other than CJK (see table 1). Size of Collections. Thirty-five librar- ies reported on the size of their collec- tions. The size of the other three collections was estimated based on the statistics gathered by the task force for annual review and survey of library re- sources of the committee for East Asian libraries in 1988. The size of mono- graphic collections ranges from 15,000 to 734,000 volumes (most respondents did not provide the number of monographic titles they have), with 21 (55%) libraries having collections of more than 100,000 volumes. Eighteen (47%) libraries have more than 1,000 serials titles. Thirteen Automation in East Asian Libraries 563 TABLEt PHYSICAL LOCATION AND LANGUAGE COVERAGE No. of Librarians % Physical locations: Separate collection 30 79 Integrated 8 21 Languages covered: Chinese 38 100 Japanese 38 100 Korean 32 86 Other languages 10 26 TABLE2 COLLECTION SIZE VS. STAFF SIZE Staff Size Pro Non pro 0-2.0 2.5-4.0 4.1-6.0 6.1-8.0 8.5+ 0-2.0 2.1-4.0 4.1-6.0 6.1-8.0 8.1-10.0 10.1+ PRO: Professional staff NONPRO: Non-professional staff (34%) libraries reported having micro- form collections. Size of Staff. The staff size of respond- ing libraries ranges from 1.26 to 35.65 full- time employees. Twenty-seven (71 %) respondents have 4 or fewer professional librarians; 2 (5%) libraries have more than 10 professional librarians. Only 7 (18%) libraries have more professional librari- ans than nonprofessional staff. Apparently, except for the two largest collections, the size of staff is not always proportionate to the size of the collection (see table 2). Public Access Mechanism. All 38 re- sponding libraries have a card catalog as a public access mechanism for CJK ma- terials. In addition to a card catalog, 8 <100 13 3 1 0 0 11 4 1 1 0 0 Collection Size (unit: 1,000 volumes) 100-299 >300 4 0 5 2 2 0 4 1 2 0 0 7 0 3 0 0 3 0 0 2 6 (21 %) libraries have a microform cata- log; 28 (74%) libraries have local online catalog systems (providing Romanized bibliographic information); 22 (58%) li- braries use either the OCLC or the RLIN CJK system; and 4 (11 %) libraries have other mechanisms, such as a printed catalog. B. Information on the Use of CJKSystems Systems Used. Eighteen OCLC CJK350 members and 20 RUN CJK members re- sponded to the questionnaire. Among 17 small collections (fewer than 100,000 volumes), 9 (24%) use OCLC CJK350 and 8 (21 %) use RUN CJK; among 12 me- 564 College & Research Libraries November1991 TABLE3 NUMBER OF CJK SYSTEM INSTALLATIONS IN LIBRARIES OF DIFFERENT SIZES, 1982-1989 Systems Date OCLC RUN N.D. 1 1 1982 0 2 1983 0 5 1984 0 3 1985 0 4 1986 7 0 1987 1 1 1988 6 1 1989 3 3 Total 18 20 • Unit: 1,000 volumes dium collections (100,000 to 299,000 vol- umes), 7 (18%) are OCLC CJK350 members and 5 (13%) are RLIN CJK members; among 9large collections (300,000 or more volumes), 2 (5%) use OCLC CJK350 and 7 (18%) use RLIN CJK. That large CJK col- lections tend to use the RLIN CJK system is not surprising because most of them are RLG members. Year of Installation. Table 3 shows the number of CJK systems installed in respondents' libraries, in correspon- dence with their collection sizes, and the year the installation took place. No sign exists of significant change in the number of installations over these past years, except for the decrease in 1987. Nevertheless, the shift of installa- tions from RLIN to OCLC in 1986, when the OCLC CJK350 was first released, is interesting. The year 1988 also was a suc- cessful year for OCLC. Although both systems had the same number of instal- lations in 1988, it is too early to predict the future. When comparing the year of the CJK system installation with the size of each collection, large collections, in general, automated their systems earlier than medium and small collections; however, this is not an absolute condi- tion. For example, four small collections installed CJK systems before 1986. Collection Size• Total <100 100-299 >300 2 2 5 3 4 7 2 7 6 38 1 0 0 0 2 1 0 4 2 1 0 1 2 1 3 4 0 1 0 1 3 3 1 5 0 17 12 9 Number of Terminals Used. Table 4 shows the number of terminals used in 36 responding libraries. Among them, 14 (39%) have one terminal, 10 (28%) have two terminals, 6 (17%) have three termi- nals, 5 (14%) have4 terminals, 1 (3%) has 6 terminals, and 2libraries did not pro- vide the information. The total number of terminals in use is at least 80, assum- ing that the two libraries that did not pro- vide data have one or more terminals. Functions CJK Systems Serve. Table 5 shows the library functions that the two CJK systems serve for 37 responding libraries (multiple answers). All re- sponding libraries use CJK systems for , cataloging;17(46%)uset~ ~og (although this nulnber is not consistent with the number reported in the above section on public access mechanisms, in which 58% of the re- spondents reported using CJK systems for public access; perhaps some respon- dents misinterpreted one of the ques- tions); 13 (35%) use them for in~.ary l<;>an purposes; 6 (16%) use them for ac- q "sitions-related t sks (e.g., preor er searching and record verification); and only 1 (3%) library reported using the CJK system for serials contrel. Benefits of CJK Systems. Originally, the question was designed to have re- TABLE4 NUMBER OF TERMINALS IN USE No. of Terminals No. of Libraries 2 3 4 6 No data TABLES 14 10 6 5 1 2 LIBRARY FUNCTIONS CJK SYSTEMS SERVE (N =37) Functions OCLC RLIN Total % Acquisitions 3 3 6 16 Cataloging 17 20 37 100 Online public catalog 6 11 17 46 Serials control 0 3 Inter-library loan 4 9 13 35 Automation in East Asian Libraries 565 labeled group A (consisting of 26 respon- dents, 70% of the total), and the rest I labeled group B. Only rank values from group A were taken for calculation of range and medians. Meanwhile, I counted the number of votes for each benefit element from both groups. As long as a library selected a benefit ele- ment, whether ranked or checked, I counted it as one vote for that benefit element. Tables 6a, 6b, and 6c show the results for the OCLC CJK350, the RLIN CJK, and the combination of both. According to tables 6a and 6b, the members of both CJK sys terns seem to be in consensus on the major benefits of CJK systems, except that RLIN CJK members ranked "access to other collections" much higher than did OCLC members. This probably is due to the RLIN CJK' s inclusion of records (for the same title) from different librar- ies, while the OCLC CJK350 only retains a unique record for a title with holding information of others attached. The RLIN CJK' s practice enables users to ac- cess more detailed holding information, spondents rank each point, from one to such as call number and editions, from five, according to its importance for in- other libraries. Another factor could be dividual libraries. However, some re- that RLIN CJK members have more spondents misunderstood the question interlibrary loan activities than OCLC and ranked them against one another; CJK350 members (see table 5). other respondents checked the points For group A of the combined CJK sys- they found to be important. As a result, tern members (see table 6c), the most the method of interpreting the results important benefit of the two systems is had to be altered. I first separated the 37 increased cataloging productivity, fol- respondents into two groups: those who lowed by authority control and access to answered according to my intention I both other libraries' and each local TABLE 6A MAJOR BENEFITS OF OCLC CJK350 Group A GroupA&B Major Benefits Med. Range No. of Votes % Increased cataloging 5 2-5 15 88 productivity Authority control 4 3-5 12 71 Access to local holdings 3 1-5 14 82 Access to other collections 3 1-5 15 88 Improved staff morale 3 1-5 14 82 Improved library image 4 1-5 14 82 n=ll n=17 566 College & Research Libraries November 1991 TABLE68 MAJOR BENEFITS OF RLIN CJK Group A GroupA&B Major Benefits Med. Range No. of Votes % Increased 5 2-5 20 100 cataloging productivity Authority control 5 1-5 17 85 Access to local 4 1-5 19 95 holdings Access to other 5 1-5 19 95 collections Improved staff 3 1-5 13 65 morale Improved library 3 1-5 16 80 image n=15 n=20 TABLE6C MAJOR BENEFITS OF OCLC & RLIN CJK SYSTEMS Major Benefits Med. Increased cataloging productivity 5 Authority control 5 Access to local holdings 4 Access to other collections 4 Improved staff morale 3 Improved library image 3 library's holding information. Con- versely, when judged by the number of votes, these benefits are ranked, starting with the most popular, as: increased cat- aloging productivity, access to local and other libraries' holding information, im- proved library image, improved staff morale, and authority control. Major Problems of CJK Systems. The question concerning major problems of CJK systems confused respondents in much the same way as did the question about CJK system benefits. I therefore used the method described earlier to an- alyze the data answering this question. Twenty-seven (75%) respondents in group A answered this question cor- rectly. The statistical results of major problems of the OCLC CJK350, the RLIN Group A GroupA&B Range No. of Votes % 2-5 35 95 1-5 25 68 1-5 33 89 1-5 33 89 1-5 27 73 1-5 30 81 n=26 n=37 CJK, and the combination of both are shown in tables 7a, 7b, and 7c. Tables 7a and 7b indicate that the members of the two systems have differ- ent concerns, although there are similar- ities also. OCLC CJK350 members ranked "searching methods" much higher than did RLIN CJK members. This is, I think, because of the QCLC CJK350's lack-o~d, s~ect, and other related searching capabillfies . Yet RLIN CJK members ranked "high costs" much higher than did OCLC CJK350 members. This is not surprising because most RLIN CJK members still use first- generation terminals; when the new MSWs are widely installed, the situation will improve significantly. What is sur- prising is that members of both the RLIN Automation in East Asian Libraries 567 TABLE7A MAJOR PROBLEMS OF OCLC CJK350 Group A GroupA&B Problems Med. Range No. of Votes % Slow response time 3 1-5 13 76 Quality of records 3 1-5 15 88 Difficult to master 1.5 1-4 13 76 Searching methods 4 1-5 16 94 Display formats 2 1-5 12 71 Inadequate for public use 3.5 1-5 13 76 System downtime 2 1-4 12 71 Hardware breakdowns 2 1-3 13 76 High costs 2 1-5 12 71 Find experienced employees 3 1-5 12 71 Support from RLG/OCLC 2 1-4 12 71 n=12 n=17 TABLE7B MAJOR PROBLEMS OF RLIN CJK Group A Problems Me d. Slow response time 2 Quality of records 3 Difficult to master 2.5 Searching methods 1 Display formats 1 Inadequate for public use 3 System downtime 2 Hard ware breakdowns 1 High costs 3 Find experienced employees 4 Support from RLG/OCLC 1 n=15 GroupA&B Range No. of Votes % 1-5 18 90 1-5 17 85 1-5 16 80 1-4 14 70 1-4 9 45 1-5 15 75 1-5 15 75 1-5 15 75 2-5 18 90 1-5 15 75 1-2 11 55 n=20 are its inadequacy for public use, its high costs, and the difficulty in finding em- ployees experienced with the system. The three most commonly chosen prob- lems are the quality of records, the high costs, and the available searching methods. C. Information on Local CJK and the OCLC CJK350 selected "in- adequate for public use" as one of the three most serious problems of the sys- tems, despite the fact that the OCLC CJK350 provides various input methods and the RLIN CJK provides various dis- play formats. This is an issue that needs further investigation. For group A of the combined CJK sys- tem members (see table 7c), the three greatest problems of the CJK systems Automated Library Systems Among the 38 respondents, 4 (11 %) do not have local automated library sys- 568 College & Research Libraries November 1991 TABLE7C MAJOR PROBLEMS OF OCLC & RLIN CJK SYSTEMS Group A GroupA&B Problems Med. Slow response time 2 Quality of records 3 Difficult to master 2 Searching methods 3 Display formats 2 Inadequate for public use 3 System downtime 2 Hardware breakdowns 1 High costs 3 Find experienced employees 3 Support from RLG/OCLC terns yet. On the other hand, 10 (26%) institutions have more than one system for different library functions. The ma- jority of respondents have integrated local automated library systems. Years of Automation. Table 8 shows the years in which 34 institutions auto- mated their library operations. For insti- tutions having more than one automated system, the earliest years are chosen as the starting years. The majority (63%) of the responding institutions automated their library functions between 1985 and 1989. In order to discover whether rela- tionships exist between the automation of general collections and CJK collec- tions, the years in which responding in- stitutions automated their general and CJK collections are drawn and tabulated in table 9. From the diagram, there is not enough evidence to support the assump- tion that those libraries that automated their general collections early also auto- mated their CJK collections early. They are largely independent events. Another surprising finding is that 16 institutions automated their CJK collections before they automated their general collections. Development of Local Systems. Table 10 shows the methods of development of responding institutions' local auto- mated library systems. The majority n=27 Range No. of Votes % 1-5 28 76 1-5 32 86 1-5 29 78 1-5 30 81 1-5 23 62 1-5 28 76 1-5 27 73 1-5 29 78 1-5 31 84 1-5 27 73 1-4 23 62 n=37 (65%) of these local automated library systems are turn-key systems or modi- fied commercial systems. Only 29% of the systems were developed in-house (most of these institutions have CJK col- lections exceeding 100,000 volumes). Functions of Local Systems. Table 11 shows the library functions local auto- mated library systems serve. Among 34 respondents, only 8 (25%) institutions have automated all functions listed. Eighteen (53%) institutions have auto- mated the four basic library functions- that is, acquisitions, circulation, online public catalog, and serials control. Nev- ertheless, the majority of respondents have online public catalog, circulation, and acquisitions systems. A few institutions' local systems also serve TABLES YEARS OF AUTOMATION Year N.D. 1970-1974 1975-1979 1980-1984 1985-1989 1990 No. of institutions 4 1 4 5 19 Automation in East Asian Libraries 569 TABLE9 YEARS OF AUTOMATING GENERAL AND CJK COLLECTIONS Year N.D. 82 83 N.D. 70 75 1 77 78 General 79 Collections 80 82 1 83 85 87 88 3 89 2 90 None other functions, such as cataloging maintenance and electronic mail service. Non-Roman Capabilities. Only 27 re- spondents answered the question con- cerning their local systems' ability to process and display non-Roman lan- guages. All answers are negative. Twenty-three (85%) respondents related the problem to hardware limitations, 22 (81 %) to software problems, 11 (41 %) to insufficient funding, and 5 (19%) to the small size of collections. No respondent thought the problem was due to lack of demand. Future Development. Twenty-nine li- braries responded to the question of whether or not their institutions plan to develop non-Roman capability. Thirteen (4~%) of them answered yes; however, no clear time frames were given, except for one library that estimated three to five years from now. As for the methods of development, 5 (17%) institutions plan to improve their existing local sys- tems; 1 (13%) institution plans to pur- chase separate hardware, but still link up to the local system; 3 (10%) institu- tions will explore more than one possi- bility; 2 (7%) institutions do not have CJK Collections 84 85 86 87 2 1 2 2 1 88 2 1 2 1 1 89 1 1 1 2 1 clear ideas on this matter yet; and an- other 2 (7%) institutions will rely on ven- dors of their local systems to do it. Sixteen (55%) respondents indicated that their institutions will not develop their local automated library systems' non- Roman capabilities. Table 12 lists these- lected reasons. · D. Relations between CJK and Local Systems Table 13 shows the statistical outcome of the relationship between CJK systems and local systems. Among 32 respon- dents to this question, 4 (13%) reported TABLE 10 METHODS OF DEVELOPMENT OF LOCAL SYSTEMS No. of Methods Institutions % Developed in-house 10 29 Turn-key systems 13 38 Modified commercial 9 27 systems Other 2 6 570 College & Research Libraries November 1991 TABLE 11 FUNCTIONS LOCAL SYSTEMS SERVE Functions No. of Institutions % Acquisition Online public catalog Circulation Serials Inter-library loan Accounting 23 34 28 20 14 19 68 100 82 59 41 56 TABLE12 REASONS NOT TO DEVELOP LOCAL SYSTEMS' NON-ROMAN CAPABILITY R~asons No. of Institutions % Hardware limitations 9 56 Software problems 9 56 Funding 7 44 Small collection 2 13 OCLC/RLIN CJK system is enough 2 13 Other (low priority) 5 31 TABLE13 RELATION BETWEEN CJK SYSTEMS AND LOCAL SYSTEMS OCLC Users RLINUsers %of Total Relation: (n=32) Parallel systems Download records from CJK systems Methods of downloading: (n=26) Directly Via tapes Other methods Character Treatments (n=26) Eliminated Stored in local systems Stored in separate tapes/files having parallel systems (a local system for Roman-language materials and a CJK system for East Asian-language ma- terials); 28 (88%) reported that they download records from the online CJK system to the local system. Among the latter, 26 answered questions regarding methods of downloading and the treat- 2 14 2 10 1 7 5 2 13 14 87 2 15 11 81 0 4 7 31 6 50 0 19 ment of CJK characters: 4 (15%) institu- tions download records directly from the CJK system to the local system; 21 (81 %) libraries download CJK records to a tape first and then to the local system (either weekly or monthly); and 1 (4%) library uses other methods. For the treatment of CJK characters during downloading, 8 (31 %) institutions automatically elimi- nate them from their local systems; 13 (50%) institutions store codings repre- senting CJK characters in their local sys- tems, although these are not accessible to the general user; and 5 (19%) institutions use methods that could be considered compromises between the first two methods-they eliminate CJK characters from their local systems and maintain archival tapes of records containing CJK characters for future use. DISCUSSION The survey results provide the follow- ing answers. Functions. The survey shows that cat- aloging is the most important function of CJK systems. Although 46% of the re- spondents reported using the CJK sys- tem as an online public catalog, most of them indicated that this function is not heavily used. Interlibrary loan (record searching and verification only) is the third most important function of the CJK system; however, it is not widely used either. More research and development are needed to enhance or expand the capabilities of these two CJK systems. Merits and Weaknesses. Although a minor problem occurred in the survey due to the misinterpretation of questions related to benefits and problems of CJK systems, I was able to resolve it by using different analysis _methods. The results from group A show the weight of im- portance of each benefit and problem element; yet the results from vote counts represent the general concern of re- sponding libraries. Although I was not surprised by the different outcomes re- sulting from these two analysis methods, I was interested to see that there are actually some consistencies between them. For ex- ample, both analysis methods indicate that the most important benefit of CJK systems is increased c~aloging produc- ~igh costs also are mcluded in both analysis results as one of the three major problems of CJK systems. A further examination of the medians of group A in tables 6c and 7c shows that responding libraries weight benefits of CJK systems higher than their problems Automation in East Asian Libraries 571 (the medians for benefits ranged from three to five, and the medians for prob- lems ranged from one to three). This in- dicates that the merits of the two systems are seen as more important than their problems-perhaps an encouraging out- come for the two systems. Relations. The majority (87%) of re- sponding libraries download records from CJK systems to their local auto- mated library systems. Because none of these local automated library systems has the capability of processing and display- ing CJK vernacular characters, 31% of the responding libraries have to strip off the vernacular fields from the bibliographic records. This situation means that only Romanized bibliographic information is downloaded, and those vernacular char- acters will never be recovered. Fortunately, the majority of libraries retain vernacular fields either in their local automated sys- tems or on separate tapes. Although they are not able to use them at present, it will be possible to use them if the non-Roman capabilities of their local automated sys- tems are developed in the future. Future Prospects. Although most li- braries wish that the non-Roman capa- bilities of their local automated systems will be further developed in the future, only 45% of them reported having plans to do so. The major barriers for the devel- opment plan are hardware and software limitations, as well as insufficient funding. Low priority also is an important factor. However, as technology progresses, the hardware and software limitations should be eliminated. The remaining barriers can be removed easily with the continuing commitment from East Asian libraries and their parent institutions. Developing the non-Roman capabilities of local systems will not only solve the problem of CJK materials, but will bene- fit other non-Roman materials. For in-house local automated library systems, the development plan might have to be carried out individually. For turn-key or modified commercial sys- tems, the cooperation among libraries using the same system and system ven- dors will be an efficient and economical solution. 572 College & Research Libraries CONCLUSIONS From this study, I conclude that the sharing of cataloging efforts (the finan- cial and human resources) among East Asian libraries has been fully served by the two CJK systems because all member libraries are involved in cataloging activ- ities using the two systems. As the use of these systems for interlibrary loan pur- poses increases, the sharing of material resources also will improve. (At present, many--interlihr.m:y_J~n _.bihli.Qgr(!phic verifi5=ation~9r old materi~Js..~ue d _ne manually using some_ major E~!, Asian libr-aries' printed catalogs, such as the Harvard-Yenching Library Catalog or the Hoo'!er Institution Library Catalog. After each individual library's retrospective conversion project is finished, the use of these two CJK systems for interlibrary loan purposes should increase greatly.) Nevertheless, the East Asian collec- tions in the United States are still not fully integrated into the main collections of their parent institutions. Most institu- tions include Romanized CJK records in their local automated library systems. However, these records do not serve many practical purposes because, due to the large number of homophones in CJK languages, Romanized CJK records are not always legible to users or even well- trained librarians. East Asian libraries still rely heavily on the card catalog for public services. The important develop- November 1991 ments of both OCLC and RLG-that is, OCLC CJK350 and RLIN CJK-are, for the most part, enjoyed only by the cata- loging staffs of East Asian libraries. However, their efforts in creating ver- nacular fields in bibliographic records are diminished Ia ter in the process of downloading from CJK systems to local systems. This is a waste of resources not only for East Asian libraries, but for the entire bibliographic community. The merits of the two systems are seen as more important than their problems-perhaps an encouraging outcome for the two systems. To bring East Asian collections and other non-Roman-language collections into the bibliographic mainstream, thus forming a complete automated public access catalog, more effort has to be ex- pended in improving local automated library systems' abilities to accommo- date non-Roman capabilities. The initia- tion and cooperation of East Asian libraries, the expertise of personnel at both OCLC and RLG regarding com- puter processing of non-Roman lan- guages, as well as the commitment of major academic institutions are the key factors in achieving this goal, which- we hope-will be achieved in another five to ten years. REFERENCES AND NOTES 1. Lucy A. Tedd, "Computer-based Library Systems: A Review of the Last Twenty-one Years," Journal of Documentation 43:145-61 (June 1987). 2. "Five Years of CJK," Research Libraries Group News 17:3 (Fall1988). 3. Wen-kai Kung, "Computerized Cataloging of East Asian Vernacular Materials in Non-RUN Libraries of North America," Association for Asian ~tudies (AAS) Committee on East Asian Libraries (CEAL) Bulletin 80:5-10 (Dec. 1986); Hee-Jung Lee, "A Study of Automated Cataloging Systems for Chinese, Japanese, and Korean Materials" M.L.S. specialization paper, (Grad. Sch. of Lib. and Info. Sci., University of California-Los Angeles, 1985); Karen T. Wei, "Current Status and Future Trends of East Asian Library Automation in North America," Information Technology and Libraries 5:140-46 (June 1986). 4. John W. Haegar, "An RLG Plan for the Inclusion of East Asian Records in the Research Libraries Information Network," International Association for Orientalists Librarians (IAOL) Bulletin 24-25:38-43 (1984), and his "RUN CJK: A Review of the First Five Years," IAOL Bulletin 30-31:18-20 (1987); Jay Lee, "Cataloging CJK Online: The ASIA Automation in East Asian Libraries 573 Experience," paper presented to subcommittee on technical processing, CEAL, AAS Conference, Mar. 16, 1989; Alan Tucker, "The East Asian Project of the Research Libraries Group," AAS. CEAL. Bulletin 69:12-24(0ct.1982);AndrewWang, "OCLCCJK Automated Library Information Network," Journal of Library and Information Science (Taipei) 11:143-53 (Oct. 1985); and Karen T. Wei, "RUN CJK vs. OCLC CJK," Journal of Educational Media and Library Science 24:82-94 (Autumn 1986), in Chinese. 5. Research Libraries Group, RLG CJK Terminal Manual (Stanford, Calif.: Research Librar- ies Group, 1983), p.25-26. 6. "RUN MultiScript Workstation Enthusiastically Received," Research Libraries Group press release, May 5, 1989. 7. "Five Years of CJK," Research Libraries Group News, p.3-11. 8. Wang, "OCLC CJK Automated Library Information Network," p.143-53, and system information prepared by him for the author (Jan. 11, 1990). 9. Ibid. 10. "The EPIC Service Is Introduced," OCLC Newsletter 183:10-16 (Jan./Feb. 1990). 11. "Use of OCLC CJK350 System Grows," OCLC Newsletter 181:31 (Sept./Oct. 1989). 12. "OCLC and RLG to Cooperate on Compatibility in Computer Linking," Research Libraries Group press release, May 7, 1990. 574 College & Research Libraries November 1991 Build Your Future with ACRL The library association that • contributes to the total professional development of over 10,000 academic and research librarians • improves service capabilities of academic and research librarians • promotes and speaks for the interests of academic and research librarianship • promotes study and research relevant to academic and research librarianship Membership benefits include free subscriptions to College & Research Libraries. ACRL's official journal. and College and Research Libraries News. ACRL's news magazine for the profession - a $50 value; reduced rates for conference registration and continuing education courses , discounts on ACRL publications -and much more . The Association of College and Research Libraries is a division of the American Library Association (ALA). ALA membership is prerequisite to ACRL membership . ~,----------------------, • I Please rush information about joining ACRL to: I I ~~ I I Address I I City State Zip I L----------------------~ Mail to: Association of College and Research Libraries • American Library Association • 50 East Huron • Chicago , IL 60611 • 312/944-6780