Professional Documents
Culture Documents
CJK Opac Busqueda
CJK Opac Busqueda
INTRODUCTION
Many North American library catalogs support title, author, and keyword
searching in non-Roman scripts. However, when it comes to subject search-
ing, these systems provide access only to controlled English-language sub-
ject headings and thesauri, such as the Library of Congress Subject Headings
(LCSH). In much of the cataloging for items in languages not written in
Roman script, English-language subject access provides neither a sufficient
description of the content nor can it ensure the retrieval of the item.
When there is no English equivalent, an English subject headings sys-
tem provides transliteration (or Romanization) of the native scripts. However,
457
458 M. El-Sherbini and S. Chen
the transliteration scheme may not be the same the user employs at time of
search. There are also concepts in non-Roman languages that are difficult
to find in English and in this case catalogers would select a controlled vo-
cabulary subject heading that is “close enough.” On the other hand, a user
who is searching for a book written in certain language about a certain sub-
ject should be able to conduct a subject search in that language if that is
preferred. The term “non-Roman/Latin script” used in this article refers to
writing systems represented in ideographs or in non-Roman alphabets, such
as Chinese characters or the Arabic script.
Although librarians in the West are paying more attention to providing
subject access to non-English language materials, especially those in non-
Roman scripts, there is very little research into either their efforts or the
preferences of end users for this kind of access. In October 2005, the Associ-
ation for Library Collections and Technical Services (ALCTS) formed the Task
Force on Non-English Access to “examine ALA’s past, present, and poten-
tial future roles in enabling access to library resources in all languages and
scripts and in addressing the needs of users of materials in all languages and
scripts through the development of library standards and practices.” The last
of the eleven recommendations proposed by the Task Force Report deals
with the need for non-Roman scripts in subject access.1
Inspired by the ALCTS Non-English Task Force recommendation, the
authors decided to investigate the issue of subject access for non-Roman
script materials. They conducted a survey to investigate user preferences
when conducting subject searches for non-Roman script materials in North
American library catalogs. This article reports the results, analyzes some
problems that respondents encountered when using controlled subject head-
ings in English for retrieving these materials, and discusses respondent
recommendations.
LITERATURE REVIEW
LCSH, for subject retrieval, but Israeli public libraries are dissatisfied with the
classified catalog.
Some authors criticized the use of Romanization tools for non-Roman
scripts and described these tools as not user friendly. They pointed out that
the Romanization system is often not known to users. Kim and Molavi ad-
dress the problems with Romanization in two different scripts. Kim analyzed
the cataloging rules for Korean materials focusing on the McCune-Reischauer
(MR) system, the Korean Romanization scheme currently used in the United
States.13 Although this system has been used for a long time in many Western
countries, and was officially adopted by the Library of Congress (LC) for use
in the cataloging of Korean language materials, it has drawbacks in searching
and retrieving materials in this language. The author conducted a survey and
confirmed that the MR system is not a user-centric tool. Molavi identified
the main issues in cataloging Persian language materials in North American
libraries, and she confirmed that these problems originated from:
Both Kim and Molavi suggested some improvements to the MR system and
the ALA-LC Romanization Table for Persian to overcome these limitations.
Creating a bilingual authority file in academic and research libraries was
the focus of research by Abdoulaye.14 Abdoulaye analyzed and described
the bilingual authority files of the main library of the International Islamic
University of Malaysia. His research is unique in that he investigated the
perceptions of catalogers and end users in relation to the bilingual authority
files. He interviewed three cataloging staff at the Department of Arabic and
Religious Resources and 25 end users. Most of the respondents felt that the
use of bilingual authority files was essential for success in searching. The end
users felt that more subject headings and bibliographic information should
be provided, while the catalogers believed that bilingualism has an effect
on bibliographic control. They also felt that AACR2 and LCSH should be
translated into the Arabic language.
As a result of the limitations of the LCSH and the weakness in the
treatment of Arabic subjects, librarians from Arab countries felt the need to
prepare their own subject headings lists based on the characteristics and
grammar of the Arabic language and the culture of the Arabic world.15 No-
table among them are the lists by Swaydan, El-Khazindar, and Khalifah and
‘A’idi.16 Because these lists were prepared by individual librarians, they do
462 M. El-Sherbini and S. Chen
not have the support of national or regional organizations and they have not
been updated. Other lists were prepared by national and regional organi-
zations, such as the Institute of Public Administration (IPA) in Riyadh, Saudi
Arabia; the Arab League Educational, Cultural and Scientific Organization
(ALECSO); and Abdul Hameed Shoman Foundation in Amman, Jordan.17
Catalogers in this region are free to use any of these lists instead of the LCSH
to assign subject headings to non-Roman script materials.
RESEARCH METHOD
The first step in studying the needs of librarians and end users for multilingual
subject access in non-Roman scripts is to look at their language and script
preferences when searching for and retrieving non-Roman materials. To
accomplish this, the authors developed six research questions:
To answer these questions, the authors drafted two questionnaires, one for
librarians (including regular library support staff members) and another for
faculty and student researchers in languages and area studies (referred to
henceforth as “end user”). The authors first tested and got feedback from
library colleagues. The final surveys consisted of twelve questions (see Ap-
pendix), including some that allowed open comments.
The authors posted the invitation to both librarians and end users
through several electronic discussion lists, including the Middle East Librar-
ians Association (MELA), the Council on East Asian Libraries (CEAL), and
the Association of Jewish Libraries (AJL), and languages and area studies
centers. The authors also posted the survey on the OCLC-Cat and AUTOCAT
electronic discussion lists. Because some of the lists were international, a few
responses were received from libraries in other countries, such as Canada
and France. The survey was posted on February 11, 2010 for three weeks and
Assessing Non-Roman Subject Access to the Library Online Catalog 463
a reminder was sent during the third week (on February 26, 2010). The au-
thors used Google Documents to conduct the survey and collect responses.
RESULTS
Demographic Data
A total of 298 respondents completed the two surveys: 155 librarians and sup-
port staff completed the librarian survey; 117 faculty and student researchers
in languages and area studies completed the end user survey. Twenty-six
respondents to the end user survey identified themselves as faculty librar-
ians, and these were excluded from analysis because their expertise and
perspective were likely different from that of end users.
Librarian participants were asked to identify their areas of work or ex-
pertise by selecting one or more functional areas. Of the 155 librarian re-
spondents, 110 worked primarily in cataloging and technical services, 43
in collection management, 53 in reference and public services, and 50 as
subject specialists. Another 15 respondents classified themselves as bibliog-
raphers or administrative staff. Most respondents in this category identified
themselves as working in more than one area or having expertise outside
their primary work area.
For the end users survey, most (79%) of the 117 end user respondents
were faculty members. Another 17 (15%) were college students; the remain-
ing 8 (7%) identified themselves as administrative staff, visiting scholars, or
independent researchers.
headings using only the English or the Romanized form. Twenty percent
indicated that they use both the English form of subject headings and the
original non-Roman scripts. Very few librarians (3%) reported using only the
original non-Roman scripts for searching subject headings.
When asked about searching by keywords in English or non-Roman
scripts, 42% of the librarians indicated that they search for non-Roman mate-
rials using only English keyword terms. Only 8% indicated that they searched
using exclusively the original non-Roman scripts and 45% searched using
keywords in both the English/Romanized form and the original non-Roman
scripts. These data suggest that librarians are more likely to use non-Roman
scripts in keyword searches and less likely to do so in subject heading
searches.
FIGURE 2 Satisfaction with Controlled English Subjects (Level 5 = Most Satisfied; Level 1 =
Least Satisfied).
FIGURE 3 Satisfaction with Using Controlled English Subject to Find Non-Roman Script Ma-
terials (Level 5 = Most Satisfied; Level 1 = Least Satisfied).
Results showed that nearly twice as many librarians as end users felt
high satisfaction (15% and 7%, respectively), while the opposite was true at
low satisfaction levels. However, a considerable number from both groups,
40% of librarians and 33% of end users, were quite satisfied with using
controlled English subject headings to locate non-Roman items.
Librarian data has been further analyzed from the point of view of
specialization and work area, while end user data also includes infor-
mation about area of specialization and academic status. Figure 5 com-
pares the responses of four groups of librarians: Cataloging/Technical
Romanized forms (21%) and searching using the English equivalent to the
non-Roman subject term (11%).
In the “Others” category of end users (8 administrative staff, visiting
scholars, and independent researchers), two indicated that it was difficult to
search using Romanization, one expressed problems using English equiva-
lents for non-Roman terms, and one person indicated that they did not find
what they wanted. Four did not specify the type of problem.
More than one-hundred write-in responses were given to this question
by both librarian and the end user participants. For easy analysis, we have
grouped them into the following categories:
ROMANIZATION PROBLEMS
Twenty-eight end users and six librarians expressed various concerns about
using Romanized terms to search non-Roman scripts. Most of the end user
comments referred to the Romanization of Chinese, Japanese, and Korean
languages. Even though standards for transliteration have been set by the
American Library Association and Library of Congress, end users are usually
not familiar with them. They indicated that they did not see any consistency
in Romanization, which led them to miss information or to get incomplete
results. Inconsistent Romanization forced end users to search by multiple
forms of terms. Some end users indicated that to obtain good results they
had to search in the original script or by International Standard Book Num-
ber (ISBN). Other end users expressed frustration with the inconsistency in
Romanization, especially in Arabic and Hebrew languages. Other concerns
related to incorrect diacritics and special characters. Romanization frequently
becomes a problem when the subject heading is a personal name, corporate
body, or a geographic name. Both the librarians and end users expressed
concern with the Romanization in library catalog and indicated that the
transliterations are often inconsistent.
LCSH LIMITATIONS
Eight end users and eight librarians expressed concerns about the general
limitations of searching by LCSH, saying the subjects are either too broad
or too biased. The end users’ concerns were related to the quality of sub-
ject access and how the subject headings were assigned by librarians. Some
end users mentioned that the subject headings seem to have been assigned
randomly and did not reflect the content of the resources. Others indicated
that sometimes subject access for non-Roman materials was too broad and
resulted in retrieval of too many unrelated materials. Librarians were con-
cerned that the limitations of LCSH that made it difficult to express subject
470 M. El-Sherbini and S. Chen
Three librarians and six end users commented on problems in finding English
equivalents for non-Roman subject terms. Librarian respondents dealt with
this issue by choosing the closest subject heading from the LCSH to create
a subject heading that in most cases was much too broad. Some librarians
suggested using the terms from the original scripts as a way of making it
easier to express the subject of non-Roman materials. End users tried using
various forms of subject headings, but still did not get the desired results. A
number of them mentioned that there is inconsistency in the Romanization
of some terms in LCSH, particularly when it comes to the use of diacritics.
Some participants from both groups (seven users and four librarians) indi-
cated that they rarely used subject headings in favor of other elements, such
as ISBN, OCLC number, and keywords, when searching for non-Roman script
materials. One end user mentioned that searching by a keyword in English
was sometimes helpful in identifying appropriate LCSH terms. Others in-
dicated that searching by author can yield better results. These responses
confirm the opinions of those respondents who find that LCSH is not com-
pletely reliable when searching for non-Roman script materials.
OTHER COMMENTS
There were other comments that did not fall under any of the aforemen-
tioned categories, but are important to mention. Five librarians indicated that
they commonly searched by subject headings in English. For example, two
librarians mentioned that they do not search by non-Roman subjects and
that it did not even occur to them to attempt such a search. Three librarians
commented that since they search for materials in a variety of languages and
cannot supply the terms in all the languages, they use the English language
heading, which organizes and unifies all the items under LCSH subject cat-
egories. One of the end users commented that he/she usually finds what
he/she is looking for but still prefers to have the original script.
should be present in all cases, because records are used differently by differ-
ent users (e.g., patrons vs. staff), and having more than a single search option
would be useful. Many staff might need to identify items for acquisition or
reference purposes in languages they do not know.
Some librarians who responded “yes” to this question also voiced some
concerns. For example, they were concerned about the lack of non-Roman
controlled subject headings. Adding subject terms is costly and labor in-
tensive, especially if librarians add them in both English and non-Roman
languages. Others respondents were concerned that local online public ac-
cess catalogs (OPACs) are not designed to accommodate controlled subject
headings from non-Roman thesauri. This, in particular, would call for a great
degree of cooperation between local libraries, catalog software vendors, and
consortia.
To address the cost of adding subject capability in non-Roman scripts,
one librarian recommended social tagging as a less costly alternative to build-
ing a controlled vocabulary system in non-Roman scripts. Another librarian
suggested adding subject searching capability in non-Roman scripts via au-
thority records rather than parallel-script subject headings in bibliographic
records. Recommendations were made to develop controlled subject heading
thesauri for languages which do not have them. Cross references, retaining
non-Roman subject headings from foreign records downloaded to OCLC,
and teaching users how to use a “good vernacular script catalog to help
themselves” were recommended.
Although many end users voiced their support for adding both English
and non-Roman subject access for non-Roman materials to the library catalog,
they also expressed their concerns about the cost and human resources
required to add such headings. Some were concerned about converting
older records. End users offered suggestions that were similar to those
offered by librarians. For example, they suggested adding cross references
to the authority records and using Unicode.
Only seven librarians commented negatively on the usefulness of adding
subject searching capability in non-Roman scripts to the library catalog. They
were concerned about the time needed to add this information and the costs
involved. They preferred to spend their time reducing backlogs and making
those materials available to users. They also mentioned that non-Roman
subject headings already exist in some of the international records loaded
to OCLC. Using LCSH is helpful for those catalogers who do not know
non-Roman language very well and who are not able to assign a subject
equivalent to the English one in the non-Roman scripts. Furthermore, non-
Roman terms are often incompatible with LCSH. Assigning subject headings
in non-Roman scripts is an additional step for catalogers and will require
learning new vocabulary for each language.
There were seventeen comments received from librarians and ten com-
ments from end users who responded “Not sure” to this question. Librarians
Assessing Non-Roman Subject Access to the Library Online Catalog 473
who responded “not sure” have reservations about library patrons using this
feature. They indicated that it is already difficult to familiarize end users with
English language subject headings. Others asked whether it is worth the cost
and staff time. Some of these librarians were not able to search non-Roman
headings in their catalog. Others questioned whether libraries will have LCSH
for every language, or have the ability to use uncontrolled vocabulary.
non-Roman scripts. Some argued that subject searches, as well as title and
author, often require contextual knowledge, and the advantage of native lan-
guage search should not be overlooked, maintaining that “most [users] want
to approach a search naturally, which means instinctively, in their native
language”; and “subject headings as well as titles [input] in its own language
would make for the ease of searching.”
Some participants, primarily those working in East Asian language areas,
expressed opposing views on this subject. One respondent who answered
yes pointed out that “Chinese names, in particular, have many homophones,”
and “searching by [their] Romanization does not eliminate these”—which
means it would increase recall but reduce the accuracy of search results.
Another who answered no to this question, however, argued that “written
Chinese is non-alphabetical, and its Romanization makes it much more con-
venient to search.” This opinion reflects the fact that most U.S. OPACs are
indexed alphabetically using the Latin alphabet and do not effectively index
other scripts, especially those that are based on characters rather than the
Latin alphabet.
Three comments revealed a preference for LCSH and a hesitance to opt
for a new system. One suggested that subject headings in non-Roman scripts
should only be provided “when the English headings inadequately express
the terms.” Two respondents were amenable to making subject headings
available in the user’s own language, even though they did not feel the need
for them personally.
East Asian
Chinese 13 (9%) 16 (16%)
Japanese 26 (18%) 15 (15%)
Korean 2 (1%) 0
Tibetan 0 3 (3%)
South Asian 2 (1%) 7 (7%)
European 18 (13%) 12 (12%)
Slavic 31 (23%) 3 (3%)
Arabic 10 (7%) 5 (5%)
Hebrew 15 (11%) 24 (24%)
Assessing Non-Roman Subject Access to the Library Online Catalog 475
daily basis.” One user brought up the idea of system supplied subject terms
for library users to use as tags. It is noteworthy that some end users preferred
librarian-assigned subject headings and did not want their replacement by
tagging. Some users, while welcoming folksonomy features in library cata-
logs, also suggested monitoring or filtering as well as user guidelines. They
were concerned about political and ethnic tagging abuses—something that
prompted other participants to reject user tagging altogether (see below).
The eleven comments made by librarian participants who rejected the
idea of end user tagging told us that these librarians were concerned about
the “dubious value” of tagging due to its subjectivity and potential for mis-
use in conveying political, social, or religious agendas. Some maintained that
end user tagging would not be at the same professional level as the library
catalog and would generate quality control problems. Some expressed con-
cern about other possible negative effects, such as the clustering of tagging
over the catalog functionality. Meanwhile, some librarians suggested other
approaches to tagging, such as author- or editor-supplied tags or systems
to allow users to create a “personalized catalog, which can be tagged and
shared among patrons” but not clustered in the public access catalog (OPAC).
Seven librarians questioned the usefulness and management of end user
tagging. The concern here was about the control of the tagging system. On
the end user side, twenty-three respondents insist that tagging cannot replace
the subject heading in controlled vocabularies. Here again, one finds support
for keeping the tagging separate from the library catalog. One respondent
suggested the wiki style to be a model of library tagging.
A minority of librarians and end users seem to agree that adding folk-
sonomy features to library catalogs is important. Allowing end user tagging
may promote the use of the library catalog. However, a significant ratio of
participants indicated that both librarians and end users are not ready for
the user tagging feature until more investigation and experiments are carried
out. Many participants were concerned with the quality of user tagging, and
subsequently suggested monitoring and filtering it. Both end users and li-
brarians support the notion of keeping end user tagging as a separate feature
that would not replace the controlled vocabulary subject headings in library
catalogs.
CONCLUSION
Several important trends emerge from the data gathered by this survey.
The first has to do with the searching habits of librarians who search for
materials published in non-Roman scripts. Results of the survey reveal that
most librarians conducting these searches use English or Romanized terms
when conducting subject searches. A smaller percentage of respondents
use both English and the original scripts to search for these materials. The
number of librarians using original script searching exclusively is statistically
insignificant. When conducting subject searches, both the librarians and end
users tend to use LCSH. The picture changes somewhat when it comes to
keyword searching. Here, nearly half of the librarians surveyed indicate that
they use non-Roman scripts in addition to English and Romanized forms.
A large portion of data gathered deals with the level of satisfaction with
English language subject searching for non-Roman script materials. Informa-
tion gathered indicates that librarians tend to be more satisfied with their
English language search results than end users. One possible explanation
for this may be that the librarians are generally more familiar with subject
heading structure and find this type of searching more effective.
Survey respondents were asked to contribute comments to a number
of questions in the survey. One of these asked them to describe problems
they experienced when doing subject searches using English or Romanized
subject terms. Comments provided by the survey respondents highlight a
number of areas of concern.
In the area of Romanization, many respondents expressed concerns
regarding its effects in searching Chinese, Japanese, and Korean language
materials. End users who are not familiar with the structure of LCSH find it
difficult to use and complain of inconsistencies in this area. Inconsistency in
Romanization was a concern of searchers in Arabic and Hebrew languages
as well. Librarians and end users were concerned with the variety of translit-
eration standards for different languages.
In general, the number of statements regarding the difficulties librari-
ans and end users encountered when they searched non-Roman items by
controlled English subject terms indicates a significant level of dissatisfaction
with the current controlled English subject heading system.
The respondents continued to use the Romanized forms in their subject
and keyword searches, due, in large part to the absence of vernacular subject
terms in U.S. library catalogs. Librarians rarely searched the catalog using
non-Roman scripts alone. Librarians and end users used English language
controlled vocabulary to find materials in non-Roman scripts and very few
480 M. El-Sherbini and S. Chen
NOTES
1. The task force advised the ALCTS “Subject Analysis Committee (SAC) working with appropriate
library organizations, to study the needs of library users for multilingual subject access in the appropriate
script(s), and to propose steps to address those needs.” This recommendation did not appear in the
September 18, 2006 Report, but was added later as an additional recommendation and was labeled “not
a high priority.” The Steering Committee on Non-English Access that was appointed at the end of 2007
to oversee the implementation of the task force’s recommendations concluded that this recommendation
Assessing Non-Roman Subject Access to the Library Online Catalog 481
was vague and broad and fell outside SAC’s scope. They made the following comments to the ALCTS
Board: “This recommendation was issued in 2007. Is it still important today? An evaluation of this
recommendation is needed; a survey of the literature to see what has been written on this topic since
2009 is needed; and forming an interest group within ALCTS to explore this issue further.”
2. James E. Agenbroad, “Romanization is not Enough,” Cataloging & Classification Quarterly 42,
no. 2 (2006): 21–34.
3. Joan M. Aliprand, “The Structure and Content of MARC 21 Records in the Unicode Environ-
ment,” Information Technology & Libraries 24, no. 4 (December 2005): 170–179.
4. Joan M. Aliprand, “Scripts, Languages, and Authority Control,” Library Resources & Technical
Services 49, no. 4 (October 2005): 243–249.
5. Ron Davies, “Models for Multilingual Subject Access in Online Library Catalogues: The ILO
Experience” (paper presented at the annual conference of the European Library Automation Group,
Bern, Switzerland, April 2–4, 2003).
6. Jung-Ran Park, “Cross-Lingual Name and Subject Access: Mechanisms and Challenges,” Library
Resources & Technical Services 51, no. 3 (July 2007): 186.
7. Ibid.
8. Hans H. Wellisch, “Bibliographic Access to Multilingual Collections,” Library Trends 29, no. 2
(Fall 1980): 223–244.
9. Ibid.
10. John Eilts, “Non-Roman Script Materials in North American Libraries: Automation and Interna-
tional Exchange,” IFLA Conference Proceedings (August 20–25, 1995), http://archive.ifla.org/IV/ifla61/61-
eilj.htm (accessed September 22, 2010).
11. Lois Mai Chan, Xia Lin, and Marcia Zeng, “Structural and Multilingual Approaches to Subject Ac-
cess on the Web,” IFLA Council and General Conference, Bangkok, Thailand (August 20–28, 1999), http://
archive.ifla.org/IV/ifla65/papers/012-117e.htm (accessed September 22, 2010).
12. Elhanan Adler, “Multilingual and Multiscript Subject Access: The Case of Israel,” IFLA
Council and General Conference, Jerusalem, Israel (August 13–18, 2000), http://archive.ifla.org/
IV/ifla66/papers/035-130e.htm (accessed September 22, 2010).
13. SungKyung Kim, “Romanization in Cataloging of Korean Materials,” Cataloging & Classification
Quarterly 43, no. 2 (2006): 53–76; Fereshteh Molavi, “Main Issues in Cataloging Persian Language Materials
in North America,” Cataloging & Classification Quarterly 43, no. 2 (2006): 77–82.
14. Kaba Abdoulaye, “Perceptions of Cataloguers and End-Users towards Bilingual Authority Files,”
Electronic Library 20, no. 3 (2002): 202–210.
15. Zahiruddin Khurshid, “Arabic Script Materials: Cataloging Issues and Problems,” Cataloging
& Classification Quarterly 34, no. 4 (2002): 67–77; Khalil Shihab, “Arabic and Multilingual Scripts
Sorting and Analysis,” Proceedings of the 6th WSEAS International Conference on Applied Informat-
ics and Communications (Elounda, Greece, August 18–20, 2006): 157–162, http://www.wseas.us/e-
library/conferences/2006elounda1/papers/537-391.pdf (accessed July 4, 2011).
16. Nasser M. Swaydan, Arabic Subject Headings, 2nd ed. (Riyadh: King Saud University Libraries,
1985); Ibrahim A. El-Khazindar, List of Arabic Subject Headings, 3rd ed. (Kuwait: Scientific Research
House, 1983); Sha’ban ‘Abd al-’Aziz Khalifah and Mohammad Awadh Al-’A’idi, Qa’imat Rous al-Maudu’at
al-’Arabiah-al-Kubra, 2nd ed. with additions and revisions (Cairo: Al-Maktaba Al-Akademiah, 1994) 2. v.
(updated by supplements).
17. Institute of Public Administration, Qa’imat Rous al-Maudu’at al-Arabi (Riyadh: IPA, 1992);
Arab League, Educational Cultural and Scientific Organization, Qa’imat Rous al-Maudu’at al-Arabi al-
Mahdoodah (Tunis: ALECSO, 1995); Abdul Hameed Shoman Foundation. . . . et al., collected and edited
by Mahmoud Itayem, Expanded Thesaurus (electronic copy) (Amman: The Foundation, 2001).
The Librarians and end users surveys were combined except for question 3
(directed only for librarians) and Question 9 (directed only for end users).
Currently most U.S.-based library catalogs only provide controlled sub-
ject access in English. At the same time, these U.S.-based libraries do have
482 M. El-Sherbini and S. Chen