LRTS 51(3)

Jung-ran Park (jung-ran.park@cis.drexel.edu) is Assistant Professor,The iSchool at Drexel, College ofInformation Science and Technology,Drexel University, Philadelphia.

The comments of the editor and anon-ymous reviewers regarding this papergreatly improved the quality of thework and I would like to express myappreciation to them. This paper isbased upon a conference presenta-tion, "Global Access to Cross-culturaland Cross-lingual Resources: CurrentResearch Trends and Their Implicationsto LIS Curricula," at the 2007 ALISEAnnual Conference, San Antonio, Texas,January 16-19, 2007.

Submitted August 14, 2006; accept-ed for publication October 17, 2006,pending revision; revision submittedDecember 12,2006, and accepted forpublication.

Cross-lingual Nameand Subject AccessMechanisms and Challenges

By Jung-ran Park

This paper considers issues surrounding name and subject access across languag-es and cultures, particularly mechanisms and knowledge organization tools (e.g.,cataloging, metadata) for cross-lingual information access. The author examinescurrent mechanisms for cross-lingual name and subject access and identifiesmajor factors that hinder cross-lingual information access. The author providesexamples from the Korean language that demonstrate the problems with cross-language name and subject access.

T oday's global information society, benefiting from rapidly advancing com-munication technologies, spans geographical, lingual, and cultural bound-

aries. Recognition of the need for knowledge organization and integration,and access to cross-cultural and cross-lingual resources has greatly increased.The 2004 ISKO International Conference on "Knowledge Organization andthe Global Information Society" and a 2004 special issue of Catalogingand Classification Quarterly ("Knowledge Organization and Classification inInternational Information Retrieval") are two examples.' International digitiza-tion projects have opened access to medieval texts as well as images and primarysources housed in libraries and institutions around the world, greatly advancingglobal access to multicultural resources.

The technological revolution that brought forth the global information soci-ety also has spurred recognition of the necessity for international collaborationaimed at multicultural education and diversity.2 Linguistic and computationallinguistic communities have collaborated in developing multilingual informationresource discovery tools, such as concept-based indexing. These are used pri-marily for cross-lingual information processing. One example is EuroWordNet,which is based on Princeton University's WordNet, a lexical database for theEnglish language.' The Open Language Archives Community (OLAC) has alsobeen engaged in archiving, disseminating, and preserving language and cul-tural resources, including language-engineering tools, through utilization of theDublin Core metadata standard.4

The challenges of accessing resources across cultures and languages suggestthis is an area of particular interest to librarians, who are responsible for descrip-tion and access. As a first step in exploring this topic, the author studied currentpractices in providing cross-cultural and cross-lingual information access. In thispaper, she identifies problem areas and suggests directions for future study. Thescope is limited to studies dealing with cataloging and metadata schemes for cross-cultural and cross-lingual information access.

180

Cross-lingual Name and Subject Access 181

Approaches to Cross-lingualInformation Access

The development of cross-lingual thesauri, subject headinglists, and name authorities, as well as the translation of theDublin Core (DC) metadata scheme into many differentlanguages, is ongoing. In addition to the activities of the DCMetadata Initiative for developing multilingual DC metada-ta, various approaches to building cross-lingual knowledgeorganization schemes have been developed with an eye tobetter access to multicultural and multilingual resources.'

Language engineering and linguistics communities havedeveloped lexical tools for cross-lingual resource discovery;these include machine translation, ontology, informationextraction, text summarization, and speech processing.Multilingual information resource discovery tools such asconcept-based ontology (e.g., EuroWordNet and GlobalWordNet Association) also have been developed.6 OLAChas been engaged in archiving, disseminating, and preserv-ing language-culture related resources by developing theOLAC Metadata standard, which defines the format usedfor the interchange of metadata within the framework of theOpen Archives Initiative (OAI).' The metadata set is basedon the complete set of DC metadata terms, but the formatallows for the use of extensions to express community-spe-cific qualifiers.

In library communities, cataloging and metadata stan-dards have been internationalized. Cross-lingual subjectaccess via conceptual mapping of Library of CongressSubject Headings (LCSH) and cross-lingual name accessthrough cross-linking of Library of Congress (LC) nameauthorities have been undertaken. The following sectionspresent a literature review and identify the challenges inher-ent in transliteration and word segmentation in nonromanscripts, with particular attention to Korean. Challenges inbuilding subject heading and name authority files for cross-lingual information access -also are discussed.

Cross-lingual Subject Access: ConceptualMapping Mechanisms

Heiner-Freiling reported the results of a survey of nationallibraries on subject headings conducted under the auspicesof the International Federation of Library Associations andInstitutions (IFLA)." According to the survey data, LCSHis predominantly used in twenty-four national libraries ofEnglish-speaking countries; in addition, a translated ormodified version of LCSH is being used in twelve othercountries. Several authors have written on the problemscaused by translated subject headings across languages andcultures.'

Subject headings of Korean collections in NorthAmerican libraries are based largely on LCSH, a transla-tion from the source language (i.e., Korean) into LCSH inEnglish. This author presented an earlier analysis of theproblems in subject headings translated between Englishand Korean.1" Problems that occur in translated subjectheadings likewise can be expected to occur in any metadatamapping process between the two languages.''

The concepts of LCSH are formulated into various syn-tactic forms-single noun, compound noun, noun phrase,and inverted phrase. The concept of a heading can beexpressed in several different fonns, leading to potentialcomplexities and inconsistencies. Partially due to the multi-ple morpho-syntactic forms used in expressing the same con-cept, cataloger inconsistencies exist even when working witha single language, such as the assignment of subject headingsin English by an English-speaking cataloger to works in theEnglish language. The translation process between two lan-guages only exacerbates such inconsistencies.

Korean subject cataloging suffers from the inevitabledrawbacks of assigning Korean concepts by employingEnglish subject headings. The conceptual mismatch anddifficulties of translation from one language to another arelargely due to different linguistic structures and socio-cul-tural norms. In the case of English and Korean, these struc-tural differences are considerable, unlike between Englishand Spanish, because English and Korean are unrelatedlanguages. For example, Korean is an agglutinative languagein which functional particles, such as case markers and func-tional affixes, are attached onto the content words as grain-matical operators. On the other hand, English and Spanishlack such characteristics. Instead, they are heavily depen-dent on word order to designate grammatical function. Themanner of conveying a semantic concept may be manifesteddifferently in Korean and English language users. Such dif-ferences in conceptual manifestation are greatly increasedin the process of translation.

The following example of a translated subject headingexemplifies these problems. The romanized Korean coi-pound phrase Hanguk nwl could be translated as:

A Korea language/The Korean languageThe language of Korea/language of KoreaKorea and a language/K)rea and languages

The Korean heading may be translated into Englishwith various forms. Major differences among these possibleheadings include the following: the prepositional phraseThe language of Korea and the conjunctional phrase Koreaand languages show indefinite and definite article variants(a versus the) and inflectional variants (language versuslanguages). Written Korean employs grammatical devices

51(3) LRTS

182 Park

such as particles (e.g., case markers denoting subject andobject) and suffixes. These can be omitted in the spokenform without causing any communicational ambiguities. Inthe written language, as with Hanguk mal in the previousexample, the omission of such functional words readily givesrise to ambiguity: Hanguk ui inal is translated as the prepo-sitional phrase The language of Korea. On the other hand,Hanguk kwa nel is translated as Korea and languages. Thus,omission of the grammatical particles ui "of" and kwa "and"creates conceptual ambiguity.

Kwasnik and Rubin examined challenges in conceptualtranslation of classification schemes across languages andcultures. 2 They assessed differences in kinship terms infourteen languages, revelating the challenges and problemsinherent in the process of translation of a classificationsystem. As a framework for culturally sensitive classifica-tion translation, certain modifications to the classificationsystem (adding or deleting terms or both) reflect individuallinguistic and cultural characteristics and are inevitable. Inthe case of one-to-two mapping, creation of cross-referencesis a practical step forward in clarification. In a similar man-ner, the use of modifiers or scope notes in order to avoidconceptual ambiguity would be advisable.

Multilingual Access to Subjects:Cross-linking Mechanisms

To date, the major project on multilingual subject headingshas been Multilingual Access to Subjects (MACS), whichaims at providing English, French, and German subjectaccess in library catalogs through cross-linking techniques.Clavel-Merrin, MacEwan, and Landry reported on this proj-ect.13 The project has been conducted by European nationallibraries-the Swiss National Library, the Bibliothýquenationale de France, The British Library, and Die DeutscheBibliothek-through international collaboration underthe auspices of the Conference of European NationalLibrarians."4

The cross-linking technique is based on conceptualmapping among the authorized headings of three subjectlists: English-LCSH, French-RAMEAU (R6pertoired'autorit6 matiere encyclop6dique et alphab6tique unifi6)and German-SWD/RSWK (Schlagwortnormdatei/Regelnftir den Schlagwortkatalog). Through a manual cross-linkingprocess, conceptually equivalent linking is established. If noequivalent concept exists across the three subject headings,the heading stands alone.

The project began with a subset of headings in theareas of theater and sports. The rationale for selecting thoseareas was to test universality and cultural variation. Thearea of sports would be expected to have a high conceptualcorrespondence across the three languages and the threesubject heading lists because the area of sports is considered

to be a less culture-bound domain; conversely, the area of'theater reflects culture-specific terms and concepts andlow correspondence across these subject headings woiildbe expected.

As expected, cross-linking in the area of sports yieldeda high degree of equivalence. MacEwan reported thatwhen comparing terms in a sample of 278 sports subjectheadings, 86 percent of headings matched across all threesubject headings lists, 8 percent of headings matched acrosstwo lists, and 6 percent of headings were unmatched.'5 Inthe more culture-bound domain of theater, the cross-link-ing match was much lower than in the less culture-hounddomain of sports. MacEwan reported that, when comparingterms in a sample of 261 theater subject headings, 60 per-cent of headings matched across all three subject headinglists, 18 percent matched across two lists, and 22 percent ofheadings were unmatched.'"

A concept realized as a word in one language can beequivalent to a linguistic morpheme (the smallest unit ofmeaning in oral and written language), word, phrase, orclause in other languages. Thus, syntactic variations areexpected to hinder the mapping process. MacEwan gave anexample of the challenge seen in creating a conceptual link-ing system across three subject headings (English, French,and German) in the following: "Track athletics-Coaches inLCSH matches with Leichtathletiktrainer in the SWD, butin RAMEAU it is only matched by adding a subdivision tothe authority record at the point of indexing a document:Athletiste-Entraineurs."'- To alleviate mapping problemscaused by such syntactic variations, links between headingsand strings are allowed. In addition, the creation of newheadings is allowed to create a conceptual mapping betweenthe subject heading lists, as long as there is literary warrantin the catalog of the user institution.

Conceptual Mismatch betweenTarget and Source Languages

The conceptual mapping process is analogous to translatingtwo or more different languages. Figure 1 illustrates somepossible conceptual mismatches in the process of semanticmapping between two languages. Precise and equivalentmapping between two languages in translation does notexist. The first and second diagrams in figure I illustratethe necessity for strategies to deal with inexact equivalencein the case of one-to-many and manv-to-one mapping. Inthe case of no conceptual equivalence, shown in the thirddiagram, the general concept in the target language mightserve as an alternative for semantic mapping. However; dueto the lack of specificity, the altemative general conceptmay not contain the original source concept, resulting in anunavoidable limitation in cross-linguistic situations.

LRTS 51 (3)

Cross-lingual Name and Subject Access 183

Owing to the dramatically different language struc-tures and cultural bases of Korean and English, translatedsubject headings involving these languages frequently arenot equivalent to the concept of the original heading. Theconcept of the translated headings is either overly broad orthe headings do not retain the original meaning. Thus, amore thorough analysis and understanding of the very dif-ferent Korean and English language structures are neededto alleviate this inevitable difficulty.

The subject heading that follows, taken from a MARCrecord describing a Korean monograph, pnmiasi, ilhlstratesthe challenges faced in conveying the original concept inthe process of mapping from the Korean concept of a word

to LCSH.

650 0651 0

Interpersonal relations.Kyonggi-do (Korea)$x social life andcustoms.

The title of the book is puniasi (exchange of services/labor) wa (and) chong (affection) ui (of) ingan (human)kwangye (relationship). The translation could be The inter-

personal relations of the exchange of labor and affection.The word pumasi describes the social structure of Korea inthe agricultural context. The pumasi is the system by whichpeople effectively provide help to one another. People whoare in need can obtain financial and other help from othersfor a short period without paving interest. They will returnthe pumasi on some other occasion when the people whogave help are themselves in need of help. This system wasoriginally developed in a traditional agricultural society andthen transferred into the urban society of modern Korea.The underlying concept of pumasi may be stated thus: soli-darity with affection in a community.

LCSH does not have a heading that is equivalent tothe pumasi system. This is because pumasi is a product ofKorean culture. In order to denote the subject heading,then, a broad and general heading such as social life and

cistonis would be employed for this monograph in thetopical subdivision of the heading (i.e., 651). As can be seen,the translated subject heading in the above record loses theoriginal concept of the Korean heading due to conceptualmismatch.

Cross-lingual Name Access throughCross-linking Mechanisms

Two major projects on cross-lingual name access throughthe cross-linking mnechanism utilizing roman script currentlyare employed. One is the Virtual International AuthorityFile (VIAF), a joint project between LC and Die DeutscheBibliothek, with OCLCs research support."s VIAF is a

Figure 1. Conceptual equivalence

Source: Jung-ran Park, "Hindrances in Semantic Mapping among MetadataSchemes: A Linguistic Perspective," Journal of InternCt Cataloging 5, no.3 (2002): 74.

single personal name authority file that combines the nanieauthorit' files of both institutions through the cross-linkingmechanism.

In the VIAF project, the anithoritv records from l)ieDeutsche Bibliothek are matched to the corresponding LCauthority records through the cross-linking mechanism.Following this linking process, maintaining the authorityfiles and providing user access to the files will be throuighthe shared OAI servers. Upon the completion of the proj-ect, each user group in the United States or Germany will beable to view personal name records established bv tihe otherinstitution and view the personal name records of each usergroup's own language.

The other project dealing with roman script is Linkingand Exploring Authority Files (LEAF), which was estab-lished in 2001 with the involvement of fifteen organizationsutilizing eight languages." Clavel reported two principalchallenges in establishing a cross-lingual authority file. 20

Both challenges are derived from linguistic variation andambiguities across languages. First are language-specificfeatures such as the order of components in compotmndnames, location of particles, and iuimnbering systein for kingsand popes. The second challenge concerns standardizationof methods for disanibiguation of honionYmns. Natural ],an-

Source concept equivalent to several target concepts:

Source Target

Two or more source concepts equivalent to one target concept:

Source Target

I"A ..- 'B

No conceptual equivalent between the source concept and the target concept:

Source 'Target

51 (3) LRTS

184 Park

guage is full of lexical ambiguities. For instance, homonymycreates ambiguity (e.g., bank [building] versus bank [river]).Homonyms have the same lexical form but manifest unre-lated meanings that are arbitrarily developed. The Anglo-American Cataloguing Rules, 2nd edition (AACR2) chapters22 through 26 present pragmatically constrained disambigu-ation techniques for the names of persons, corporate bod-ies, and places by differentiating contexts.2' For example,to disambiguate identical names, birth or death dates (orboth) are added (e.g., John Q. Smith [1904-1972] versusJohn Q. Smith [1905-]). In the case of ambiguous corporatebody names, a qualifier is added-e.g., John Smith (firm).According to Clavel, the addition of academic and nobil-ity titles is generally standardized for disambiguating hom-onyms.22 The specification of profession or activity, however,is much less standardized. Accordingly, this creates prob-lems in cross-linking of authority files across languages.

Several authors have looked at nonroman scripts (par-ticularly East Asian languages such as Korean, Japanese, andChinese) and have found that transliteration causes cross-lingual name access problems, because of the nature of thelanguage. 23 Names in the Korean, Chinese, and Japaneselanguages utilize Chinese ideographs owing to a commonhistory; thus, variant forms of names are represented inthese languages. For example, in the case of the Koreanname, Hangul (Korean vernacular script), Chinese ideo-graph and the transliterated form are all used.

When discussing Korean, one must take into accountthe differences in transliteration schemes between thosebased on phonetic structure and those based on morphemicstructure. Differences in transliteration schemes are alsoapplicable to other nonroman scripts.

For instance, LC's relatively recent adoption of thePinyin transliteration scheme from the Wade-Giles schemein transcribing Chinese language materials illustrates thecomplex issues surrounding the differences in translitera-tion schemes even involving the same language. Arsenaultreported on an experiment in retrieval efficiency amongmonosyllabic Pinyin, polysyllabic Pinyin, and Wade-Gileswhile searching known item exact title and keywords intitle.24 The findings of the study demonstrate that the poly-syllabic Pinyin system, which transcribes Chinese accordingto syntactic unit (i.e., word by word), significantly increasesretrieval efficiency compared to monosyllabic Pinyin andWade-Giles, which share the feature of transcribing Chinesemorpheme by morpheme.

Naito presented a variety of ways of transcribing thesame Japanese name, such as phonetic transcription inHiragana and phonetic transcription in Katakana, transcrip-tion in simple form, and Chinese scripts.2" Table 1 (fromNaito) illustrates this.

This author presented issues relating to the Koreantransliteration scheme.2 6 In South Korea, no unified trans-

literation scheme is used. Different transliteration schemesare employed in different sectors for varying uses. Forexample, libraries and publishing industries employ theMcCune-Reischauer (MR) system in publication and bib-liographic records." The Yale system is uniformly used bylinguists within Korea and abroad.2" Lastly, governmentdocuments, including street signs and road maps, employthe Ministry of Education system.29

The differences among these schemes reflect the lin-guistic representation of sound systems. The MR systemand Ministry of Education system are based on the pho-netic structure of Korean. Transliteration based on phoneticstructure encodes words in the manner in which they arepronounced. For example, in English the word two is tran-scribed phonetically as [tu].

The Yale system is based on morphemic structure.Morphemic structure-based transliteration transcribes thebase form of a word regardless of sound changes. Korean isa language that employs rich morpho-phonemic complexity.The base form of a word changes according to the adjacentsound environment. Most agglutinative languages, includingJapanese, fall into this category. They are all very compli-cated morpho-phonemically. For example, the form of theKorean word mul (water) is changed into muri when thesubject case particle -i is attached to it. Morphemic struc-ture-based transliteration is not reflective of sound changeas is the phonetic type of transliteration utilized in the MRscheme; instead, it reflects the base form.

The current cataloging system dealing with Koreanmaterials employs the MR transliteration scheme. One ofthe major drawbacks of the use of the MR system is that itcauses semantic loss. This is especially critical in the area ofname access. Transliteration of words following the way inwhich they are pronounced has the potential of representing

Table 1. Japanese personal name

,9 R • •Form he usedX X Simplified character of "•IF< 62 #5% £ Phonetic description in

Hiraganaý7 1 r7 7" •Phonetic description in

Katakana"7147 79 Phonetic description in

Katakana by computer half-width character still in use

Kurosawa Akira Romanized form

KUROSAWA Akira Family name + Given name

Akira Kurosawa English form (?)

Akira KUROSAWA Given name + Family name

Source: Eisuke Naito, "Names of The Far East: Japanese, Chinese andKorean Authority Control," Cataloging & Classification Quarterlv 38, no.3 (2004): 257.

LRTS 51 (3)

Cross-lingual Name and Subject Access 185

a name ambiguously. For example, the Korean name KimSok-min becomes Kim Song-min according to the MR sys-tem. With author names transliterated according to the MRsystem, ambiguity becomes almost inevitable. The linguistRamsey noted that "This information loss becomes espe-cially critical when all cataloging work is done by computer,and so it is perhaps time to give some thought as to howappropriate McCune-Reischauer is in cases where precisedata processing is required."3'

The MR system, based as it is on phonetic structure,does not disambiguate different meanings of homographs(i.e., same words but different meanings), one of the pri-mary causes of semantic ambiguity. This phenomenon canbe illustrated by an example in English: two, to, too. If thesethree lexical items are transcribed according to the pronun-ciation [tul, the resulting semantic ambiguity can be clearlyseen. This happens frequently with the MR scheme. Suchambiguities inevitably cause significant impediments in theprocess of information retrieval.

In addition, the MR system results in variations in thecreation of bibliographic records. When catalogers tran-scribe words according to pronunciation, they can createinconsistent and arbitrary records. This is based on thefact that the pronunciation of words can vary according tospeech style. If a cataloger pronounces a word or phraseusing careful speech style, the resulting transcription wouldbe different from that of a transcription based on casualspeech style. The creation of differing bibliographic recordsis thus entirely possible, either by the same cataloger or dif-ferent catalogers transcribing identical material.

The following bibliographic record illustrates thisproblem.

100 1 Kim, Young-un,$1927-245 10 Ceh-2 kAonggungnon: $bkungmin

kukka Aui wansAong Aul wihayAo/$cKim Yong-un.

246 3 Ch"io"an.260 SAoul T"AukpyAolsi :$Chisik SanAopsa,

$c1998.

The portion of the title field (245) in bold,kAonggungnon, reflects the casual speech style. If thecataloger who created this record had pronounced it usingcareful speech, the final consonant of the first syllable(i.e., kon) remains as a nasal sound, as indicated in bold:kAongungnon, as opposed to kAonggungnon. In casualspeech, however, the nasal sound [n] becomes assimilatedinto the following velar sound [ng].

The MR transliteration scheme contains inherentinconsistencies that can have a significant impact on infor-mation organization and retrieval. Semantic ambiguity,inconsistency, and semantic loss are critical issues hinder-

ing information retrieval and sharing bibliographic records.Consequently, the goals of bibliographic control are notachieved.

Problems of Word Segmentation

Difficulty in word segmentation occurs in agglutinativelanguages such as Japanese and Korean because of theirinherent morpho-syntactic flexibility. Agglutinative lan-guages allow functional particles such as case markers andinflectional affixes to be attached onto the content words asgrammatical operators. For example, the word muli [water +subjective case affix] is composed of the content word (i.e.,mul: water) and the functional affix (i.e., i: subjective casemarker). This creates flexible word segmentation betweenfunctional and content words. Such flexibility of wordsegmentation in Korean creates inconsistent and arbitrarypractice in word division; such inconsistency can be found ineven the most authoritative Korean dictionaries. Accordingto Yi Sung-u, word segmentation errors appear in 29 percentof Korean standard books in the school system.' This high-lights the difficulty in conducting word segmentation in thewritten Korean form.

Arbitrary word segmentation does not cause commu-nication problems in everyday language use, since com-municative ambiguities stemming from inconsistent wordsegmentation can be resolved through contextual cues.However, such flexibility in word segmentation is a criti-cal factor in hindering information sharing and discoveryin the digital environment, which does not provide contex-tual cues.

The Library of Congress ALA-LC Romanization Tablesprovides rules specifying word segmentation and offerfour basic underlying principles.,2 The first basic principleis "Each word or lexical unit (including particles) is to beseparated from other words." :"3 The following Korean bib-liographic record illustrates this principle.

245 00 Ynoksa sok Auij in'gan kwa ehisAong

Aul t"amgu handa /$c Kim Chae-yong

... I et al.] p"yAon.250 Che l-p"an.260 SAOul :$bHan'gilsa,$c1998.

The title field (245) can be segmented in the followingway: Yoksa sokA uiA in'ganA kwaA chisongA ulA t"a11guA

handa. The segmentation is denoted by the mark A, desig-nating a total of eight word divisions. This principle followsone of the suggestions presented at the 1981 workshopconference on Korean transliteration, held at the Universityof Hawaii under the auspices of the Korean StudiesCenter, and reported by Austerlitz.3 4 The main aim of the

51(3) LRTS

LRTS 51(3)

conference was to examine the Korean transliteration sys-tem (i.e., MR system) to produce consistent guidelines fortransliterating Korean language.

This principle creates problems when users search abibliographic record because word division following theLC principle is not utilized by Korean users; it is contraryto conventional practices of the language. The previousexample title consists of only three word divisions in theKorean written form: YoksasokuiA in'gankwaA chisongualA

tCamguhanda. Moreover, this rule presents another intrinsicdifficulty. It applies only to case particles of a noun phrase,not to affixes of verb phrases. Thus, the word division prin-ciple is not applied to entire units of the sentence.

The MR transliteration scheme based on phoneticstructure has critical drawbacks because it causes seman-tic loss, semantic ambiguity, and cataloging inconsistency.A transliteration scheme based on morphemic principleshas substantial merit because it significantly contributes toresolving semantic ambiguity and inconsistency. One of theprincipal advantages of basing transliteration on morphemicprinciples is that the need for diacritical symbols also issubstantially reduced, in contrast to a transliteration schemebased on phonetic principles, which increases the employ-ment of diacritical symbols.

Word segmentation in agglutinative languages is veryflexible. Even though guidelines and rules for word divisionexist, inconsistent and arbitrary practices are inevitable. Anautomatic parser of word segmentation based on linguisticprinciples is critically needed to ensure consistency of bib-liographic records.

Linguistic Universality and Relativityacross Language Structures

Impediments to enhancing access to cross-cultural andcross-lingual resources are largely derived from the coinplexities and variation of linguistic structures across lan-guages. Linguistic and cultural approaches in developingcross-lingual and cross-cultural knowledge organizationsystems are critically needed.

The facility of natural language, in all its complexity,variability, and richness, is the defining aspect of humanity.This very complexity of expression and richness of lexical-ization and linguistic structures becomes problematic inthe electronic environment of information retrieval. Eventhough natural language possesses some characteristics thatare independent of a specific language, many more lan-guage-specific characteristics exist. Such language-specificcharacteristics demonstrate that the structure of languageis so closely intertwined with its source culture and societythat it is inseparable from it. Natural language is not justmere arrangements of words, but the mirror of culture.

Combinations and arrangements of words do not reflectspecific cultural and pragmatic meanings that are inherentcharacteristics in any given language structure.

Language-specific variations and differences in lexical-ization patterns can be found easily in everyday language

uses such as naming conventions, kinship terms, addressforms, numbering systems, color terms, and names for bodyparts. For example, in Anglo-American society, building des-ignations (e.g., LeBow College of Business), brand names(e.g., Ford), and even common reference nouns (e.g., miav-erick, boycott, lynch) originating from family names or titlesare common. Conversely, this phenomenon is nonexistentin Korean language and society. Thus, one can say that thisEnglish-specific naming convention manifests the culturaltrait of Anglo-American society.

Collectivist-oriented cultural and social norms, basedon hierarchical structure, are closely reflected in the Koreanlanguage. This can be especially seen in the sophisticatedhonorific system and in the employment of various linguisticdevices, such as lexical items existing in both plain and hon-orific form (e.g., na/cho [plain/honorific form] 'F, nai/yonsey[plain/honorific form] 'age', chada/chumnusida [plain/honor-ific form] 'sleep: verb), to name a few. It is also seen in syn-tactic structures (e.g., honorific agreement in subject/object,predicate, and case markers). Such variant lexical forms aremerely one illustration of a synonymy phenomenon that isnot found in English, as shown in table 2.

The Need to Develop InteroperableGuidelines for Cross-linking Names and

Subjects and Conceptual Mapping

A critical need for the development of common guidelinesfor cross-linking of names (e.g., person, place, corporatebody) across languages exists. Development of such interop-erable cross-linking guidelines should be guided by theexamination of morpho-syntactic variations across languagestructures, especially for the structures of names.

Word segmentation and transliteration schemes dealingwith nonroman scripts also play a part in limiting access tocross-lingual and cross-cultural resources. Standardizationof such transliteration schemes and development of mecha-nisms geared toward consistent word segmentation also arecritically needed. Specifically, reexamination of translitera-tion schemes and development and application of a morpho-syntactic parser based on linguistic principles for automaticword segmentation are vital conditions for cross-lingualinformation access.

Development of knowledge organization schemes forcross-lingual subject access also is hindered by the lack ofcommon conceptual mapping criteria that are interoperal)leacross languages and cultures. Semantic mapping, involv-

186 Park

Cross-lingual Name and Subject Access 187

Table 2. Korean lexical honorific system

English equivalent Plain form Honorific form

age: noun nai vonse

house: noun chip taek

sleep: verb chada chumusida

eat: verb mokta tusida

Source: Jung-ran Park, "Hindrances in Semantic Mapping among MetadataSchemes: A Linguistic Perspective," Journal a/Internet Cataloging 5, no.

3 (2002): 63.

ing inetadata and subject heading lists across languages,is one of the most critical issues in resource discoverv andinformation exchange. Without achieving interoperability ofsemantic mapping, application of cross-lingual knowledgeorganization tools for the retrieval of networked resourceswill be significantly hindered. In order to develop interop-erable conceptual mapping guidelines across languages andcultures, identification of lexicalization patterns based onsemantic, syntactic, and pragmatic linguistic analysis is criti-cally needed.

Cross-linguistic differences result in conceptual andlexical gaps and overlaps between target and source lan-

Are multiple subscription sources

EBSCO brings simplicity and efficiency to the management ofsubscriptions, from order placement to renewal. As your single-sourcesupplier for print, e-journal, e-book and database subscriptions, EBSCOhandles all ordering and countless access details on your behalf.

Our proven approach saves your staff time and your library money byeliminating the need to communicate with multiple contacts and dealvvith assorted reporting methods. EBSCO's subscription managementservice provides you with realistic, helpful solutions.

guages that present themselves during the inapping process.Conceptual mapping between languages presents a varietyof lexical gaps and overlaps including inexact equivalence,partial equivalence, nonequivalence, and single-to-multipleequivalence. Culture-specific language characteristics sug-gest that, in order to overcome problems in the develop-ment of cross-lingual knowledge organization tools (e.g.,subject headings, thesauri, metadata) and to ensure interop-erability among these tools cross-linguistically, language-specific characteristics must be taken into account.

Conclusion

Complexities and variations of linguistic stnictures acrosslanguages and cultures have a significant effect on name andsubject access across languages. Thus, study offinguistic andcultural approaches to developing cross-lingual and cross-cul-tural knowledge organization systems is critically needed. Themajor research gaps in current literature concern addressingissues in relation to developing interoperable guidelines forcross-linking of names and developing common conceptualmapping criteria that are interoperable across languages andcultures for cross-lingual subject access.

INFORMATION SERVICES

www.ebsco.com

51(3) LRTS

188 Park

This underlies the necessity of future studies in morpho-syntactic variation across languages for cross-lingual nameaccess and an examination of lexicalization patterns basedon semantic, syntactic, and pragmatic linguistic analysis forcross-lingual subject access. Drawbacks in word segmenta-tion and transliteration schemes dealing with nonroman lan-guages also call for reexamination of transliteration schemesand for the development of a morpho-syntactic parser forautomatic word segmentation.

References

1. International Societyfor Knowledge Organization, "KnowledgeOrganization and the Global Information Society," 8thInternational IKSO Conference. www.ucl.ac.uk/isko2004/index.htm (accessed Aug. 4, 2006); J. Williamson and ClareBeghtol, eds., "Knowledge Organization and Classificationin International Information Retrieval," Cataloging &Classification Quarterly, 37 no. 1/2 (2003); Alvaro Quijano-Solis, Pilar Maria Moreno-Jimenex, and Reynaldo Figueroa-Servin, "Automated Authority Files of Spanish-LanguageSubject Headings," Cataloging & Classification Quarterly29, no. 1/2 (2000): 209-23; Reynaldo D. Figueroa-Servinand Berta Enciso, "Subject Authority Control at El Colegiode Mexico's Library: The Whats and Hows of a Project,"Cataloging & Classification Quarterly 32, no. 1 (2001):65-80; Jung-ran Park, "Hindrances in Semantic Mappingamong Metadata Schemes: A Linguistic Perspective," Journalof Internet Cataloging 5, no. 3 (2002): 59-79; Barbara H.Kwasnik and Victoria L. Rubin, "Stretching ConceptualStructures in Classifications across Languages and Cultures,"Cataloging & Classification Quarterly 37, no. 1/2 (2003): 33-47; Zahiruddin Khurshid, "Arabic Script Materials: CatalogingIssues and Problems," Cataloging & Classification Quarterly34, no. 4 (2002): 67-77; Annarita Sanso, "Ancient ItalianState: An Authority List Project," Cataloging & ClassificationQuarterly 39, no. 1/2 (2004): 577-84.

2. John Agada and Brenda Hough, "The Emporia-Nigeria Project:Building Global Partnerships to Support Development of CivilSociety in Nigeria," unpublished paper presented at the 2005ALISE Annual Conference, Jan. 11-14, 2005, Boston; SergioChaparro-Univazo, "Some Issues on LIS Education andCollaboration in Latin America," paper presented at the 2005ALISE Annual Conference, Jan. 11-14, 2005, Boston. http://dlist.sir.arizona.edu/701 (accessed Mar. 24, 2006); BharatMehra, "Cross-Cultural Perspective of International DoctoralStudents: 'Two-way' Learning to Further Internationalizationin LIS Education," unpublished paper presented at the 2005ALISE Annual Conference, Jan. 11-14, 2005, Boston.

3. Cognitive Science Laboratory, Wordnet: A Lexical Databasefor the English Language. http://wordnet.princeton.edu(accessed Dec. 1, 2006); EuroWordNet. www.illc.uva.nl/EuroWordNet (accessed Dec. 1, 2006).

4. Christiane Fellbaum, ed., WordNet: An Electronic LexicalDatabase (Cambridge: MIT Pr., 1998); George A. Miller etal., "Introduction to WordNet: An On-Line Lexical Database,"International Journal of Lexicography 3, no. 4 (1990): 235-44;

LRTS 51(3)

Eduard Hovy et al. Multilingual Infornation Management:Current Levels and Future Abilities. A report comnnissionedby the U.S. National Science Foundation and also deliv-ered to the European Commission's Language EngineeringOffice and the U.S. Defense Advanced Research ProjectsAgency, Apr. 1999. www-2.cs.cinu.edu/-ref/mlimn/index.shtml (accessed Aug. 4, 2006); Pick Vossen, "EuroWordNet:Building a Multilingual Database with Wordnets for EuropeanLanguages," ELRA Newsletter 3, no. 1 (1998): 7-10.

5. Jung-ran Park, "Language- Related Open Archives: Impacton Scholarly Communities and Academic Librarianship,"E-JASL: The Electronic Journal of Academic and SpecialLibrarianship 5, no. 2/3 (2004), http://southernlibrarianship.icaap.org/content/v05n02/park jOl.htm (accessed Aug. 4,2006); Jung-ran Park, "Human Language: Resources fromLinguistics and Beyond," College & Research Libraries News66, no. 3 (2005): 173-76, 228.

6. EuroWordNet, www.illc.uva.nl/EuroWordNet (accessed Mar.24, 2006); Global WordNet Association, www.globalwordnet.org (accessed Mar. 24, 2006).

7. OLAC: Open Language Archives Community. www.language-archives.org (accessed Aug. 4, 2006); OLAC Metadata[standard], www.language-archives.org/OLAC/metadata-20070405.html (accessed Mar. 24, 2006).

8. Magda Heiner-Freiling, "Survey on Subject HeadingLanguages Used in National Libraries and Bibliographies,"Cataloging & Classification Quarterly 29, no. 1/2 (2000):189-98.

9. Quijano-Solis, Moreno-Jimenex, and Figueroa-Servin,"Automated Authority Files of Spanish-Language SubjectHeadings," 209-23; Figueroa-Servin and Enciso, "SubjectAuthority Control at El Colegio de Mexico's Library," 65-80;Park, "Hindrances in Semantic Mapping," 59-79; Kwasnikand Rubin, "Stretching Conceptual Structures," 33-47;Khurshid, "Arabic Script Materials," 67-77; Sanso, "AncientItalian State," 577-84.

10. Park, "Hindrances in Semantic Mapping," 59-79.11. Jung-ran Park, "Semantic Interoperability across Digital

Image Collections: A Pilot Study on Metadata Mapping,"in CAIS/ACSI 2005 Data, Information, and Knowledge in aNetworked World, ed. Liwen Vaughan. Proceedings of the2005 Annual Conference of the Canadian Association forInformation Science held with the Congress of the SocialSciences and Humanities of Canada (University of WesternOntario, London, Ontario, June 2-4, 2005). www.cais-acsi.ca/proceedings/2005/park- -2005.pdf (accessed Mar. 24,2006).

12. Kwasnik and Rubin, "Stretching Conceptual Structures,"33-47.

13. Genevieve Clavel-Merrin, "MACS (Multilingual Accessto Subjects): A Virtual Authority File Across Languages,"Cataloging & Classification Quarterly 39, no. 1/2 (2004):323-30; Andrew MacEwan, "Crossing Language Barriersin Europe: Linking LCSH to Other Subject HeadingLanguages," Cataloging & Classification Quarterly 29, no.1/2 (2000): 199-207; Patrice Landry, "Multilingual SubjectAccess: The Linking Approach of MACS," Cataloging &Classification Quarterly 37, no. 3/4 (2004): 177-91.

Cross-lingual Name and Subject Access 189

14. MACS: Multilingual Access to Subjects. https://macs.vuib.ac.be/pub (accessed Mar. 24, 2006).

15. MacEwan, "Crossing Language Barriers in Europe," 199-207.

16. Ibid.17. Ibid., 203.18. OCLC Research Projects, "Virtual International Authonity

File." www.oclc.org/research!/projects/viaf (accessed Aug. 4,2006).

19. Institute of Inforniation Systems & Information Management,Joanneuni Research, "LEAF-Linkingand ExploringAuthorityFiles," www.joanneum.at/index.php?id=358&L=1&L=1 (ac-cessed Mar. 24, 2006); Jutta Weber, "'LEAF: Linking andExploring Authority Files," Cataloging & ClassificationQuarterly 38, no. 3/4 (2004): 227-36.

20. Pierre Clavel, "LEAF," paper presented at ELAG 2003: CrossLanguage Applications and the Web (27th Library SystemsSeminar, Switzerland, Apr. 2-4, 2003). http://elag.kb.nl/elag2003/www.elag2003.ch/pres/pres-clavelp.pdf (accessedMar. 24, 2006).

21. Anglo-American Cataloging Rules, 2nd ed., 1998 rev. (Ottawa:Canadian Library Assn.; London: Library Assn. Publishing;Chicago: ALA, 1998).

22. Clavel, "LEAF."23. Eisuke Naito, "Names of the Far East: Japanese, Chinese,

and Korean Authority Control," Cataloging & ClassificationQuarterly 38, no. 3 (2004): 251-68; Jung-ran Park,"Information Retrieval of Korean Materials Using the CJKBibliographic System: Issues and Problems," in Proceedingsof the Second KSAA Biennial Conference: Korean Studiesat the Dawn of the Millennium, Young-A Cho, ed. (MonashUniversity, Melbourne, Australia, Korean Studies Associationof Australasia, Sept. 24-25, 2001), 245-55. wNxv.arts.inonash.edu.au/koreanl/ksaa/coinference/2 ljungranpark.pdf (accessedMar. 24, 2006); Hee-sook Shin, "Quality of Korean CatalogingRecords in Shared Databases," Cataloging & ClassificationQuarterly 36, no. 1 (2003): 55-90; Jian Wang, "Chinese Serials:

History, Characteristics, and Cataloging Considerations,"Cataloging & Classffication Quarterly :36, no. 1 (200.3):41-54.

24. Cl6nient Arsenault, "Word Division in the Transcription ofChinese Script in the Title Fields of Bibliographic Becords,"Cataloging & Classification Qaarlerly 32, no, 3 (2001):109-27.

25. Naito, "Names of the Far East," 251-68; Park, "InforinationRetrieval of Korean Materials."

26. Park, "Information Retrieval of Korean Materials," 245-55.27. George M. McCune and Edwin 0. Reisehauer, "'I'le

Romanization of' the Korean Language: Based upon ItsPhonetic Structure," Transactions of the KYora Brane'h of thl'Royal Asiatic Society 29 (1940): 1-55.

28. Yale Romanization: Encyclopedia. http://e(xperts.a)o(uit.coiii/e/y/ y,aYale_Roiiianization.htm (accessed Aug. 4, 2006); YaleSystem of Korean Romanization. wwv.tifs.ac.jl)/ts/persoial/choes/Eyale.htnd (accessed Aug. 4, 2(X)6).

29. Ministrn of Culture and Tourism of the Republic of' China,The Ro,utaization of Korean (Seoul: Ministrs of Cultireand Tourism, 1979); Munhwa Kang, wangbu kosi (TheMinistry of Culture and Tourism of the Republic of KoreaAnnouncement). Ministry of Culture and Tourism of theRepublic of China, Sac Roinacha pyjouipojp (New Svstem ofRomanization for the Korean Language, 2000).

30. S. Robert Ramsey, "Writing Korean \vith Roman Letters,"Korean Journal 22, no. 8 (1982): 33.

31. Sung-u Yi, Sac Machninpop kica Ky!/jowg ui Sirclhes (Seoul:Omungak, 1993).

32. Library of Congress and the American l,ibrary Association,ALA-LC Ronianization Tables: Transliterat'ion sc'ieicsfo)r Non-Roman Scripts, (Washington, D.C.: CatalogingDistribution Service, 1991).

33. Ibid., 82.

34. Robert Austerlitz et al., "Report of the Workshop Conferenceon Korean Ronianization," Korean Studies 4 (1980): 111-25.

51(3) LRTS

COPYRIGHT INFORMATION

TITLE: Cross-lingual Name and Subject Access: Mechanisms andChallenges

SOURCE: Libr Resour Tech Serv 51 no3 Jl 2007

The magazine publisher is the copyright holder of this article and itis reproduced with permission. Further reproduction of this article inviolation of the copyright is prohibited. To contact the publisher:http://alastore.ala.org/