Policy: Numeric Character References


Numeric Character References or NCRs are common markup constructs used in markup languages like HTML and XML, where a sequence of characters will be rendered as a single character. NCRs are structured as ampersand ( & ), pound sign ( # ), lowercase letter x, four-position Unicode character code, and a trailing semicolon ( ; ). For example, च. This policy is about the use of NCRs in MARC cataloging records in OCLC (and hence Alma).


Catalogers most often use NCRs in the context of non-Latin scripts. Catalogers may supply parallel non-Latin fields only for scripts supported by OCLC, in accordance with our Network Level Bibliographic Mandate.

Non-Latin scripts supported by OCLC:

  • MARC-8 scripts (subsets of UTF-8 characters, so they also compatible with UTF-8 Unicode): Arabic, CJK (Chinese, Japanese, Korean), Cyrillic (within the MARC-8 character set), Greek, or Hebrew scripts.
  • UTF-8 Unicode only scripts: Armenian, Bengali, Cyrillic (outside the MARC-8 character set), Devanagari, Ethiopic, Syriac, Tamil, or Thai scripts. These scripts are not included in MARC-8.

Notice that Cyrillic is the one script that straddles both categories. In general, modern Slavic languages using Cyrillic (like Russian, Bulgarian, Ukrainian) are within MARC-8, and Old Church Slavic and non-Slavic languages using Cyrillic (like Kazakh, Uzbek and Mongolian) use additional Unicode characters.

If any non-MARC-8 scripts are exported in MARC-8 data format, the non-MARC-8 characters are saved in Numeric Character Reference (NCR) format (see Numeric Character Reference for more information). However, all settings for Alma should be UTF-8 Unicode.

NCRs should NOT be used to create non-Latin scripts for scripts not supported by OCLC. Non-Latin fields using NCRs to render nonsupported scripts should be deleted from OCLC WorldCat records when cataloging.

However, some scripts/characters that were not supported in the past are supported by OCLC now, so non-Latin fields using NCRs in those cases will need to be upgraded.

Old record using NCRs, should be upgraded when used for cataloging (Mongolian example):

Cataloging example in Mongolian

Upgraded record using newly supported characters (Kazakh example):

Cataloging example in Kazakh

Connexion Client Help advises that non-Latin characters that are not supported by OCLC can be entered and exported to a local system using Unicode, but then deleted from the OCLC WorldCat record (NOTE: this violates Alliance policy, don’t do it); or you can enter the name of the character within square brackets using the Unicode standard name (for example, enter [schwa]), or for CJK characters, enter the reading of the character (for example, enter [yin]). NOTE: Client Help does not advise using NCRs. But AskUs@oclc.org told us via email: “However, for the stray unsupported character that appears in a supported script, then the NCR is appropriate. If the script is not supported, it should not be represented in the database, but rather be transliterated.”

Exceptions to this policy may be made in the case of large record sets provided by vendors, but Alliance members must make a commitment to using the available records that most closely adhere to this policy in such cases.

See OCLC’s International Cataloging Help Pages for more details.


Software: Alma

Current phase: Phase 5: Approved

Written by: Collaborative Technical Services Team

Approved by: Collaborative Technical Services Team on 8/15//2014; Reaffirmed with no changes by Technical Services Working Group on 10/25/2016; Minor update approved by Technical Services Working Group on 11/29/2106

Last updated: 11/26/2016

Nature of last update: minor update

Document History: Minor update approved on 11/29/2106 included removing the following phrase: “Examples include Georgian, Khmer, and anything else not listed above.”