eduKate Singapore

VIEW THIS AS

Auto mode follows the Route Engine until you choose a viewpoint.

YOU ARE HERE

ROUTE CHECK

CONNECTED TO

WHAT NEXT

Use the canonical route for this room, or HELP if you are unsure.

Next useful routeCanonical hubHELP

How People Translate Quickly | MARC 21 Catalogue Translation QA: Localize Discovery Summaries Without Rewriting the Book

—

by

eduKate Asia

MARC 21 catalogue translation becomes faster when a library separates the language of a work from the language used to describe that work to readers. To localize a bibliographic record responsibly, start with a clearly identified discovery field, such as a public summary, and preserve the transcribed title, record identifiers, controlled access points and language evidence. A fluent new description cannot compensate for a catalog entry that now names the wrong edition or gives an incorrect impression of the languages inside the item.

For teams trying to create multilingual library catalogue summaries, the practical job is to add or revise a target-language discovery explanation while keeping MARC 21 field semantics and local cataloging policy intact. Fields such as 520, 245, 041 and 880 are not four interchangeable boxes for translated prose. They perform different tasks: summarizing content, recording title information, coding languages associated with the item, and representing data in alternate scripts. Understanding their mechanisms prevents a translation from becoming accidental recataloging.

This article follows one fictional education-library record from source description to reviewed multilingual display. It shows how to distinguish an original title from an editorial summary, handle linked scripts without treating them as translation slots, trace authority-controlled subjects, and test what patrons actually see. The dominant reader job is deliberately specific: produce a trustworthy translated summary for discovery, not a new cataloging code or a replacement of the source record.

1. The catalogue answers two different reader questions

A reader may ask, “What is this book called?” and then ask, “What is it about?” Those questions can be answered in different languages without becoming the same question. The title of an item is evidence from a particular manifestation or a cataloging source. A public summary is a description written or selected to help people decide whether that item is useful. Translating the summary can improve access. Replacing a transcribed title with an invented translated title can obscure what the library actually holds.

Imagine a bilingual school library holding an English-language book called River Weather: A Student’s Field Notebook. Its original title appears on the title page. The library wants parents who read Chinese to understand what the book teaches. A new Chinese summary can explain that it introduces rainfall observation, field measurements and safe interpretation of river conditions. That does not mean the original volume suddenly has a Chinese title page, or that the physical item has become a Chinese-language edition.

MARC 21 is a structured description system in which field numbers, indicators and subfields communicate different information to catalog software. One field may contain the transcribed title, another a summary, and another coded language data. A machine can use those distinctions to sort, index and display results. If a translator is given a flat dump with the numbers removed, they may produce polished language in the wrong place. The first quality requirement is therefore semantic placement, not literary elegance.

The result matters beyond one search screen. A school may export records to a union catalogue, share them with a library consortium, or reindex them in a new discovery layer. If local translated descriptions overwrite source evidence, the error can travel to systems that interpret the same field differently. A responsible workflow preserves the distinction between work, edition, local annotation and display translation so that another catalogue can make sense of the record later.

This also makes a valuable educational example: language is not a transparent label pasted onto an object. Metadata makes claims about an object. Every translated claim should be traceable to the item or to a clearly marked editorial source. The translator’s job is to improve readers’ understanding without changing the claims that the catalog has authority to make.

2. Examine field functions before choosing editable text

The Library of Congress MARC 21 reference defines field 245 as the title statement. Its subfields can distinguish the title proper from other title information and statements of responsibility. The indicators also serve cataloging functions, including whether a title added entry is made and the number of nonfiling characters. A title in this field may be required to reflect the prescribed source of information and applicable cataloging rules. It should not be treated as a marketing headline for another locale.

Field 520 contains summary and related descriptive information about content. It is the most obvious candidate for a reader-facing summary translation, subject to local practice, provenance, copyright and discovery-layer capabilities. A summary might be concise or detailed; it might be a human-written abstract, a publisher-supplied annotation or a cataloger’s note. Before translating it, ask who authored it and whether the library has permission to reuse or translate that text. Not every supplied blurb belongs to the library in the same way.

Field 041 expresses languages associated with the described item in a more detailed way than the fixed-field language code alone. It may distinguish the language of textual content, the original language of a translation, or the language of a summary or abstract in the resource. Those are facts about what the held item contains. Merely adding a Chinese-language catalog note to describe an English book does not magically make the book a Chinese-language text. Do not alter coded language data unless the bibliographic facts themselves warrant it under cataloging policy.

Field 880 represents data from an associated field in a different script and uses subfield 6 to link representations. It is important for multiscript records, but it is not a universal second-language summary field. A different-script representation may be a transliteration, an original-script title or other properly linked data according to the record’s construction. A translation that changes meaning and an alternate script representation that changes writing system can have different cataloging status. Conflating them is a subtle but consequential category error.

Controlled access points add another layer. Subject headings, classification numbers, standard numbers and authority-linked names are not merely decoration. They support consistent retrieval across records. A free translation of a subject heading might remove the match to an authority file or create a concept that sounds broader than the controlled term. A discovery interface can show local-language labels for concepts, but that display decision should not silently replace a controlled source string with an unapproved free-text phrase.

3. Choose the narrowest lawful translation job

Before opening the record in an editor, write a sentence describing exactly what will change. A sound brief might say: “Create one reviewed Chinese-language content summary for the reader interface, derived from a verified English 520 note, while retaining the original English note and all bibliographic identity fields.” That brief is narrow enough to test. The vague instruction “translate the MARC record into Chinese” invites the translator to modify every readable string, including material that should remain evidence.

Determine where the translation will live. Some library systems support multilingual note displays or separate locale-specific metadata layers. Others allow repeated notes but have no standard way to label the language of a locally added translation in their catalog interface. A local extension, companion record or discovery-layer field may be more appropriate than pretending the base MARC specification has a one-size-fits-all target-language slot. The librarian responsible for cataloging policy should approve the storage choice and its export behaviour.

This distinction prevents standards theater. It is easy to write a perfectly plausible MARC-looking example that a particular library system cannot interpret correctly. A reliable article must admit that MARC fields have defined meanings while multilingual presentation practices also depend on local profiles, cataloging rules and software. A translated 520 note can be a valid local objective; its exact representation, language labelling and export semantics require implementation-specific review.

Preserve a source snapshot and record which bibliographic item it describes. Save the stable local record identifier, relevant control numbers, holdings reference, original description, and target-language review request. Copy only the approved source note into the translator’s workspace. Mark all transcribed, coded, authority-controlled and linkage fields as read-only. A translator should not need to decide whether changing the publication date is acceptable; that field should never have been in the editable queue.

A well-designed brief also identifies audience. A summary for primary-school learners differs from one for specialist researchers, even when both are accurate. But adapting reading level must not introduce content not found in the book. If the source note says that the work discusses how to record rainfall, the summary must not promise lessons on flood forecasting unless the book actually covers them. Audience adaptation is a presentation decision under evidence constraints.

4. A worked record: describe the item, do not invent another one

Take the fictional River Weather: A Student’s Field Notebook. The title page gives the English name; the item teaches observation and recording of local weather and water conditions. The original catalog has an English content summary. A Chinese-language discovery interface wants to offer a short, accurate description so parents can decide whether the book suits a child learning environmental science.

An editor might be tempted to replace the title proper with an appealing Chinese headline meaning “The Secret Life of Rivers.” That sounds attractive, but it is not a faithful transcribed title and may suggest a different book. Another editor might instead mark the book’s text as Chinese in the language code because the summary is Chinese. That makes the catalog promise a language the item does not contain. Both errors are understandable if the task was expressed as “translate this record” rather than “translate this discovery note.”

A sound source-target note pair keeps the claim stable. The English might say, “Introduces pupils to recording rainfall, stream levels and weather patterns through short field activities.” A natural Chinese-language note would preserve the learner audience, the three observed phenomena and the practical field-activity format. If the edition contains tables but no experiments, avoid translating “activities” as formal laboratory experiments. The surrounding catalog and the item itself can resolve that nuance.

A simplified illustrative MARC display follows. It is not a complete cataloging record and should not be imported as one. The 520 translation placement must be adapted to the library’s approved language-note and discovery-layer profile; the example makes the difference between original title and public description visible rather than prescribing a universal multilingual encoding.

245 10 $a River weather : $b a student's field notebook.
520 ## $a Introduces pupils to recording rainfall, stream levels and weather patterns through short field activities.
041 0# $aeng

Approved target-language discovery summary (stored according to local policy):
面向学生介绍如何通过简短的户外观察活动记录降雨量、溪流水位和天气变化。

Unchanged structural facts: original title, item-language code, local record ID, holdings linkage.

Now conduct a two-direction check. From the target-language summary, translate the claims back into a neutral English list: student audience, rainfall measurements, stream-level records, patterns of weather, short field activities. Compare that list with the source note and, when available, the item. Do not demand word-for-word back-translation. Instead ask whether each fact is present and whether any new fact appeared. This detects additions and omissions that a grammar review would miss.

The review should also check that the note identifies the correct edition. One edition may contain downloadable worksheets while an earlier one does not. If the source summary was copied from a later edition, a beautiful translation reproduces the wrong claim. Include edition and holdings context before translation begins, and return the question to catalog staff when source evidence is uncertain.

5. Treat the title as bibliographic evidence

There are legitimate reasons to provide translated titles in discovery. A library might offer a reader-facing translation in a local display field, use an authorised parallel title printed on the item, or record a variant title under applicable cataloging rules. These are not the same operation as overwriting field 245 with an invented translation. The precise choice depends on what appears on the manifestation and the library’s cataloging rules. Use a cataloger to decide whether a title is transcribed, parallel, supplied or simply offered as an explanatory interface label.

Even small changes to a title statement can affect search. Punctuation and nonfiling characters influence how certain systems index or sort titles. A change that makes a phrase natural in Chinese might require an entirely different filing treatment, but that should follow a deliberate metadata rule, not a translator’s guess. The safest default for a summary-localization project is to leave the title fields untouched and expose the translation through an explicit reader-facing mechanism.

Names of authors and organisations can introduce another trap. A familiar Chinese form of an English author name may be appropriate in a localized display, but changing the authorised name access point can disrupt authority matching. Transliteration and translation are distinct operations. A name authority may include references across scripts; the library should use that structure rather than treating the access point as a free-form text field. A well-meaning editor who replaces a controlled name with a convenient local spelling may reduce discoverability rather than improve it.

Preserve subtitles and part numbering when explaining the book. A book with Volume 2 on its title page should not be described as the introductory first volume. If a summary omits the volume because the reader already sees it elsewhere, make sure the display actually shows that information. Localization review should consider the complete result card, not only the isolated summary. A description that is unambiguous in the full record may be misleading in a cropped mobile card.

Finally, document title policy separately from the summary translation. Once an organisation starts a multilingual catalog program, editors may request localized titles later. Keeping that future possibility in a policy queue prevents accidental changes during the current task. One dominant reader job per workflow is not bureaucratic minimalism; it makes quality evidence interpretable.

6. Language coding: the distinction most likely to disappear

A language code in a bibliographic record communicates information about the described material. In MARC field 041, subfield a commonly relates to the language of textual content or sound track, subject to material-type context. Subfield h may record an original language associated with a translated work. Subfield b may record languages of summaries or abstracts contained in the described resource. These uses are not interchangeable with the language in which a librarian writes an explanatory catalog note.

Suppose an English edition of a scientific field guide includes a Spanish-language abstract printed near the back. The language of that abstract can be a bibliographic fact relevant to appropriate 041 coding. But when the library staff creates a Chinese note for its website, the physical guide has not acquired a Chinese abstract. Using a catalog note as evidence that the item itself contains Chinese would misinform readers looking for a book in that language.

The distinction becomes more important with translated editions. A French original published as an English translation can legitimately require original-language and translation-language facts in appropriate codes, depending on the record and cataloging rules. A new local-language catalogue explanation is a different layer entirely. A translator who sees both “English translation” and “Chinese catalog summary” must keep those layers separate. Otherwise the record may appear to describe a non-existent trilingual edition.

Language codes may look like ordinary three-letter or two-letter words, but they are controlled values. Do not translate the code tokens into full language names inside a coded field. A user interface can display “English” or its localized equivalent based on those codes. The fixed code itself provides interoperability. If the source record has an incorrect language code, correct it through a separate cataloging process with appropriate review, not as an incidental side effect of translating the summary.

A practical QA routine therefore has two language columns: “languages evidenced in the held item” and “languages available for catalog display.” The first belongs to bibliographic description; the second belongs to a multilingual discovery service. They may overlap, but neither implies the other. Teach staff to name the column explicitly whenever they discuss language. That small habit prevents a great deal of metadata confusion.

7. Alternate scripts are not simply alternate translations

Field 880 exists so that a field’s data may be represented in another script with an explicit relationship to the associated field. Its subfield 6 linkage is not decoration; it identifies the association and supports multiscript processing. A system that discards the link might still display readable characters, but lose the intended structural pairing. This is why editing an 880 field requires familiarity with linked fields and the library’s representation model.

For example, a work originally titled in Japanese may have information represented in Japanese script and in a romanised form. Those are often two representations of the same title, not a translated title in a different language. A Chinese explanatory title might describe the meaning of that title, but it is not automatically the right value to insert into the 880 field. Translation changes language; transliteration changes script representation according to a mapping convention. Sometimes both can be useful to readers, but they should not be confused in the machine record.

If the library’s summary-localization project encounters a record with 880 linkages, the immediate rule is conservative: do not rewrite or delete them. Treat their integrity as an invariant. Inspect whether the discovery interface draws displayed text from the linked field and whether any local enhancement layers interfere with it. A new Chinese summary should not make a Japanese title disappear or cause a name to display twice without context.

This is another case in which a translator can be linguistically correct while technically wrong. The temptation to “make everything in the record Chinese” can erase information that enables multilingual readers to search by original script. A respectful catalogue allows original-script evidence, transliteration and target-language explanation to coexist according to clearly assigned functions.

Because multiscript indexing varies among implementations, test actual searches. Search by the original script, by its approved romanisation and by any newly added descriptive Chinese terms. Confirm that records remain linked to the same holdings and do not split into accidental duplicates. That test is not only about how a field looks; it is about whether readers can still discover the same item by different access routes.

8. Controlled subjects, names and the risk of a plausible replacement

A catalog’s subject headings can look like descriptive English phrases. The difference is that controlled headings point into a vocabulary maintained for consistent access. A free translation may be understandable to one reader but may no longer map unambiguously to that authority concept. For example, a heading for environmental observation could be broader or narrower than the expression a translator chooses. Replacing it directly can interfere with subject browsing and linked-data functions.

A safer multilingual discovery layer can retain the controlled heading or authority identifier and, where available, display an approved label in the target language. The source concept remains stable; its presentation changes. This parallels the identity-versus-label distinction seen in other structured systems, but in cataloging the stakes include scholarly provenance and conceptual precision. A local label should not assert a formal equivalence unless the concept mapping has been reviewed.

Author and corporate headings deserve the same care. A personal name may have variant-script references in an authority record. A publisher name may have an official translated brand form. Those are cataloging and authority decisions, not automatic sentence-translation decisions. If a summary uses a translated organisation name for readability, that does not authorise altering the controlled access point. The display note and search key may serve different functions.

Classification numbers, international standard numbers, record control numbers and holdings location codes are likewise structural. Some may contain hyphens, letters or meaningful-looking fragments; none should be reworded because a translator thinks their components have natural-language equivalents. Locking them is not distrust of translators. It is a recognition that language specialists should be given a task whose boundaries match their expertise.

During review, compare the proposed summary against the subject headings as a consistency clue without treating headings as a substitute for reading the source evidence. If the summary claims a book teaches marine chemistry while the catalog and contents show general water observation, investigate the discrepancy. Either the summary or the underlying cataloging may need correction. Do not simply rewrite one to match the other without examining the item.

9. Source provenance: whose summary are we translating?

A 520 note may come from a cataloger, a publisher, an external supplier or an earlier imported catalog. Those origins affect permission, reliability and editorial responsibility. An approved target-language summary should record its source and, where policy requires, the translator or reviewer and the date of review. It is tempting to assume that public catalogue display means all text is freely reusable. That is not a safe assumption for longer supplied descriptions.

Ask whether the source note is descriptive or promotional. A publisher blurb may use persuasive claims such as “the definitive introduction.” A library summary for educational discovery should generally prioritize verifiable content. If local policy permits translating a publisher blurb, retain its attribution and do not quietly convert it into an apparently independent library recommendation. A reader should be able to distinguish factual scope from marketing language when that distinction matters.

Check source edition and chronology. A catalog imported five years ago may contain a summary of a revised edition, a missing companion booklet or a chapter that was later removed. Translation accelerates the spread of such errors if a copied source is treated as authority. A source audit is especially important when translating a short summary that will be reused widely across mobile apps, branch catalogs and school learning portals.

When the original note is vague, ask for the physical or digital item’s contents rather than inventing helpful detail. The correct target-language wording can sometimes be more explicit about relationships already present in the source, but it should not infer educational level or curriculum alignment without evidence. For example, a book that contains graphs is not automatically suitable for a specific school examination. Assertions about grade suitability should be traced to the library’s reviewed classification or a separate advisory source.

Keep the translated note as a versioned editorial artifact. If a librarian edits the English summary later, the system should flag the target-language note for review rather than silently preserving it as current. A metadata translation without provenance quickly becomes orphan text: readable, persuasive and difficult to verify. A small source reference and review status make it maintainable.

10. Encoding, punctuation and the reader’s actual search result

MARC records may be exchanged in machine-readable representations and displayed through library software that indexes strings, labels and punctuation according to its own configuration. A Chinese-language note can look right in an editor but break after export or indexing. Character encoding and import rules need testing. In particular, do not let an export pipeline replace non-Latin characters with question marks or silently drop them because an intermediate process assumes the wrong character set.

Punctuation is not a cosmetic issue in title statements. Certain punctuation marks may reflect cataloging conventions or distinguish title and subtitle information. A summary note has more editorial freedom, but translated punctuation should still be tested against the discovery layer’s rendering and search tokenisation. Quotation marks, proper names and numerals should retain their intended meaning, especially in short notes where a small formatting change can alter which phrase the reader associates with the title.

Test both detail pages and condensed result lists. A long target-language sentence may be truncated after a fixed number of characters, leaving an unfinished clause that accidentally reverses the meaning. Chinese and other scripts differ from English in space usage and line-break behaviour. A character-count threshold is not equivalent to a readable-summary threshold. Check the semantic information left visible in a mobile preview, then adjust the display copy without changing the underlying bibliographic facts.

Search behavior also needs multilingual testing. Can a patron find the item with a key concept from the new summary? Does the catalog tokenise it adequately? Does the original title still retrieve the same item? Does a search by the author’s authority-controlled name still return it? Some discovery systems index summary fields separately or omit them from default search. A translation may improve display but not keyword retrieval unless the index configuration is reviewed.

Avoid treating every unsuccessful search as a translation error. The target summary may be accurate while the index has not been rebuilt or the field is not indexed. Separating editorial defects from discovery-configuration defects makes troubleshooting faster and helps the right team own the repair.

11. Practical workflow from export to reviewed multilingual display

Start with a baseline MARC export or record snapshot and a clear list of approved editable note fields. Capture record IDs, title statements, edition information, language codes and note provenance as read-only context. Create a narrow source-to-target workspace for the summaries. A tool that gives the translator access to entire MARC records may be convenient, but it multiplies accidental-change risk. A structured export with locked surrounding fields usually improves both speed and accountability.

Group records by subject area and audience rather than translating a random alphabetical list. A bilingual educational reviewer can maintain terminology across several related science books when those items are considered together. For each summary, mark source uncertainty, proper-name questions, reading-level choice and evidence gaps. The reviewer should not have to infer which editorial liberties a translator took. Give them a short change rationale for any deliberate paraphrase that might look broader than the source wording.

Run a content comparison against the frozen source. Extract claims: topic, audience where supported, method, included formats, geographic scope and edition-specific features. Compare target claims to the evidence list. Then run a separate structural comparison: the control fields, transcribed title, fixed language data, controlled access points, identifiers and holdings connections are unchanged. Those checks answer different questions and should produce separate pass or fail results.

Import or stage the approved target summary through the library’s supported mechanism. The exact technical placement must follow local cataloging and discovery policy; do not invent a universal MARC encoding for all multilingual summaries. After import, verify that the source description still appears where expected, that the new note is correctly labelled and that other language layers are not replaced. Test export as well as local display if the record will be syndicated to another system.

Finally, prepare a release packet containing the source record ID, note versions, approvals, change history and screenshots of representative discovery views. If a source note changes, the target-language note should be flagged for re-review. Treat translations as maintained catalog data. They should not become permanent text whose connection to its source has been forgotten.

12. Five failure modes worth rehearsing with the team

The first failure is title substitution. A new headline is written in the original title field because the translator thinks all visible English should disappear. The remedy is to restore bibliographic transcription and present explanatory local-language wording through an approved display feature. The diagnostic question is simple: could a patron use this catalog title to locate the exact edition on the shelf or in a publisher’s listing? If not, the title claim may have been changed.

The second failure is coded-language inflation. The library adds a Chinese summary to the catalog and then updates the item-language code as though the book itself were Chinese. The remedy is to distinguish item languages from catalog interface languages and correct the bibliographic codes through cataloging review if needed. A search filter for Chinese-language books should not return an English book solely because its catalogue has a Chinese note.

The third failure is broken alternate-script linkage. An edit removes or alters the subfield linkage in a multiscript record, producing duplicated or missing titles. The remedy is to restore the original linked-field structure from the baseline and retest original-script and romanised searches. A general-purpose translation spreadsheet may not expose these relationships, so the ingestion system needs structural safeguards.

The fourth failure is uncontrolled subject translation. A free equivalent replaces a controlled authority heading. The display now reads naturally, but the patron cannot browse consistently with other records. Restore the authorised concept and add an approved multilingual label in the appropriate layer. Do not solve the problem by creating a second authority concept with a superficially similar English gloss.

The fifth failure is orphaned description. A translated summary remains on the site after the library updates the underlying note for a new edition. It may make an outdated promise about worksheets, diagrams or supplementary materials. Tie summaries to source versions, detect changes, and require editorial reapproval where facts differ. A language review that never checks edition drift is not complete catalog QA.

13. Worked case: the summary is accurate but the item is not

Imagine a publisher releases two editions of River Weather. The first introduces observations with worksheets; the second adds a chapter on tide gauges. The catalog’s English summary was imported from the second edition, but the library holds the first. A translator receives the English note and carefully renders every detail into Chinese. The result is excellent translation and poor bibliographic description because it promises content missing from the held item.

The discovery tester notices that patrons are searching for tide-gauge explanations, then reporting that they cannot find them in the book. This is not solved by changing one Chinese noun. The team must compare the source record with holdings and edition evidence, repair the English description through cataloging review, and only then update the Chinese summary. Reusing the translation unchanged would preserve the problem. Translation QA must therefore include provenance, not just linguistic equivalence.

Now add a second complication. The publisher markets the revised edition under a different subtitle. The local discovery layer shows a translated explanatory title as the most prominent line and hides the original English title on mobile. A student finds what appears to be the revised title, borrows the older volume, and becomes confused. The fix involves presentation hierarchy: preserve the transcribed title visibly enough to identify the edition, and make any explanatory translated title clearly subordinate or separately labelled.

This scenario teaches a general rule: local-language discoverability should increase transparency about the item, not replace source evidence with an appealing promise. The best translation is sometimes less dramatic than the marketing copy. It tells the reader exactly what is known and leaves uncertain claims out until someone can verify them.

14. Transfer exercise: a bilingual exhibition catalogue

Consider a bilingual museum library holding an exhibition catalogue printed with an English title and Japanese text on selected pages. Its MARC record has a transcribed title, language information, original-script data and a publisher-provided English summary. The library wants a Mandarin discovery summary for visiting families. A well-meaning editor proposes to translate every visible field and to switch the language of the whole item to Chinese because the new interface will be Chinese.

Your task is to design a safer scope. Which field should serve as the main source for the Mandarin discovery text? Which data must remain evidence about the item? How will you represent the Japanese script already in the record? What source permissions and display behaviours must be checked before publishing? Finally, explain how a Chinese-language filter for books should treat this item, and why a Chinese catalog summary does not settle that question.

A good answer uses the authorised English summary as one source of descriptive claims, verifies its content and rights, and creates a Mandarin note through the library’s approved multilingual mechanism. The original title, edition, control numbers, holdings, language codes and linked alternate-script data remain stable unless a qualified cataloger independently finds an existing error. The Japanese-script representation is preserved according to its linkages rather than replaced with a Mandarin translation. Search indexing is tested across original title, author and new summary concepts.

The final filter question exposes the heart of the problem. An item containing English and Japanese is not transformed into a Chinese-language resource because a librarian writes a Chinese description. The catalog can welcome Chinese-reading patrons without falsely describing the language of the held material. That principle transfers to museum collections, institutional repositories and archival systems even though their exact metadata schemas differ.

15. A release checklist based on evidence, not optimism

Before approving a multilingual summary, check that the item and edition were verified, the source note has an identified provenance, and the target note contains no unverified feature or audience claim. Check that the chosen storage mechanism is permitted by the library’s cataloging and discovery policy, and that the translation is labelled or displayed in a way readers can understand. A sentence may be beautifully written but still fail these conditions.

Run the structural diff. Fields containing bibliographic title transcription, language coding, authority headings, controlled numbers, record identifiers and alternate-script linkages should remain unchanged in this summary-only task. Any difference is an exception requiring explicit cataloger review. Confirm that source and target versions are linked editorially for future updates, without pretending that the base schema automatically enforces that link.

Then test what patrons see. Search by original title and translated summary concepts, inspect result cards on a small screen, try the relevant language and material-type filters, and confirm that the record still leads to the correct holdings. Check fonts, line breaking, punctuation and original-script rendering. A catalog record is successful only when users can discover the correct item and form an accurate expectation of its content.

Document the outcome with a short approval record: source item ID, summary source, target language, translator, catalog reviewer, permission status, import method, index update, spot-test evidence and unresolved questions. A future editor should be able to re-create the decision without interviewing the entire original project team. That is what makes a translation program maintainable when staffing changes.

The essential result is modest and powerful: readers who use another language can learn what a library holds, while the catalogue continues to describe the same real object. Keeping description, transcription and controlled access apart is the mechanism that makes speed safe.

Frequently asked questions

May I replace MARC field 245 with a translated title for local readers?

Not as a routine summary-localization step. Field 245 carries title-statement semantics and may be based on transcription. Consult the cataloging rules and use an approved separate display mechanism when offering an editorial translation.

Is MARC 520 always the right place for a translated summary?

It is the core MARC field for summary information, but multilingual storage and language labelling vary by local profile and discovery software. Confirm local practice rather than assuming a universally portable target-language subfield.

Should I change 041 $a after adding a Chinese summary to an English book’s catalog?

No, not solely because the catalog now displays a Chinese note. Language codes describe languages associated with the item under the field’s rules, not every language used by the catalog interface.

Does field 041 $b describe the language of the cataloger’s own note?

It is defined for language of a summary or abstract associated with the described resource. Do not automatically treat it as a code for any newly added public catalogue explanation.

Can an 880 field hold any translation into a different language?

No. It has alternate-script representation and subfield-linkage semantics. Translation and alternate graphic representation are not identical tasks; preserve existing linkage and follow qualified cataloging policy.

What is the safest way to translate a controlled subject heading?

Keep the authority-controlled concept or identifier stable and, where the system supports it, display an approved target-language label. Free replacement of the heading can break consistent subject access.

Can I shorten the translated note for mobile display?

Yes, if the shortened version preserves the item’s decisive claims and remains consistent with the source evidence. Test what the reader actually sees after truncation rather than trusting character count alone.

How do I know whether the original summary is safe to translate?

Check authorship, source, permitted reuse and relevant licensing or catalog policy. A public publisher blurb is not automatically unrestricted material.

What if the original English summary contains an edition error?

Do not translate the error as though it were a fact. Flag it for cataloging review, establish the correct description of the held item, and then prepare the target-language note.

Does a translated summary make the record searchable in that language?

Not necessarily. The discovery index must include the relevant field and support appropriate tokenisation. Test actual queries and index refresh behaviour.

Authoritative sources and specification checks