Translate | ISO 25964-1:2011 Multilingual Thesauri, Preferred Terms, Concept Relationships and Search Retrieval

ISO 25964-1:2011 translation helps librarians, archive teams, terminology managers, museum curators, knowledge-graph editors and search engineers answer a practical search question: how to translate multilingual controlled vocabulary labels without merging distinct concepts or breaking cross-language search. A correct translation must do more than reproduce technical nouns. It needs to preserve what the source author observed, promised, requested, measured, authorised or left unresolved. When readers search for how to translate these records correctly across languages, the answer starts with the underlying decision, not a dictionary. This guide shows a reproducible method with original examples and carefully defined review criteria.

ISO 25964-1:2011 has a particular scope: ISO 25964-1:2011 gives recommendations for building and maintaining monolingual and multilingual thesauri for information retrieval, including a data model and exchange format. As checked in November 2026, a replacement was under publication, so edition status needs rechecking before new adoption. Translation of specialist documents in this field is therefore about recording accurate relationships among terminologist, indexer, subject librarian, curator, language reviewer, metadata owner and search engineer. Distinguish the words used by the source, the evidence available to its author and the rules of the local organisation. A sentence can be grammatically perfect while changing a recipient’s authority or changing the stage of a process; the correction must restore that boundary rather than merely sound smoother.

Quick answer: map every actor and decision, preserve exact identifiers and qualification language, build a bilingual concept register, check source evidence before asserting equivalence, and require an independent domain review before publication. In particular, Translation of a label is not automatic concept equivalence. Preserve concept IDs, scope notes and hierarchy relationships before selecting a multilingual label or mapping. The 24 worked cases below are fictional educational examples, not quotations from the protected ISO publication and not official certification guidance. They teach readers how to detect a material translation error without inventing the requirements of the standard.

At a glance: solve the reader’s immediate problem

• The specific question: how to translate multilingual controlled vocabulary labels without merging distinct concepts or breaking cross-language search
• Who must understand the translation: librarians, archive teams, terminology managers, museum curators, knowledge-graph editors and search engineers
• Parties and roles to distinguish: terminologist, indexer, subject librarian, curator, language reviewer, metadata owner and search engineer
• Documents to audit: concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record
• Data that must survive unchanged: concept IDs, locale tags, hierarchy depths, version numbers, mapping references and retrieval test counts
• Public scope: ISO 25964-1:2011 gives recommendations for building and maintaining monolingual and multilingual thesauri for information retrieval, including a data model and exchange format. As checked in November 2026, a replacement was under publication, so edition status needs rechecking before new adoption.
• Red-line risk: Translation of a label is not automatic concept equivalence. Preserve concept IDs, scope notes and hierarchy relationships before selecting a multilingual label or mapping.
• Approval: bilingual language review plus an appropriately qualified domain owner

Contents

1. Scope and method
2. Twelve core translation principles
3. Controlled terminology
4. 24 worked translation cases
5. 10-stage review workflow
6. Four-week learning plan
7. Frequently asked questions
8. Sources and further learning

Read the standard’s scope before translating the associated records

ISO 25964-1:2011 gives recommendations for building and maintaining monolingual and multilingual thesauri for information retrieval, including a data model and exchange format. As checked in November 2026, a replacement was under publication, so edition status needs rechecking before new adoption. It is essential to distinguish the public catalogue’s description from the complete copyright-protected publication. A translator should not claim the present article reproduces official clauses, supplies a certification checklist or determines all applicable legal duties. Obtain the authorised source and any controlled contract or procedure for actual operational decisions.

Set up an evidence map with four questions. What is the source document’s purpose? Who is entitled to make a decision? Which record supports the factual claim? Which qualifiers restrict that claim? A separate field should capture uncertainty that originates in the source. If a record is missing or contradictory, do not conceal that fact with polished target-language prose. Query the competent owner, identify the affected sentence and keep the unresolved status visible until a decision is documented.

The most revealing bilingual test is to cover the source page and ask a colleague to explain what they would do after reading the translation. Repeat that exercise with the source language. A difference in action indicates a possible material error, even when an automated similarity tool reports high agreement. Protect the source’s purpose first; edit for readability once the permission, condition, quantity and chronology are secure.

Twelve core translation principles

1. Concept versus word

A concept may have several lexical representations. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

A literal translation can point to a different subject category. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use Concept ID and scope note. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

2. Preferred label versus synonym

One term may be selected for indexing while variants aid search. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

A variant must not automatically become another preferred concept. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use Controlled term record. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

3. Broader and narrower links

A hierarchy models concept scope rather than sentence grammar. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

Reversing broader and narrower breaks navigation. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use Hierarchy graph. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

4. Related versus narrower

Association need not mean a child relationship. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

Do not turn related-to into a type-of claim. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use Relation field. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

5. Homographs

One spelling can refer to completely different concepts. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

Financial bank must not absorb riverbank objects. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use Disambiguation note. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

6. Unequal language boundaries

Languages and cultures divide domains differently. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

A convenient bilingual word pair may be only approximate. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use Subject expert assessment. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

7. Scope notes

Definitions for indexing help explain intended collection usage. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

A narrow note must not become a universal dictionary definition. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use Indexing examples. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

8. Domain-specific terms

The same label changes meaning across knowledge domains. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

Seal in marine biology is not a document security device. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use Facet metadata. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

9. Deprecation and migration

Labels change while identifiers and archival meaning may persist. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

Do not delete old search entry points casually. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use Version and redirect map. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

10. Precision and recall

Search expansion can retrieve more results while increasing irrelevant matches. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

Fluency alone is not a retrieval quality metric. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use Evaluation queries. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

11. Import and export

Interchange must preserve language and relationship fields. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

Dropping language tags can scramble a multilingual vocabulary. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use Round-trip validation. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

12. Replacement edition

The 2011 edition was published while another edition was moving to publication. For ISO 25964-1:2011, this is not a cosmetic distinction: a reader may use the sentence to decide what to record, what to accept, or what remains to be done. The translator’s first responsibility is to identify the event or classification implied by the original, including who is involved and any time or scope limitation. If a source sentence compresses several events, the target language can explain them more clearly, but it must not add an event that the source did not record.

Do not cite a draft as if it were a released standard. Consider the reader who knows the target language well but has never seen the original document. That person is entitled to rely on its explicit wording. Test the translated phrase by asking whether it would cause a different submission, handoff, decision or conclusion. When the answer is yes, revise the passage even if the initial rendering is idiomatic. Preserve all conditions such as only, subject to approval, not yet, estimated and within the stated period. These small expressions protect practical meaning.

Verification should use ISO catalogue status. The reviewer can place three columns beside the translated segment: source evidence, the claimed action or status, and the target-language interpretation. If any column does not agree, hold that segment and request a source-owner answer. Keep the resolution in a dated bilingual glossary, because the same conceptual distinction may recur in notices, labels, databases and handover records. This also turns one correction into institutional learning rather than a private stylistic preference.

Controlled terminology: meanings and traps

The following compact register provides original working explanations, not reproduced ISO definitions. A useful multilingual glossary records more than a literal equivalent: it stores concept, context, approved label, rejected label, evidence and reviewer. For ISO 25964-1:2011, review these entries against the actual source publication and documents before adoption. A term that is safe in one domain may be misleading in another.

1. concept

Working meaning: unit of meaning represented by labels. The important mistranslation to avoid is not just spelling. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

2. preferred term

Working meaning: selected indexing label for a concept. The important mistranslation to avoid is not any synonym. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

3. alternative term

Working meaning: nonpreferred lexical search access. The important mistranslation to avoid is not unrelated entry. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

4. nonpreferred label

Working meaning: valid term not selected as canonical. The important mistranslation to avoid is not to be erased. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

5. scope note

Working meaning: instruction for how concept is used in collection. The important mistranslation to avoid is not universal meaning. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

6. broader concept

Working meaning: concept of larger scope. The important mistranslation to avoid is not grammatically longer word. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

7. narrower concept

Working meaning: concept with smaller subject scope. The important mistranslation to avoid is not parent. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

8. related concept

Working meaning: association not inherently hierarchical. The important mistranslation to avoid is not synonym. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

9. homograph

Working meaning: same form with different referents. The important mistranslation to avoid is not same indexed category. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

10. polysemy

Working meaning: related senses requiring evaluation. The important mistranslation to avoid is not automatic identity. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

11. language tag

Working meaning: locale designation for lexical form. The important mistranslation to avoid is not cosmetic metadata. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

12. concept ID

Working meaning: stable identity of classified meaning. The important mistranslation to avoid is not translated title. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

13. deprecated label

Working meaning: old label no longer selected. The important mistranslation to avoid is not automatically unsearchable. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

14. exact mapping

Working meaning: claimed equivalence of defined concept scopes. The important mistranslation to avoid is not plausible word match. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

15. search recall

Working meaning: relevant items retrieved divided by relevant items. The important mistranslation to avoid is not precision. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

16. search precision

Working meaning: relevant results divided by returned results. The important mistranslation to avoid is not recall. This word should be mapped using the source document’s actual context and any established terminology register. Ask whether the target label would place the item in the same category, give the same person the same authority, and indicate the same stage of work. Record a short example of permitted use and an explicit counterexample. If the target language requires a descriptive phrase rather than a single word, use it: clarity and conceptual fidelity take priority over word-for-word symmetry.

Twenty-four worked translation cases

Every case is a fictional training example. Each pairs a source-style note with a plausible but wrong reading, a correction and a type of evidence to consult. No sample should be passed off as an ISO clause, an authentic case file or professional legal advice. Work through each case by identifying the exact word that changes the reader’s likely action.

Case 1: Bank ambiguity

Source-style note: “Bank labels financial institutions.” An attractive but wrong translation would say “River banks grouped with finance.” The repair is to Retain distinct concepts. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the collection examples before approval. This is primarily an event sequence problem. First reconstruct what happened before and after the event. A fluent target sentence is unacceptable if it changes the sequence, compresses two events into one or makes a pending step appear finished. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Bank labels financial institutions”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 2: Seal ambiguity

Source-style note: “Seal means archive fastener.” An attractive but wrong translation would say “Marine mammal merged.” The repair is to Separate domain meanings. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the facet field before approval. This is primarily an decision authority problem. Ask who can request, review, approve and receive. Those are distinct forms of involvement, and institutional labels alone do not tell you whether a particular person had authority. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Seal means archive fastener”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 3: Preferred car label

Source-style note: “Automobile preferred and car alternate.” An attractive but wrong translation would say “Two unrelated concepts created.” The repair is to Keep synonym under concept. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the thesaurus entry before approval. This is primarily an source evidence problem. Locate the record that supports the claim. When the record is unavailable, flag the segment rather than replacing uncertainty with an attractive but unsupported phrase. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Automobile preferred and car alternate”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 4: Regional spelling

Source-style note: “Colour and color locale variants.” An attractive but wrong translation would say “Separate concept IDs generated.” The repair is to Preserve regional labels. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the language tags before approval. This is primarily an scope boundary problem. Inspect small but consequential words: only, except, may, pending, subject to, at most and approximately. Preserve the narrowest actual promise. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Colour and color locale variants”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 5: Fruit hierarchy

Source-style note: “Citrus fruit narrower than fruit.” An attractive but wrong translation would say “Citrus placed above fruit.” The repair is to Keep direction. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the hierarchy check before approval. This is primarily an status distinction problem. Map the passage to proposed, performed, reviewed and resolved states. Do not infer a later state merely because the source reports an earlier one. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Citrus fruit narrower than fruit”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 6: Archive association

Source-style note: “Archives related to records management.” An attractive but wrong translation would say “Every archive item made child of management.” The repair is to Keep associative relation. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the relation file before approval. This is primarily an exact identifiers problem. Numbers, reference strings, dates, codes and units must be compared character by character. A clean-looking translated sentence can still contain a destructive data error. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Archives related to records management”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 7: Unmatched term

Source-style note: “Local heritage category lacks exact target label.” An attractive but wrong translation would say “Everyday word asserted exact.” The repair is to Use qualified approximate mapping. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the curator review before approval. This is primarily an claim strength problem. Keep the distance between observation and conclusion visible. A statement that an outcome is possible is not equivalent to asserting it is confirmed. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Local heritage category lacks exact target label”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 8: Narrow scope

Source-style note: “Use label for painted ceramic vessels.” An attractive but wrong translation would say “Use for all ceramics.” The repair is to Keep technique qualifier. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the scope note before approval. This is primarily an reader action problem. Imagine a reader acting solely on the target-language sentence. If the source and target would trigger different actions, reopen the translation choice. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Use label for painted ceramic vessels”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 9: Legacy alias

Source-style note: “Historical label still occurs in records.” An attractive but wrong translation would say “Alias removed from retrieval.” The repair is to Keep redirect path. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the migration map before approval. This is primarily an event sequence problem. First reconstruct what happened before and after the event. A fluent target sentence is unacceptable if it changes the sequence, compresses two events into one or makes a pending step appear finished. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Historical label still occurs in records”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 10: Institution labels

Source-style note: “School maps to several local institution forms.” An attractive but wrong translation would say “All institutions merged.” The repair is to Represent differentiated concepts. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the education authority metadata before approval. This is primarily an decision authority problem. Ask who can request, review, approve and receive. Those are distinct forms of involvement, and institutional labels alone do not tell you whether a particular person had authority. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “School maps to several local institution forms”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 11: Duplicate entry

Source-style note: “Two synonyms refer to same defined concept.” An attractive but wrong translation would say “Keep false duplicates with divergent IDs.” The repair is to Merge only after approval. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the version history before approval. This is primarily an source evidence problem. Locate the record that supports the claim. When the record is unavailable, flag the segment rather than replacing uncertainty with an attractive but unsupported phrase. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Two synonyms refer to same defined concept”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 12: Locale error

Source-style note: “English term tagged en-GB.” An attractive but wrong translation would say “Imported as fr-FR.” The repair is to Correct language tag. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the export sample before approval. This is primarily an scope boundary problem. Inspect small but consequential words: only, except, may, pending, subject to, at most and approximately. Preserve the narrowest actual promise. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “English term tagged en-GB”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 13: Identifier preservation

Source-style note: “Concept ID museum:0042.” An attractive but wrong translation would say “Translated title used as key.” The repair is to Keep stable ID. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the schema before approval. This is primarily an status distinction problem. Map the passage to proposed, performed, reviewed and resolved states. Do not infer a later state merely because the source reports an earlier one. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Concept ID museum:0042”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 14: Cross-collection ward

Source-style note: “Ward has distinct administrative senses.” An attractive but wrong translation would say “Marked exact equivalent.” The repair is to Document qualified relationship. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the scope notes before approval. This is primarily an exact identifiers problem. Numbers, reference strings, dates, codes and units must be compared character by character. A clean-looking translated sentence can still contain a destructive data error. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Ward has distinct administrative senses”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 15: Proposal state

Source-style note: “Microplastics heading proposed.” An attractive but wrong translation would say “Published before approval.” The repair is to Keep proposal state. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the editorial queue before approval. This is primarily an claim strength problem. Keep the distance between observation and conclusion visible. A statement that an outcome is possible is not equivalent to asserting it is confirmed. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Microplastics heading proposed”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 16: Retired heading

Source-style note: “Deprecated term appears in old catalogue.” An attractive but wrong translation would say “Historical link deleted.” The repair is to Maintain migration. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the index history before approval. This is primarily an reader action problem. Imagine a reader acting solely on the target-language sentence. If the source and target would trigger different actions, reopen the translation choice. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Deprecated term appears in old catalogue”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 17: Transliteration

Source-style note: “Script variant of personal name used.” An attractive but wrong translation would say “Original authority ID discarded.” The repair is to Keep authority record. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the names register before approval. This is primarily an event sequence problem. First reconstruct what happened before and after the event. A fluent target sentence is unacceptable if it changes the sequence, compresses two events into one or makes a pending step appear finished. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Script variant of personal name used”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 18: Broad expansion

Source-style note: “Search expands to many vague synonyms.” An attractive but wrong translation would say “False positives become dominant.” The repair is to Test precision. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the retrieval results before approval. This is primarily an decision authority problem. Ask who can request, review, approve and receive. Those are distinct forms of involvement, and institutional labels alone do not tell you whether a particular person had authority. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Search expands to many vague synonyms”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 19: Low recall

Source-style note: “Variant spelling retrieves relevant items.” An attractive but wrong translation would say “Variant removed.” The repair is to Restore alternate access. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the query test before approval. This is primarily an source evidence problem. Locate the record that supports the claim. When the record is unavailable, flag the segment rather than replacing uncertainty with an attractive but unsupported phrase. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Variant spelling retrieves relevant items”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 20: Birds category

Source-style note: “Birds mapped to all vertebrates.” An attractive but wrong translation would say “Hierarchy expands incorrectly.” The repair is to Keep agreed narrower grouping. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the subject scheme before approval. This is primarily an scope boundary problem. Inspect small but consequential words: only, except, may, pending, subject to, at most and approximately. Preserve the narrowest actual promise. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Birds mapped to all vertebrates”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 21: Cultural craft

Source-style note: “Local handcraft category has no exact translation.” An attractive but wrong translation would say “Tourist phrase declared equivalent.” The repair is to Preserve nuance and note. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the curatorial memo before approval. This is primarily an status distinction problem. Map the passage to proposed, performed, reviewed and resolved states. Do not infer a later state merely because the source reports an earlier one. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Local handcraft category has no exact translation”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 22: Multiword label

Source-style note: “Cold chain logistics treated as controlled phrase.” An attractive but wrong translation would say “Words split into unrelated categories.” The repair is to Keep multiword concept. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the industry glossary before approval. This is primarily an exact identifiers problem. Numbers, reference strings, dates, codes and units must be compared character by character. A clean-looking translated sentence can still contain a destructive data error. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Cold chain logistics treated as controlled phrase”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 23: Export tags

Source-style note: “Language tags missing after exchange.” An attractive but wrong translation would say “All imports labelled English.” The repair is to Validate round trip. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the export validation before approval. This is primarily an claim strength problem. Keep the distance between observation and conclusion visible. A statement that an outcome is possible is not equivalent to asserting it is confirmed. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Language tags missing after exchange”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Case 24: Synonym collision

Source-style note: “Alternate term belongs to two concepts.” An attractive but wrong translation would say “Search silently selects first.” The repair is to Disambiguate mapping. Notice what the error changes: sometimes the performer, sometimes the scope, sometimes an action’s state. A reviewer should not correct only grammar and leave that operational difference intact. State in a short translator’s note why the replacement expresses the same evidence and authority as the source.

Check the search log before approval. This is primarily an reader action problem. Imagine a reader acting solely on the target-language sentence. If the source and target would trigger different actions, reopen the translation choice. A useful exercise is to ask a second reviewer, without seeing the original, what has already happened, what is still pending, who is accountable and whether the described evidence supports the conclusion. Compare their answers with the source owner. If they disagree, retain the issue in the correction log; do not silently substitute a confident synonym.

To practise this case, write three separate sentences in the target language: one for the observed or requested fact, one for its limiting condition and one for the next legitimate step. Translate the result back into the source language as a diagnostic, not as proof of correctness. Check whether the actor, event, time and state survive. If an implication appears that was absent from “Alternate term belongs to two concepts”, revise it. This is how the exercise moves from memorising terms to preserving an operational decision.

Ten-stage quality review workflow

Successful ISO 25964-1:2011 translation needs staged quality assurance, not a single final spellcheck. The following sequence is an original review method, not a purported set of clauses from the ISO document. Work from document identity to actor authority, then figures, states, bilingual comprehension and controlled release. The sequence can be adapted to the size and risk of the communication.

Stage 1. Scope and ownership

Identify the issuer, the audience, the source edition and the specific document’s purpose. An educational overview should not be mistaken for a licence, contract or standard. For this particular topic, test the result against concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record. Each stage needs a concrete acceptance criterion: the reviewers should be able to show an unaltered original, the proposed translation and the record supporting their judgement. If disagreement is about missing evidence, do not ask the language editor to invent it. If the disagreement is about subject-matter practice, route it to an accountable practitioner; if it concerns applicable law, obtain appropriate local legal review. Record the agreed wording and any approved exceptions in the document history.

Stage 2. Terminology control

Approve terms by concept and context; retain rejected candidates with reasons so that the next translator does not repeat a known semantic error. For this particular topic, test the result against concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record. Each stage needs a concrete acceptance criterion: the reviewers should be able to show an unaltered original, the proposed translation and the record supporting their judgement. If disagreement is about missing evidence, do not ask the language editor to invent it. If the disagreement is about subject-matter practice, route it to an accountable practitioner; if it concerns applicable law, obtain appropriate local legal review. Record the agreed wording and any approved exceptions in the document history.

Stage 3. Actor audit

Mark every performer, approver and recipient. Where a passive sentence hides the actor, do not invent one; seek clarification when the distinction matters. For this particular topic, test the result against concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record. Each stage needs a concrete acceptance criterion: the reviewers should be able to show an unaltered original, the proposed translation and the record supporting their judgement. If disagreement is about missing evidence, do not ask the language editor to invent it. If the disagreement is about subject-matter practice, route it to an accountable practitioner; if it concerns applicable law, obtain appropriate local legal review. Record the agreed wording and any approved exceptions in the document history.

Stage 4. Modality review

Search for may, must, should, might, only, unless and subject to. Compare obligations, permissions and uncertainty in both languages. For this particular topic, test the result against concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record. Each stage needs a concrete acceptance criterion: the reviewers should be able to show an unaltered original, the proposed translation and the record supporting their judgement. If disagreement is about missing evidence, do not ask the language editor to invent it. If the disagreement is about subject-matter practice, route it to an accountable practitioner; if it concerns applicable law, obtain appropriate local legal review. Record the agreed wording and any approved exceptions in the document history.

Stage 5. Evidence and figures

Verify strings, quantities, units, dates and version labels independently from the surrounding prose. Preserve source formatting where required. For this particular topic, test the result against concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record. Each stage needs a concrete acceptance criterion: the reviewers should be able to show an unaltered original, the proposed translation and the record supporting their judgement. If disagreement is about missing evidence, do not ask the language editor to invent it. If the disagreement is about subject-matter practice, route it to an accountable practitioner; if it concerns applicable law, obtain appropriate local legal review. Record the agreed wording and any approved exceptions in the document history.

Stage 6. State tracing

Check that a request has not become an approval, a proposal a decision, a review a final acceptance, or a suspected problem a proven fault. For this particular topic, test the result against concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record. Each stage needs a concrete acceptance criterion: the reviewers should be able to show an unaltered original, the proposed translation and the record supporting their judgement. If disagreement is about missing evidence, do not ask the language editor to invent it. If the disagreement is about subject-matter practice, route it to an accountable practitioner; if it concerns applicable law, obtain appropriate local legal review. Record the agreed wording and any approved exceptions in the document history.

Stage 7. Linguistic cold read

Have a bilingual reviewer explain the target text without seeing the original. Compare that interpretation with the decision intended by the source. For this particular topic, test the result against concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record. Each stage needs a concrete acceptance criterion: the reviewers should be able to show an unaltered original, the proposed translation and the record supporting their judgement. If disagreement is about missing evidence, do not ask the language editor to invent it. If the disagreement is about subject-matter practice, route it to an accountable practitioner; if it concerns applicable law, obtain appropriate local legal review. Record the agreed wording and any approved exceptions in the document history.

Stage 8. Domain sign-off

Ask a professional user to test the translation as though it were the only document available. Record any likely wrong action. For this particular topic, test the result against concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record. Each stage needs a concrete acceptance criterion: the reviewers should be able to show an unaltered original, the proposed translation and the record supporting their judgement. If disagreement is about missing evidence, do not ask the language editor to invent it. If the disagreement is about subject-matter practice, route it to an accountable practitioner; if it concerns applicable law, obtain appropriate local legal review. Record the agreed wording and any approved exceptions in the document history.

Stage 9. Release control

Record approving reviewers, version dates and distribution routes. Ensure readers can identify the current translation and find superseded versions when necessary. For this particular topic, test the result against concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record. Each stage needs a concrete acceptance criterion: the reviewers should be able to show an unaltered original, the proposed translation and the record supporting their judgement. If disagreement is about missing evidence, do not ask the language editor to invent it. If the disagreement is about subject-matter practice, route it to an accountable practitioner; if it concerns applicable law, obtain appropriate local legal review. Record the agreed wording and any approved exceptions in the document history.

Stage 10. Correction and rollback

When an error is found, correct affected copies, retain an audit trail and notify the operational owners who may have acted on the earlier text. For this particular topic, test the result against concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record. Each stage needs a concrete acceptance criterion: the reviewers should be able to show an unaltered original, the proposed translation and the record supporting their judgement. If disagreement is about missing evidence, do not ask the language editor to invent it. If the disagreement is about subject-matter practice, route it to an accountable practitioner; if it concerns applicable law, obtain appropriate local legal review. Record the agreed wording and any approved exceptions in the document history.

Four-week practice plan

Week 1 — learn to read the source

Collect ten fictional or properly authorised examples involving concept identifier, preferred term, alternative label, scope note, broader/narrower relation, related-concept link, language tag and migration record. Identify who wrote each passage, the intended readers, the documented event and any qualifier. Map terminologist, indexer, subject librarian, curator, language reviewer, metadata owner and search engineer to the actions they can legitimately perform. Do not translate every word immediately: first paraphrase the event in simple source-language prose. Highlight uncertain source statements and ask the document owner to explain the missing information. A source record understood poorly cannot become accurate merely by using a sophisticated translation tool.

Week 1 deliverable: a dated worksheet with five examples, reviewer observations, corrected passages and the rationale for each change. Evaluate understanding by checking whether the learner preserves meaning in new cases rather than by how quickly they type. This plan does not confer a professional qualification or certify compliance with ISO 25964-1:2011.

Week 2 — build a bilingual decision glossary

Draft a small table of the sixteen terminology entries above, adding the target-language wording, a disallowed misleading choice and one application sentence. Compare the entries with the real organisation’s terminology. Then translate five short passages and have a colleague read them cold. Ask the colleague which events are complete and which remain pending. Record disagreements separately as language, evidence, permission or process-stage errors; each category needs a different kind of correction.

Week 2 deliverable: a dated worksheet with five examples, reviewer observations, corrected passages and the rationale for each change. Evaluate understanding by checking whether the learner preserves meaning in new cases rather than by how quickly they type. This plan does not confer a professional qualification or certify compliance with ISO 25964-1:2011.

Week 3 — stress-test error cases

Work through twelve of the 24 fictional cases without looking at the repair. Mark the actor, event, limitation and supporting record. Introduce a deliberate mistake to a copy, then ask an independent reviewer to find it. Focus especially on concept IDs, locale tags, hierarchy depths, version numbers, mapping references and retrieval test counts. These values are not improved by elegant phrasing: they require exact verification. After the review, create a defect ledger with the incorrect statement, likely consequence, corrected wording and prevention control.

Week 3 deliverable: a dated worksheet with five examples, reviewer observations, corrected passages and the rationale for each change. Evaluate understanding by checking whether the learner preserves meaning in new cases rather than by how quickly they type. This plan does not confer a professional qualification or certify compliance with ISO 25964-1:2011.

Week 4 — simulate publication and rollback

Prepare an educational multilingual pack with controlled source, approved terminology, translated notices, reviewer initials and version date. Give it to a professional colleague who did not see the drafting process. Require them to explain what actions it authorises and which statements are uncertain. Simulate finding a material error after release: identify affected copies, prepare a corrected version and communicate what changed. Close the exercise only when recipients would now act on the same intended information.

Week 4 deliverable: a dated worksheet with five examples, reviewer observations, corrected passages and the rationale for each change. Evaluate understanding by checking whether the learner preserves meaning in new cases rather than by how quickly they type. This plan does not confer a professional qualification or certify compliance with ISO 25964-1:2011.

Frequently asked questions

Can an online translator handle this whole task?

Automatic tools can draft low-risk wording, but a domain practitioner and bilingual editor must review consequential roles, codes, statuses and permissions. A high fluency score is not an authority check. Apply this principle to the particular project document, and retain the relevant source, reviewer and change history. If a conflict cannot be settled from evidence available to the translator, mark it unresolved and escalate it before publishing high-consequence instructions. The target reader should not inherit greater certainty than the original author had.

Should identifiers be translated?

Identifiers, hashes, codes, version labels and record references should normally remain exact. Translate their surrounding descriptions while preserving the source-to-target mapping. Apply this principle to the particular project document, and retain the relevant source, reviewer and change history. If a conflict cannot be settled from evidence available to the translator, mark it unresolved and escalate it before publishing high-consequence instructions. The target reader should not inherit greater certainty than the original author had.

What if two target-language equivalents seem correct?

Choose using the source concept, intended audience, existing controlled vocabulary and an explicit test sentence. Record why one candidate is accepted and the other rejected. Apply this principle to the particular project document, and retain the relevant source, reviewer and change history. If a conflict cannot be settled from evidence available to the translator, mark it unresolved and escalate it before publishing high-consequence instructions. The target reader should not inherit greater certainty than the original author had.

Can the article be used as the official standard?

No. These examples are independently written educational nonfiction based on a public scope; organisations need the authorised standard and their own governing documents. Apply this principle to the particular project document, and retain the relevant source, reviewer and change history. If a conflict cannot be settled from evidence available to the translator, mark it unresolved and escalate it before publishing high-consequence instructions. The target reader should not inherit greater certainty than the original author had.

What if the source itself is ambiguous?

Open a question to the source owner, preserve evidence of the ambiguity and avoid turning a guess into a new obligation or conclusion. Apply this principle to the particular project document, and retain the relevant source, reviewer and change history. If a conflict cannot be settled from evidence available to the translator, mark it unresolved and escalate it before publishing high-consequence instructions. The target reader should not inherit greater certainty than the original author had.

How should multiple countries be handled?

Preserve the original meaning and document references, then have local specialists check operational adaptation. Do not claim that one translation creates universal legal validity. Apply this principle to the particular project document, and retain the relevant source, reviewer and change history. If a conflict cannot be settled from evidence available to the translator, mark it unresolved and escalate it before publishing high-consequence instructions. The target reader should not inherit greater certainty than the original author had.

How can practical quality be tested?

Ask independent readers to reconstruct who may act, what is decided, what remains open and which evidence supports the answer. Compare source and target responses. Apply this principle to the particular project document, and retain the relevant source, reviewer and change history. If a conflict cannot be settled from evidence available to the translator, mark it unresolved and escalate it before publishing high-consequence instructions. The target reader should not inherit greater certainty than the original author had.

When is a translation ready to publish?

Only when the terminology, numerical values, roles, qualifications and version controls have passed their appropriate reviews and unresolved high-risk issues are closed. Apply this principle to the particular project document, and retain the relevant source, reviewer and change history. If a conflict cannot be settled from evidence available to the translator, mark it unresolved and escalate it before publishing high-consequence instructions. The target reader should not inherit greater certainty than the original author had.

Official source and connected learning

The authoritative public scope for ISO 25964-1:2011 is available from the official ISO catalogue entry. Edition and published status were checked in November 2026. Standards are copyrighted and this article provides original educational scenarios, not the standard’s text. Use authorised documents and appropriate technical or legal review for real implementation.

For the general translation framework, return to Master Art of Translation (EDKSG-TRANS-MASTER-WORLD-000). To strengthen concept-level words, use the protected Vocabulary Learning Hub. For interpreting modality, agency, emphasis and meaning, use How English Works: Translation and Meaning Across Languages. These existing pages remain the conceptual owners; this page answers one narrower professional question.

Final practical criterion: people reading the source and translated versions should be able to name the same roles, reconstruct the same event sequence, identify the same open decisions and cite the same evidence. If they cannot, improve the target wording before real readers depend on it. Good translation protects decisions and preserves reality, not simply a smooth-looking page.