Translate | ISO 21801-1:2020 Cognitive Accessibility, Plain-Language Instructions, Forms and Confirmation Messages

Translate | ISO 21801-1:2020 Cognitive Accessibility, Plain-Language Instructions, Forms and Confirmation Messages: This world-facing translation guide shows localization teams, public-service writers, accessibility reviewers, translators, form designers and educators how to translate cognitive-accessibility instructions, multi-step forms and plain-language confirmation messages without changing the user’s real decision or introducing new barriers. It begins with what a reader needs to do when translating an actual document: identify the document’s function, specify its audience, decide which measurements and permissions must remain unchanged, and flag every unsupported interpretation. The aim is a useful target-language text whose real-world consequences match the source, rather than a sentence that merely sounds fluent.

ISO 21801-1:2020 provides general guidelines for designing cognitively accessible systems, products, services and built environments. It addresses cognitive accessibility rather than every aspect of physical or sensory accessibility, and was confirmed as current in 2025. The reference is used here to explain one carefully bounded translation problem, not to paraphrase a copyrighted specification clause by clause. The central challenge is not simply choosing shorter words. It is preserving the task, choices, consequences and points of reassurance while reducing unnecessary mental work. A short mistranslation can cause a person to miss an essential step or mistake a request for consent for information about consent. Technical translations should preserve explicit numbers and conditional words, but they also need to preserve the hidden connections between a claim and its evidence, between a step and its prerequisite, and between a label and the action it triggers.

Quick answer and reading route: First define the field and who will act on it. Second separate observation, possibility, requirement, warning, status and target. Third translate the claim while retaining its evidence and constraints. Finally test the wording on a realistic decision, document the reviewer’s corrections, and release only the approved version. Below are twelve connected lessons, twenty-eight original scenarios, a terminology register, reader questions and a practical four-week course. The guide complements the existing translation master rather than creating a replacement hub.

• Reader: localization teams, public-service writers, accessibility reviewers, translators, form designers and educators
• Core task: translate cognitive-accessibility instructions, multi-step forms and plain-language confirmation messages without changing the user’s real decision or introducing new barriers
• Immediate action: Preserve the named quantity, condition, status and scope before polishing prose.
• Decision check: Can the target reader act or interpret the finding exactly as the original permits?
• Limit: The examples are original instructional scenarios, not reproduced clauses from ISO. They do not establish formal accessibility certification. Real users, including people with varied cognitive needs, must help test the localized experience.

Contents and the logic of the guide

1. Separate easier reading from missing information
2. Preserve a single action per instruction
3. Distinguish required from optional fields
4. Keep yes and no choices symmetric
5. Use examples without changing rules
6. Represent time limits plainly
7. Differentiate alert, warning and error
8. Keep navigation labels consistent
9. Reduce memory burden without altering identity
10. Support recovery when something fails
11. Explain what a confirmation actually confirms
12. Test readability with real user tasks
13. Worked translation decisions
14. Controlled terminology
15. Bilingual quality and governance
16. Questions readers actually ask
17. Four-week learning and review plan

What is being translated—and what is not?

A professional translator does not need to reproduce an entire standard to help a reader understand documents associated with it. A product screen, a public information page, an inspection report or a data dictionary is a different genre from a normative specification. First identify which genre is in front of you. A public explanation should be understandable without suppressing caveats; a data field must remain stable enough for systems to exchange it; and a procedure must keep the order and conditions of its actions. These are different tests of faithfulness, even when they refer to the same subject.

This guide uses purpose-written, fictional examples to teach decisions instead of claiming to quote standard clauses. The published scope is linked near the end so readers can verify the precise edition. A national adoption, legal requirement, or contractual reference may impose stricter obligations than a generic educational page can establish. When the source is incomplete, the correct next step is to ask the competent owner for clarification rather than make a plausible-sounding addition.

1. Separate easier reading from missing information

Start with a real reading problem. A public office converts a three-page application guide into a one-page translation. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. Shorter text accidentally deletes the eligibility exception. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Keep the exception near the action to which it applies. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Check whether a first-time reader can state both the rule and the exception. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for separate easier reading from missing information: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

2. Preserve a single action per instruction

Start with a real reading problem. A device manual combines login, email verification and password change. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. A translator renders the three acts as a single vague instruction. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Number the activities and preserve the original dependency. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Watch a reader perform the steps in the intended order. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for preserve a single action per instruction: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

3. Distinguish required from optional fields

Start with a real reading problem. An intake form marks emergency contact optional but identity required. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. Polite words make the optional field sound obligatory. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Use controlled equivalents for required, optional and conditional. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Compare every field label with the validation rules. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for distinguish required from optional fields: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

4. Keep yes and no choices symmetric

Start with a real reading problem. A consent form offers yes and no radio choices with explanatory help. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. Negation shifts and a translated yes appears to agree to a refusal. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Test the exact proposition to which yes and no refer. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Back-translate both choices and test their consequences. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for keep yes and no choices symmetric: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

5. Use examples without changing rules

Start with a real reading problem. A form provides one example postal address beside a blank field. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. A target-language example looks like a valid mandatory address. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Mark the sample explicitly and avoid country assumptions. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Ask whether a reader could confuse sample and actual data. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for use examples without changing rules: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

6. Represent time limits plainly

Start with a real reading problem. A booking reminder says a link expires after 48 hours. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. The translation turns a duration into a calendar date or vice versa. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Keep duration, start event and timezone if given. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Check with a concrete simulated deadline. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for represent time limits plainly: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

7. Differentiate alert, warning and error

Start with a real reading problem. A kiosk reports that an input cannot be accepted. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. The translation suggests irreversible account loss. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Name what went wrong and what the person can do next. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Confirm status with developers before altering severity. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for differentiate alert, warning and error: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

8. Keep navigation labels consistent

Start with a real reading problem. A mobile app alternates Continue, Next and Submit. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. All three become one target-language verb. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Preserve the distinction between moving screens and finalizing. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Map button labels to the screens and backend actions. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for keep navigation labels consistent: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

9. Reduce memory burden without altering identity

Start with a real reading problem. A government portal refers to earlier documents by ambiguous pronouns. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. A localized instruction makes the reader recall which certificate is meant. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Repeat the exact document name when needed for comprehension. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Test whether the reader can locate the required document. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for reduce memory burden without altering identity: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

10. Support recovery when something fails

Start with a real reading problem. A payment form says try again but does not say if payment went through. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. The translation adds a promise that money was not deducted. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Describe uncertainty and a safe status-check route. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Verify against actual transaction-state design. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for support recovery when something fails: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

11. Explain what a confirmation actually confirms

Start with a real reading problem. A healthcare booking page displays Request received. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. The translator writes Appointment confirmed. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Keep request acknowledgment separate from final allocation. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Ask customer service to identify the status transition. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for explain what a confirmation actually confirms: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

12. Test readability with real user tasks

Start with a real reading problem. A team approves readable language solely through automated scores. A translator who approaches this as isolated word replacement will miss the relationship between the label and the action it governs. The audience may be comparing alternatives, following instructions or interpreting an audit result. Decide which of those tasks the original author expected before making the sentence smoother. In this situation, the most useful initial question is: what would a reader believe or do differently if this phrase were translated too broadly?

Why meaning can slip. A short passage is still confusing to its intended readers. The problem is not only an unfortunate synonym. The misleading wording rearranges information about what is permitted, known, measured, required or still undecided. Locate the evidence field, status control or process step that establishes the boundary. If the source lacks such a field, mark the issue as unresolved and retain the source’s uncertainty in the translation. A document that is easy to read but more confident than the evidence is not a successful educational translation.

Recommended editorial move. Pair language checks with task-based observation and accessible testing. Write a first version that names the relevant object, the condition under which the statement holds and any important exception. Only after this semantic check should you adjust sentence length, word order and tone for the target readership. If the target language normally places a condition after the main clause, consider whether readers could miss it; proximity may matter more than preserving the source sentence structure. The aim is neither literalism nor free rewriting, but an equivalent practical decision.

How to verify it. Document misunderstanding, correction and retest. Prepare one deliberately misleading alternative and ask a second reader to choose between interpretations without seeing your preferred answer. Record which expression they misunderstood and what contextual information corrected it. Then verify identifiers, dates, counts, quantitative units and any status-dependent permission directly against the source record. An approved terminology table should store not only the preferred expression but also the tempting expression that was rejected and the reason.

Translate this chapter into a usable practice

Make a two-column working note for test readability with real user tasks: on the left write the source wording and its operational meaning; on the right write what the target reader must conclude. Then test the document in the situation described above rather than in a dictionary alone. If the reader’s answer differs, determine whether the root cause is a false friend, missing precondition, ambiguous reference, imprecise statistical label or an omitted qualification. Correct only the component that creates that misunderstanding, so the rest of the document remains traceable.

Twenty-eight worked translation decisions

The examples below are original mini-cases. Each has a small source fragment, an attractive but unsafe interpretation, and a better meaning-preserving brief for translation into any target language. They are not quotations from an ISO publication and they are not representative measurements. Treat them as diagnostic exercises: the right answer is the one that keeps the reader’s choices and inferences within the original evidence.

Case 1. Registration screen

Illustrative source: “Create an account before booking.” Tempting but unsafe rendering: “Book now.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: An account must first be created. Reason: The action is prerequisite, not a completed reservation. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Create an account before booking” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 2. Eligibility notice

Illustrative source: “Applicants aged 16 or above may apply.” Tempting but unsafe rendering: “All applicants may apply.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: The age condition remains attached to eligibility. Reason: Removing a condition changes who can proceed. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Applicants aged 16 or above may apply” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 3. Form field

Illustrative source: “Preferred contact method (optional).” Tempting but unsafe rendering: “Contact method required.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: The preference is optional. Reason: False obligation can make someone abandon a form. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Preferred contact method (optional)” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 4. Agreement control

Illustrative source: “I do not want updates.” Tempting but unsafe rendering: “I want updates.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: A negative opt-out statement. Reason: Yes or tick must not invert the choice. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “I do not want updates” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 5. Device label

Illustrative source: “Hold the button for 3 seconds.” Tempting but unsafe rendering: “Press button three times.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: One continuous three-second press. Reason: Duration is not the number of presses. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Hold the button for 3 seconds” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 6. Instruction panel

Illustrative source: “Scan the code, then enter the number shown.” Tempting but unsafe rendering: “Enter any number.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: The number derives from the scanned code. Reason: Sequence is part of the meaning. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Scan the code, then enter the number shown” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 7. Service status

Illustrative source: “We received your request.” Tempting but unsafe rendering: “Your service is approved.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Receipt of a request only. Reason: Acknowledgment is not a decision. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “We received your request” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 8. Error message

Illustrative source: “This code has expired.” Tempting but unsafe rendering: “This account has expired.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: A particular access code is unusable. Reason: Do not transfer failure to the whole account. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “This code has expired” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 9. Help text

Illustrative source: “Use the same name as on your ID.” Tempting but unsafe rendering: “Use a nickname.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: The name must match the identity document. Reason: Name matching is a verification requirement. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Use the same name as on your ID” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 10. Scheduling button

Illustrative source: “Save draft.” Tempting but unsafe rendering: “Submit application.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Store incomplete work without final submission. Reason: Draft and final action differ. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Save draft” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 11. Payment notice

Illustrative source: “Payment status unavailable; check history.” Tempting but unsafe rendering: “Payment failed.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: The result is not yet known. Reason: Unknown is not a negative transaction result. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Payment status unavailable; check history” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 12. Checklist

Illustrative source: “Bring one of the following documents.” Tempting but unsafe rendering: “Bring all documents.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Alternative evidence choices, not a cumulative list. Reason: Quantifier determines workload. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Bring one of the following documents” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 13. Application limit

Illustrative source: “You can edit before final submission.” Tempting but unsafe rendering: “You can edit after submission.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Change is allowed only up to final submission. Reason: Time boundary controls the permission. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “You can edit before final submission” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 14. Accessibility label

Illustrative source: “More information about fees.” Tempting but unsafe rendering: “Pay fees now.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Opens explanation, not a payment action. Reason: A label must match the control’s behavior. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “More information about fees” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 15. Upload tip

Illustrative source: “Maximum file size 10 MB.” Tempting but unsafe rendering: “Minimum file size 10 MB.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Upper limit of ten megabytes. Reason: Direction of inequality must survive. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Maximum file size 10 MB” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 16. Confirmation page

Illustrative source: “We will email a decision within five working days.” Tempting but unsafe rendering: “Decision arrives within five days.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Working-day timeframe, not all calendar days. Reason: Weekend assumptions alter expectation. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “We will email a decision within five working days” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 17. Queue status

Illustrative source: “You are number 6 in the queue.” Tempting but unsafe rendering: “Six minutes remaining.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Place in queue, not a duration estimate. Reason: Ordinal position is not elapsed time. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “You are number 6 in the queue” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 18. Clinic reminder

Illustrative source: “Do not eat for six hours before this test.” Tempting but unsafe rendering: “Do not eat for six hours after.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: The restriction precedes the test. Reason: Temporal direction must not flip. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Do not eat for six hours before this test” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 19. Emergency help

Illustrative source: “If you feel unsafe, contact local emergency services.” Tempting but unsafe rendering: “Call this specific foreign number.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: A location-appropriate referral is required. Reason: Do not transplant an emergency phone number across countries. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “If you feel unsafe, contact local emergency services” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 20. Cancellation text

Illustrative source: “Cancel the appointment request.” Tempting but unsafe rendering: “Cancel a confirmed appointment.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Withdraw the request, whose approval is pending. Reason: Object and status matter. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Cancel the appointment request” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 21. Data consent

Illustrative source: “Sharing is off by default.” Tempting but unsafe rendering: “Sharing starts automatically.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Opt-in default is off. Reason: Privacy default must be exact. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Sharing is off by default” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 22. Address hint

Illustrative source: “Example only: 12 Sample Road.” Tempting but unsafe rendering: “Your address: 12 Sample Road.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: The address is an illustration, not user data. Reason: Example data must not look prefilled. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Example only: 12 Sample Road” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 23. Password rule

Illustrative source: “At least 12 characters.” Tempting but unsafe rendering: “Exactly 12 characters.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Minimum length is twelve. Reason: Minimum and exact length are different requirements. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “At least 12 characters” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 24. Return navigation

Illustrative source: “Back to previous step.” Tempting but unsafe rendering: “Restart application.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Move back one step without resetting. Reason: Navigation cannot silently destroy work. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Back to previous step” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 25. Human review

Illustrative source: “Your documents are being reviewed.” Tempting but unsafe rendering: “Your documents are verified.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Review is in progress. Reason: Process and outcome are distinct. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Your documents are being reviewed” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 26. Language selector

Illustrative source: “Switch display language.” Tempting but unsafe rendering: “Translate my submitted information.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Only interface language changes. Reason: Interface language does not imply data transformation. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Switch display language” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 27. Public counter

Illustrative source: “Please take a ticket.” Tempting but unsafe rendering: “Please buy a ticket.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Obtain a queue number. Reason: Polysemy of ticket needs service context. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “Please take a ticket” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Case 28. Technical issue

Illustrative source: “The form was not sent. Try again.” Tempting but unsafe rendering: “Your information was deleted.” A bilingual reviewer should not judge these only on grammar or elegance. Ask what concrete situation the source describes and which part of the tempting rendering changes the decision. The difference may look small on a screen, but the target reader sees only the finished message, not the translator’s intentions.

Meaning to preserve: Submission did not succeed. Reason: Failure to send is not proof of deletion. This distinction should be entered in a terminology or issue log with the document version, the intended audience, and the reviewer who has authority to confirm the interpretation. If the source itself contains ambiguity, an editorial note or author query is more defensible than an invented fact. A localized message can use natural grammar and locally intelligible syntax while preserving the operational boundary expressed here.

Apply the test: Put the unsafe and improved interpretations on separate cards, show them to two readers, and ask what they would report, do or conclude. Compare responses against “The form was not sent. Try again” and the underlying record. Where a button, metric or field is involved, test the adjacent label and help text too. Record whether the mistake would affect understanding, consent, timing, a measured result or a decision. That classification determines how urgently the translation needs correction before public release.

Controlled terminology: twenty distinctions to review

A glossary should be a set of tested concepts rather than an alphabetical collection of favorite words. For each entry record the source-language definition, intended target-language equivalent, situations where it is not appropriate, its data type or action where relevant, an example sentence and an approval owner. The following twenty entries should be distinguished before a public release in this field.

1. cognitive accessibility: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Separate easier reading from missing information”, especially where shorter text accidentally deletes the eligibility exception. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
2. plain language: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Preserve a single action per instruction”, especially where a translator renders the three acts as a single vague instruction. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
3. required field: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Distinguish required from optional fields”, especially where polite words make the optional field sound obligatory. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
4. optional field: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Keep yes and no choices symmetric”, especially where negation shifts and a translated yes appears to agree to a refusal. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
5. conditional field: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Use examples without changing rules”, especially where a target-language example looks like a valid mandatory address. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
6. confirmation: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Represent time limits plainly”, especially where the translation turns a duration into a calendar date or vice versa. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
7. acknowledgment: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Differentiate alert, warning and error”, especially where the translation suggests irreversible account loss. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
8. consent: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Keep navigation labels consistent”, especially where all three become one target-language verb. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
9. opt in: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Reduce memory burden without altering identity”, especially where a localized instruction makes the reader recall which certificate is meant. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
10. opt out: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Support recovery when something fails”, especially where the translation adds a promise that money was not deducted. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
11. error state: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Explain what a confirmation actually confirms”, especially where the translator writes appointment confirmed. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
12. warning: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Test readability with real user tasks”, especially where a short passage is still confusing to its intended readers. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
13. instruction: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Separate easier reading from missing information”, especially where shorter text accidentally deletes the eligibility exception. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
14. navigation: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Preserve a single action per instruction”, especially where a translator renders the three acts as a single vague instruction. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
15. submit: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Distinguish required from optional fields”, especially where polite words make the optional field sound obligatory. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
16. save draft: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Keep yes and no choices symmetric”, especially where negation shifts and a translated yes appears to agree to a refusal. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
17. eligibility: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Use examples without changing rules”, especially where a target-language example looks like a valid mandatory address. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
18. validation message: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Represent time limits plainly”, especially where the translation turns a duration into a calendar date or vice versa. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
19. timeout: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Differentiate alert, warning and error”, especially where the translation suggests irreversible account loss. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.
20. recovery step: Define it for the particular artifact before selecting an equivalent. Test it against the distinction in “Keep navigation labels consistent”, especially where all three become one target-language verb. Record whether it is a label, a measured value, a permission, an item state or an interpretive concept. A preferred translation without that role description is liable to be reused in the wrong setting.

A bilingual quality review that can be audited

Begin with an inventory of every source file and every surface on which the translation will appear: title, caption, chart axis, hover message, form field, accessibility description, email, mobile notification and downloadable attachment. Inconsistent wording across these surfaces is more damaging than an imperfect phrase in a single paragraph. Assign a document owner and an accountable technical reviewer. The translator owns language choices; the technical owner confirms claims that depend on systems, measurements or professional judgment.

Then build a claims table with five columns: source phrase, type of claim, evidence or system field, proposed target meaning and verification result. Label each claim as factual observation, definition, future target, permission, instruction, limitation or unknown. This forces the team to notice a frequent translation failure: a fluent expression that changes a future intention into a present achievement, a proposal into an approval, or a measurement into a guarantee. Test every number and classification in the table rather than trusting general proofreading.

Run a structured reverse-reading session. Give a target-language reader an extracted sentence or screen without the source. Ask for the action or inference it permits, then compare their answer with the source record. Do not treat back-translation as a substitute for user testing; it is one diagnostic method among several. For important content, repeat the test with people who have different levels of background knowledge. Keep examples fictional during training, and secure any real personal or restricted data.

Before release, compare the approved glossary with every occurrence of the term, verify links and codes, and check that translations do not break text layout or reading order. Keep a dated release note naming the source version, target locale, reviewer, outstanding questions and any limitations users need to see. After publication, monitor misunderstandings and support requests rather than assuming a completed proofreading cycle means the language is permanently correct.

Questions readers actually ask

Does plain language mean deleting legal detail?

The first step is to define the relevant object and what evidence supports the statement. In this topic, shorter text accidentally deletes the eligibility exception. A correct translation makes that boundary visible rather than appealing to a familiar but overbroad synonym. Check how the original author uses the term in neighboring fields and whether the target audience would infer a different condition, status or outcome. If the point affects a technical judgment or an actual obligation, refer it to the responsible subject specialist.

A practical way to decide is to test the sentence as a small user task. Show the proposed text without explanation, ask the reader to identify the action or conclusion, and compare it with the source. Follow the review method in Chapter 1, which addresses “Separate easier reading from missing information”. The answer should be recorded with the decision owner so that another translator working next month does not reopen a solved ambiguity or accidentally restore the rejected wording.

Can readability scoring establish cognitive accessibility?

The first step is to define the relevant object and what evidence supports the statement. In this topic, polite words make the optional field sound obligatory. A correct translation makes that boundary visible rather than appealing to a familiar but overbroad synonym. Check how the original author uses the term in neighboring fields and whether the target audience would infer a different condition, status or outcome. If the point affects a technical judgment or an actual obligation, refer it to the responsible subject specialist.

A practical way to decide is to test the sentence as a small user task. Show the proposed text without explanation, ask the reader to identify the action or conclusion, and compare it with the source. Follow the review method in Chapter 3, which addresses “Distinguish required from optional fields”. The answer should be recorded with the decision owner so that another translator working next month does not reopen a solved ambiguity or accidentally restore the rejected wording.

What makes a translated button label dangerous?

The first step is to define the relevant object and what evidence supports the statement. In this topic, a target-language example looks like a valid mandatory address. A correct translation makes that boundary visible rather than appealing to a familiar but overbroad synonym. Check how the original author uses the term in neighboring fields and whether the target audience would infer a different condition, status or outcome. If the point affects a technical judgment or an actual obligation, refer it to the responsible subject specialist.

A practical way to decide is to test the sentence as a small user task. Show the proposed text without explanation, ask the reader to identify the action or conclusion, and compare it with the source. Follow the review method in Chapter 5, which addresses “Use examples without changing rules”. The answer should be recorded with the decision owner so that another translator working next month does not reopen a solved ambiguity or accidentally restore the rejected wording.

Should examples be translated literally?

The first step is to define the relevant object and what evidence supports the statement. In this topic, the translation suggests irreversible account loss. A correct translation makes that boundary visible rather than appealing to a familiar but overbroad synonym. Check how the original author uses the term in neighboring fields and whether the target audience would infer a different condition, status or outcome. If the point affects a technical judgment or an actual obligation, refer it to the responsible subject specialist.

A practical way to decide is to test the sentence as a small user task. Show the proposed text without explanation, ask the reader to identify the action or conclusion, and compare it with the source. Follow the review method in Chapter 7, which addresses “Differentiate alert, warning and error”. The answer should be recorded with the decision owner so that another translator working next month does not reopen a solved ambiguity or accidentally restore the rejected wording.

Can a translator simplify a form without developer approval?

The first step is to define the relevant object and what evidence supports the statement. In this topic, a localized instruction makes the reader recall which certificate is meant. A correct translation makes that boundary visible rather than appealing to a familiar but overbroad synonym. Check how the original author uses the term in neighboring fields and whether the target audience would infer a different condition, status or outcome. If the point affects a technical judgment or an actual obligation, refer it to the responsible subject specialist.

A practical way to decide is to test the sentence as a small user task. Show the proposed text without explanation, ask the reader to identify the action or conclusion, and compare it with the source. Follow the review method in Chapter 9, which addresses “Reduce memory burden without altering identity”. The answer should be recorded with the decision owner so that another translator working next month does not reopen a solved ambiguity or accidentally restore the rejected wording.

How do we test consent wording?

The first step is to define the relevant object and what evidence supports the statement. In this topic, the translator writes Appointment confirmed. A correct translation makes that boundary visible rather than appealing to a familiar but overbroad synonym. Check how the original author uses the term in neighboring fields and whether the target audience would infer a different condition, status or outcome. If the point affects a technical judgment or an actual obligation, refer it to the responsible subject specialist.

A practical way to decide is to test the sentence as a small user task. Show the proposed text without explanation, ask the reader to identify the action or conclusion, and compare it with the source. Follow the review method in Chapter 11, which addresses “Explain what a confirmation actually confirms”. The answer should be recorded with the decision owner so that another translator working next month does not reopen a solved ambiguity or accidentally restore the rejected wording.

Does ISO 21801-1 replace usability research?

The first step is to define the relevant object and what evidence supports the statement. In this topic, shorter text accidentally deletes the eligibility exception. A correct translation makes that boundary visible rather than appealing to a familiar but overbroad synonym. Check how the original author uses the term in neighboring fields and whether the target audience would infer a different condition, status or outcome. If the point affects a technical judgment or an actual obligation, refer it to the responsible subject specialist.

A practical way to decide is to test the sentence as a small user task. Show the proposed text without explanation, ask the reader to identify the action or conclusion, and compare it with the source. Follow the review method in Chapter 1, which addresses “Separate easier reading from missing information”. The answer should be recorded with the decision owner so that another translator working next month does not reopen a solved ambiguity or accidentally restore the rejected wording.

What should be recorded in a localization change log?

The first step is to define the relevant object and what evidence supports the statement. In this topic, polite words make the optional field sound obligatory. A correct translation makes that boundary visible rather than appealing to a familiar but overbroad synonym. Check how the original author uses the term in neighboring fields and whether the target audience would infer a different condition, status or outcome. If the point affects a technical judgment or an actual obligation, refer it to the responsible subject specialist.

A practical way to decide is to test the sentence as a small user task. Show the proposed text without explanation, ask the reader to identify the action or conclusion, and compare it with the source. Follow the review method in Chapter 3, which addresses “Distinguish required from optional fields”. The answer should be recorded with the decision owner so that another translator working next month does not reopen a solved ambiguity or accidentally restore the rejected wording.

Four-week application course

Week 1: Map the original concepts

Collect a sample packet of twelve short artifacts, identify their audiences, and classify every instruction, number, status and condition. Construct a multilingual brief showing why a reader needs each field. Distinguish the source’s words from the reality that the words report. Use three examples from the worked-cases section and one new case created by your team. Keep rejected translations in the training record with explanations, because an error that has been understood is more valuable for future reviewers than a silent correction.

Week 2: Draft controlled translations

Translate the packet into a target language, create a twenty-entry terminology register, and perform deliberate contrast tests between similar labels. A partner should identify which terms refer to decisions, which to evidence and which to interface actions. Use three examples from the worked-cases section and one new case created by your team. Keep rejected translations in the training record with explanations, because an error that has been understood is more valuable for future reviewers than a silent correction.

Week 3: Test real interpretation

Run supervised target-reader tasks. Show people the translated notices, metrics or records and ask them what has happened, what remains uncertain, and what they must do next. Log mistakes by severity, correct root causes and rerun the tasks. Use three examples from the worked-cases section and one new case created by your team. Keep rejected translations in the training record with explanations, because an error that has been understood is more valuable for future reviewers than a silent correction.

Week 4: Release with evidence

Perform an independent bilingual review, check status and numerical fields against authoritative records, issue a versioned terminology list, and write a short change report. Establish a route for future amendment and reader feedback. Use three examples from the worked-cases section and one new case created by your team. Keep rejected translations in the training record with explanations, because an error that has been understood is more valuable for future reviewers than a silent correction.

Sources, limits and further reading

The scope and edition were checked against the official published record on 5 November 2026: ISO official ISO catalogue. This article provides original educational examples and does not supply a conformity assessment, licensed standards text or professional authorization. The examples are original instructional scenarios, not reproduced clauses from ISO. They do not establish formal accessibility certification. Real users, including people with varied cognitive needs, must help test the localized experience.

This is a specialist translation application, not a general translation curriculum or a competing hub. For the overall knowledge architecture continue to The Master Art of Translation; to develop precise word meanings, use the Vocabulary Learning Hub; and to understand how grammar and semantic relations carry meaning between languages, use How English Works. The three routes form a connected study path: vocabulary clarifies the concept, sentence structure clarifies the proposition, and specialist review checks the practical decision.