Master Art of Translation | The Negative-Question Answer Fidelity System — How Yes, No, Echo Replies and Corrections Keep Their Meaning

How to translate negative questions and yes-or-no answers without reversing the answer

To translate negative questions correctly, do not translate a bare yes or no in isolation. First identify the proposition being answered—such as the parcel arrived—then determine whether the respondent confirms that proposition, rejects it, agrees with a negative assumption, or is uncertain. In English, the exchange “Didn’t the parcel arrive?” / “Yes, it did” commits to arrival; “No, it didn’t” commits to non-arrival. A different language can conventionalise the answer particle differently, making a literal yes/no swap hazardous.

People searching how to translate yes and no in negative questions, negative question answer translation, how to translate ‘isn’t it?’ correctly, or why Japanese, Korean, Mandarin and English yes/no answers differ usually face the same hidden problem: the words visible on the page do not by themselves tell the translator which truth value the speaker accepts. Confirmation bias, question form, echo answers, politeness and the questioner’s expectation can pull in different directions. The solution is to track the answer’s explicit clause, its conversational target, and the action it authorises.

This world-facing translation guide supplies a practical diagnostic method for negative yes-no questions, confirmation questions, contradiction particles, short answers, echo replies, question tags and multilingual response polarity. It is an advanced child of Master Art of Translation | The Complete System for Moving Meaning Between Languages, not a replacement for the root. The worked material uses invented illustrations and semantic English glosses, not purported native-language translations; a language-specific reviewer must validate any real target-language formulation.

Immediate answer: translate the respondent’s commitment, not the apparent surface polarity of the reply token. Keep the difference between ‘the event happened’, ‘the event did not happen’, ‘I agree with the speaker’s negative guess’, and ‘I do not yet know’. If a yes/no token could reasonably flip the decision, replace it with a full-clause confirmation or ask for clarification before publication.

Meaning first: rewrite the proposition as a simple affirmative statement, then mark true, false, unknown or conditional.
Response second: examine the answer’s verb phrase, echo, correction, follow-up and context, not merely a response particle.
Language convention third: check how the particular target variety answers negative questions and whether it has special contradiction markers.
Consequence fourth: test what a reader or operator would actually do after reading the translation.
Review fifth: use independent bilingual review for high-risk consent, safety, eligibility, dates or financial decisions.
Boundary: never silently supply certainty, assent or a completed action where the speaker supplied none.

Scope and relationship to earlier chapters
The hidden problem: one short reply, several logical readings
The answer-fidelity meaning model and truth table
Thirty-two mechanisms that change how answers are interpreted
The repeatable twenty-eight-gate translation workflow
Seventy-two worked case studies with diagnostics
Thirty failure modes and practical repairs
Twenty-four transfer laboratories for real workflows
Practice, calibration and a four-week plan
Frequently asked questions, terminology and sources
Next routes and protected hubs

Scope: response interpretation, not general negation or grammar teaching

This chapter belongs to architecture EDKSG-TRANS-MASTER-WORLD-1610. Its protected distinction is answerhood: how a speaker resolves, rejects, postpones or reframes the alternative presented by a question, especially when the question is negative. It does not reproduce the previous discourse-marker function chapter or retell the general treatment of negation, presupposition and speech acts. It connects those ideas at a single high-risk junction: what the recipient of an answer thinks has been confirmed.

A negative question is not automatically a negative answer. “Hasn’t the inspection finished?” is a question, not a report that the inspection has failed. “Yes, it has” confirms completion. “No, not yet” denies completion at the relevant time. “I thought it had” describes a belief and leaves the present outcome open. “Only the first stage” expresses a qualified result. They are not interchangeable, even though a hurried translator might map every short response onto a pair of antonyms.

The method applies to literary dialogue, interviews, subtitles, support messages, community information, survey questionnaires, software interfaces, logistics, publishing, service records and everyday conversation. It does not provide legal, medical or safety approval; those contexts need qualified domain review. It also does not assert that one yes/no convention describes every dialect, register or discourse setting. Real speech is context-sensitive, and even speakers of the same language sometimes repair ambiguous negative-question answers by adding a clause.

The hidden problem: a reply can agree with a sentence but deny its proposition

Propositional polarity

A questioner’s wording can contain a negation while the respondent is answering whether the underlying event occurred. ‘Didn’t the package leave?’ raises the proposition that the package left while presenting a negative formulation. The response may align with the negative wording or with the positive proposition, depending on the language and conversational design.

Diagnostic consequence. A translation that matches particles but flips the event changes what the reader believes occurred. Write the event as a separate yes/no fact before selecting any target-language response token. Read the translation aloud as if you knew nothing of the source. If it implies a different action or a different state of affairs, it is not an acceptable rendering merely because the individual words look familiar.

Agreement with the questioner

Someone may say ‘Exactly’ to endorse a negative expectation rather than to confirm a positive event. If the expectation is ‘the report was not approved’, agreement can mean non-approval. But ‘Exactly’ after ‘isn’t it approved?’ in another tone may confirm approval.

Diagnostic consequence. Use the whole preceding utterance and any following explanation. If two readings remain plausible, translate the uncertainty rather than manufacturing a yes/no commitment. Read the translation aloud as if you knew nothing of the source. If it implies a different action or a different state of affairs, it is not an acceptable rendering merely because the individual words look familiar.

Answer to an alternative

The question ‘Wasn’t it Tuesday rather than Wednesday?’ contains two possible dates and a negation of one. A bare ‘No’ might challenge the presupposed correction or deny that the event was Tuesday. Replacing it with a target-language ‘No’ without an event statement loses the referent.

Diagnostic consequence. List each candidate date and the answer’s overt correction; preserve what is established and what is still open. Read the translation aloud as if you knew nothing of the source. If it implies a different action or a different state of affairs, it is not an acceptable rendering merely because the individual words look familiar.

Expectation and bias

Negative questions often carry an expectation: ‘Didn’t you receive my note?’ may suggest the questioner expected receipt. The respondent could be correcting the assumption, giving a factual reply, or reacting to a mild accusation. An accurate translation tracks both the informational answer and interpersonal force.

Diagnostic consequence. Do not strengthen the expectation into a charge or erase the speaker’s disappointment when it affects the conversation. Read the translation aloud as if you knew nothing of the source. If it implies a different action or a different state of affairs, it is not an acceptable rendering merely because the individual words look familiar.

Echo construction

The clause ‘I did’ repeats a positive verb commitment. It can be safer than ‘yes’ across response systems because its content is visible. Nevertheless, ellipsis requires recovering the correct verb and time frame. ‘I did’ may answer sending, opening or reading depending on the last question.

Diagnostic consequence. Reconstruct the omitted predicate from the nearest legitimate question under discussion, not from a plausible but different activity. Read the translation aloud as if you knew nothing of the source. If it implies a different action or a different state of affairs, it is not an acceptable rendering merely because the individual words look familiar.

Conditional response

A reply such as ‘If the payment clears’ is neither a yes nor a no. It states a condition under which the requested action may happen. The common translation shortcut ‘yes, after payment’ may turn a conditional possibility into a promise.

Diagnostic consequence. Record both the condition and the absent commitment. A reviewer should be able to explain what event remains contingent. Read the translation aloud as if you knew nothing of the source. If it implies a different action or a different state of affairs, it is not an acceptable rendering merely because the individual words look familiar.

Two-stage action

A process can include request, approval, execution and confirmation. Answering ‘Didn’t they approve it?’ with ‘They reviewed it’ names a different stage. Translating review as approval is a polarity-adjacent status error with practical consequences.

Diagnostic consequence. Define the exact verb and workflow state the question concerns; exclude related but non-equivalent states. Read the translation aloud as if you knew nothing of the source. If it implies a different action or a different state of affairs, it is not an acceptable rendering merely because the individual words look familiar.

Embedded denial

An answer may deny knowing rather than deny the event: ‘No, I don’t know whether it arrived’. This is a negative epistemic answer, not evidence of non-arrival. A compressed subtitle ‘No, it didn’t arrive’ crosses the knowledge boundary.

Diagnostic consequence. Annotate the scope of negation: is it the event, speaker’s knowledge, authority, willingness or ability that is denied? Read the translation aloud as if you knew nothing of the source. If it implies a different action or a different state of affairs, it is not an acceptable rendering merely because the individual words look familiar.

Question tag orientation

‘It shipped, didn’t it?’ has an affirmative main clause and negative tag; ‘It didn’t ship, did it?’ has a negative main clause and affirmative tag. The respondent can confirm either the positive event or negative non-event. Tag polarity is not an instruction to invert the answer.

Diagnostic consequence. Separate the main assertion, the confirmation request and the respondent’s full proposition. Read the translation aloud as if you knew nothing of the source. If it implies a different action or a different state of affairs, it is not an acceptable rendering merely because the individual words look familiar.

High-stakes ambiguity

In low-risk small talk, a brief ambiguous reply may be repairable later. In a consent or emergency flow, the same ambiguity can trigger irreversible action. Translation quality is therefore partly a risk-allocation problem, not merely a smoothness problem.

Diagnostic consequence. When interpretation determines permission or a safety action, use explicit propositions, request a read-back, and require independent review. Read the translation aloud as if you knew nothing of the source. If it implies a different action or a different state of affairs, it is not an acceptable rendering merely because the individual words look familiar.

A precise meaning model: propositions, commitments and response direction

Let P denote the affirmative proposition under discussion, such as P = ‘the access code works’. The four operational states are P affirmed, P denied, P not established, and P conditionally established. ‘Yes’ and ‘no’ are surface forms; the model does not treat them as stable international truth-value labels. The question may present P, not-P, or a biased alternative, while the answer may refer to the positive proposition, the preceding negative sentence or the questioner’s belief.

Question or reply in English	Affirmative proposition P	Speaker’s recoverable commitment	Safe English meaning note
Didn’t the code work? — Yes, it did.	The code worked.	P affirmed.	The code worked.
Didn’t the code work? — No, it didn’t.	The code worked.	P denied.	The code did not work.
Didn’t the code work? — I haven’t tested it.	The code worked.	Not established.	Testing has not happened; no outcome is claimed.
Didn’t the code work? — Only on the older device.	The code worked universally.	Restricted proposition.	Works in one tested condition, not all conditions.
Isn’t the form complete? — Almost.	The form is complete.	Near completion, not completion.	At least one item remains.
You didn’t sign, did you? — I signed yesterday.	The person signed.	P affirmed despite a negative main clause.	The signature was made yesterday.
You didn’t sign, did you? — Correct.	The person signed.	Context-dependent agreement.	Do not infer a signature without clarification.
Won’t you join us? — I wish I could.	The person will join.	Not affirmed; often inability or regret.	Do not turn a courteous reply into acceptance.

A translator should distinguish the question under discussion, the answer’s antecedent, and the resulting commitment. An utterance may also contain a separate action request. ‘Don’t you want to confirm?’ can request a response, encourage an action or imply disappointment, depending on context. That interpersonal feature belongs in the review but must not obscure whether confirmation was actually given.

Cross-linguistic evidence provides the reason for this discipline. Research on English and Korean negative questions reports that superficially corresponding yes/no answers may have opposite interpretations; research involving Mandarin and English likewise documents a conventional response difference. Linguistic work on polarity particles explores how responses relate to questions and assertions in discourse. These are findings about particular linguistic systems and contexts, not a licence to assume every speaker always follows a simple one-line rule. For exact varieties, consult an expert and test authentic examples.

Thirty-two mechanisms that affect how an answer travels between languages

1. Underlying proposition

Source-side diagnostic: Didn’t the courier leave? / Yes, they did. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: Choose the factual event ‘courier left’ as P; distinguish it from the wording ‘courier did not leave’. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: A target reply that confirms the negative wording when the speaker affirms departure reverses the practical status. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

2. Negative-question bias

Source-side diagnostic: Haven’t you finished the draft? / Not yet. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The question carries an expectation of completion. The answer denies completion at the current time, not the speaker’s future intention. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Translating ‘not yet’ as a refusal suggests unwillingness, which is a different claim. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

3. Truth-based response convention

Source-side diagnostic: Didn’t the device start? / It started on the second attempt. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: Preserve the concrete outcome rather than importing a foreign answer-particle mapping. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Language-specific particles may align with the truth of a negative statement, so an English ‘yes’ cannot be mechanically copied. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

4. Agreement-based response convention

Source-side diagnostic: Isn’t the route closed? / That’s right; it is closed. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: Agreement targets the negative proposition that the route is closed, rather than the positive question of openness. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Mistaking agreement for ‘open’ would give the reader an unsafe route assumption. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

5. Dedicated contradiction marker

Source-side diagnostic: The sensor isn’t active? / On the contrary, it is. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: Some languages can use a special particle to contradict negative propositions; English often needs a full counterstatement. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Replacing a contradiction with ordinary assent may lose the reversal. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

6. Short auxiliary echo

Source-side diagnostic: Didn’t you upload it? / I did. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The auxiliary repeats an affirmative predicate; the correct proposition is ‘I uploaded it’. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: If a previous turn also mentioned downloading, the echo’s referent must be recovered carefully. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

7. Verb phrase ellipsis

Source-side diagnostic: Didn’t you copy the last page? / Only the first two. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The fragment concerns which pages were copied. It must not be expanded to an entire document copy. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: A translator who adds ‘I copied it’ falsely reports completion. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

8. Scope of not

Source-side diagnostic: Don’t you think the test passed? / I don’t think so. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: ‘I don’t think’ weakens epistemic endorsement; it may not be equivalent to ‘the test failed’ asserted directly. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Keep the difference between belief and verified outcome. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

9. Negative concord

Source-side diagnostic: Nobody didn’t respond is not always a straightforward double cancellation across varieties. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: Translate the intended negative meaning in the particular language, rather than assuming two visible negatives always cancel. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Do not generalise from formal English double-negation exercises to languages with grammatical negative concord. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

10. Question-tag anchoring

Source-side diagnostic: It wasn’t submitted, was it? / No, it wasn’t. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The affirmative P ‘it was submitted’ is denied by the respondent, while the question seeks confirmation. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: A translated positive tag can be natural without changing the underlying answer. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

11. Choice-question structure

Source-side diagnostic: Was it not approved, or merely delayed? / Delayed. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: A choice answer selects a state, rather than simply affirming or rejecting a yes/no proposition. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Flattening to ‘no’ destroys the selected alternative. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

12. Temporal anchor

Source-side diagnostic: Didn’t the permit expire? / It expires next month. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: Present validity and future expiry must be separated; the answer indicates it has not expired yet. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: The wrong tense converts future risk into an already completed event. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

13. Completion versus initiation

Source-side diagnostic: Didn’t the transfer go through? / It was initiated. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The transfer began but successful completion is not confirmed. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Equating initiation with receipt creates a false financial claim. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

14. Ability versus willingness

Source-side diagnostic: Can’t you attend? / I can, but I prefer not to. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The speaker rejects presumed inability while declining the event. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: ‘No, I can’t’ would invent incapacity and lose deliberate refusal. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

15. Permission versus possibility

Source-side diagnostic: Can’t I use the archive? / You can request access. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: Permission to request is not permission to use immediately. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Translating an application path as granted access changes authority. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

16. Deontic ‘should’

Source-side diagnostic: Shouldn’t we verify first? / We should. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: A question that recommends a precaution is accepted; the recommended action is verification. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: If the answer becomes ‘we should not’, the workflow is inverted. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

17. Offer and invitation

Source-side diagnostic: Won’t you have a seat? / Thanks, I’ll stand. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The negative grammatical form may perform a courteous offer rather than express expected refusal. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Literal negation can create a rude or confusing invitation. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

18. Complaint and accusation

Source-side diagnostic: Didn’t you read my message? / I saw it but not the attachment. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The answer separates reading the message from reading the attached file. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: A translated unconditional ‘yes’ wrongly accepts the broader accusation. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

19. Presupposition challenge

Source-side diagnostic: Why didn’t you approve it? / I never received it. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The respondent rejects the assumed opportunity to approve; this is not a simple answer about refusal. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Do not turn missing receipt into deliberate non-approval. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

20. Source of knowledge

Source-side diagnostic: Wasn’t the gate locked? / That’s what the log says. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: A report of evidence is weaker than firsthand confirmation. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: If the target states ‘it was locked’ without attribution, the commitment escalates. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

21. Quote versus endorsement

Source-side diagnostic: Didn’t the witness say no? / She did, but the recording is unclear. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: A witness’s reported answer is not necessarily the current speaker’s endorsement of its meaning. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Keep source ownership and uncertainty visibly distinct. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

22. Speaker repair

Source-side diagnostic: Didn’t you cancel? / No—sorry, yes, I cancelled. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The correction changes the operative proposition; the self-repair must survive. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Subtitles that delete the correction can preserve the wrong earlier particle. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

23. Overlapping speech

Source-side diagnostic: Didn’t the alarm ring? / No— [overlap] yes, the second alarm. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: Speaker overlap can cause a stray ‘no’ to attach to the wrong question. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Use speaker separation and nearby turns before resolving the answer. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

24. Partial coverage

Source-side diagnostic: Didn’t all teams submit? / Three did; one hasn’t. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The response splits universal completion into a partial and outstanding set. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Rendering ‘yes’ erases the non-submitting team. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

25. Quantity and threshold

Source-side diagnostic: Didn’t at least ten people sign? / Nine signed. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The answer negates a numerical threshold while reporting a nearby count. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: A small number change can affect eligibility or process triggers. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

26. Counterfactual response

Source-side diagnostic: Wouldn’t the plan have worked? / Only if the network had stayed online. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The reply offers a conditional hypothetical, not a historical success. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Treating it as an actual success misstates what happened. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

27. Status change over turns

Source-side diagnostic: Didn’t you say it was blocked? / It was, but it is open now. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: Past and present statuses are both true at different times. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Discarding one time anchor produces a contradiction that did not exist. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

28. Irony and incredulity

Source-side diagnostic: Oh, you didn’t forget again? / I did, unfortunately. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: The question’s tone may be incredulous, but the answer admits forgetting. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Translating sarcasm as fact must not change the admitted event. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

29. Polite indirect refusal

Source-side diagnostic: Won’t you stay for the discussion? / I have another commitment. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: A reason can pragmatically decline an invitation without a direct no. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Mark a probable refusal for ordinary dialogue but seek explicit acceptance where booking depends on it. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

30. Backchannel versus answer

Source-side diagnostic: You haven’t checked, right? / Mm-hm. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: A short acknowledgement may signal listening, agreement or a request to continue. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Do not promote an unanchored backchannel into a factual yes/no. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

31. Translation memory contamination

Source-side diagnostic: Didn’t they approve the third draft? / Not yet. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: A stored response from an affirmative question may have the same target particle but a different truth condition. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Retest cached segments when the source question changes polarity. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

32. Reader action invariant

Source-side diagnostic: Isn’t the machine safe to restart? / We have not cleared it. A translator should first name the positive underlying proposition, then examine whether the answer makes that proposition true, false, unresolved or conditional. This discipline prevents a surface-form interpretation from masquerading as the speaker’s actual commitment.

How to apply it: Whatever sentence is chosen, a reader must not conclude restarting is authorised. Write a full-clause semantic paraphrase before choosing any abbreviated target form. Compare the paraphrase with neighbouring turns and with the action the reader will take. When the source is an interview or dialogue, the next response is often the best available evidence of how the original participants understood it.

Failure check: Compare the operational action a reader will take, not only lexical similarity. To diagnose the error, ask a bilingual reader to answer two separate questions: “What did the speaker say happened?” and “What should the recipient do next?” A fluent translation fails if it sends the reader to the opposite branch of either decision.

The repeatable twenty-eight-gate answer-fidelity method

Gate 01: Recover the local exchange

Read the full question, response, prior prompt and immediate follow-up. Do not begin with an isolated yes/no token; capture the conversational move that created the question. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

An efficient reviewer tests this gate with a counterreading: imagine a reader who knows only the target text and is trying to decide whether the event occurred. If that reader can reasonably arrive at the opposite fact, improve the wording with an overt subject and verb. Where the source genuinely leaves a matter unresolved, record that uncertainty rather than force an apparently tidy answer.

Gate 02: Assign speaker IDs

Mark each speaker or narrator clearly. In recorded speech, ambiguous diarisation can attach a short ‘no’ to someone else’s question. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 03: Write affirmative P

Convert the question’s core content to one simple affirmative proposition. ‘Didn’t the item ship?’ becomes P = ‘the item shipped’. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 04: Mark grammatical negation

Identify whether not negates the main verb, a subordinate belief, an ability modal, a quantity, a condition or the preceding assertion. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 05: Mark question type

Distinguish neutral yes/no enquiry, negative confirmation question, offer, invitation, complaint, rhetorical question, tag, alternative and echo question. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 06: Identify conversational bias

Record whether the questioner appears to expect P, expect not-P, or remain neutral. Treat this as contextual information, not a substitute for the answer. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 07: Extract the response clause

If the answer has ‘yes, it did’, ‘no, not yet’ or ‘only if’, give the clause greater weight than a bare token without assuming it is flawless speech. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 08: Detect response-system convention

Consult target-language grammar and representative speaker practice; do not simply copy the literal yes/no forms from the source. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 09: Normalize the truth state

Classify P as affirmed, denied, unresolved, mixed or conditionally true. State what evidence would justify the classification. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 10: Record time of claim

A process may be pending now but complete later. Identify reference time before translating completion and status predicates. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 11: Record entity and scope

Specify which file, shipment, person, user, clause or version is being discussed. A negative reply about one object must not generalise to all. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 12: Record epistemic source

Separate direct observation, database status, hearsay, expectation and inference. An answer about knowledge must not be upgraded to an answer about the world. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 13: Identify contradiction particle

If the language uses a particle specifically to reverse a prior negative, translate its corrective function rather than mechanically translating its dictionary gloss. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 14: Check particles versus echoes

An echoed verb phrase may express the commitment more clearly than a response token; retain both when necessary for a natural and unambiguous target. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 15: Preserve the questioner’s face

A polite correction may maintain social tact. Avoid rewriting a respectful answer into a confrontational or accusatory sentence. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 16: Preserve refusal versus inability

Do not conflate ‘will not’, ‘cannot’, ‘may not’ and ‘has not’. These differ in willingness, capability, permission and event status. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 17: Trace alternatives

For either/or questions, map each candidate choice separately. A response that names one alternative must remain an alternative choice, not a blanket no. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 18: Trace restrictive quantifiers

Check some, all, none, at least, only, and not yet; a negative answer to an all-question often permits some affirmative cases. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 19: Keep repair sequences

When a speaker changes an answer, decide with the style brief whether to show the correction or provide an explicitly edited transcript; never silently leave the obsolete claim. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 20: Check implied promises

An invitation or request can invite a polite non-answer. Do not translate a socially positive response as a contractually affirmative acceptance. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 21: Draft a target-language full clause

Before using a natural short reply, draft a target sentence that makes the outcome explicit. This is the review anchor. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 22: Draft a natural short reply

Use a particle, auxiliary echo or clause consistent with the target language’s real conventions, without changing the full-clause meaning. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 23: Back-paraphrase the target

Ask an independent reader to restate the target in a simple affirmative proposition and a truth state. Compare that result with P. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 24: Run a decision test

For a booking, access, permission, approval or workflow status, ask what action a recipient would now take. Compare with the source’s action branch. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 25: Run an ambiguity test

If two reasonable readings survive, do not invent resolution. Flag the uncertainty, preserve the ambiguity where appropriate, or request clarification. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 26: Run an accessibility test

Check that the reader can distinguish yes/no choice labels and complete answer text without relying only on visual colour, tone or voice. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 27: Record review evidence

Save the source question, explicit P, target version, reviewer decision and any unresolved alternative. This record supports consistent future revisions. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Gate 28: Approve only a stable commitment

The final gate passes when proposition, agency, time, certainty and authorised action match the source within the brief. Otherwise revise and retest. Evidence to retain: a short written note containing the answer’s positive proposition, the respondent’s commitment and the specific risk if the rendering were reversed. In practice, this note can be a comment alongside a bilingual segment, a spreadsheet cell or a review ticket; it need not slow routine low-risk translation when the context is obvious.

Seventy-two original case studies: source, false move, safe meaning and diagnostic

The English micro-dialogues below are constructed examples, not archival quotations or certified translations. A ‘safe meaning’ is a semantic anchor for a target-language translation, not a suggestion to replace every natural short reply with unnatural English-like syntax. Each case isolates a different combination of commitment, scope, time, agency, response convention and consequence.

Case 01: File exchange

Source dialogue: “Didn’t you attach the final file?” — “Yes, I attached version three.” Recoverable commitment: The final file, version three, was attached. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “No, I didn’t attach it.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The recipient might request a file already present. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The final file, version three, was attached.” Then check whether a shorter, natural target expression independently conveys that commitment. Inspect the attachment name and version, not the yes token alone. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Transfer exercise: Imagine the next turn is “So should I proceed?” Decide whether the answer authorises the requested action, denies it, or leaves the matter open. Write a target-language follow-up that asks only for the missing fact, not for a repetition of everything already established. If two responsible readers disagree, retain the ambiguity and escalate it for clarification.

Case 02: Missing enclosure

Source dialogue: “Didn’t the message include the schedule?” — “No, only the agenda was included.” Recoverable commitment: The schedule was absent; the agenda was present. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Yes, the schedule was included.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The recipient may stop searching for a document that is missing. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The schedule was absent; the agenda was present.” Then check whether a shorter, natural target expression independently conveys that commitment. Separate the two attachments by exact title. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 03: Draft status

Source dialogue: “Haven’t you completed the draft?” — “Almost; the conclusions are still open.” Recoverable commitment: The draft is not complete. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Yes, it is complete.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A reviewer might publish prematurely. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The draft is not complete.” Then check whether a shorter, natural target expression independently conveys that commitment. Test whether unfinished sections remain. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 04: Version approval

Source dialogue: “Wasn’t the third version approved?” — “The second was approved; the third is pending.” Recoverable commitment: Version two approved, version three pending. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The third version is approved.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The wrong revision might be released. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Version two approved, version three pending.” Then check whether a shorter, natural target expression independently conveys that commitment. Maintain version-specific status identifiers. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 05: Delivery confirmation

Source dialogue: “Didn’t the parcel arrive?” — “Yes, it arrived this morning.” Recoverable commitment: The parcel arrived this morning. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “No parcel has arrived.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A team might unnecessarily send a duplicate. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The parcel arrived this morning.” Then check whether a shorter, natural target expression independently conveys that commitment. Verify delivery date against the response. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 06: Delivery exception

Source dialogue: “Didn’t the parcel arrive yesterday?” — “No, it is due tomorrow.” Recoverable commitment: It did not arrive yesterday; it is expected tomorrow. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “It arrived yesterday.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The reader might stop monitoring a pending delivery. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “It did not arrive yesterday; it is expected tomorrow.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep past non-arrival and future expectation separate. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 07: Partial shipment

Source dialogue: “Haven’t all the boxes arrived?” — “Only four of six arrived.” Recoverable commitment: Four boxes arrived; two remain outstanding. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “All six boxes arrived.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Inventory reconciliation may close early. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Four boxes arrived; two remain outstanding.” Then check whether a shorter, natural target expression independently conveys that commitment. Retain denominator and unreceived count. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 08: Dispatch versus receipt

Source dialogue: “Wasn’t the package delivered?” — “It was dispatched, not delivered.” Recoverable commitment: Dispatch happened; delivery is unconfirmed. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Delivery was completed.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A status email would overpromise. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Dispatch happened; delivery is unconfirmed.” Then check whether a shorter, natural target expression independently conveys that commitment. Check which milestone the verb names. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 09: Service booking

Source dialogue: “Didn’t you confirm the appointment?” — “I requested it; they haven’t confirmed.” Recoverable commitment: Request submitted; appointment not confirmed. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The appointment is confirmed.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Someone might travel without a booking. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Request submitted; appointment not confirmed.” Then check whether a shorter, natural target expression independently conveys that commitment. Do not collapse request into confirmation. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 10: Appointment refusal

Source dialogue: “Won’t you take the earlier appointment?” — “I can’t; I have a conflict.” Recoverable commitment: Earlier appointment not accepted due to a conflict. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Earlier appointment accepted.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A scheduler could assign the wrong slot. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Earlier appointment not accepted due to a conflict.” Then check whether a shorter, natural target expression independently conveys that commitment. Treat the refusal as outcome, not the invitation’s polite form. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 11: Deadline ambiguity

Source dialogue: “Isn’t the deadline tomorrow?” — “No, it is the day after tomorrow.” Recoverable commitment: The deadline is one day later than suggested. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The deadline is tomorrow.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A team may rush or miss the actual date. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The deadline is one day later than suggested.” Then check whether a shorter, natural target expression independently conveys that commitment. Preserve the corrected date explicitly. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 12: Time-zone question

Source dialogue: “Didn’t registration close at eight?” — “Eight in the organiser’s time zone, not ours.” Recoverable commitment: The time requires its correct time zone before comparison. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Registration closed at eight local time.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: People may miss access due to a zone mismatch. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The time requires its correct time zone before comparison.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep zone label with the clock time. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 13: Signature question

Source dialogue: “You didn’t sign, did you?” — “I signed the amended page yesterday.” Recoverable commitment: The amended page was signed yesterday. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The document was not signed.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A form may be rejected incorrectly. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The amended page was signed yesterday.” Then check whether a shorter, natural target expression independently conveys that commitment. Identify the signed page and date. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 14: Consent uncertainty

Source dialogue: “Haven’t they consented?” — “They asked for more information; no decision yet.” Recoverable commitment: Consent has not been given or refused. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “They consented.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A process might begin without permission. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Consent has not been given or refused.” Then check whether a shorter, natural target expression independently conveys that commitment. Do not infer agreement from a question. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 15: Withdrawal request

Source dialogue: “Didn’t the participant refuse?” — “They declined the recording, not the interview.” Recoverable commitment: Recording declined; interview status separate. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The interview was declined entirely.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A team could cancel a permitted conversation. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Recording declined; interview status separate.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep each permission separately scoped. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 16: Invitation courtesy

Source dialogue: “Won’t you join our discussion?” — “Thank you, but I need to leave.” Recoverable commitment: The invitation was politely declined. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “They accepted the invitation.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A host may reserve an unnecessary place. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The invitation was politely declined.” Then check whether a shorter, natural target expression independently conveys that commitment. Look for concrete acceptance, not friendly wording. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 17: Meeting attendance

Source dialogue: “Aren’t you coming to the meeting?” — “I can attend the first part only.” Recoverable commitment: Partial attendance was offered. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Full attendance confirmed.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A chair may rely on someone for the whole event. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Partial attendance was offered.” Then check whether a shorter, natural target expression independently conveys that commitment. Preserve duration and partiality. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 18: Agenda correction

Source dialogue: “Didn’t we agree to postpone the debate?” — “We agreed to postpone the vote, not the debate.” Recoverable commitment: Only the vote was postponed. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The debate was postponed.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The wrong event could be removed from the agenda. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Only the vote was postponed.” Then check whether a shorter, natural target expression independently conveys that commitment. Track exact noun phrase after ‘postpone’. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 19: Committee decision

Source dialogue: “Wasn’t the proposal rejected?” — “The vote was delayed; no decision was made.” Recoverable commitment: Neither acceptance nor rejection occurred. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The proposal was rejected.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A public summary may misreport the decision. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Neither acceptance nor rejection occurred.” Then check whether a shorter, natural target expression independently conveys that commitment. Distinguish deferral from rejection. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 20: Minutes check

Source dialogue: “Didn’t the minutes say no?” — “They recorded an abstention, not a negative vote.” Recoverable commitment: Abstention was recorded. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “A negative vote was recorded.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The tally could become inaccurate. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Abstention was recorded.” Then check whether a shorter, natural target expression independently conveys that commitment. Do not force abstain into yes/no. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 21: Survey positive

Source dialogue: “Don’t you find the instructions clear?” — “Yes, the instructions are clear.” Recoverable commitment: The respondent reports clarity. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The respondent reports confusion.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Survey statistics may reverse satisfaction. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The respondent reports clarity.” Then check whether a shorter, natural target expression independently conveys that commitment. Anchor to full-clause meaning. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 22: Survey negative

Source dialogue: “Don’t you find the instructions clear?” — “No, the last step is confusing.” Recoverable commitment: The respondent reports a clarity problem. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The respondent reports clarity.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Usability issues may be hidden. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The respondent reports a clarity problem.” Then check whether a shorter, natural target expression independently conveys that commitment. Use explicit scales or full statements. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 23: Survey not applicable

Source dialogue: “Haven’t you used this feature?” — “I have never used it.” Recoverable commitment: No prior use is reported; opinions untested. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The feature was tried and rejected.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Invalid feedback could be classified as negative experience. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “No prior use is reported; opinions untested.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep never-used distinct from disliked. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 24: Evaluation scale

Source dialogue: “Isn’t the service excellent?” — “It’s adequate, not excellent.” Recoverable commitment: The evaluation is moderate and corrects the scale. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The response is enthusiastic approval.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A summary would overstate satisfaction. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The evaluation is moderate and corrects the scale.” Then check whether a shorter, natural target expression independently conveys that commitment. Retain the contrastive adjective. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 25: Technical test

Source dialogue: “Didn’t the test pass?” — “It passed locally, but failed remotely.” Recoverable commitment: Results differ by environment. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “All tests passed.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A release could ignore environment-specific failure. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Results differ by environment.” Then check whether a shorter, natural target expression independently conveys that commitment. Name both environments. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 26: Regression status

Source dialogue: “Hasn’t the bug been fixed?” — “The patch is written but not deployed.” Recoverable commitment: A patch exists; no deployed fix confirmed. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The bug is fixed in production.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Users may be told a defect is resolved prematurely. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “A patch exists; no deployed fix confirmed.” Then check whether a shorter, natural target expression independently conveys that commitment. Distinguish coding, deployment and validation. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 27: Account access

Source dialogue: “Isn’t the account active?” — “The profile exists, but access is suspended.” Recoverable commitment: Profile exists; access is not currently allowed. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The account can be used.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A user may repeatedly attempt denied access. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Profile exists; access is not currently allowed.” Then check whether a shorter, natural target expression independently conveys that commitment. Separate existence from authorisation. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 28: Password reset

Source dialogue: “Didn’t the reset link work?” — “It opened, but the token expired.” Recoverable commitment: The link opened; reset could not complete. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Password was reset.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Support might close the ticket without recovery. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The link opened; reset could not complete.” Then check whether a shorter, natural target expression independently conveys that commitment. Track opening versus successful completion. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 29: Data deletion

Source dialogue: “Didn’t you delete the record?” — “I sent a deletion request; it is queued.” Recoverable commitment: Deletion requested, not confirmed executed. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The record was deleted.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A privacy report may claim erasure without evidence. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Deletion requested, not confirmed executed.” Then check whether a shorter, natural target expression independently conveys that commitment. Separate request and completion. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 30: Audit evidence

Source dialogue: “Wasn’t the audit signed off?” — “Reviewers finished checking; sign-off is pending.” Recoverable commitment: Review finished; approval not yet granted. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Audit approval granted.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Compliance state may be misstated. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Review finished; approval not yet granted.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep review and authorisation separate. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 31: Invoice payment

Source dialogue: “Didn’t the invoice get paid?” — “The transfer was initiated this afternoon.” Recoverable commitment: Payment initiated; receipt unverified. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Payment received.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Accounts may incorrectly close an outstanding item. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Payment initiated; receipt unverified.” Then check whether a shorter, natural target expression independently conveys that commitment. Use financial milestone precision. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 32: Refund request

Source dialogue: “Haven’t they refunded you?” — “They approved a refund; it hasn’t arrived.” Recoverable commitment: Refund approved but not received. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Refund received.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Someone may expect available funds that are absent. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Refund approved but not received.” Then check whether a shorter, natural target expression independently conveys that commitment. Distinguish approval from settlement. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 33: Price inclusion

Source dialogue: “Isn’t delivery included?” — “Only standard delivery is included.” Recoverable commitment: Standard service included, premium delivery excluded. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “All delivery options included.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A customer may order on false cost assumptions. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Standard service included, premium delivery excluded.” Then check whether a shorter, natural target expression independently conveys that commitment. Retain modifier ‘standard’. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 34: Contract condition

Source dialogue: “Won’t the guarantee apply?” — “Only if maintenance records are complete.” Recoverable commitment: Coverage is conditional on records. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Coverage is guaranteed unconditionally.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A reader may assume protection not established. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Coverage is conditional on records.” Then check whether a shorter, natural target expression independently conveys that commitment. Preserve the explicit condition. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 35: Eligibility decision

Source dialogue: “Didn’t my application qualify?” — “It qualified for review, not admission.” Recoverable commitment: Screening passed; admission unconfirmed. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Admission awarded.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A person may act as if the place is secured. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Screening passed; admission unconfirmed.” Then check whether a shorter, natural target expression independently conveys that commitment. Identify each eligibility gate. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 36: Waitlist status

Source dialogue: “Am I not on the list?” — “You’re on the waiting list, not the confirmed list.” Recoverable commitment: Waitlisted, not confirmed. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Confirmed acceptance.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A customer may miss a needed alternative. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Waitlisted, not confirmed.” Then check whether a shorter, natural target expression independently conveys that commitment. Mark list type. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 37: Opt-out setting

Source dialogue: “Didn’t I opt out?” — “You opted out of emails, not system alerts.” Recoverable commitment: Email preferences changed; alerts remain. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “All notifications disabled.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Unexpected alerts may continue. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Email preferences changed; alerts remain.” Then check whether a shorter, natural target expression independently conveys that commitment. Scope the opt-out by channel. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 38: Notification delivery

Source dialogue: “Haven’t the alerts stopped?” — “Push alerts stopped; emails continue.” Recoverable commitment: One channel stopped; the other remains. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “All alerts stopped.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A troubleshooter might close the wrong issue. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “One channel stopped; the other remains.” Then check whether a shorter, natural target expression independently conveys that commitment. List channel-level states. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 39: Inspection completion

Source dialogue: “Didn’t the inspection finish?” — “The site visit finished; the report is pending.” Recoverable commitment: Visit complete, report incomplete. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The inspection process is fully complete.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A project could move ahead too soon. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Visit complete, report incomplete.” Then check whether a shorter, natural target expression independently conveys that commitment. Track report issue separately. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 40: Equipment clearance

Source dialogue: “Isn’t it safe to restart?” — “No, clearance has not been given.” Recoverable commitment: Restart has not been authorised. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Restart is permitted.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A user may take an unsafe action. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Restart has not been authorised.” Then check whether a shorter, natural target expression independently conveys that commitment. Prefer explicit no-restart status in safety copy. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 41: Weather observation

Source dialogue: “Wasn’t the route open?” — “It was open earlier; it is closed now.” Recoverable commitment: Earlier openness does not imply current access. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Route open now.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Travellers could rely on stale conditions. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Earlier openness does not imply current access.” Then check whether a shorter, natural target expression independently conveys that commitment. Anchor status to time of observation. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 42: Warning revision

Source dialogue: “Didn’t the warning expire?” — “The previous notice expired; a new one is active.” Recoverable commitment: One notice ended; another applies. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “No warning applies.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The reader could underestimate present risk. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “One notice ended; another applies.” Then check whether a shorter, natural target expression independently conveys that commitment. Identify notice version and effective period. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 43: Research result

Source dialogue: “Didn’t the experiment confirm the theory?” — “It supports one prediction, not the whole theory.” Recoverable commitment: Evidence is partial, not conclusive. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Theory fully confirmed.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Readers may overstate evidence. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Evidence is partial, not conclusive.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep the precise tested prediction. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 44: Replication issue

Source dialogue: “Wasn’t the result replicated?” — “The method was reused; the outcome differed.” Recoverable commitment: Method repeated, result not replicated. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Original finding replicated.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Scientific reports may mislead. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Method repeated, result not replicated.” Then check whether a shorter, natural target expression independently conveys that commitment. Separate procedural replication from same outcome. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 45: Statistical threshold

Source dialogue: “Didn’t the rate exceed ten percent?” — “It reached exactly ten percent.” Recoverable commitment: Rate is equal to the threshold, not above it. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Rate exceeded ten percent.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A rule keyed to ‘greater than’ may fire wrongly. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Rate is equal to the threshold, not above it.” Then check whether a shorter, natural target expression independently conveys that commitment. Preserve strict versus inclusive comparison. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 46: Sample count

Source dialogue: “Haven’t twenty volunteers enrolled?” — “Nineteen completed registration; one is pending.” Recoverable commitment: Nineteen registered; one pending. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Twenty completed enrollment.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A study could begin without required enrollment. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Nineteen registered; one pending.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep count and status separate. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 47: Archive document

Source dialogue: “Wasn’t the letter dated in spring?” — “The envelope was; the letter is undated.” Recoverable commitment: Envelope dated; letter itself lacks a date. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Letter dated in spring.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Catalogues may invent provenance. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Envelope dated; letter itself lacks a date.” Then check whether a shorter, natural target expression independently conveys that commitment. Protect the distinction between artefact and enclosure. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 48: Interview testimony

Source dialogue: “Didn’t the witness identify the driver?” — “They described a coat, not a person.” Recoverable commitment: No personal identification in the stated evidence. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Driver identified.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A report may imply stronger testimony. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “No personal identification in the stated evidence.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep observation and identification separate. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 49: Transcript omission

Source dialogue: “Didn’t she answer yes?” — “She said ‘I think so’, then paused.” Recoverable commitment: Tentative agreement only; certainty unestablished. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Unqualified yes.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The transcript may erase uncertainty. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Tentative agreement only; certainty unestablished.” Then check whether a shorter, natural target expression independently conveys that commitment. Retain hedge and pause where relevant. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 50: Audio overlap

Source dialogue: “Didn’t the second person disagree?” — “Their ‘no’ overlaps a different question.” Recoverable commitment: The reply’s referent is unresolved. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Firm disagreement with the current claim.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A transcript could misattribute dissent. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The reply’s referent is unresolved.” Then check whether a shorter, natural target expression independently conveys that commitment. Review timing and diarisation. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 51: Subtitle timing

Source dialogue: “Didn’t you say you’ll return?” — “I said I might return.” Recoverable commitment: Possibility expressed, no firm promise. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Definite return promised.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Viewers may misread character intention. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Possibility expressed, no firm promise.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep modal contrast in compressed subtitles. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 52: Character irony

Source dialogue: “You didn’t forget my birthday again?” — “I did. That’s why I’m here early.” Recoverable commitment: The speaker admits forgetting while trying to compensate. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The speaker remembered.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A joke or conflict changes meaning. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “The speaker admits forgetting while trying to compensate.” Then check whether a shorter, natural target expression independently conveys that commitment. Use next-turn admission to interpret the question. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 53: Rhetorical complaint

Source dialogue: “Can’t anyone fix this?” — “We can try tomorrow.” Recoverable commitment: Attempt offered, successful fix unpromised. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “A successful repair is guaranteed.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Readers may infer an assurance absent from dialogue. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Attempt offered, successful fix unpromised.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep attempt versus result distinct. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 54: Conditional travel

Source dialogue: “Won’t you travel with us?” — “If the tickets become affordable.” Recoverable commitment: Participation is conditional on price. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Participation confirmed.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Organisers may count a conditional attendee. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Participation is conditional on price.” Then check whether a shorter, natural target expression independently conveys that commitment. Mark condition before planning. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 55: Invitation deferral

Source dialogue: “Aren’t you joining the tour?” — “I will decide after seeing the schedule.” Recoverable commitment: Decision deferred. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Invitation rejected.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The host could discard a potential attendee. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Decision deferred.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep future decision open. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 56: Household task

Source dialogue: “Didn’t you wash the dishes?” — “I washed the cups, not the plates.” Recoverable commitment: Cups washed, plates not washed. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “All dishes washed.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Someone may assume a task fully done. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Cups washed, plates not washed.” Then check whether a shorter, natural target expression independently conveys that commitment. Preserve part-whole contrast. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 57: Repair request

Source dialogue: “Haven’t you repaired the chair?” — “I tightened one screw; the leg still moves.” Recoverable commitment: One step completed; repair unresolved. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Repair complete.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A user might sit before stability is checked. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “One step completed; repair unresolved.” Then check whether a shorter, natural target expression independently conveys that commitment. Track safety-relevant residual defect. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 58: Food restriction

Source dialogue: “Doesn’t the meal contain nuts?” — “The sauce may contain traces.” Recoverable commitment: Allergen absence cannot be confirmed. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The meal contains no nuts.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A person may make a high-risk dietary decision. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Allergen absence cannot be confirmed.” Then check whether a shorter, natural target expression independently conveys that commitment. Do not collapse uncertainty into safe/unsafe assurance. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 59: Ingredient sourcing

Source dialogue: “Didn’t you remove the ingredient?” — “I removed it from the recipe, not the stock.” Recoverable commitment: Recipe changed; existing stock unchanged. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “All existing products altered.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Customers may infer current inventory is reformulated. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Recipe changed; existing stock unchanged.” Then check whether a shorter, natural target expression independently conveys that commitment. Mark batch and time. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 60: Education feedback

Source dialogue: “Didn’t the learner understand the concept?” — “They solved the example but could not explain why.” Recoverable commitment: Procedural success, conceptual understanding unverified. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Concept mastered.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A teacher may advance prematurely. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Procedural success, conceptual understanding unverified.” Then check whether a shorter, natural target expression independently conveys that commitment. Test explanation separately from answer. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Offer the target text to a reader who has not seen the original. Ask them to state what is known, what remains uncertain and which action is permitted.

Case 61: Assessment completion

Source dialogue: “Haven’t you submitted every answer?” — “One answer is still blank.” Recoverable commitment: Assessment incomplete. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “All answers submitted.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The learner may lose an opportunity to correct. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Assessment incomplete.” Then check whether a shorter, natural target expression independently conveys that commitment. Preserve blank item count. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 62: Translation check

Source dialogue: “Didn’t the translator resolve that ambiguity?” — “They marked both readings for review.” Recoverable commitment: Issue documented, not resolved. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Ambiguity resolved.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Editors may publish an unapproved interpretation. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Issue documented, not resolved.” Then check whether a shorter, natural target expression independently conveys that commitment. Treat annotation as open status. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 63: Terminology approval

Source dialogue: “Isn’t this term approved?” — “It is proposed, awaiting domain validation.” Recoverable commitment: Proposed term, not authorised term. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Term officially adopted.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The glossary may propagate an error. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Proposed term, not authorised term.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep authority and workflow state. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Remove the questioner’s expectation and read the respondent’s clause on its own. If the resulting truth claim changes, the translation is borrowing unsupported meaning from the question.

Case 64: Machine-translation output

Source dialogue: “Didn’t the model translate ‘no’ correctly?” — “The clause is right, but the reply token flips commitment.” Recoverable commitment: Meaning must be judged at dialogue level. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “The translated answer is fully correct.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Readers could follow the wrong decision branch. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Meaning must be judged at dialogue level.” Then check whether a shorter, natural target expression independently conveys that commitment. Back-paraphrase event truth, not token appearance. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 65: Bilingual review

Source dialogue: “Haven’t both reviewers agreed?” — “One agrees; the other flagged a contradiction.” Recoverable commitment: Review disagreement remains. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Consensus reached.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: An error may bypass arbitration. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Review disagreement remains.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep dissenting review notes. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 66: Publication status

Source dialogue: “Wasn’t the correction published?” — “It is drafted and awaiting final check.” Recoverable commitment: No public correction yet. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Correction is live.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Readers may rely on information not yet corrected. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “No public correction yet.” Then check whether a shorter, natural target expression independently conveys that commitment. Check actual publication state. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Draw a two-column status table: confirmed facts on the left, missing decisions on the right. Reject a rendering that moves an item across columns without evidence.

Case 67: Press statement

Source dialogue: “Didn’t the organisation deny the allegation?” — “It declined to comment.” Recoverable commitment: No denial was made in the response. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “A categorical denial occurred.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Coverage may attribute words never spoken. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “No denial was made in the response.” Then check whether a shorter, natural target expression independently conveys that commitment. Distinguish silence from contradiction. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 68: Performance report

Source dialogue: “Didn’t the costs fall?” — “Unit costs fell; total spending rose.” Recoverable commitment: Two metrics moved in opposite directions. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “All costs fell.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A board may infer a false overall trend. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Two metrics moved in opposite directions.” Then check whether a shorter, natural target expression independently conveys that commitment. Keep denominator and metric labels. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 69: Equipment condition

Source dialogue: “Isn’t the battery charged?” — “It shows a charge indicator, but capacity is untested.” Recoverable commitment: Indicator on; usable capacity unverified. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Battery fully functional.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A user could depend on unavailable power. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Indicator on; usable capacity unverified.” Then check whether a shorter, natural target expression independently conveys that commitment. Distinguish displayed state from tested performance. Conversational cooperation is not the same as formal assent. A friendly tone, apology or reason can accompany non-acceptance and must not be treated as a completed transaction. Change the question into a positive one but keep the answer’s factual content. The verified event or non-event should remain stable across the two formulations.

Case 70: Membership lapse

Source dialogue: “Haven’t you renewed?” — “I paid the fee, but the renewal isn’t confirmed.” Recoverable commitment: Payment made, membership status pending. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Membership fully active.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: The member may assume access prematurely. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Payment made, membership status pending.” Then check whether a shorter, natural target expression independently conveys that commitment. Confirm account-state update. Partial results require an explicit domain and denominator. If one subgroup differs, the translation should not turn the entire set into a universal statement. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 71: Translation quote

Source dialogue: “Didn’t the client approve the translation?” — “They approved the quotation, not the delivered text.” Recoverable commitment: Commercial quote approved; text review open. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Translation content approved.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A translator might skip revision. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Commercial quote approved; text review open.” Then check whether a shorter, natural target expression independently conveys that commitment. Specify which object received approval. Time of utterance and time of event are different coordinates. Review whether ‘already’, ‘not yet’, ‘previously’ or ‘tomorrow’ changes the operative status. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 72: Service cancellation

Source dialogue: “Didn’t the booking get cancelled?” — “The request was sent; the cancellation hasn’t been acknowledged.” Recoverable commitment: Cancellation pending confirmation. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Booking definitely cancelled.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A customer may incur a fee unknowingly. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Cancellation pending confirmation.” Then check whether a shorter, natural target expression independently conveys that commitment. Preserve acknowledgement state. A respondent sometimes corrects the question’s premise rather than choosing a presented yes/no answer. Preserve that correction even when it requires an expanded clause. Compare before-and-after workflow states rather than individual words. A translated response is wrong if it closes a step that the source left pending.

Case 73: Public notice

Source dialogue: “Haven’t the doors reopened?” — “The lobby reopened; the hall remains closed.” Recoverable commitment: Partial reopening with restricted area. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Entire facility reopened.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: Visitors might attempt entry to closed areas. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Partial reopening with restricted area.” Then check whether a shorter, natural target expression independently conveys that commitment. Name physical spaces separately. The consequential distinction is between the truth of the asserted event and the stance implied by the question. A translator must not let a fluent short form override a clearer full clause. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Case 74: Decision escalation

Source dialogue: “Didn’t the manager say no?” — “They said ‘not until we have the figures’.” Recoverable commitment: Approval is conditional and deferred, not categorically rejected. This is the fact pattern the translator should preserve even when the natural target form uses a different answer particle, an echo verb or a compact sentence.

Tempting but false rendering: “Permanent refusal.” That wording would let the recipient infer a different proposition, time, scope or level of commitment. Why it matters: A team may abandon a viable request. The mismatch is not a matter of stylistic preference: a real user could choose a different next step from it.

Repair and quality test: Translate the answer first as an explicit clause: “Approval is conditional and deferred, not categorically rejected.” Then check whether a shorter, natural target expression independently conveys that commitment. Preserve future condition and decision ownership. Where a process has multiple milestones, each verb names a different state; the answer should not be strengthened into a later milestone simply because that is what the questioner expected. Reverse the surface reply token in your imagination. If the same event reading still survives only because of the full clause, the clause is the reliable translation anchor.

Thirty failure modes: diagnostic symptoms and repairs

Failure 1: Token substitution

What goes wrong: A yes token is carried across languages without checking how the target language responds to a negative question. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Use the verb phrase and proposition status as the anchor; select the target particle by local convention. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A reader concludes the opposite of what happened. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 2: Negation stacking

What goes wrong: Two visible negative expressions are treated as automatically cancelling each other in every language variety. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Check whether the source grammar uses negative concord and where negation takes scope. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A negative fact is accidentally affirmed. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 3: Assumed confirmation

What goes wrong: The translator treats ‘exactly’ as agreement with the positive underlying event without reading the prior negative statement. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Determine what claim ‘exactly’ picks up and make that claim overt if needed. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: An asserted non-event becomes a completed event. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 4: Courtesy as consent

What goes wrong: A polite thanks or apology is treated as accepting a request, invitation or contract. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Retain the indirect refusal or uncertainty and use explicit acceptance only when given. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A participant is counted as having consented. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 5: Question as fact

What goes wrong: The expectation in ‘haven’t they finished?’ becomes a report that they finished. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Keep question force and the distinction between presumed and verified status. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A team receives false progress information. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 6: Confirmation marker overload

What goes wrong: A question tag is translated as the logical negation of the main assertion. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Resolve the main assertion and intended confirmation direction before rendering the tag. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: The answer branch is inverted. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 7: Lost antecedent

What goes wrong: A bare ‘no’ is attached to the wrong earlier question in a multi-turn dialogue. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Trace adjacency and repair through the transcript or surrounding text. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A correct reply becomes a contradiction. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 8: Speaker swap

What goes wrong: An overlapping short response is attributed to the wrong voice. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Check speaker timing, turn boundaries and diarisation before choosing a target reply. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: The wrong person appears to approve or reject. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 9: Deictic drift

What goes wrong: A reply about ‘tomorrow’ is taken as tomorrow relative to publication, not the conversation. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Store the conversation’s reference date and adapt only where the brief permits. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: Deadlines shift silently. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 10: Past-to-present collapse

What goes wrong: A state valid yesterday is translated as current without an update. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Keep the past qualifier and current status separately. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A reader relies on expired information. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 11: Intention as action

What goes wrong: The respondent says they plan to submit but the target says the submission happened. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Separate future intention from completed action. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A workflow closes prematurely. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 12: Attempt as success

What goes wrong: An attempt or initial step is treated as proof of a desired outcome. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Use verbs such as attempted, requested or initiated when that is all the source supports. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: Completion is claimed without evidence. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 13: Not-knowing as denial

What goes wrong: The speaker says they do not know whether P is true; the translation says P is false. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Represent epistemic uncertainty, not a factual negative. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: The reader thinks an untested event was disproved. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 14: Not-yet as never

What goes wrong: A temporary absence is interpreted as a permanent refusal or impossibility. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Preserve aspectual markers and relevant future possibility. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A process is abandoned unnecessarily. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 15: Part as whole

What goes wrong: An answer about one unit is applied to an entire shipment, group or document. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Write explicit counts, subsets, items and exceptions. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: Outstanding work disappears. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 16: Wrong threshold

What goes wrong: Equal-to is converted into greater-than or vice versa when answering a numerical question. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Retain the mathematical comparison and its inclusivity. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: Eligibility or triggers change. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 17: Qualification dropped

What goes wrong: Words such as only, perhaps, usually or under test conditions disappear from a short reply. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Compare the target against the restricted proposition, not an expanded claim. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: Readers treat a narrow outcome as universal. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 18: Source attribution erased

What goes wrong: Reported evidence becomes the current speaker’s own certainty. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Mark who asserts, suspects, observed or recorded each fact. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A rumour becomes endorsed evidence. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 19: Unresolved correction

What goes wrong: A first answer survives after the speaker explicitly corrected it. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Apply a transparent transcript policy and preserve the final operative statement. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: Users follow the superseded response. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 20: Rhetorical misreading

What goes wrong: A complaint is read as a neutral factual request or vice versa. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Analyse speech act, intonation and neighbouring turns while preserving actual event claims. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: Tone and practical intention drift. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 21: Alternative-set collapse

What goes wrong: An ‘either A or B?’ question is reduced to a plain yes/no question. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: List each possible answer and retain which option the reply selects. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A chosen alternative is lost. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 22: Modal confusion

What goes wrong: Cannot, may not, will not and should not are collapsed into generic ‘no’. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Distinguish ability, permission, willingness, advice and obligation. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: The reason and authority for rejection change. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 23: Politeness intensification

What goes wrong: A mild corrective reply becomes a sharp contradiction in the target. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Choose a natural construction that preserves disagreement without inventing hostility. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A relationship or negotiation is damaged. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 24: Mixed answer simplification

What goes wrong: A partial ‘yes, but’ is shortened to an unconditional ‘yes’. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Retain the restrictive clause or restate it in a compact full answer. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: The exception is hidden. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 25: Quoted reply endorsement

What goes wrong: A witness’s ‘yes’ is treated as the narrator’s endorsement of the witness’s claim. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Maintain reported-speech boundaries and evidence status. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: The narrative falsely certifies the event. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 26: Machine translation hallucination

What goes wrong: A fluent model output adds ‘confirmed’, ‘cancelled’ or ‘completed’ absent from the dialogue. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Use a state-transition checklist during post-editing. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: Users act on invented certainty. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 27: Glossary overreach

What goes wrong: A bilingual glossary maps yes and no to fixed words independently of negative question structure. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Treat response particles as constructions with contextual conditions. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: Correct dictionary entries produce wrong discourse meaning. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 28: Accessibility omission

What goes wrong: A user interface offers colour-coded yes/no responses without explicit text or contextual wording. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Expose full accessible labels and the underlying proposition in control text. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: Assistive users cannot determine their choice. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 29: Reviewer anchoring

What goes wrong: A reviewer checks against the draft translation rather than the original dialogue. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Run an independent source-first paraphrase and compare target action consequences. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: Both stages perpetuate the same error. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Failure 30: False certainty at release

What goes wrong: The team publishes an unresolved answer as a definitive status to meet deadline. This is a failure of semantic alignment rather than an innocuous difference in word order. A competent review must distinguish what the questioner expected from what the respondent actually asserted.

Concrete repair: Mark ambiguity, request a source query and document any justified conservative handling. Record the source answer as an affirmative proposition with an explicit true, false, unknown or conditional value. Then test the target against that record, especially where the translated sentence invites a yes/no decision.

Reader-impact test: A public claim is stronger than the evidence. Ask the reviewer to imagine two readers acting on the opposite interpretations. If their actions would differ, the wording must be made sufficiently clear for the stakes and medium. If both versions imply the same action but differ in tone, inspect social force in a separate pass rather than confusing tone with truth.

Twenty-four transfer laboratories: moving the method into real production

Laboratory 01: Customer support

Production situation. A customer asks whether a replacement is not yet available; the representative says the order is approved but no item has shipped. The focal translation problem concerns status, approval, shipment and delivery. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Write an explicit answer that keeps approval true and shipment false; test whether the customer expects a package now. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Counterexample and test. Imagine an editor deletes all qualifiers and retains only a yes/no particle because the layout is constrained. Then imagine a second editor rewrites the particle as a full unconditional statement. Which version would cause a different next action, and why? Have a reviewer who has not seen the source report the event state, the speaker’s confidence and the action they would take. A mismatch in any one of those fields triggers revision. This laboratory is passed only when both the fluent version and the explicit review paraphrase lead to the same decision.

Transfer to another medium. Convert the result into a short subtitle, a customer-service message and a screen-reader-friendly label. Each medium may distribute information differently: subtitles require compactness, support messages can state a condition, and interfaces need action-labelled choices. The underlying proposition and outcome must remain fixed across all three. Document the one fact that would change the decision if translated incorrectly, and include it in the acceptance checklist.

Laboratory 02: Multilingual survey

Production situation. An item asks whether the respondent does not disagree with a proposal; the reply options are simply yes and no. The focal translation problem concerns double negation, agreement and response-system conventions. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Rewrite as a positive single proposition before localising response labels; conduct a cognitive interview. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 03: Software localization

Production situation. A confirmation dialog asks ‘Don’t you want to keep this setting?’ with buttons Yes and No. The focal translation problem concerns question wording, button action and accessibility. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Replace ambiguous buttons with ‘Keep setting’ and ‘Discard setting’, then test screen-reader announcements. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 04: Recorded interview

Production situation. The respondent first says no, then corrects to yes when asked about a different date. The focal translation problem concerns repair, transcription policy and reference time. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Preserve the correction, note the changed date and do not attach the old response to the new question. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 05: Live captioning

Production situation. A speaker asks ‘Didn’t it work?’, then another responds while a third interrupts. The focal translation problem concerns speaker separation and temporal alignment. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Check audio timing, speaker labels and a full-clause restatement; avoid a guessed categorical answer. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 06: Literary dialogue

Production situation. A character says ‘Won’t you stay?’ and receives ‘I have to leave before dawn’. The focal translation problem concerns invitation versus answer and implied refusal. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Convey the polite decline without inventing a literal negative statement that damages voice. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 07: Technical QA

Production situation. A defect is patched but not deployed; a manager asks whether it has not already been fixed. The focal translation problem concerns staging, deployment and verification. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Distinguish written code from tested release and set the system status to pending. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 08: Scientific reporting

Production situation. A question assumes a result was replicated; the answer distinguishes repeated method from different outcome. The focal translation problem concerns evidence quality and replication. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Report what was repeated and what did not match; preserve the limit in summaries. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 09: Financial communication

Production situation. A payment initiation is mistaken for the creditor receiving funds. The focal translation problem concerns transaction stages and verified receipt. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Preserve initiation as a distinct state and never certify settlement without evidence. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 10: Consent workflow

Production situation. A respondent says they are willing to learn more; a coordinator asks if this means agreement. The focal translation problem concerns interest, information and permission. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Translate willingness to discuss, not consent to a defined action. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 11: Travel information

Production situation. A route was open earlier and closed at the time of the reply. The focal translation problem concerns time anchors and public-facing action. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Show the current closure explicitly and preserve the older status only as history. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 12: Educational feedback

Production situation. The learner can perform a routine exercise but cannot explain the reasoning. The focal translation problem concerns performance versus understanding. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Translate the partial evidence without claiming conceptual mastery. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 13: Eligibility notice

Production situation. Preliminary screening was successful; final acceptance has not been issued. The focal translation problem concerns status taxonomy and reader expectations. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Keep the two gates distinct in both headline and body text. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 14: Food-safety query

Production situation. The meal may have traces of a substance; the question presumes the meal contains none. The focal translation problem concerns negative question, uncertainty and safety. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Avoid an unconditional ‘yes, safe’; render the uncertainty and seek qualified confirmation. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 15: Document archive

Production situation. An enclosure has a date but the main letter does not. The focal translation problem concerns object referents and inferred metadata. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Do not impute the enclosure’s date to the letter; annotate the provenance gap. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 16: Public meeting minutes

Production situation. A member abstains after a negative confirmation question about a proposal. The focal translation problem concerns yes/no vote versus abstention. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Keep abstention outside the binary tally and preserve who held which position. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 17: Translation-memory reuse

Production situation. A saved ‘yes’ translation is reused after a question changes from affirmative to negative. The focal translation problem concerns segment context and polarity switching. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Invalidate or retest the reused answer when the conversational antecedent changes. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 18: Machine-translation post-editing

Production situation. A model produces fluent text but a short reply reverses an outcome. The focal translation problem concerns semantic invariant and source-first review. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Run an independent truth-state comparison for every negative-question pair. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 19: Subtitling length limit

Production situation. The subtitle budget tempts deletion of ‘only if’ from a conditional answer. The focal translation problem concerns condensation and mandatory qualifiers. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Prefer a shorter explicit conditional to a smooth but unconditional yes. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 20: Community instructions

Production situation. A user is told they may request access, not that access has been granted. The focal translation problem concerns permission scope and service action. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Translate the request pathway, and make the ungranted access state explicit. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 21: Emergency handover

Production situation. A question suggests a device is cleared, but the responder says final clearance has not been given. The focal translation problem concerns safety authorization and no-action state. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Prefer a clear prohibition on proceeding, within the source’s actual instruction and approved brief. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 22: International meeting

Production situation. Participants use different response systems and switch languages mid-discussion. The focal translation problem concerns cross-linguistic turn coordination. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Have the chair repeat the agreed proposition, not merely tally spoken yes tokens. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 23: Research questionnaire

Production situation. The response ‘I don’t know’ is incorrectly coded as ‘no’. The focal translation problem concerns unknown, refusal and negative answer. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Create an uncertainty code so absence of knowledge does not become negative evidence. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Laboratory 24: Editorial fact-checking

Production situation. An interviewee denies approving wording but confirms reviewing it. The focal translation problem concerns authorship, responsibility and action stages. Before translating any response particle, the reviewer must formulate the proposition that a user would believe if the answer were read as positive, and the separate proposition they would believe if it were read as negative.

Working method. Preserve the denial of approval while acknowledging the review that occurred. Prepare two short target-facing paraphrases: one that states the established fact and one that states what remains uncertain or unapproved. Compare these paraphrases against the source exchange. The right choice is not necessarily the shortest; it is the shortest natural wording that preserves the relevant truth conditions and does not add an unsupported operational guarantee.

Exercises, scoring rubric and four-week practice plan

A useful exercise is not simply to replace a word with a foreign-language equivalent. Start with a source dialogue. Rewrite its proposition in positive form. Label the response P, not-P, unknown, partial or conditional. Then produce a full target sentence, a natural short version and a back-paraphrase. Invite a second reader to infer the next action. Record disagreements as evidence of where additional context is needed.

Exercise 1: Negative-yes polarity

Prompt: “Didn’t the alert arrive? / Yes, it did.” Write the affirmative underlying proposition (P = alert arrived) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: arrival confirmed; do not translate as missing alert. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 2: Negative-no polarity

Prompt: “Didn’t the alert arrive? / No, it didn’t.” Write the affirmative underlying proposition (P = alert arrived) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: arrival denied; retain relevant time frame. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 3: Unknown status

Prompt: “Didn’t the alert arrive? / I haven’t checked.” Write the affirmative underlying proposition (P = alert arrived) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: outcome unknown; lack of checking is not lack of arrival. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 4: Partial completion

Prompt: “Didn’t everyone reply? / Only two replied.” Write the affirmative underlying proposition (P = everyone replied) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: universal proposition not established; some responses received. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 5: Early completion

Prompt: “Isn’t this late? / We submitted yesterday.” Write the affirmative underlying proposition (P = submission late) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: questioner’s assumption corrected by a dated action. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 6: Permission distinction

Prompt: “Can’t I enter? / You can apply for access.” Write the affirmative underlying proposition (P = entry currently permitted) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: application possible, current permission not granted. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 7: Invitation versus acceptance

Prompt: “Won’t you come? / I’ll try.” Write the affirmative underlying proposition (P = respondent will come) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: attempt intended, attendance not certain. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 8: Positive tag

Prompt: “The files are complete, aren’t they? / Three are missing.” Write the affirmative underlying proposition (P = all files complete) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: completion denied with a count. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 9: Negative tag

Prompt: “The files aren’t complete, are they? / They are now.” Write the affirmative underlying proposition (P = files complete now) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: current completeness affirmed. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 10: Contradiction

Prompt: “The door isn’t open, right? / Actually, it is.” Write the affirmative underlying proposition (P = door open) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: correction affirms openness. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 11: Temporal revision

Prompt: “Didn’t it close? / It closed last week, then reopened.” Write the affirmative underlying proposition (P = currently closed) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: current reopening established, previous closure retained. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 12: Quote ownership

Prompt: “Didn’t she deny it? / She said ‘I wasn’t there’.” Write the affirmative underlying proposition (P = she issued a denial) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: reported denial, not independent proof of absence. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 13: Conditional state

Prompt: “Isn’t the trial approved? / Only after ethics review.” Write the affirmative underlying proposition (P = trial currently approved) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: approval condition not yet shown satisfied. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 14: Alternative answer

Prompt: “Wasn’t the invoice cancelled or amended? / Amended.” Write the affirmative underlying proposition (P = invoice cancelled) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: amendment selected, cancellation not affirmed. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 15: Uncertain reassurance

Prompt: “Isn’t this safe? / It has not been tested.” Write the affirmative underlying proposition (P = item safe) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: safety unestablished, not affirmed. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 16: Repaired response

Prompt: “Didn’t you post it? / No—wait, I did.” Write the affirmative underlying proposition (P = item posted) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: final correction affirms posting. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 17: Knowledge scope

Prompt: “Didn’t it break? / I don’t know.” Write the affirmative underlying proposition (P = item broke) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: no truth-value commitment. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 18: Modal distinction

Prompt: “Can’t you help? / I could, but not today.” Write the affirmative underlying proposition (P = ability to help) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: general ability affirmed, present availability denied. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 19: Numerical threshold

Prompt: “Didn’t more than twelve register? / Exactly twelve.” Write the affirmative underlying proposition (P = count greater than twelve) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: threshold not exceeded. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 20: Subgroup restriction

Prompt: “Didn’t the team succeed? / The night shift did.” Write the affirmative underlying proposition (P = whole team succeeded) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: one subgroup success, whole-team inference unsupported. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 21: Unconfirmed transaction

Prompt: “Didn’t the money arrive? / Bank says processing.” Write the affirmative underlying proposition (P = funds received) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: processing stage, receipt unknown. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 22: Deferred decision

Prompt: “Aren’t you accepting? / I’ll decide next week.” Write the affirmative underlying proposition (P = acceptance now) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: neither acceptance nor refusal yet. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 23: Indirect refusal

Prompt: “Won’t you join us? / I’ve another appointment.” Write the affirmative underlying proposition (P = participation committed) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: conversational refusal is likely, without inventing a formal rejection statement. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Exercise 24: Editorial review

Prompt: “Didn’t the author approve the quote? / They checked the spelling.” Write the affirmative underlying proposition (P = quote approved) and decide which part of the response supplies the most reliable commitment. Produce two English semantic anchors: one that could be safely expanded into a full target-language clause, and one short line suitable for an interface or subtitle. Keep every date, status, contrast and condition the response actually provides.

Answer key and rationale: Expected result: spelling review does not entail approval. The presence of negative words in a question is not evidence for an answer’s truth value. A reviewer should be able to show how this outcome is inferred from the respondent’s actual clause, not from an assumed universal dictionary mapping for ‘yes’ or ‘no’. If there are reasonable alternative readings, describe what additional information would distinguish them.

Translation acceptance rubric

Dimension	0: unsafe	1: needs work	2: acceptable
Propositional commitment	Opposite or invented fact	Ambiguous status	Correct truth/uncertainty state
Scope and referents	Wrong object or whole group	Scope recoverable with effort	Exact objects and subset limits
Time and completion	Future or partial action made complete	Some temporal ambiguity	Correct stage and time
Source and confidence	Hearsay asserted as certainty	Attribution weak	Knowledge source preserved
Social force	Refusal portrayed as consent	Tone awkward	Intent and politeness natural
Reader action	Wrong operational step	Extra clarification likely	Expected next step preserved

Score each dimension from zero to two. A zero in propositional commitment or reader action is a release-blocking defect even if the total score looks good. For high-stakes material, a qualified reviewer should also check domain-specific law, safety instructions and accessibility requirements; the rubric is an editorial tool, not a certification standard. Document the exact source line, the proposed correction and the independent back-paraphrase.

Week one: recognise answer systems

Study positive questions, negative questions, question tags, auxiliary echoes and full-clause answers. Classify fifteen small dialogues without translating them. Record P and not-P explicitly for each, and note when the source lacks evidence.

Daily review record: original question, full answer, affirmative proposition, commitment class, target wording, reader interpretation and one unresolved issue. At the end of the week, inspect all error types rather than relying on a single accuracy percentage. A credible improvement is fewer inverted commitments and fewer unapproved actions, not simply more fluent wording.

Week two: produce contrastive versions

Take fifteen dialogues in two registers—spoken conversation and formal notices. Translate each first into an unambiguous full clause, then into a natural short form. Compare whether the short form changes the respondent’s commitment.

Week three: stress-test decisions

Use mixed-status examples involving requests, approvals, execution and receipt. Assign one person to write a risky but fluent mistranslation and another to diagnose it from the event-state table. Document the repair.

Week four: independent reader tests

Give unseen target-only versions to independent readers, ask them to state the established facts and next actions, and analyse disagreements. Keep a small, context-rich translation memory of validated negative-question patterns rather than isolated yes/no tokens.

Forty frequently asked questions about negative-question translation

FAQ 1: Can yes mean no in translation?

The same surface response particle may attach to the truth of a proposition or to agreement with a negative statement in different languages. The exact mapping is language- and context-dependent. Translate the speaker’s underlying claim, not the particle alone.

Practical check: identify the proposition, its truth or uncertainty status, and the next action a target-language reader would infer. If this question involves a real product, safety system, consent decision or policy requirement, validate the final wording with a qualified reviewer familiar with the target language and domain.

FAQ 2: How should I translate ‘Didn’t you do it?’

Identify the event ‘you did it’ as the positive proposition. ‘Yes, I did’ affirms that event in ordinary English; ‘No, I didn’t’ denies it. Then choose an appropriate target construction for that result.

FAQ 3: Why do negative questions feel confusing?

They combine a question with a negative formulation and often a prior expectation. The listener may interpret a short response against the positive event or the negative assumption. A full response clause reduces the ambiguity.

FAQ 4: Are English negative-question answers always unambiguous?

No. Bare particles can be conversationally unclear, especially after complex negations, alternative questions or unusual intonation. A careful translator can use follow-up context or flag the uncertainty.

FAQ 5: Do Japanese and English use the same yes/no convention?

Not in every negative-question environment. Linguistic research finds cross-language differences in negative-question verification and response systems. Validate the precise Japanese construction with a qualified speaker rather than using a blanket conversion chart.

FAQ 6: What about Korean negative-question answers?

Research comparing Korean and English documents different interpretations of short yes/no replies to negative questions. Interpret the full response and its antecedent in the exact Korean variety and context.

FAQ 7: Does Mandarin use the same rules as English?

Research on Mandarin-speaking learners of English identifies cross-language differences in conventional responses to negative questions. Do not assume one fixed word-for-word mapping; use full clauses when confusion would matter.

FAQ 8: Does French have a special corrective yes?

French ‘si’ can be used to contradict a negative statement or question in relevant contexts. It is not an invitation to replace every English ‘yes’ with ‘si’. Preserve whether the speaker corrects a negative assumption.

FAQ 9: What does German ‘doch’ do?

Among its uses, German ‘doch’ can contradict a negative proposition. Its function depends on syntax, context and discourse. A bilingual reviewer should choose a natural construction that expresses the actual corrective commitment.

FAQ 10: What is a polarity particle?

A small response expression such as English ‘yes’ or ‘no’ that relates a turn to a proposition or question. Research treats the relationship as more complex than a pair of dictionary equivalents.

FAQ 11: What is an echo reply?

A response that repeats or recovers an auxiliary or predicate, such as ‘I did’. It often makes the answer’s polarity clearer than a bare token, but the omitted action still needs a correct antecedent.

FAQ 12: Should I always expand yes and no into full sentences?

Not always. Natural conversational translations can remain short when the target convention is clear and the surrounding dialogue anchors the answer. Use expansion when ambiguity or consequential decisions justify it.

FAQ 13: How can I handle a bare ‘correct’?

Determine which preceding claim is being confirmed. A question may contain a negative expectation, an alternative or several subclaims. If the antecedent cannot be identified, query the source or preserve uncertainty.

FAQ 14: Is ‘not yet’ a negative answer?

It denies completion at the relevant present time while commonly leaving future completion possible. Do not transform it into ‘never’, ‘will not’ or an unconditional refusal.

FAQ 15: How should I translate ‘I don’t know’?

As lack of knowledge about the event, not necessarily as a negative event claim. This matters in surveys, reports, investigations and customer support.

FAQ 16: Can a polite response be a refusal without no?

Yes. A reason such as ‘I have another commitment’ may pragmatically decline an invitation. Preserve the social action, but avoid treating indirect speech as formal consent or an official decision without sufficient context.

FAQ 17: How do I translate ‘won’t you’?

Inspect whether it functions as an invitation, offer, appeal, complaint or neutral question. The negative form alone does not determine the social intention.

FAQ 18: Are question tags just grammar?

No. Tags can request confirmation, challenge an assumption or indicate confidence. Their polarity should be read alongside the main proposition and actual response.

FAQ 19: What if two people answer at the same time?

Establish speaker identity and timing first. In a transcript, a response particle attached to the wrong question can invert the reported position. Keep uncertainty where the audio is unresolved.

FAQ 20: What if the respondent repairs their answer?

Preserve the corrected operative answer according to the transcription or editorial brief. Do not silently choose the first particle if the later words clearly correct it.

FAQ 21: Can translation memory store ‘yes’ as a segment?

It can, but uncontextualised storage is risky. For negative questions, a translation memory should preserve the question form, antecedent, response system and validated full-clause reading.

FAQ 22: Is machine translation reliable for yes/no answers?

Fluency alone is insufficient. Run question-answer pairs through a source-first truth-state check and an independent target-only paraphrase, especially where short replies control workflow decisions.

FAQ 23: How should surveys phrase negative questions?

Where editorial authority permits, prefer direct, single-proposition questions and action-specific response labels. Test respondent interpretation in the target language before fielding the survey.

FAQ 24: How do I localise confirm dialogs?

Use buttons that state their actions, such as ‘Keep changes’ or ‘Discard changes’, when that matches the interaction design. Context-free Yes and No labels may become hazardous after a negative prompt.

FAQ 25: What does it mean to preserve answerhood?

The translation should preserve which alternative the respondent selects, whether they reject the premise, and whether the issue remains open. The target may express this using different grammatical resources.

FAQ 26: What is the question under discussion?

The current issue participants are trying to resolve. A response may target that issue rather than the nearest isolated clause; identifying it helps avoid wrong antecedents.

FAQ 27: How do I preserve conditional answers?

Keep conditions such as ‘if the network holds’ or ‘after approval’. A conditional answer does not become a present unconditional yes merely because it sounds cooperative.

FAQ 28: Can ‘maybe’ count as acceptance?

Not as a definitive commitment. It may express uncertainty or willingness to consider, but an action requiring consent or confirmation needs an explicit, appropriate commitment.

FAQ 29: How are negative answers different from denials of knowledge?

A negative event answer says the event did not happen. A denial of knowledge says the speaker cannot establish whether it happened. Different follow-up actions are appropriate.

FAQ 30: How do I handle all-versus-some replies?

Track the quantified set. ‘Three of five’ does not affirm ‘all five’, and may not deny that some items succeeded. Keep counts and denominators.

FAQ 31: Should translator notes be visible to readers?

Usually not unless the publishing brief calls for a note. Reviewers should document ambiguity internally; public-facing text should be clear and faithful without exposing unnecessary production jargon.

FAQ 32: Can a correction particle replace a full explanation?

Sometimes in natural dialogue, but not always in risk-sensitive material. If the intended reader may not identify the contradicted claim, a short explanatory clause is preferable.

FAQ 33: How do I verify a translated refusal?

Ask a target-only reader whether the respondent agreed, declined, deferred or made a conditional offer. Compare that answer to the source and its social context.

FAQ 34: Are there universal response rules for all languages?

No simple two-word chart can safely account for every language, variety and discourse condition. Comparative research supports careful system-specific analysis rather than broad stereotypes.

FAQ 35: What is the minimum safe test for negative questions?

Write P as an affirmative proposition, classify the source response and back-paraphrase the translation. If the response changes P’s truth status, revise.

FAQ 36: Do I need to translate every hesitating sound?

Not necessarily. Preserve hesitation where it materially affects certainty, timing, testimony or character voice. The decision depends on genre, accessibility and the translation brief.

FAQ 37: Does punctuation change answer direction?

Punctuation can clarify speech boundaries and emphasis but cannot replace propositional analysis. A question mark or quotation mark is not proof of the answer’s factual commitment.

FAQ 38: How can I prevent SEO copy from overstating the method?

Describe it as a structured quality-review method, not a universal guarantee of equivalence. Link to credible linguistic sources and distinguish your illustrative examples from published research.

FAQ 39: Is ISO 17100 a rulebook for yes/no particles?

No. ISO 17100 addresses translation-service processes and relevant requirements; it does not prescribe a universal answer system. Use it for process framing, not as proof of specific language behaviour.

FAQ 40: Where should I continue learning?

Return to the translation root for the entire meaning-transfer system, use the preceding discourse-marker chapter for neighbouring conversational signals, and consult the language or vocabulary hubs for related but independently owned topics.

Terminology glossary and research notes

Answerhood: The relationship between a response and the issue or alternatives raised by a question. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Affirmative proposition P: The positive event or state whose truth can be affirmed, denied or left open. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Negative polar question: A yes/no question containing negation or a negative polarity construction. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Response particle: A short expression such as yes or no whose meaning depends on response conventions and discourse. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Truth-based response: An analytic label for systems whose particle interpretation is related to the truth of a prior proposition or statement; implementations vary. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Polarity-based response: An analytic label for response systems oriented toward affirmative and negative propositional alternatives. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Agreement-based response: A response linked to accepting or rejecting a previous assertion, which may itself be negative. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Contradiction marker: A construction or particle marking a reversal of a negative claim or expectation in languages where available. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Auxiliary echo: A short reply repeating an auxiliary to expose the chosen polarity and person or tense. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Negative concord: A grammatical pattern in which multiple negative expressions can contribute to one semantic negation. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Question under discussion: The conversational issue that the speaker is currently addressing. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

At-issue content: The proposition directly under consideration rather than merely suggested background assumptions. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Epistemic uncertainty: An unresolved state of knowledge, not automatically a negative fact. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Deontic modality: Language of permission, duty, prohibition or recommendation. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Response bias: The influence of context or expectation on how a question is framed or answered. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Back-paraphrase: An independent target-language reader’s explicit explanation of what a translation claims. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Decision invariant: The requirement that a reader’s authorised next action match the source despite linguistic changes. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Release blocker: An error serious enough to withhold publication or operational use until corrected. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Semantic scope: The part of a clause to which a negation, quantifier, condition or modal applies. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Reader-action test: A check of what a target-only reader would do after receiving the translated message. In this chapter, the term is used to describe a reviewable distinction, not to prescribe a universal morphological form for every language. When an example is underspecified, document the missing context before selecting a target construction.

Research foundations and responsible use

The explanations are informed by current publicly accessible reference material checked in November 2026: Cambridge Grammar on yes-no and negative yes-no questions describes affirmative questions, negative-question bias, offers and complaints; British Council LearnEnglish covers English question and negative forms, while its short-answer reference explains auxiliary echoes and related uses.

For cross-linguistic interpretation, consult Park and Dubinsky (2019), The syntax and semantics of negative questions and answers in Korean and English; Zhang and Vanek (2021), From “No, she does” to “Yes, she does”; and Akiyama (1992), Cross-linguistic contrasts of verification and answering among children. They establish documented differences but do not justify unverified example translations into other languages.

For a formal perspective on response systems, see Roelofsen and Farkas, Polarity particle responses as a window onto the interpretation of questions and assertions (Language, 2015), and Kim (2024), Parametric differences between English and Korean response particles. Their analyses provide theories to compare rather than a single mandatory editorial rule.

For translation-service process and review specifications, ISO 17100:2015 describes requirements for translation services and their processes; it is not an interpreting-services specification or a free source of linguistic equivalence tables. A future revision of this standard remains distinct from the current publication. The workflow above is an eduKateSG editorial method informed by these sources, not a claim of certification by any standards body.

Next routes: preserve distinct owners and continue the architecture

For the full architecture, return to Master Art of Translation | The Complete System for Moving Meaning Between Languages. The immediately preceding Discourse Marker Function Fidelity chapter explains neighbouring cues such as well, actually, right and you know without making them substitutes for proposition-level answer checking. Read this child as a specialist answer-fidelity method, not as a new owner for the earlier chapters.

For word-level learning, visit the existing Vocabulary Learning Hub; it remains the owner of vocabulary methods and word development. For English-specific grammar and expression, use the protected How English Works | Translation and Meaning Across Languages route. This article points outward only when those topics illuminate answer translation. Their titles, URLs, contents and navigational roles remain independent.

A next reserved-tens child in the broader translation architecture should be selected only after a collision and ownership scan. This chapter itself makes no claim to reserve or publish that future page. Its enduring rule is narrow and testable: before translating a negative-question answer, establish what the respondent actually commits to, retain every consequential condition, and verify the target-language reader reaches the same outcome.